Embed Size (px)
Citation preview
FLORIDA PUBLIC HURRICANE LOSS PROJECTION MODEL
Engineering Team Final Report
submitted to:
Dr. Shahid Hamid
Director
LABORATORY FOR INSURANCE, FINANCIAL, AND ECONOMIC RESEARCH
INTERNATIONAL HURRICANE RESEARCH CENTER
FLORIDA INTERNATIONAL UNIVERSITY
Volume I
Exposure and Vulnerability Components of the Florida Public Hurricane Loss Projection
Model
March 2005
i
ENGINEERING TEAM
Principal Investigators Kurtis Gurley, Ph. D. Associate Professor Civil and Coastal Engineering Department University of Florida Jean-Paul Pinelli, Ph.D., P.E. Team leader Associate Professor Civil Engineering Department Florida Institute of Technology Chelakara Subramanian, Ph.D., P.E. (UK) Professor Aerospace Engineering Department Florida Institute of Technology Graduate Research Assistants Anne Cope, Ph. D. Currently with Reynolds, Smith, and Hills Florida Liang Zhang, M.S. Currently with Risk Management Solutions California Joshua Murphree, M.S. Arnoldo Artiles Graduate Research Assistant Civil Engineering Department Florida Institute of Technology
ii
Pranay Misra Graduate Research Assistant Civil Engineering Department Florida Institute of Technology Consultants: Dr. Sneh Gulati Associate Professor Department of Statistics Florida International University Dr. Emil Simiu NIST Fellow Building and Fire Research Laboratory National Institute of Standards and Technology
iii
Abstract
The need to predict hurricane-induced losses for $1.5 trillion worth of existing
structures exposed to potential hurricane devastation in the state of Florida has
prompted the Florida Department of Financial Services (FDFS) to charge a group
of researchers to develop a Public Hurricane Loss Projection Model Project.
The Project calls for the efforts of several functional teams: A meteorological team
is in charge of developing a hurricane wind model; an engineering team develops
the building vulnerability and exposure model; an actuarial team translates the
building physical damage to insurance loss and a computer team builds a user-
friendly and stable computer platform. This report is a part of the work of the
engineering team.
The report focuses on single-family residential buildings and manufactured homes.
It covers three topics. First, the development of a unique component approach
vulnerability model is introduced. Second, the report presents the results of a
building exposure study, namely, the determination of the most common structural
types for four Florida regions. Finally, the report proposes a cost calculation model
combining the results of the previous two topics.
The component approach vulnerability model is built upon the following procedure.
Five basic damage modes, i.e. breakage of openings, loss of roof cover; loss of roof
sheathing; roof to wall connection failure, and wall failure, were defined for a
generic residential structural type. Each of these five damage modes was assigned
four levels of intensity: no damage, light, medium and heavy damage. All these
partial damage modes can then be combined in 217 possible damage states. For
iv
each of these combined damage states, their probability of occurrence for different
wind speed intervals can be estimated though Monte Carlo simulations, for each
identified structural type. The results are damage or vulnerability matrices for each
particular structural type.
The selection of the particular structural types is the result of the exposure study
presented here. To study the building exposure of Florida, the author analyzed nine
Florida counties’ property tax appraiser databases. As a result, for those nine
counties, the report presents the detailed distributions of major building
characteristics: roof type; roof cover; exterior wall structure, area, and year built.
The results are then used to define the most common structural types and their
probability of occurrence over four Florida regions: North, Central, Southeast, and
the Keys.
The results of the vulnerability modeling and building exposure studies are
combined in a cost estimation procedure. Based on an estimate of the cost
associated with each damage states, two types of damage can be estimated: the
average annual cost of hurricane induced damage for a given return period; or the
damage cost for a given hurricane scenario.
v
Table of Contents
Abstract ................................................................................................ iii
Table of Contents.................................................................................. v
List of Figures ..................................................................................... xii
List of Tables...................................................................................... xix
Acknowledgments............................................................................ xxiv
Chapter 1. Introduction ....................................................................... 1
1.1 Background.....................................................................................................1
1.2 Goals and Objectives .....................................................................................4
1.3 Scope................................................................................................................6
Chapter 2. Literature Review.............................................................. 9
2.1 Introduction....................................................................................................9
vi
2.2 Post Damage Investigations.........................................................................10
2.3 Damage Prediction Models .........................................................................12
2.4 Building Component Studies ......................................................................15
2.4.1 Roof Systems ..........................................................................................16
2.4.2 Roof to Wall Connections.......................................................................20
2.4.3 Wall Systems...........................................................................................20
2.4.4 Openings .................................................................................................22
2.4.5 Wind Pressure Capacity ..........................................................................22
2.4.6 Impact Resistance....................................................................................23
2.4.7 Summary of Building Component Resistance ........................................25
2.5 Manufactured Houses..................................................................................25
2.6 Building Load Studies..................................................................................26
2.6.1 External Pressure Field ...........................................................................27
2.6.2 Internal Pressure......................................................................................28
2.7 Structural Classification..............................................................................29
vii
2.8 Conclusions of Literature Review ..............................................................30
Chapter 3. Vulnerability Model ........................................................ 32
3.1 Introduction..................................................................................................32
3.2 Basic Damage Modes ...................................................................................33
3.3 Combined Damage States............................................................................36
3.4 Damage Matrix.............................................................................................42
3.5 Summary.......................................................................................................42
Chapter 4. Structural Classification................................................. 43
4.1 Introduction..................................................................................................43
4.2 Sources of Information ................................................................................44
4.2.1 Florida Hurricane Catastrophe Fund Exposure Database .......................44
4.2.2 HAZUS Manual ......................................................................................55
4.2.3 Tax Appraisers’ Databases......................................................................59
4.3 Information Gathered in the Structural Survey .......................................62
4.3.1 Roof Cover ..............................................................................................63
viii
4.3.2 Roof Type................................................................................................66
4.3.3 Exterior Wall ...........................................................................................68
4.3.4 Year Built ................................................................................................70
4.3.5 Number of Stories ...................................................................................72
4.3.6 Building Areas ........................................................................................72
4.4 Counties Statistics ........................................................................................73
4.4.1 Brevard County .......................................................................................73
4.4.2 Hillsborough County...............................................................................82
4.4.3 Pinellas County .......................................................................................89
4.4.4 Escambia County ....................................................................................96
4.4.5 Leon County..........................................................................................101
4.4.6 Walton County ......................................................................................105
4.4.7 Broward County ....................................................................................112
4.4.8 Palm Beach County...............................................................................118
4.4.9 Monroe County .....................................................................................124
ix
4.4.10 Define the Most Common Structural Types .......................................130
4.4.11 Distribution of the Most Common Structural Types Per County .......132
4.5 Florida Regions ..........................................................................................136
4.5.1 Introduction ...........................................................................................136
4.5.2 Distribution of Structural Types Per Region.........................................139
4.5.3 Result Analysis......................................................................................141
4.5.4 Error Estimation ....................................................................................142
Chapter 5. Statistical Analysis of Florida Manufactured Homes 145
5.1 Introduction................................................................................................145
5.2 Sources of Information ..............................................................................147
5.2.1 Florida Hurricane Catastrophe Fund Exposure Database .....................147
5.2.2 HAZUS Manual ....................................................................................151
5.2.3 Tax Appraiser’s Databases....................................................................152
5. 3 Information Gathered in the Survey.......................................................153
5.4 Counties Statistics ......................................................................................154
x
5.4.1 Brevard County .....................................................................................155
5.4.2 Pinellas County .....................................................................................157
5.4.3 Hillsborough County.............................................................................160
5.4.4 Broward County ....................................................................................162
5.4.5 Palm Beach County...............................................................................162
5.4.6 Monroe County .....................................................................................164
5.4.7 Escambia, Leon and Walton .................................................................166
5.5. Most Common Manufactured Homes Types in Florida........................167
5.6 Result Analysis ...........................................................................................169
Chapter 6. Cost Estimation ............................................................. 171
6.1 Introduction................................................................................................171
6.2 Average Annual Damage...........................................................................171
6.3 Numerical Example....................................................................................173
Chapter 7. Conclusions and Recommendations ............................ 178
7.1 Summary and Conclusion .........................................................................178
xi
7.2 Uncertainties...............................................................................................180
7.3 Research Underway and Recommendations ...........................................182
References.......................................................................................... 184
xii
List of Figures
Figure 1. Public Hurricane Loss Prediction Model Project Layout: Four
Components and Their Functions ..............................................................................4
Figure 2. Qualitative Representation of Relative Damage to Building Component in
Hurricane Andrew....................................................................................................12
Figure 3. Qualitative Knowledge of Resistance Capacity of Building Components
..................................................................................................................................25
Figure 4. Components of a Single Family Home.....................................................34
Figure 5. Venn Diagram for the Basic Damage Modes of a Masonry Home..........37
Figure 6. Venn Diagrams of the Combined Damage States (subsets of Figure 5) ..39
Figure 7. Total Dollar Value Distribution in Florida ...............................................46
Figure 8. Ratio of Residential Buildings in Florida.................................................47
Figure 9. Ratio of Commercial Buildings in Florida ...............................................47
Figure 10. Ratio of Home Owner in Florida ............................................................48
xiii
Figure 11. Ratio of Renters in Florida .....................................................................48
Figure 12. Ratio of Frame Exterior Wall in Florida.................................................49
Figure 13. Ratio of Joisted Masonry Homes in Florida...........................................49
Figure 14. Ratio of Non-Combustible in Florida.....................................................50
Figure 15. Ratio of Masonry Non-combustible in Florida.......................................50
Figure 16. Ratio of Modified Fire Resistive in Florida............................................51
Figure 17. Ratio of Heavy Timber Joisted Masonry in Florida ...............................51
Figure 18. Ratio of Superior Non-combustible in Florida .......................................52
Figure 19. Ratio of Superior Masonry Non-combustible in Florida........................52
Figure 20. Ratio of Masonry Veneer Distribution ...................................................53
Figure 21. Ratio of Masonry Distribution................................................................53
Figure 22. Ratio of Semi-wind Resistive Distribution.............................................54
Figure 23. Ratio of Wind Resistive Distribution .....................................................54
Figure 24. Four Regions with Sample Counties Highlighted ..................................61
Figure 25. Distribution of Year Built for Single Family Homes in Brevard County
..................................................................................................................................71
Figure 26. Brevard County Single Family Area Range Distribution (unit: ft2) .......73
xiv
Figure 27. Brevard County Roof Material Detail Distribution................................74
Figure 28. Brevard County Roof Type Detail Distribution .....................................76
Figure 29. Brevard County Exterior Wall Material Detail Distribution ..................77
Figure 30. Brevard County Roof Height Distribution .............................................79
Figure 31. Brevard Year Built Distribution .............................................................80
Figure 32. Brevard County Area Distribution..........................................................81
Figure 33. Hillsborough County Roof Cover Detail Distribution ...........................82
Figure 34. Hillsborough County Roof Type Detail Distribution .............................84
Figure 35. Hillsborough County Exterior Wall Detail Distribution ........................86
Figure 36. Hillsborough County Year Built Distribution ........................................88
Figure 37. Hillsborough County Area Distribution .................................................89
Figure 38. Pinellas County Roof Cover Detail Distribution....................................90
Figure 39. Pinellas County Roof Type Distribution ................................................92
Figure 40. Pinellas County Exterior Wall Detail Distribution.................................93
Figure 41. Pinellas County Year Built Distribution.................................................94
Figure 42. Pinellas County Area Distribution..........................................................95
Figure 43. Escambia County Roof Cover Material Detail Distribution ..................96
xv
Figure 44. Escambia Roof Type Detail Distribution ...............................................97
Figure 45. Escambia County Exterior Wall Material Detail Distribution ...............98
Figure 46. Escambia County Year Built Distribution............................................100
Figure 47. Leon County Roof Cover Distribution .................................................101
Figure 48. Leon County Roof Type Distribution...................................................102
Figure 49. Leon County Exterior Wall Material Detail Distribution.....................103
Figure 50. Leon County Year Built Distribution ...................................................105
Figure 51. Walton County Roof Cover Material Detail Distribution ....................106
Figure 52. Walton County Roof Type Distribution ...............................................108
Figure 53. Walton County Exterior Wall Material Detail Distribution .................110
Figure 54. Walton County Year Built Distribution................................................112
Figure 55. Broward County Roof Cover Material Detail Distribution ..................113
Figure 56. Broward County Roof Type Detail Distribution ..................................114
Figure 57. Broward County Exterior Wall Distribution ........................................115
Figure 58. Broward County Year Built Distribution .............................................117
Figure 59. Broward County Area Distribution ......................................................118
Figure 60. Palm Beach Roof Cover Material Detail Distribution..........................119
xvi
Figure 61. Palm Beach Roof Type Detail Distribution..........................................120
Figure 62. Palm Beach Exterior Wall Material Detail Distribution.......................121
Figure 63. Palm Beach County Year Built Distribution ........................................123
Figure 64. Palm Beach Area Range Distribution...................................................124
Figure 65. Monroe County Roof Cover Material Detail Distribution ...................125
Figure 66. Monroe County Roof Type Detail Distribution....................................126
Figure 67. Monroe County Exterior Wall Detail Distribution...............................127
Figure 68. Monroe County Year Built Distribution...............................................129
Figure 69. Monroe County Area Range Distribution.............................................130
Figure 70. Four Regions with Sample Counties Highlighted and Name Marked .137
Figure 71. Ratio of Manufactured Housing ...........................................................147
Figure 72. Ratio of Manufactured Home – Fully Tied Down Pre 7/13/1994. .......148
Figure 73. Ratio of Manufactured Home – Fully Tied Down Post 7/13/1994 ......148
Figure 74. Ratio of Manufactured Home – Partially Tied Down ..........................149
Figure 75. Ratio of Manufactured Home – not Tied Down...................................149
Figure 76. Ratio of Manufactured Home – Unknown ...........................................150
Figure 77. Most Common Type of Manufactured Tie Downs ..............................153
xvii
Figure 78. Brevard County Manufactured Home Roof Material Distribution.......155
Figure 79. Brevard County Manufactured Home Roof Type Distribution............155
Figure 80. Brevard County Manufactured Home Exterior Wall Material
Distribution ............................................................................................................156
Figure 81. Brevard County Manufacture Homes Year Built Distribution.............156
Figure 82. Brevard County Manufactured Home Area Range Distribution ..........157
Figure 83. Pinellas County Manufactured Home Roof Cover Material Distribution
................................................................................................................................157
Figure 84. Pinellas County Manufactured Home Roof Type Distribution ............158
Figure 85. Pinellas County Manufactured Home Exterior Wall Material
Distribution ............................................................................................................158
Figure 86. Pinellas County Manufactured Home Area Range Distribution ..........159
Figure 87. Pinellas County Manufactured Home Year Built Distribution ............159
Figure 88. Hillsborough County Manufactured Home Roof Cover Material
Distribution ............................................................................................................160
Figure 89. Hillsborough County Manufactured Home Roof Type Distribution....161
Figure 90. Hillsborough County Manufactured Home Year Built Distribution ....161
Figure 91. Broward County Manufactured Home Year Built Distribution ...........162
xviii
Figure 92. Palm Beach County Manufactured Home Roof Type Distribution......163
Figure 93. Palm Beach County Manufactured Home Exterior Wall Material
Distribution ............................................................................................................163
Figure 94. Palm Beach County Manufactured Home Year Built Distribution......164
Figure 95. Monroe County Manufactured Home Roof Cover Material Distribution
................................................................................................................................164
Figure 96. Monroe County Manufactured Home Roof Type Distribution ............165
Figure 97. Monroe County Manufactured Home Exterior Wall Distribution .......165
Figure 98. Monroe County Manufactured Home Area Range Distribution ..........166
xix
List of Tables
Table 1. Structural Characteristics for HAZUS Vulnerability Modeling ................14
Table 2. Assumed Probabilities of Occurrence of Sub-damage Modes Oi,
Conditional on Wind Speeds Intervals Associated with the Wind Speeds v. ..........35
Table 3. Example of Estimated Probabilities of Damage States, Conditional on
Wind Speed Intervals Associated with the Speeds v ...............................................42
Table 4. ISO Construction Classification.................................................................45
Table 5. Percentage of the Types of Roof Covering in the Existing Populations of
Residential Buildings Including Apartments and Condominiums...........................58
Table 6. Percentage of the Types of New and Re-roofing Systems Being Installed
on Residential Buildings ..........................................................................................58
Table 7. Sample Manufactured Home Materials in HAZUS Manual......................59
Table 8. Criteria of Defining Manufactured Homes Types in HAZUS...................59
Table 9. Contact Building Officials in Each of the Sample Counties......................61
Table 10. Three Major Building Type Distributions ...............................................62
xx
Table 11. Broward County Single Family Houses Roof Cover Type Distribution. (1)
..................................................................................................................................64
Table 12. Broward County Single Family Houses Roof Cover Type Distribution (2)
..................................................................................................................................65
Table 13. Broward County Single Family Houses Roof Cover Type Distribution .66
Table 14. Comparison of Gable Roof and Hip Roof ...............................................67
Table 15. Gambrel Roof and Mansard Roof............................................................67
Table 16. Brevard County Single Family Buildings Exterior Wall Materials.........70
Table 17. Percentage of Number of Story Distribution of Pinellas County ............72
Table 18. Brevard County Combined Roof Material Distribution ..........................75
Table 19. Brevard County Combined Roof Type Distribution................................76
Table 20. Brevard County Combined Exterior Wall Distribution ...........................78
Table 21. Hillsborough County Combined Roof Cover Distribution......................83
Table 22. Hillsborough County Combined Roof Type Distribution .......................85
Table 23. Hillsborough County Exterior Wall Combined Distribution...................87
Table 24. Pinellas County Roof Cover Combined Distribution ..............................91
Table 25. Pinellas County Roof Type Distribution..................................................92
Table 26. Pinellas County Exterior Wall Distribution.............................................93
xxi
Table 27. Escambia County Combined Roof Cover Material Distribution.............97
Table 28. Escambia Roof Type Combined Distribution..........................................98
Table 29. Escambia County Exterior Wall Material Distribution............................99
Table 30. Escambia Number of Story Distribution..................................................99
Table 31. Leon County Roof Cover Distribution...................................................102
Table 32. Leon County Combined Roof Type Distribution ..................................103
Table 33. Leon County Exterior Wall Material Distribution.................................104
Table 34. Walton County Combined Roof Cover Material Distribution...............107
Table 35. Walton County Combined Roof Type Distribution...............................109
Table 36. Walton County Exterior Wall Material Distribution .............................111
Table 37. Broward County Roof Cover Material Distribution ..............................114
Table 38. Broward County Roof Type Distribution ..............................................115
Table 39. Broward County Exterior Wall Distribution..........................................116
Table 40. Palm Beach Combined Roof Cover Material Distribution ....................120
Table 41. Palm Beach Roof Type Distribution......................................................121
Table 42. Palm Beach Exterior Wall Material Combined Distribution.................122
Table 43. Monroe County Roof Cover Material Distribution................................126
xxii
Table 44. Monroe County Roof Type Combined Distribution..............................127
Table 45. Monroe County Exterior Wall Combined Distribution .........................128
Table 46. Simplified Structural Characteristic Types............................................130
Table 47. Assumed Structural Type Definitions....................................................131
Table 48. Central Region Counties Model Distribution ........................................133
Table 49. Northern Region Counties Model Distribution......................................134
Table 50. Southern Region Counties Model Distribution......................................135
Table 51. The Key Region – Monroe County Model Distribution........................135
Table 52. Number of Counties and Population in Each Region ............................138
Table 53. Probability of Occurrence of Structural Type for 3 Regions. ................140
Table 54. Probability of Occurrence of Structural Types for the Key Region ......141
Table 55. p̂ Values and Error for Each Type for 3 Regions .................................144
Table 56. Sample Manufactured Home Materials in HAZUS Manual..................151
Table 57. Criteria of Defining Manufactured Homes in HAZUS Manual ............151
Table 58. Probability of Occurrence of Manufactured Home Types for 3 Regions.
................................................................................................................................168
Table 59. Probability of Occurrence of Manufactured Home Types in Key Regions.
................................................................................................................................168
xxiii
Table 60. p̂ Values and Error for Each Type for 3 Regions .................................169
Table 61. Damage Matrix for Three Structural Types...........................................174
Table 62. Sample Wind Field Model Data.............................................................174
Table 63. Replacement Cost of Three Damage States...........................................174
xxiv
Acknowledgments
This research is a component of Public Hurricane Loss Projection Model Project
sponsored by the Florida Department of of Financial Services through the
International Hurricane Research Center at Florida International University. Dr.
Shahid Hamid at Florida International University is the project manager. Many
experts of the multidisciplinary team contributed to the work reported in this report.
They include Dr. Jean-Paul Pinelli and Dr. Chelakara Subramanian from the
Florida Institute of Technology, Dr. Kurtis Gurley from University of Florida, Dr.
Sneh Gulati from the Florida International University and the computer group from
the Florida International University lead by Dr. Shu-ching Chen. Dr. Emil Simiu
from National Institute of Standard Technology was the catalyst for the formulation
of the damage prediction model. Finally graduate students Isabelle Reisse from
Florida Institute of Technology and Anne Cope from University of Florida
provided valuable input. The report itself is adapted from Ms. Liang Zhang
graduate thesis.
1
Chapter 1. Introduction
1.1 Background
Each year, strong wind can cause billions of property damage worldwide. Within
the U.S., windstorms are one of the costliest natural hazards, far outpacing
earthquakes in total damage (Wind Engineering, 1997; Landsea et al, 1999).
According to statistics published by the Munich Re Group for the year 2001,
windstorms were responsible worldwide for 55 % of the $36 billion in economic
losses and 88% of the $11.5 billion in insured losses due to all natural disasters
combined. Similar percentages were recorded for the U. S.
Over half of the hurricane-related damage in the U. S. occurs in the state of Florida,
which has $1.5 trillion in existing building stock currently exposed to potential
hurricane devastation. With approximately 85% of the rapidly increasing
population situated on or near the 1200 miles of coastline, Florida losses will
continue to mount in proportion to coastal population density. It is therefore critical
for the state of Florida to be able to estimate expected losses due to hurricanes and
their measure of dispersion. Among all the industries strut by hurricane, insurance
industry inevitably pay the heavy price in terms of the economic loss. For this
reason the Florida Department of Insurance asked a group of researchers to develop
a public hurricane loss projection model. This project is referred as the Public
Hurricane Loss Prediction Model Project, or in short the Project.
2
The goal of the Project is to develop and maintain a computer model to assess
hurricane risk, and to project annual expected insured residential losses for specific
sites, zip codes, counties and regions in Florida. These losses can be estimated for
both individual property and for entire portfolios of residential properties. The
proposed model shall also project insured losses for user defined scenarios and
historical events.
While the Project is funded by the state of Florida and the Florida Department of
Insurance (DOI), it will be developed predominantly by academic experts
contracted by the International Hurricane Center at Florida International University.
These experts come from the various universities in Florida, the National Oceanic
and Atmospheric Administration and National Institute of Standards Technology.
The Project is being developed by the best available methodologies, techniques,
theories and scientific principles. The proposed model will be developed without
bias and will be non-proprietary and transparent. It will be subject to external
review and will comply with the standards set by the Florida Commission on
Hurricane Loss Projection Methodology (FCHLPM). It is expected the model and
its components will be available to the insurance and reinsurance industry.
The proposed model can be used to (a) provide assistance to the Florida
Department of Insurance and the insurance industry in the rate making process; (b)
provide a state of the art non-proprietary wind field model for public use; (c)
provide a check on the assumptions, analysis and results generated by the
proprietary models; (d) help evaluate reinsurance risk; (e) assess the efficacy of
disaster mitigation strategies.
The proposed risk assessment model has meteorological, engineering, GIS,
financial and actuarial components in accordance to four modules: wind, exposure,
vulnerability and loss. It draws upon the expertise of a team of meteorologists,
3
wind and structural engineers, statisticians, computer scientists, actuaries, and
financial experts from Florida International University, from the State University
System and elsewhere.
Meteorologists from National Oceanic and Atmospheric Administration (NOAA)
and Florida State University are responsible for developing simulated hurricane
sets with various characteristics and wind field simulation. Wind and Structural
engineers are responsible for developing the exposures and vulnerability model to
estimate the physical damage for different exposure and wind speed. And then the
loss module comprising actuaries and financial experts will compute the economic
losses for a given insurance exposure, taking into account specific coverage terms
like limits and deductibles, etc. Finally, the computer platform for the model and its
components is developed. The computational architecture and programming
language are selected. Object oriented analysis of the model components is
conducted. The components are going to be properly integrated. The process
includes the development of databases for input and output quantities, the technical
and visual specifications, user interface, algorithms and computer codes for the
simulation model according to accepted scientific, simulation, and software
engineering principles. The process of translating the models, algorithms, and
design data into computer programs using appropriate programming languages will
be specified. The following Figure 1 illustrates the four components and their
functions.
4
Figure 1. Public Hurricane Loss Prediction Model Project Layout: Four
Components and Their Functions
The meteorological group comprises experts from the Hurricane Research Division,
NOAA, and the Florida State University Department of Meteorology. The
wind/structural engineers come from the Department of Civil Engineering at the
Florida Institute of Technology, Department of Civil Engineering at the University
of Florida and the National Institute of Standards and Technology. Financial and
actuarial experts are from the Department of Finance and International Hurricane
Center at the Florida International University. Computer scientists are mainly from
Department of Computer Science in Florida International University. The project
statistician is from Department of statistics at the Florida International University.
1.2 Goals and Objectives
This report, as a component of wind and structural engineering group’s work,
focuses on two of the four modules of the Project ---- exposure and vulnerability
modules. In any hazard assessment model, exposure can be defined as
Meteorology group to develop wind field simulations model
Engineering group to develop vulnerability model and exposure study
Financial group to compute insurance loss considering specific coverage terms
Computer group to build up computer flatform for the whole project
5
representation of the geographical and structural attributes of the properties, such as
terrain roughness, building size, type of overall structural, roof shape and wall
materials. In our model, we narrow down the exposure definition to structural
attributes of the buildings only and leave the geographical exposure to wind field
modeling group.
Structural exposure’s importance lies in its ability to show the built-in factors that
affect the wind resistance of a structure. The structural classification survey that
we conducted in 9 Florida counties reveals the statistical distributions of the most
important structural exposure characteristics, such as roof type, exterior wall and
area, etc, so that viewer of the classification results can get the building information
of a certain area at a glance.
In our Project, the structural classification serves two functions: first of all, it
provides the most common structural types of for Monte Carlo simulation to
generate typical building vulnerability matrices. Secondly, the statistical
distribution of each structural type will be used in annual hurricane loss prediction
calculations.
Vulnerability can be defined as the relationship between hazard and consequent
physical damage or an estimate of the cost to repair the damage. In our model, the
damage is defined as the ratio of replacement over the total replacement cost of the
structure. There are many imponderables involved in defining the wind
vulnerability of buildings like wind speed, wind direction, storm duration, building
size and geometry, roof shape, terrain conditions, shielding by surrounding
structures, construction quality, building codes and their implementation, etc. In
our Project, we use 3-second peak gust wind speed in meter per second in the
center of a zip code area as a linkage of the hurricane to the damage of structure. A
6
Monte Carlo simulation model will be run to link wind speed to structural damage,
thus to building vulnerability matrix.
The results of vulnerability model and structural classifications will be used in the
calculation of annual prediction of wind-induced loss. The prediction calculations
will be conducted in steps: First of all, the probable damage for a certain type of
structure in the zip code area subjected to a wind speed in the interval {v- v/2, v +
v/2} m/s is the sum of all the relative possible damage states for speed v in that
interval multiplied by their probabilities of occurrence. And then a similar
procedure applies to each of the wind speeds to generate the damage of this certain
type. After the damage for this certain type is obtained. The damage of each
different structural type is to be calculated. And after the damages for each
structural types are multiplied by the respective relative frequency of those types in
the building population of the area, the average damage of all the buildings the area
will be obtained. The process is repeated for each area, and the results for each area
are added to obtain the estimated expected hurricane-induced annual damage to
buildings for the entire state.
1.3 Scope
The report comprises four major parts arranged by the inner-logical relationships.
First of all, a thorough literature search results are introduced. The literature search
comprehensively summarizes the most recent progress and work done in the
hurricane risk assessment field. It contains 5 topics: available research and results
in post damage investigation; summary of available damage prediction model;
detail summary of available building component study; studies on manufactured
home; available results in building load studies. The literature survey is in terms of
both descriptive and numerical data. The results of research conducted by different
7
people and groups are compared and analysis, the limitations of some research are
also pointed out.
Based on the study and analysis of existing vulnerability model, the author
proposes a component approach vulnerability model and discusses its development.
The component approach explicitly accounts for both the resistance capacity of the
various building components, such as roof or wall and the load effects produced by
wind events to predict damage at various wind speeds. In the component approach
the resistance capacity of a building can be broken down into the resistance
capacity of its components and the connections between them. Damage to the
structure occurs when the load effects from wind or flying debris are greater than
the component’s capacity to resist them. Once the strength capacities, load
demands, and load path(s) are identified and modeled, the vulnerability of a
structure at various wind speeds can be estimated, which enable the prediction of
wind-induced damage and of corresponding repair/replacement. This approach
makes use of probabilistic information on basic damage model.
A nine-county-structural classification survey has been completed. The procedure
and results are introduced. Through collecting and processing nine Florida county
property tax appraisers’ database, the author defines the most common types for
each county. Starting from this point, a development of methodology to translate
county’s most common type into region’s most common types of structures has
been conducted. The uncertainties and errors are discussed.
Finally, the annual cost estimation model based on the outcomes obtained from
vulnerability model and structural statistics is introduced. And a numerical example
is presented. It is also introduced that the cost estimation model can be used to
generate loss prediction for any number of residential buildings or an individual
wind-storm event. Limitation and uncertainties of the report is discussed.
8
Estimates are affected by uncertainties regarding on one hand the behavior and
strength of the various components and, on the other, the load effects produced by
hurricane winds. The recommendations for future research are also discussed at the
end of the report.
9
Chapter 2. Literature Review
2.1 Introduction
As mentioned before, the engineering team’s goal is to develop models to predict
the amount of physical damage to buildings, given the wind field. The first task of
the engineering team was to review the relevant literature and define the state of the
art in vulnerability studies. An extensive literature includes potentially useful
papers and reports in the following areas: post-disaster investigations, damage
prediction models, building component studies, manufactured homes, building
exposure investigation and building loading studies. The current knowledge base is
summarized in the following sections.
Because hurricanes or severe wind storms have a great impact on the insurance
industry, many insurance companies develop risk assessment models to assess the
financial risk to their portfolios due to the hurricanes. With the help from
catastrophe scientists or modelers, they have been on the forefront of examining the
impact of severe winds on property, and routinely perform analysis of the wind
induced economic losses to insurance portfolios based on vulnerability models.
However, because of the proprietary nature of such models, such effort is not
available for public use or scrutiny. So our literature research is only restricted to
the materials published in the public domain.
10
The results presented in this chapter included also the findings of two other
graduate students: Isabelle Reisse from Florida Institute of Technology and Anne
Cope from the University of Florida.
2.2 Post Damage Investigations
Numerous papers are available on damage from hurricanes Alicia, Andrew, Hugo,
Iniki, and Opal. Much of the evidence they contain is anecdotal, but they
incorporate useful information on types of failures commonly encountered and
recommendations to prevent similar failures in future events. For example, the
damage to buildings in the Houston-Galveston area during Hurricane Alicia was
attributed to the lack of adequate hurricane resistant construction, rather than to the
severity of the storm (Kareem 1985, 1986). A similar conclusion was reached on
damage to buildings during Hurricane Hugo (Sill and Sparks 1990).
A reliability analysis of roof performance during Hurricane Andrew found actual
performance to be better than predicted by the governing building code at the time,
although the authors stress the need for further research to quantify statistically
both construction characteristics and damage due to storms (NAHB 1999). Phang
in 1999 also offers a gathering of observations of the damage on low rise buildings
caused by Andrew. He observed that plywood sheathing performed remarkably
better than board sheathing, diagonal bracing was critical at gable ends, and gable
roofs showed much more structural damage than hip roofs.
Some researches have also been conducted in Australia by Mahendren (1995) who
gives an overview of the typical damages encountered by low rise buildings in the
tropics, subjected to either hurricanes or severe storms. In addition, he and the
Australian scientific community also stress the fact that full scale testing is
11
necessary to better predict the behavior of the entire building system when
subjected to those high speed winds.
While these studies are extremely valuable for the development of safer housing,
they do not offer a sufficient basis from which to draw reliable quantitative
conclusions. In particular, damage versus wind speed curves cannot be predicted
with reasonable accuracy from the information presented in most post-disaster
reports. The information obtained from these studies does, however, provide a
means of validating the results of a detailed component approach. One would
expect the most common types of failures detailed in post-disaster reports to be the
same as the types of failure typically obtained from Monte Carlo simulations of
hurricane force winds on typical structures.
Some of these studies also provide a means of estimating the distribution of the
building stock in Florida cities. The most comprehensive studies, undertaken by the
National Association of Home Builders (NAHB) following Hurricanes Andrew and
Opal, include information on the sample size and types of homes investigated
(NAHB 1993, 1996). This information, in combination with census data and other
resources, could be useful for predicting typical sizes and types of homes in other
Florida areas. Furthermore, the storm damage reports help to set priorities for
research efforts by identifying the building components that experience most
damage.
A qualitative representation of the relative damage to major building components
during Hurricane Andrew is provided in Figure 2 (courtesy of Anne Cope). The
relatively comprehensive information on the occurrence of damage to both roof
cover and openings is in stark contrast with the paucity of information available on
the resistance capacities of those two types of components. This will be
demonstrated in the following sections.
12
Relative Damage to Building Components in Hurricane Andrew
Roof Cover
Openings
Roof Sheathing
Walls
Roof to WallConnectionLow High
Occurrence of Damage
Figure 2. Qualitative Representation of Relative Damage to Building Component in Hurricane Andrew
2.3 Damage Prediction Models
Only a handful of studies available in the public domain have been found that
predict damage for hurricane prone areas. A few studies use post-disaster
investigations to fit damage versus wind speed curves. For example, the
relationship between home damage from insurance data and wind speed was found
for Typhoons Mireille and Flo (Mitsuta et al 1996). A similar relationship was
established by Bhinderwala in his Master thesis (1995). Bhinderwala’s thesis used
limited insurance loss data from Hurricane Andrew to set up a relationship between
gradient wind speed and loss ratio for zip code sized areas. His research shows the
linear increase of loss ratio (insurance loss/insurance policy) in lower range of wind
speed and a large increase in average loss ratios occurred in higher wind speed
range. The preciseness of his research is constrained by the lack of consistent
13
quality and quantity of insurance file. Besides the data used in his research are
maintained by the insurance company which are generally non-technical in nature.
The information obtained from insurance data is of some interest, but they are
mostly inapplicable to residential construction in Florida.
A study by Holmes (1996) presents the vulnerability curve for a fully-engineered
building with lognormally-distributed strength, but clearly indicates the need for
more thorough post disaster investigations to better define damage prediction
models (Holmes 1996). Since residential buildings are not fully-engineered, this
information is also not readily applicable.
A method for predicting the percentage of damage within an area as a function of
the gradient wind speed, gust factor, average value of the buildings, and two
parameters which govern the rate of damage increase with wind speed is also
available (Sill and Kozlowski 1997). The difficulty with this process is that it
requires determining the values of the damage rate parameters for the region in
question. Knowledge of the rate of damage with increasing wind speed is currently
scarce, and predictions of these coefficients would lack any reasonable accuracy.
And predictions for public spending have been estimated (Boswell et al 1999).
Such predictions are not applicable to the current project because they deal only
with the public costs of emergency management and recovery. Losses to individual
homeowners are not taken into account. Huang et al. (2001) presented a risk
assessment model based on an analytical expression for the vulnerability curve.
The expression is obtained by regression techniques from insurance claim data for
Hurricane Andrew and Hugo. With the wind field model and regression analysis
vulnerability curves, one can achieve event-based results. Although such an
approach is simple, it is highly dependent on the type of construction and
construction practices common to the areas represented in the claim data. Recent
14
changes in building codes would not be reflected by such curves. A.C.Khanduri et
al (2003) also presented a similar method of assessment of vulnerability and a
methodology to translate a known vulnerability curves from one region to another
region.
Lastly, a damage prediction model that incorporates elements of a component
approach is being implemented for the HAZUS project. It collected structural
statistics information.
Based on the results of these statistical surveys, HAZUS provides vulnerability
curves for residential buildings with the most prevalent characteristics in South
Florida, listed in Table 1. Each vulnerability curve corresponds to a building with
asphalt shingle roof and a combination of the following parameters:
Table 1. Structural Characteristics for HAZUS Vulnerability Modeling
Building Size Different floor plan areas: from1200 to 1800 sqft
Roof shape Hip or Gable Wall construction Wood, Un-reinforced, Reinforced masonry
Roof sheathing attachment 6d or 8d
Roof-wall connections Strapped or toe nailed Garage Yes or No Number of stories one or two
15
The total number of possible combinations of parameters that can be considered
and therefore used in the simulation is 96, each with different characteristics. All
the 96 possibilities have been tested in HAZUS, but all the results are not presented
in the manual. Damage state vs. maximum peak gust wind speed curves are
presented in the manual for 12 buildings with some of these properties: one and
two story high, un-reinforced masonry and wood walls, toe nailed roof trusses or
strapped roof trusses, hip or gable roofs and no garage. The curves were obtained
for fenestration, roof cover, roof sheathing and roof/wall damage. Every
vulnerability curve was developed for various surface roughness lengths: 0.03m
(open terrain), 0.35m (typical suburban terrain), 0.7m (suburban terrain with some
trees or densely spaced homes) and 1.0m (treed suburban terrain). But due to the
proprietary property of this project, detail modeling procedure is not available for
the public scrutiny.
In summary, the information currently available in the public domain is not
sufficient to accurately predict damage to Florida buildings in the event of a
hurricane strike. Given the information currently available, using damage to wind
speed vulnerability curves from existing post-disaster investigation reports and
damage prediction models would require the acceptance of wide uncertainty
bounds.
2.4 Building Component Studies
This and the following two sections is essentially the contribution from Anne Cope
from the University of Florida. They are included here for the sake of completeness.
A detailed component approach to determining the vulnerability of buildings to
damage at various wind speeds involves analyzing both the resistance capacity of
the building and the load effects produced by wind events. HAZUS is an example
of a damage prediction model that uses this approach (Minor and Schneider 2001).
16
An advantage of using this engineered approach is its flexibility. The resistance
capacity of a building can be broken down into the resistance capacity of its
components and the connections between them. The most vulnerable types of
components and connections for residential low-rise structures, which make up the
overwhelming bulk of Florida structures, include the roof system, roof to wall
connections, wall systems, wall-to-foundation connections, openings, and in the
case of manufactured home the anchors into the ground. Damage to the structure
occurs when the load effects from wind or flying debris are greater than the
component’s capacity to resist them. It was found from post-disaster investigations
that component failures allowing penetration of the building envelope cause
damage to the contents of the structure and lead to further structural damage
(NAHB, 1993). Literature obtained to quantify the resistance capacity of each
building component is summarized in the following paragraphs.
2.4.1 Roof Systems
The roof system is made up of three subcomponents, the roof covering, sheathing,
and trusses or rafters. Damage from hurricane events occurs from the loss of roof
cover, the loss of sheathing, or, though less likely, the loss of the entire roof. The
resistance capacities of each subcomponent are discussed in the following sections.
2.4.1.1 Roof Covering
One portion of the resistance capacity of the roof system to wind uplift includes the
ability of the shingles, tiles, or other roof covering to stay attached to the roof
sheathing. The loss of covering, though not vital to the structural integrity of the
building, can contribute significantly to the damage of the contents of the structure
and increase insurance losses. Also, the loss of roof covering creates debris that
17
will impact structures in the area, potentially breaking windows and doors and
increasing damage.
In general, there is limited information available about the uplift capacity of
shingles and tiles. Most building codes (with the exception of the National
Building Code published by the Building Officials and Code Administrators
International, Inc.) do not indicate a minimum wind resistance capacity for roof
coverings (Devlin 1996). Though contractors are not required to specify roof
coverings that meet their standards, Factory Mutual (FM), Underwriters
Laboratories (UL), and the American Society for Testing and Materials (ASTM)
have developed test methods for commercial and residential roof coverings. These
test methods do not model the pressure developed on both sides of the roof surface
that occurs in high wind events and do not represent the wide variety of roof slopes
and shapes that exist. The current UL test method for asphalt shingles, developed
over 30 years ago, involves blowing steady 60 mph winds over a sample section.
The choice of wind speed was determined by the fan capacity at the time and has
not been changed (Devlin 1996). Furthermore, the steady fan does not model the
turbulent conditions under which the shingles will have to perform, which will
influence roof covering performance (Hosoya, Cermak and Dodge, 1999). The
Metro Dade Building Code (MDBC) Compliance Office does require that asphalt
shingles withstand both the sample section fan test at 110 mph instead of 60 mph
and a wind driven rain penetration test. Tiles must withstand a static uplift test
performed by pulling on a concrete anchor or epoxy bolt driven in to the tile
(Devlin 1996). This is the only region in the United States known to require roof
covering to meet minimum criteria.
Since the structural capacity of the roof is generally thought to be independent of
the covering or lack of covering on the sheathing, studies on this subject are limited.
One study provides an approach for estimating the wind action on shingles, but
18
does not predict failure loads, citing the unknown capacity of the adhesive (Peterka
et al 1997). Another study, “Fixing Studies for MRTI Normal Weight Tiles,”
conducted by Cherry (1991) has not yet been located, though the conclusions can
be inferred from the HAZUS report (Minor and Schneider 2001). Results of the
study include the uplift capacity for various types of normal weight tile roofs that
are attached with nails. Since most Florida tile roofs are mortar set, not nailed, the
results of the study are not readily applicable to Florida homes. The study will
serve as a baseline from which to infer performance of the mortar set roofs. Since
only two studies have been found in the literature review, there is a strong need for
further research and continued literature search.
In summary, the capacity of roof coverings is currently very limited. While the
asphalt shingle fan tests and tile uplift tests are certainly a step in the right direction,
they do not provide information about the failure capacity of these building
materials. Also, the tests do not generally represent the conditions these
components would face in hurricane events. Homes of different roof slope, shape,
orientation to the wind, and having different surroundings will have different
pressure fields on the roof surface. Currently, there is a great deal of uncertainty in
the prediction of these pressure fields. The fan test simulates only one of the many
possibilities. Limited research has been conducted in this area, and further study is
needed. For these reasons, the literature search will continue for articles
concerning roof covering. Until further information can be gathered, a reasonable
starting point for estimating the resistance capacity of roof covering would be to
assume that products in South Florida have a mean resistance slightly greater than
required by the MDBC, and products elsewhere have a lower mean resistance
capacity.
19
2.4.1.2 Sheathing
A second portion of the wind uplift resistance capacity of the roof system includes
the ability of the sheathing to remain fastened to the trusses. A considerable body
of research has been conducted in this area following the Hurricane Andrew strike
in Southern Florida, primarily by the American Plywood Association (APA) and by
Clemson University. Conclusions drawn from these studies indicate the capacity of
sheathing panels is best represented by treating the panel as a whole, rather than
evaluating the capacity of individual fasteners (Mizzel 1994). Findings show that
the tributary area method falls short of determining the strength of the weakest link
individual fastener due to the load sharing which takes place between fasteners.
Results from the studies include mean failure pressures and Coefficients of
Variation (COVs) for panels of different wood species with different fastener sizes
and schedules (Cunningham 1993, Mizzel 1994, Rosowsky et al 1998). Sufficient
information exists in these studies to determine confidence intervals on the uplift
capacity of a panel on a typical Florida home, given knowledge of the species of
wood most widely available at the time of construction and the most typical
fastener size used in the area.
2.4.1.3 Trusses and Rafters
The last subcomponent of the roof system, the trusses or rafters, is less important to
the prediction of damage. Individual trusses or rafters will not fail in uplift before
massive damage has already occurred from the loss of sheathing. The contribution
of the trusses or rafters to the overall capacity of the building occurs in the
resistance to the loss of the entire roof as a whole unit. While this type of failure is
less likely to occur than the loss of sheathing, it does happen and is usually due to
the failure of the roof to wall connection (NAHB 1993). The capacity of this
connection is discussed in the next section.
20
2.4.2 Roof to Wall Connections
Post disaster studies have found that the roof to wall connection is another vital
characteristic of the overall resistance of the home to hurricane force winds (NAHB
1993). The uplift capacities of several types of roof to wall connections for light
frame wood construction are available (Reed et al 1997, Canfield et al 1991). The
study conducted by Reed further investigated the potential for load sharing between
rafter connections and found that load sharing existed in nailed connections, but not
in hurricane strap connections. Meanwhile, one paper investigates the possible
modifications on conventional nailed roof-to-wall connections that would be
affordable and available to home owners. Indeed, Conner, Gromala and Burgess
(1987) showed that adding a lag bolt (8 in) to the conventional nailing would
increase the wind resistance of the roof to withstand 99% of all hurricanes. No
studies have been located that investigate the possible differences in uplift capacity
for roof to masonry wall connections, though masonry structures make up more
than 80% of the South Florida residential building stock (NAHB 1993).
Confidence intervals for the uplift capacity of roof to wall connections in a typical
Florida home can be determined from the information currently available, given the
most widely used hurricane strap connection in the area and making the assumption
that the same capacity exists for both masonry and light frame wood construction.
A study of the capacity of roof to wall connections for typical masonry walls and
typical hurricane strap connections is currently being proposed at the University of
Florida to add to the current knowledge base.
2.4.3 Wall Systems
Wall failures are much less commonly cited in post damage reports than roofing
system failures. In many cases, wall failures could be attributed to improper
installation of connections (NAHB 1993). Furthermore, the damage to wood frame
21
walls in Hurricane Andrew was found to be dependent on the structural integrity of
the roof system (NAHB 1993). Fastener schedules for wood frame cladding are
based on shear or out of plane loading, and limited testing has been conducted for
wall sections subject to uplift loads from roof suction. One study determines the
failure uplift loads for various types of wall panels and inter-story details (Striklin
1996). Also, preliminary shear wall load results are available from a full-scale
testing program of a light frame wood house (Foliente et al 2000). While these
results are preliminary, future results from this study are expected to provide
detailed information on the load sharing and capacities of wood frame substructures.
The failure of wall systems in typical Florida wood framed homes can be
determined using the research (Yancey et al, 1988) and building code data currently
available. The possibilities of failure in shear, bending, and uplift must all be
considered as limit states.
Damage to masonry walls was less prevalent than damage to wood frame walls,
and masonry walls were less dependent on the integrity of the roof system for their
survival (NAHB 1993). However, damage surveys (Pinelli, 1999) have shown that
un-reinforced masonry might be a weak link in the structural system. After failure
of an opening, the increased internal pressure can lead to the collapse of masonry
walls, which triggers the collapse of the whole structure. Dawe and Aridru (1993)
predicted the failure pressure for simply reinforced and pre-stressed wall sections.
The failure pressure for typical Florida homes could be reasonably predicted from
the results of this study. No data is currently available on the uplift capacity of
masonry walls. In spite of the fact that the majority of Florida homes are masonry
construction, the uplift capacity of these walls is not a critical research need
because this type of failure has not been observed in post damage studies.
Further information on the impact resistance of typical walls is also needed. The
larger problem with predicting whether failure will occur lies not only in the
22
prediction of capacity, but in the prediction of impact by debris as well. A method
for determining the probability of debris strike and the impact loading of different
types of debris must be developed. The following section covers this issue with
more details.
2.4.4 Openings
The capacity of windows, doors, garage doors and other openings to wind pressure
is the subject of great debate. The penetration of openings causes damage to homes
in two ways. First, the penetrated opening allows rain and wind to enter the
structure and damage the contents. Secondly, and most importantly to the
structural integrity, openings allow wind to enter and create additional internal
pressure which contributes to the uplift on the roof, causing failure. Penetration
failure may occur as the result of wind pressure overloading the component or as
the result of debris impacting the component and causing a hole. Either case causes
additional damage to the structure.
2.4.5 Wind Pressure Capacity
A vast amount of choices exist for windows, doors, and other openings. Since it
would be nearly impossible to quantify the failure pressure and impact load
resistance of each type of window, door, and garage door, some means of
estimation must be used. To date, no research has been located in the area of
failure pressure for windows and doors. Further literature search will continue in
this area. Until more detailed information can be gained, failure pressure for
typical Florida home openings could be estimated from the current Southern
Building Code Congress International (SBCCI) requirements, or from manufacturer
ratings. Both of these methods introduce a significant amount of uncertainty to the
model.
23
2.4.6 Impact Resistance
In addition to wind pressure resistance, openings must also resist impact from
flying debris generated in a storm. Many homeowners use plywood, storm shutters,
or other materials to cover glass windows and doors in preparation for a storm.
Following the damage inflicted by Hurricane Andrew, ASTM, SBCCI and the
South Florida Building Code Commission (SFBC) developed impact resistance
testing protocols to ensure that the protection used by homeowners will adequately
prevent missile penetration (IBHS 2000). The tests are similar, with the most
stringent requiring materials for openings less than 30 feet above the ground to
withstand the impact of a 9 lb. 2x4 traveling at 50 feet per second. Once the
specimen has been impacted, it must further withstand the effects of a cyclic
pressure test. Impact load studies on typical panels have recently been identified
from Texas Tech University, as presented at the America’s Conference on Wind
Engineering (ACWE) June 4-6, 2001. When available, these studies will be
reviewed to determine if the results can be applied to typical Florida home
openings. It must be noted that even though storm shutters and plywood are
available for home protection, not every homeowner will use them or anchor them
securely. It should also be noted that the use of a 2x4 as a typical piece of debris
was the subject of debate at the recent ACWE 2001. The flight speeds of various
pieces of potential debris were estimated by B. Lee of the University of Portsmouth
as part of a recent study on high-rise building vulnerability. When this study is
available, the results will be analyzed to determine the applicability of this
information to predicting damage of typical Florida homes.
In addition to the resistance capacity against missile penetration, information is also
needed to predict the likelihood that a debris missile will actually strike the opening,
or covering placed on the opening. To date, no public domain information has
been located to accurately predict the likelihood of missile strike on openings.
24
Simiu, E. and Cordes (1980 and 1983) developed a computer program for the
probabilistic assessment of tornado-borne missile speeds, which might be of some
use if the program is still available.
Predicting the damage to typical Florida homes would require the development of a
model to predict debris missile strikes. A starting point for this endeavor is to
investigate post disaster photographs and reports for types of debris found in
damaged areas. From this information, the likelihood of debris impact could be
determined for different types of surroundings. This approach has been used in
some proprietary damage prediction models.
In summary, the information currently available does not adequately predict the
failure probability of openings, covered or not, in hurricane events. Failure
pressures and failure impact loadings can be estimated for openings designed to
resist impact according to the ASTM, SBCCI, or SFBC standards. Failure pressure
for traditional types of windows and doors not protected by shutters or plywood
covering can be estimated from non-impact SBCCI standards or from manufacturer
information. Since these types of openings are not designed to withstand impact of
flying debris, it can be assumed that they will fail if impacted at all. Further
information on both the pressure capacity and impact resistance of typical openings
and opening covers is needed, but until that information can be gained, these
assumptions can be applied. The larger problem with predicting whether failure
will occur lies not in the prediction of capacity, but in the prediction of impact by
debris. A method for determining the probability of debris strike and the impact
loading of different types of debris must be developed. Further investigation and
continued literature search is needed in this area.
25
2.4.7 Summary of Building Component Resistance
A qualitative representation of the relative amount of knowledge currently available
is provided in Figure 3 (courtesy of Anne Cope). When this qualitative summary
of the research conducted to date is compared to the relative occurrence of damage
presented earlier, areas where information is critically needed are indicated. For
example, the high occurrence of damage to both roof cover and openings in
contrast to the low amount of information available about the resistance capacities
of those two components indicates a high target area for future research.
Relative Knowledge About Resistance Capacity of Building Components
Roof Cover
Openings
Roof Sheathing
Walls
Roof to WallConnection (Wood)
Roof to WallConnection (Masonry)Low High
Knowledge
Figure 3. Qualitative Knowledge of Resistance Capacity of Building Components
2.5 Manufactured Houses
Great attention has also been paid to the particular case of manufactured homes
during the last two decades, especially because of the severe damage they
encounter during storms or hurricanes. The statistics obtained from post disaster
26
surveys speak for themselves: 97% of the manufactured homes in Dade County
were destroyed by Andrew (Mafi 1992). Although their number has decreased
slightly since the 1970’s, their affordability is still very appealing to the retired and
immigrant population despite the high risk that they represent for their safety and
welfare. Several post disaster reports indeed showed that such residences are a
source of concerns in terms of economical losses for insurance companies and for
their owners as well (Marshall, 1994; NISTIR 5189, 1993)
Most of these reports seem also to agree on the fact that anchorage failure is one of
the primary damage mechanisms. In order to have a better understanding of the
behavior of this type of housing when facing strong winds, full-scale tests were
conducted to study the anchoring system, the overall drag, and lift forces (NBSIR
77-1289). Further investigations unfortunately showed that in the majority of the
cases, when installed on poor soils, the load resisting capacity of auger anchor
installations stands significantly below the requirements in the installation
standards (Pearson, Longinow & Meinheit 1996). Some reports also focus on the
comparisons of the performances of manufactured homes built after various
standards: mainly Manufactured Home Construction and Safety Standards
(MHCSS) and ASCE 7-98 (NISTIR 5370, Marshall 1992). With the exception of
the post disaster reports that can be used to validate the vulnerability model that is
to be created, there is an unfortunate lack of studies and information on
manufactured homes.
2.6 Building Load Studies
The second half of a detailed component approach to determining the vulnerability
of buildings to damage at differing wind speeds involves accurately predicting the
wind loads on a building. Loads induced by wind take the form of both external
and internal pressures.
27
2.6.1 External Pressure Field
To accurately predict loads on the building at different wind directions, the pressure
field on the roof and walls must be well defined at those directions. The design
pressure zones on the roof and walls of buildings (ASCE 7-02) encompass all
possible wind directions, not the particular direction the wind might be blowing at
any given moment during a hurricane event. Since that is the case, some means
other than ASCE 7-02 must be used to predict the loads acting on the roof and
walls. A considerable body of research exists for wind pressures on low-rise
buildings, and current efforts are underway by Cope and others at the University of
Florida to summarize existing data and accurately define the pressure field on the
roof of low-rise structures.
Comparisons of full-scale pressure data to wind tunnel model pressure data have
proven that wind tunnel studies adequately predict pressures for low-rise structures
in open country terrain (Cope 1997, Peterka et al 1998). Enough information is
available to develop reasonable predictions of average, or time independent,
pressure fields on low-rise structures, given the wind speed and direction, for
buildings in open country terrain (Ginger and Letchford 1995, Marwood and Wood
1997, Kumar and Stathopoulos 1998, Pettit et al 1999). Limited information exists
on the probability distribution or spectral content of wind pressure on low-rise
buildings (Ho et al 1995, Peterka et al 1998, Cope and Gurley 2001). Further
investigation is needed in this area. Also, the spatial correlation of time dependent
pressure is needed to accurately determine aggregate loading conditions on
sheathing panels. Current research is underway to determine both the probability
distribution and spatial correlation for gable roof structures, which make up the
majority of Florida homes (Cope and Gurley 2001). The effects of shielding from
the presence of other buildings or trees must be taken into account. A recent study,
conducted by Chang and Meroney and presented at ACWE 2001, indicates the
28
extent to which nearby buildings shield others. When available, this paper will be
reviewed to determine the extent to which the results can be applied. Information
about the effects of openings in suburban terrain is also available (Young 1997).
In summary, information exists to adequately predict average and peak pressures
on low-rise residential structures in open country terrain. Further information is
needed to translate this data into adequate predictions of pressure on the building
surface at any given time during a hurricane event. First, the effect of terrain other
than open country must be evaluated. Though limited comparisons have been
made between open country and suburban terrains in wind tunnel studies, the
effects of shielding by neighboring buildings can be accounted for using
engineering judgment and information available in the Change and Meroney and
Young papers. Second, the probability content and spatial correlation must be
investigated to determine the aggregate pressure distribution on sheathing panels at
any given time and wind direction. Efforts are currently underway to address these
issues.
2.6.2 Internal Pressure
A considerable body of research also exists for the internal pressure of buildings as
it relates to the external pressure (Vickery 1986, Harris 1990, Vickery 1994, Beste
and Cermak 1997). Unfortunately, most of this information exists only for a
limited range of structures with certain types of openings and will not adequately
model the changes in internal pressure experienced by a low-rise building damaged
in a hurricane event. Further investigation and continued literature review is
needed. In the absence of sufficient information, the assumption that the internal
pressure is the value provided by ASCE 7-98 unless an opening has been
penetrated is a reasonable baseline assumption. Once the building envelope has
been penetrated, the assumption that the interior section will become pressurized to
29
the value of static pressure at the opening should be used until further information
is available.
2.7 Structural Classification
Structural classification exposure has been investigated by the HAZUS group. The
structural classification in the HAZUS manual is representative of the building
stock in South Florida (Dade, Broward and Palm Beach counties) Although our
project intends to study residential buildings, the HAZUS manual also provides
some useful information about both commercial and residential buildings.
To obtain a statistical distribution of the roof type cover and number of stories for
residential buildings, the HAZUS team collected information from several sources
and averaged the results. From aerial photographs used to estimate the damage to
the roofs of 1633 homes in the Miami area it was estimated that 21% were two-
story homes with gable roofs. The remaining homes were one-story houses, with
23% having hip roofs and the remainder having gable roofs. The statistics from
investigation of aerial photographs reveal an approximate ratio of 3:1 for hip roof
vs. gable roof for single family residential building.
Additional data was obtained from the HUD post-Andrew damage survey and the
building shape database developed during the Residential Construction and
Mitigation Program (HAZUS Manual). Observations gathered for these studies
show a similar distribution of the roof cover types and building height. The
Residential Construction and Mitigation Program data is representative of a sample
of 1103 homes of which 29% had hip roofs, 56% had gable roofs, 10% had a
combination of both and 5% were of other types. 85% of the buildings were single
story homes. The ratio between hip vs. gable roof for all buildings is roughly 3:1.
30
On the other hand, the HUD post disaster survey has been based on a sample of 466
homes; 80% of which were single story homes (80% with gable roofs). Of the
remaining two story homes, 95% had gable roofs.
The roof cover material distribution found in the HAZUS manual is a result of a
survey sent to twenty roof contractors located in Dade, Broward and Palm Beach
counties. 90% of residential roof covers were either tiles or shingle. Detail results
will be presented in Chapter 4.
2.8 Conclusions of Literature Review
Given the data currently available from post damage reports and damage prediction
models, a detailed component approach has the flexibility to accommodate changes
in construction practice and can be tailored to represent typical Florida homes and
their surroundings. And the effectiveness of a component approach is dependent on
having the necessary detailed information in the input portfolio files. Though a
considerable body of research exists in the area of low-rise structure capacity and
loading, more information is needed to form accurate predictions of those
capacities and loadings during hurricane events. The following subjects require
further investigation:
1. Roof Covering: Limited information exists concerning the uplift capacity
of shingles, tiles, or other roof covering systems. Given a lack of failure
capacity information, failure could be predicted based on the estimation that
most products in the South Florida area would meet the Dade County
requirements and products in other areas would have a lower mean capacity.
2. Roof to Wall Connections: No information exists for the uplift capacity of
roof to masonry wall connections. Currently, the assumption could be made
that the capacity is the same for masonry walls as wood frame walls, but
31
further research is necessary to make an accurate prediction. Since the
majority of Florida residences are masonry wall structures, it is a critical
need and should be listed as a top priority for future research.
3. Walls: Limited information exists on the capacity of masonry walls.
Further study is needed to accurately predict the uplift capacity and failure
pressure of these walls, as well as the wall to foundations connection
capacities. Since this type of failure is not cited often in post-disaster
investigations, the priority of this type of research would be secondary to
that of roof to wall connections and debris impact.
4. Debris Impact: Investigations into the probability of missile strikes on
residential structures and the impact loading they deliver is vital to the
prediction of opening failures. Also, more detailed information is needed
on the capacity of typical openings and covers or shutters used to protect
them during storms.
5. External Pressure: A large body of information exists on the average and
peak pressures on low-rise structures, but limited information exists to
relate wind speed and direction to time dependent forces on the building
envelope.
But current literature survey didn’t provide any detail information on the building
statistics and component-approach vulnerability modeling. The building statistics
information contained in HAZUS manual and other sources are based on small
samples and most of them are qualitative, thus they can not serve as major sources
of building stock statistics.
32
Chapter 3. Vulnerability Model
3.1 Introduction
This chapter introduces the unique component vulnerability module of the public
hurricane prediction model. Most commercial loss projection models use post-
disaster investigations of available claim data to fit damage versus wind speed
vulnerability curves. The high dependence on the type of construction practices
common to the areas represented in the claim data in the approach undermines its
preciseness and advantage of simplicity. Obviously, past data cannot be adequately
reflects the recent changes in building codes or construction practices. In addition,
because very often few reliable wind speed data are available, damage curves
obtained by regression from observed data can be misleading.
In contrast, a component vulnerability approach explicitly accounts for both the
resistance capacity of the major building components and the load effects produced
by wind events to predict damage at various wind speeds. In the component
approach the resistance capacity of a building can be broken down into the
resistance capacity of its components and the connections between them. In a wind
engineering context we will define vulnerability as a measure of the susceptibility
to damage, expressed as a function of the wind speed. Damage to the structure
occurs when the load effects from wind or flying debris are greater than the
33
component’s capacity to resist them. Once the strength capacities, load demands,
and load path(s) are identified and modeled, the vulnerability of a structure at
various wind speeds can be estimated. Estimates are affected by uncertainties
regarding on one hand the behavior and strength of the various components and, on
the other, the load effects produced by hurricane winds.
This chapter presents and illustrates the principle of a probabilistic component
approach to the prediction of wind-induced damage. In our approach, damage is
defined as the ratio of replacement costs over total over replacement costs of the
building. Our approach makes use of probabilistic information on basic damage
modes. This information is used to calculate probabilistic information on combined
damage states. The latter consist of combinations of basic damage modes,
determined by engineering judgment, post-disaster observations, and/or analysis.
The chapter discusses basic damage modes and their probabilistic characterization.
Then, combined damage states and the derivation of their probabilistic
characteristics are considered. Once this is done it is possible to estimate
repair/replacement costs associated with building damage induced by windstorms,
which are discussed in the Chapter 6.
3.2 Basic Damage Modes
The report is focused on residential low-rise structures of different types that make
up the overwhelming bulk of the Florida building stock. For purposes of illustration,
this chapter presents the approach for a building belonging to a specified type: an
unreinforced masonry house with timber gable roof covered with shingles. Its most
vulnerable types of components are shown in Figure 4. They correspond to the
following five significant basic damage modes: (1) breakage of openings (O); (2)
loss of tiles (T); (3) loss of roof or gable end sheathing (S); (4) roof to wall
34
connection damage (C); and (5) masonry wall damage (W). For a specified wind
speed v the building will either not experience damage, or experience several of
these five basic damage modes. Some damage modes are independent of each other
(e.g., loss of tiles and breakage of openings); others are not (e.g., given that the
building has experienced window breakage, the probability of its losing tiles
increases)
Openings - O
Roof Sheathing Roof Cover - T
Roof to Wall Connections - C
Walls - W
Roof Sheathing - S
Roof to Wall
Figure 4. Components of a Single Family Home
The model is further refined by dividing each basic damage mode into several sub-
damage modes (e.g., Oi, i =0,1,2,3) according to the degree of damage: no damage,
light, moderate, or heavy damage. For example, we can define O0 as zero loss of
opening (no damage), O1 as loss of less than 25% of openings (light), O2 as loss of
25% to 50% of openings (moderate), and O3 as loss of in excess of 50% of
openings (heavy). Sub-damage modes can similarly be defined for the other basic
damage modes, denoted by Tj, Sk, Cl,,Wm, (j,k,l,m =0,1,2,3). The sub-damage
modes corresponding to a damage mode must be so defined that they are mutually
exclusive. For example, the union of the sub-events Oi (i = 1,2,3) is equal to the
event O, and the sum of their probabilities is equal to the probability of O.
35
The wind speeds v considered in the study are 3 sec average gust wind speeds at
10m. For purposes of illustration, Table 2 lists assumed probabilities of occurrence
of the sub-damage modes for the case of opening failures (see Figure 3),
conditional on wind speeds belonging to 5m/s intervals centered on values of v
varying from 45 to70 m/s. For example, the fifth column-second row cell states
that for v in the interval 57.5m/s < v ≤ 62.5m/s, P(O2|v)=35% is the probability that
a building will experience moderate opening damage, and the fifth column- third
row cell P(O3|v)= 60% is the probability that the building will experience heavy
opening damage. To simplify the notation we may omit the notation “|v” in all
subsequent developments, that is, we will use the shorthand notation P(x) in lieu of
P(x|v).
Table 2. Assumed Probabilities of Occurrence of Sub-damage Modes Oi,
Conditional on Wind Speeds Intervals Associated with the Wind Speeds v.
v(m/s) 45 50 55 60 65 70 P(O1|v) 6% 10% 5% 5% 5% 0% P(O2|v) 4% 30% 40% 35% 20% 10% P(O3|v) 0% 10% 40% 60% 75% 90%
For each damage mode the event “no damage” (i=0) is unity minus the sum of the
probabilities of the three sub-damage modes (i=1,2,3). For example, for v=40 m/s
the probability for no opening damage is P(O0)=1-P(O)=1-[P(O1)+P(O2)+P(O3)] =
100%-5%=95%.
The choice of basic damage modes is in general determined by practical
considerations such as the type of structure, the format of the requisite probabilistic
information and the extent to which it is available, the need for keeping the model
reasonably simple, and the requisite accuracy of the loss estimation. The
methodology is independent of the basic damage modes being considered in the
calculations.
36
3.3 Combined Damage States
When a windstorm causes damage to a structure, it will usually cause different
damage modes to different components at the same time. We shall refer to these
combinations of damage modes as combined damage states. Since the resulting
combined damage states require not only theoretical but also architectural and
structural engineering scrutiny, it is appropriate to use an engineering approach to
their definition. The damage states being considered must satisfy the following
requirements:
• They must be combinations of the damage modes described previously.
• They must be chosen with a view to enabling damage estimates to be
made correctly, in the sense that no possible damage state is omitted,
and no double or multiple counting of damage states occurs.
• They must make sense from an architectural and structural engineering
point of view. For example, for a building covered by conventional
sheathing, it may be assumed that wall damage will not occur without
some loss of sheathing. Similarly, although tile and opening failures do
not necessarily cause roof-to-wall connection damage, it is reasonable to
assume that no roof to wall connection damage will occur without some
tile loss and opening breakage.
37
O C
T
S W
Figure 5. Venn Diagram for the Basic Damage Modes of a Masonry Home
The basic damage modes are represented in the diagram of Figure 5. The partial or
total overlap of the basic damage modes is based on engineering judgment.
Figure 5 is the point of departure in the process of defining combined damage
states. Associated with the basic damage modes O, T, S, W and C are events --
combined damage states – whose union represents the total damage universe shown
in Figure 5. Combined damage states can similarly be considered that involve sub-
damage modes. We consider the events associated with the occurrence of the
following combinations of sub-damage modes:
Event 1. O0T0 (no damage). See hatched area in Fig. 6a. Since it is assumed that
all damage involves first some opening breakage and/or tile loss, the lack of both of
these is equivalent to no damage.
Events 2, 3, 4. Oi T0 i=1, 2, 3 (opening failure and no tile loss). The hatched area
in Fig. 6b represents the sum of events 2,3,4 or (O T0). Recall that each area Oi is a
subset of the set O; for convenience this is not shown in any of the graphs of
38
Fig.6.The probabilities of these sub-states will help to estimate the cost of repair of
homes that have only opening failures.
Event 5, 6, 7. O0 Tj S0 -i=1, 2, 3 (tile failure and no opening or sheathing loss).
The hatched area in Fig. 6c represents the sum of events 5,6,7 or (O0 T S0). The
probabilities of these sub-states will help to estimate the cost of repair of homes
that have only roof cover failures (e.g., homes with effectively boarded openings
and strong garage doors).
Events 8 through 16. Oi Tj S0 -i,j=1,2,3 (opening and tile failure and no sheathing
loss). The hatched area in Fig. 6d represents the sum of events 8 to 16 or (O T S0).
Events 17 through 25. O0 Tj Sk -j, k=1,2,3 (tile and sheathing failure and no
opening failure). The hatched area in Fig. 6e represents the sum of events 17 to 25
or (O0 T S).
Events 26 though 52. Oi Tj Sk W0 C0 (opening, tile and sheathing loss and no wall
and connection failure) i, j, k =1, 2, 3. The hatched area in Fig. 6f represents the
sum of events 26 to 52 or (O T S W0 C0).
Events 53 through 133. Oi Tj Sk Cl W0 -i,j,k,l=1,2,3 (opening, tile, sheathing and
connection failure and no wall failure). The hatched area in Fig. 6g represents the
sum of events 53 to 133 or (O T S C W0).
Events 134 through 214. Oi Tj SkWm C0 -i,j,k,m=1,2,3.(opening, tile, sheathing and
wall failure but no connection failure). The hatched area in Fig. 6h represents the
sum of events 134 to 214 or (O T S W C0).
Events 215 through 457. Oi Tj Sk Cl Wm –i,j,k,l,m=1,2,3.(opening, tile, sheathing,
wall, and connection failure). The hatched area in Fig. 6i represents the sum of
events 215 to 457 or (O T S W C).
39
a
b c
d
e f
g
h i
Figure 6. Venn Diagrams of the Combined Damage States (subsets of Figure 5)
There are a total of 457 damage state events. However, not all of these events are
of interest from a damage estimation point of view. Engineering considerations
allow the elimination of a number of events. There are several scenarios:
• When roof cover damage (T) and sheathing damage (S) occur at the same
time, the damaged area of the roof cover must be larger than the damaged
area of sheathing. We can therefore eliminate all the damage states which
40
pertain to damaged area of roof cover equal to or smaller than the damaged
area of sheathing, i.e. eliminate events that contain TjSk when j<k.
• When roof cover damage (T) or sheathing damage (S) or opening damage
(O) occur together with wall damage (W) or connection damage (C ), the
level of damage for T or S or O should be larger than for W or C. That is,
there is only a small probability that a wall would suffer heavy damage
while the roof cover has suffered light damage. Thus we can eliminate all
the damage states which contain lower levels of roof covering damage and
decking damage and opening damage than wall damage and connection
damage. i.e. eliminate events containing Oi, Sk, Tj, Wm, and Cn when i, j,
k < m, n. In particular, when severe wall damage and severe roof to wall
connection damage occur together, the whole structure collapses. So if roof
to wall connection and wall damage are both heavy (i.e., if W3 and C3
occur), the only significant damage event will be O3T3S3W3C3, so that we
can eliminate all events OiTjSkW3C3 for which i, j, k =1,2.
These engineering considerations allow the elimination of 240 damage states,
leaving 217 damage states. Note from Fig. 5 that, for any specified wind speed,
any two distinct damage states are mutually exclusive. For example, a structure
cannot experience both the state Oi Tj Sk W0 C0 and the state Oi Tj Sk W0 Cl.
The implementation of this model is currently under way. The determination of the
probabilities of occurrence the combined damage state events rely upon the use of a
41
component-based Monte Carlo simulation engine, which is developed at University
of Florida and detailed described in [1]. The simulation relates estimated
probabilistic strength capacities of building components to 3 sec average gust wind
speeds through a detailed wind and structural engineering analysis that includes
effects of wind-borne missiles.
The component approach taken in the Monte-Carlo simulation explicitly accounts
for both the uncertain resistance capacity of the various building components and
the load effects produced by wind events to predict damage at various wind speeds.
The resistance capacity of a building is broken down into the resistance capacity of
its components and the connections between them.
Damage to the structure occurs when the load effects from wind or flying debris are
greater than the component’s capacity to resist them. The probabilistic strength
capacities, load paths, and load sharing between components are identified and
modeled. The probable damage to a particular building class at a given wind speed
is estimated through many simulations of the same structure at that wind speed,
randomly sampling component capacities between each simulation. Changes in
loading on components as a result of component/connection failure within a single
simulation are accounted for. Each simulation represents an instance of a storm of
same intensity. The wind speed is then increased to the next discrete value and the
process repeated. The end result is a table describing the probability of combined
damage levels to the components, conditioned upon peak wind speeds.
No duration effect is included in the model. It is assumed that the wind speed
represents the highest 3 sec peak gust speed over the duration of a storm at a
particular location, and when applied to the building model it produces the
maximum loading for that storm. Since the component strengths are set, if they do
not fail at the highest 3 second gust, they won’t fail at all. Other average wind
42
speeds could be considered for the simulation (e.g. 1 min or 10 min average wind
speed), in which case, the pressure coefficients to compute the pressure
distributions would be adjusted accordingly, resulting in the same maximum
pressures. The 3 sec average was adopted because it is the standard in the U.S.A.
3.4 Damage Matrix
A possible resulting damage matrix generating from Monte Carlo simulation is
summarized in Table 3. The resulting probabilities of the 217 combined damage
states will be assessed as necessary to ascertain the extent to which they are
physically realizable. This will provide useful guidance to the future development
of the simulation procedure. For example, the fourth column-third row cell states
that for v in the interval 57.5m/s < v ≤ 62.5m/s, combined degree of damage states
P(O T0) has a 35% probability of occurrence.
Table 3. Example of Estimated Probabilities of Damage States, Conditional on
Wind Speed Intervals Associated with the Speeds v
v (m/s) 45 50 55 60 65 70
P (O0T0/v) 82% 38.% 6.0% 0.0% 0.0% 0.%
P(O T0/ v) 9.1% 40.% 54.% 20.0% 0.0% 0.%
P(O0 T S0/ v) 7.6% 10.% 3.7% 0.0% 0.0% 0.%
P(O0 T S/ v) 0.5% 0.5% 0.3% 0.0% 0.0% 0.%
P(O T S0/ v) 0.4% 2.0% 6.3% 30.% 20.% 0.%
P(OTSW0C0/ v) 0.5% 0.5% 0.33% 1.0% 10.% 0.%
P(OTSCW0/ v) 0.% 3.% 3.87% 6.0% 0.% 0.%
P(OTSWC0/ v) 0.% 0.% 3.57% 9.0% 0.% 0.%
P(OTS CW/ v) 0.% 6.% 21.9% 34.% 70.% 100%
3.5 Summary
This Chapter presented the component approach vulnerability modeling for the
Project. This resulting damage probability matrix is used in the damage cost
estimation discussed in Chapter 6.
43
Chapter 4. Structural Classification
4.1 Introduction
In order to estimate the hurricane losses over a large geographical area with a
building population composed of different structural types, it is important to
identify the predominant structural types that make up the majority of the building
stock in the different regions of Florida. This chapter will discuss the methodology
and results of defining the predominant structural types, or the most common
structural types.
The reason for searching structural classification information from outside
resources other than insurance portfolio files is that insurance company’s concern is
more focused on fire safety issues; the classification used in portfolio files does not
directly reflect the structural characteristics of the buildings but rather their fire
resistance. Because of this reason, insurance portfolio files can not be our main
source of getting structural classification information; however, it can provide
cross-check to our results obtained from other sources.
To ensure that the damage predictions based on the simulation are accurate and
reliable, it is critical to define the right structural type model. For the efficiency of
the model, it is also critical to limit the total number of models to be simulated to
44
those that are statistically significant in any given area. Therefore, the
determination of the right mix of structural types or models that characterize the
building population in any given area is another important component of the
methodology. This chapter presents the procedure and results of the statistical
analysis of the building population of Florida. All the assumptions will be also
introduced.
4.2 Sources of Information
The author used several sources of information for getting Florida up-to-date
building structural types information. The most important source is the Florida
counties property tax appraisers’ databases which are the main resource of
gathering building structural information. Another source is the Florida Hurricane
Catastrophe Fund (FHCF) database. And we also obtained the limited information
Hazus report of Miami – Dade County as the validation of southern counties like
Broward and Palm Beach.
4.2.1 Florida Hurricane Catastrophe Fund Exposure Database
The 1998 Florida Hurricane Catastrophe Fund Industry Data Guide contains a
descriptive list of the various construction types used to characterize the structures
of each insured property in an insurance portfolio file. They are referred to as the
ISO construction classes shown in Table 4. They are fire related and are based on
characteristics such as combustible and non-combustible materials.
45
Table 4. ISO Construction Classification
Construction Type Description
Frame Buildings where the exterior walls are wood or other combustible materials, including construction where the combustible materials are combined with other materials such as brick veneer, stone veneer, wood iron-clad and stucco on wood
Joisted Masonry Buildings where the exterior walls are constructed of masonry materials such as adobe, brick, concrete, gypsum block, hollow concrete block, stone, tile or similar materials and where the floor and roof are combustible.
Non-Combustible Buildings where the exterior walls and the floors and roof are constructed of, and supported by, metal, asbestos, gypsum or other non-combustible materials. (Other than constructions defined by the description for Code 8.
Masonry Non-Combustible
Buildings where the exterior walls are constructed of masonry materials as described in Code 2 with the floors and roof of metal or other non-combustible materials. (Other than constructions defined by the description for Code 9.
Modified Fire Resistive
Buildings where the exterior walls and the floors and roof are constructed of masonry or fire resistive materials with a fire rating of one hour or more but less than two hours
Fire Resistive Buildings where the exterior walls and floors and roof are constructed of masonry or fire resistive materials having a fire resistance rating of not less than two hours
Heavy Timber Joisted Masonry
Joisted Masonry construction where the following additional conditions exist: Roof Deck has a minimum thickness of 2 inches with roof supports having a minimum dimension of 6 inches, or roof assembly is documented to have a UL wind uplift classification of 90 or equivalent
Superior Non-Combustible
Non-combustible construction where the following additional conditions exist: Floors and roof constructed of 2 inches of masonry on steel supports, or documented to be constructed of 22 gauge metal or heavier on steel supports, or documented to have a wind uplift classification of 90 or equivalent.
Superior Masonry Non-Combustible
Masonry non-combustible construction where the following additional conditions exist: Floors and roof constructed of 2 inches of masonry on steel supports, or documented to be constructed of 22 gauge metal or heavier on steel supports, or documented to have a wind uplift classification of 90 or equivalent.
Masonry Veneer Buildings with exterior walls of combustible construction veneered with brick, masonry, or stone.
Unknown Unknown commercial or residential construction Mobile Home - Fully Tied Down, manufactured before 7/13/94
Mobile/Manufactured Housing which has anchors and tie-downs as required by Section 320.8325, Florida Statutes.
Mobile Home - Fully Tied Down, manufactured on or after 7/13/94
Mobile/Manufactured Housing which has anchors and tie-downs as required by Section 320.8325, Florida Statutes.
Mobile Home – Partially Tied Down
Mobile/Manufactured Housing which is exempt from Section 320.8325, Florida Statutes, and has anchors and tie-downs. These units shall meet the Department of Highway Safety and Motor Vehicles’ minimum standards for the manufacture or installation of anchors, tie-downs, over-the-roof ties, or other reliable methods of securing when over-the-roof ties are not suitable due to factors such as unreasonable cost, design, or potential damage to the unit. These devices, when properly installed, shall cause the unit to resist overturning or sliding from the force of wind.
Mobile Home - Not Tied Down
Known that the mobile home is not tied down.
Mobile Home - Unknown
Unknown if the mobile home is tied down, or nature of the tie-downs is unknown.
46
The information contained in the 1998 Florida Hurricane Catastrophe Fund
Exposure Report can be represented graphically as shown in Figure 7 to 23. These
figures were prepared by a consultant hired by NIST, as part of the Project. Each
map shows the geographic distribution across Florida of the ISO construction
classes of residential buildings.
Exposure (Dollars)889304 - 496312096496312096 - 10964963841096496384 - 17650000641765000064 - 26709171202670917120 - 5088393216
Total Exposure in Zip
Figure 7. Total Dollar Value Distribution in Florida
47
Ratio of zip0.111 - 0.4930.493 - 0.7390.739 - 0.8460.846 - 0.9250.925 - 1
Ratio of Residential in Zipcode
Figure 8. Ratio of Residential Buildings in Florida
Ratio of zip0 - 0.0390.039 - 0.1050.105 - 0.2230.223 - 0.4180.418 - 0.889
Ratio of Commercial in Zipcode
Figure 9. Ratio of Commercial Buildings in Florida
48
Ratio of zip0 - 0.50.5 - 0.810.81 - 0.9270.927 - 0.9760.976 - 1
Ratio of Homeowners in Zipcode
Figure 10. Ratio of Home Owner in Florida
Ratio of zip0 - 0.0240.024 - 0.0720.072 - 0.1750.175 - 0.4380.438 - 1
Ratio of Renters in Zipcode
Figure 11. Ratio of Renters in Florida
49
Ratio of zip0 - 0.0880.088 - 0.1940.194 - 0.3110.311 - 0.4650.465 - 0.866
Ratio of Frame in Zipcode
Figure 12. Ratio of Frame Exterior Wall in Florida
Ratio of zip0 - 0.3040.304 - 0.4830.483 - 0.6520.652 - 0.8030.803 - 0.976
Ratio of Joisted Masonry in Zipcode
Figure 13. Ratio of Joisted Masonry Homes in Florida
50
Ratio of zip0 - 0.0030.003 - 0.010.01 - 0.0280.028 - 0.0630.063 - 0.128
Ratio of Non-Combustible in Zipcode
Figure 14. Ratio of Non-Combustible in Florida
Ratio of zip0 - 0.0170.017 - 0.0530.053 - 0.1280.128 - 0.2520.252 - 0.78
Ratio of Masonry Non-Combustible in Zipcode
Figure 15. Ratio of Masonry Non-combustible in Florida
51
Ratio of zip0 - 0.0070.007 - 0.0240.024 - 0.0540.054 - 0.1410.141 - 0.6
Ratio of Modified Fire Resistive in Zipcode
Figure 16. Ratio of Modified Fire Resistive in Florida
Ratio of zip000 - 0.0010.001 - 0.0010.001 - 0.001
Ratio of Heavy Timber Joisted Masonryin Zipcode
Figure 17. Ratio of Heavy Timber Joisted Masonry in Florida
52
Ratio of zip0 - 00 - 0.0010.001 - 0.0030.003 - 0.0040.004 - 0.006
Ratio of Superior Non-Combustible in Zipcode
Figure 18. Ratio of Superior Non-combustible in Florida
Ratio of zip0 - 0.0030.003 - 0.010.01 - 0.0230.023 - 0.0470.047 - 0.074
Ratio of Superior Masonry Non-Combustible in Zipcode
Figure 19. Ratio of Superior Masonry Non-combustible in Florida
53
Ratio of zip0 - 0.3350.335 - 0.5120.512 - 0.6750.675 - 0.8250.825 - 0.989
Ratio of Masonry Veneer in Zipcode
Figure 20. Ratio of Masonry Veneer Distribution
Ratio of zip0 - 0.0180.018 - 0.0490.049 - 0.110.11 - 0.2520.252 - 0.78
Ratio of Masonry in Zipcode
Figure 21. Ratio of Masonry Distribution
54
Ratio of zip0 - 0.0070.007 - 0.0230.023 - 0.0550.055 - 0.1420.142 - 0.6
Ratio of Semi-Wind Resistive in Zipcode
Figure 22. Ratio of Semi-wind Resistive Distribution
Ratio of zip0 - 0.0270.027 - 0.0960.096 - 0.2240.224 - 0.4180.418 - 0.994
Ratio of Wind Resistive in Zipcode
Figure 23. Ratio of Wind Resistive Distribution
55
The above figures provide rough distribution information of some useful structural
characteristics, for example distribution of timber frame and masonry wall, shown
in the Figure 12 and Figure 21. The distribution of timber frame (Figure 12) is a
very important figure that helped define the 4 Florida regions described in section
4.5.
However, most information gathered here are for the fire resistance concerns,
which does not directly reflect structural characters such as roof shape, and
structure areas etc., which are crucial to evaluate the wind resistance. Since the
FHCF database can not provide useful structural classifications for structural
vulnerability modeling, we have to find other sources to define the most common
structural types.
4.2.2 HAZUS Manual
Another data source is the HAZUS manual, which contains mainly data on roof
types and story height for 3 counties of South Florida.
The structural classification in the HAZUS manual is representative of the building
stock in South Florida (Dade, Broward and Palm Beach counties) and is divided
into two main sections. One section is a statistical survey of the distribution of
each roof type and covers. The other section is a survey of typical houses found in
South Florida. Although our project intends to study residential buildings, the
HAZUS manual also provides some useful information about both commercial and
residential buildings.
The following is a summary of the information obtained form HAZUS.
56
To obtain a statistical distribution of the roof type cover and number of stories for
residential buildings, the HAZUS team collected information from several sources
and averaged the results.
4.2.2.1 Roof types (gable, hip)
From aerial photographs used to estimate the damage to the roofs of 1633 homes in
the Miami area it was estimated that 21% were two-story homes with gable roofs.
The remaining homes were one-story houses, with 23% having hip roofs and the
remainder having gable roofs. The statistics from investigation of aerial
photographs reveal an approximate ratio of 3:1 for hip roof vs. gable roof for single
family residential building investigated.
Additional data was obtained from the HUD post-Andrew damage survey and the
building shape database developed during the Residential Construction and
Mitigation Program (HAZUS Manual). Observations gathered for these studies
show a similar distribution of the roof cover types and building height. The
Residential Construction and Mitigation Program data is representative of a sample
of 1103 homes of which 29% had hip roofs, 56% had gable roofs, 10% had a
combination of both and 5% were of other types. 85% of the buildings were single
story homes. The ratio between hip vs. gable roof for all buildings is roughly 3:1.
On the other hand, the HUD post disaster survey has been based on a sample of 466
homes; 80% of which were single story homes (80% with gable roofs). Of the
remaining two story homes, 95% had gable roofs. As a result, the following
simplified default building stock for single family residential construction in South
East Florida is suggested
60% Single story gable
20% Single story hip
20% Two story gable
57
Note: No damage to flat roof homes was observed in the aerial photographs. A
combination of a hip section and a flat section was classified as a hip roof building.
Based on the above information, an approximate ratio of hip vs. roof in Southern
Florida counties from the HAZUS and other post-disaster report is 3:1. The author
will use this ratio to check against the one obtained from tax appraiser’s database.
4.2.2.2 Roof Covers (shingles, tiles…)
The roof cover material distribution found in the HAZUS manual is a result of a
survey sent to twenty roof contractors located in Dade, Broward and Palm Beach
counties. The following is an average of the total amount of data collected through
this survey:
90% of residential roof covers were either tiles or shingle (tiles being more
common than shingle), 3.3% were wood shakes and 2.3% metal pan.
The following tables represent the most important questions asked in the survey as
well as the average answer given.
58
Table 5. Percentage of the Types of Roof Covering in the Existing Populations of
Residential Buildings Including Apartments and Condominiums
Cover type Average response in percentage Asphalt shingles 41.3 Cement fiber 0.3 Metal panel, architectural 2.3 metal panel, structural 0.3 metal shingles 0.4 Slate 0.3 Tile 49.5 Wood shakes/shingles 3.3 other 2.5
Table 6. Percentage of the Types of New and Re-roofing Systems Being Installed
on Residential Buildings
Roof cover type Average response in percentage Asphalt shingles 39 Cement fiber 0 Metal panel, architectural 2.1 metal panel, structural 0.1 metal shingles 0.3 Slate 0.1 Tile 54 Wood shakes/shingles 1.6 Other 2.8
4.2.2.3 Manufactured homes
No comprehensive statistical survey is presented in the HAZUS manual on
manufactured homes. Other sources of information should therefore be considered
to obtain a distribution of the different types of roof covers and materials used for
the siding.
59
The manual provides, however, a rough list of the materials commonly used in
manufactured housing construction.
Table 7. Sample Manufactured Home Materials in HAZUS Manual
Roof membrane systems Metal skin attached directly to the roof trusses or roof sheathing attached to the roof trusses
Siding of older mobile homes Wood, vinyl or metal panels
There is no information about siding of new manufactured homes in the manual.
The basic models used in the damage simulation are presented in the following
table:
Table 8. Criteria of Defining Manufactured Homes Types in HAZUS
Tiedowns Yes or No
Construction types Pre-HUD, HUD
A total of 28 damage state curves were presented in the manual. Combinations of
the possibilities listed in the previous table were tested for several surface
roughness: 0.03m, 0.1m, 0.35m representing respectively a typical open terrain, a
relatively open terrain and a typical suburban terrain.
4.2.3 Tax Appraisers’ Databases
Taxes are sometimes classified as either specific or ad valorem. Property taxes are
almost invariably ad valorem. Ad valorem taxes are based on a fixed proportion of
the value of the property with respect to which the tax is assessed. They require an
appraisal of the taxable subject matter's worth. Ad valorem property taxes are based
on ownership of the property, and are payable regardless of whether the property is
used or not and whether it generates income for the owner (although these factors
60
may affect the assessed value). Property tax is usually the primary source of
revenue for local counties. Property is based on the evaluation of fair market value
and usually assessed at the county level by a county board of tax assessors and
appraisers.
The property appraiser is primarily responsible for identifying, locating, and
valuing all property within the county for ad valorem tax purposes. Being
custodian of certain county records, the property appraiser of each county must
maintain property record cards, subdivision plats, ownership maps, sales data
records and prior tax rolls. From these records one can determine building sizes, a
description of their components, and property characteristics. The author used
these data to find the most common structural types of each county.
Property appraiser’s database are the most comprehensive and accurate information
of building structural characteristics accessible at the present time. However, each
county has its own property appraiser team, thus the database’s formats and
contents vary dramatically from county to county. It should be noted that some
properties are tax exempted by law and therefore not included in the appraiser
databases. But due to the small number of property exemptible, this issue could be
overlooked in terms of statistical significance.
Although the databases’ contents and format vary county to county, most of them
contain some useful structural information to define the most common structural
types in each county. The engineering group has tried to access up to about fifteen
counties databases. Some counties refused to provide assistance and some counties
databases didn’t match the layout explanation file along with them. The counties
that we tried access to but couldn’t get through include Santa Rosa, Alachua,
Marion, Orange, Duval, Okaloosa, Osceola, Bay and Flagler. At last, nine counties
databases were finally accessed and processed properly. They are Brevard,
61
Hillsborough, Pinellas, Walton, Escambia, Leon, Broward, Palm Beach and
Monroe County. These counties will be referred as sample counties. They are
chosen simply because their tax appraiser’s databases are available. Their
geological distribution is shown in the map of Figure 24.
Figure 24. Four Regions with Sample Counties Highlighted
During the process of acquiring and processing the database, the engineering group
has established cooperative relationships with building officials in each county. The
available contacts for different county are listed in the following Table 9.
Table 9. Contact Building Officials in Each of the Sample Counties
County name Contact person Telephone number Fax number
Hillsborough Anita Meneudes 813-276-8811 813-272-5519
Pinellas Laurine Petrosky 727-464-3831
Broward Jack Large (computer) Tonny Hudge
954-357-6827 954-357-8474
Monroe Elizabeth Harvey 407-671-8808
Leon Chris Lewis 850-488-6102
Brevard Greg John Steruaged
321-264-6707 321-264-6938
Escambia Chris Jones 850-434-2735 850-435-9526
Palm Beach Barbara Monticello 561-355-2866
Walton Kevin Laird 850-267-4500
62
4.3 Information Gathered in the Structural Survey
Tax appraisers' databases contain large quantity of building information, and it is
necessary to extract those characteristics related to the vulnerability of the buildings
to wind. Tax appraiser’s databases keep record of all buildings taxable, including
residential and commercial. We only investigated residential buildings in this
research. Residential buildings can be classified into three categories by their usage:
single family residential buildings, condominiums and manufactured house (mobile
homes). The structural characteristics of these three types of buildings vary
dramatically, which requires separate models to be assigned to each category.
Based on this philosophy, all the residential buildings in each county database were
first divided into three major categories: single family residential buildings,
condominiums, and manufactured homes. The Table 10 shows the distribution.
Table 10. Three Major Building Type Distributions
Single Family Residential Manufactured Homes Condominium
Total number Number % Number % Number %
Brevard County 175596 146799 84% 13069 7% 15728 9% Pinellas 364610 250826 69% 11859 3% 101925 28% Hillsborough 342693 315232 92% 6633 2% 20828 6% Broward 372502 344508 92% 4280 1% 23714 6% Palm Beach 494428 222918 45% 4116 1% 267394 54% Escambia 105804 102289 97% 2109 2% 1406 1% Walton 93613 Leon 67705 60294 89% 6398 9% 1013 1% Monroe 37733 23480 62% 6077 16% 8176 22% Note: 1. Walton County: Database combine mobile home and condominiums as single family or multi-family residential buildings.
2. Broward County: Statistics shown are calculated after eliminate blank records
63
Under each category, the author chose to extract information on 6 critical building
characteristics for statistical distribution analysis. They are roof cover, roof type,
exterior wall material, number of story, year built and building area.
The reasons of choosing these 6 structural characteristics and major assumptions
used in the process of data are discussed in the rest of this section, the sample result
tables are presented for each characteristic for illustration purposes.
4.3.1 Roof Cover
The resistance capacity of a roof system to wind uplift usually includes the capacity
of roof cover, sheathing and trusses or rafters. Roof cover is exposed to weather
and usually is in the form of asphalt shingles or tiles. When wind uplift force
exceeds the capacity of the roof cover materials or their connection, the loss of roof
cover will occur. As the immediate consequence, sheathing will be exposed to
strong wind, and often heavy rain. Most plywood sheathing will severely weaken
without the protection of the roof cover material, which renders, more likely, the
loss of sheathing. Loss of sheathing will then reduce the overall integrity of the
building structures by either increasing the internal pressure or weakening the
connection of roof truss system.
The roof cover types and distributions were collected and processed for all 9
counties in the survey. For example, Table 11 shows the distribution of roof cover
for single family buildings in Broward County.
64
Table 11. Broward County Single Family Houses Roof Cover Type Distribution. (1)
Roof material Percentage
Wood Shingle 35%
asbestos Shingle 1%
composite Shingle 21%
Cement Tile 27%
Barrel Tile 7%
Others (including blank record) 9 %
100%
(Note: Broward County’s database contains a number of data with blank record.
This problem occurs in other database too, but Broward County was the most
serious case).
Shingles can be made from wood, asphalt, asbestos and other composite materials.
Shingle roofs are the most commonly used on residential properties in North
America. Early shingles were made by saturating rag-felts with asphalt and by
coating each side of the saturated felt with an asphalt-mineral filer-coat, covering
the top surface of the shingle with mineral granules to prevent sunlight and increase
weather resistance and coating the bottom surface with a material to prevent shingle
from sticking together in storage or shipment. More recently many manufacturers
began producing shingles using a fiberglass mat to replace the asphalt-mineral filer-
coat. The fiberglass mat is thought to have better tear resistance, possibly slightly
better fire resistance, as the shingle getting thinner, it is easier to transport and
install.
Tiles are also important roofing materials. Tiles can be made from clay or concrete.
Tile roofs have advantages of being durable, capable of withstanding fire, wind,
hail, earthquakes, snow and intense heat. For this reason, tile roofing
manufacturers offer much longer warranties than shingles. Tiles also provide
65
architectural aesthetics and versatility because of the vast array of tile styles and
colors. Concrete or clay tile roofing are claimed to have high resistance to severe
storms and hurricane-force winds. Its design and construction provide high air
permeability, which helps relieve wind stress and allow for a substantial flexible
airspace between the roof deck and the tile. And when concrete or clay roof tiles
interlock, their inherent flexibility relieves the forces generated by wind as they
move. Actually, the wind resistance of a tile roof is highly dependent on the type
and quality of the installation. Tile can also be an important source of debris.
Lamentably, there is very little data requiring the wind resistance of tile roof.
In order to depict the building materials distribution, the author combined similar
structural characteristics together. For roof cover material category, different type
of shingles such as wood shingle, asphalt shingle, asbestos shingle, wood shake etc.
are classified into “Shingle”. Accordingly, clay tile, comment tile, barrel tile and
slate, etc. are classified as “Tile”. This combination reduced the roof cover
materials’ types down to 3 major types “Shingle”, “Tile” and “Others”. Table 12
shows the simplified Broward County roof cover materials distribution results.
Table 12. Broward County Single Family Houses Roof Cover Type Distribution (2)
Roof material Percentage Shingle 57% Wood Shingle asbestos Shingle composite Shingle Tile 34% Cement Tile Barrel Tile Others 9% 100%
The building material types will be used in the Monte Carlo simulations. At this
stage; there isn’t enough information to characterize differently shingles and tiles in
66
the simulation. Therefore shingle and tile are combined together. But they can be
modeled separately anytime in the future since the statistical information is
available. When shingles and tiles are combined together we have the following
Table 13.
Table 13. Broward County Single Family Houses Roof Cover Type Distribution
Roof material Percentage Shingle/Tile 91% Others 9%
This statistics agree closely with the result from HAZUS manual detailed in
Section 4.2.2.2.
4.3.2 Roof Type
Different roof types have different capacity to resist strong winds as shown by
many post-disaster surveys and test results. The majority of roof types for single
family houses are gable, hip.
Gable roofs can be simply described as two pitched roofs with a gable at each end.
On the other hand, hip roof can be described as gable roof with the gable ends
brought together at the same pitch as the rest of the roof. The sketches shown
below in Table 14 are typical gable and hip roofs.
67
Table 14. Comparison of Gable Roof and Hip Roof
Gable Roof Hip Roof
Some small portion of structures uses “mansard roof” and “gambrel roof” shown in
Table 15.
Table 15. Gambrel Roof and Mansard Roof
Gambrel roof Mansard roof
The author classifies the gambrel roof into gable roof and mansard roof into hip
roof based on their structural similarities. Gable and hip roofs have different
structural characteristics which causes distinctive aerodynamic effects around roofs.
The past disaster survey (Phang in 1999 post-hurricane Andrew), shows gable roofs
showed much more structural damage than hip roofs. It is critical to find the
proportions of gable and hip in the building stock. However, most counties’
databases combine gable and hip together in their record. There are only two
counties, Brevard and Escambia, distinguishing between hip and gable roofs in
68
their databases. In both cases, the ratio between gable roof and hip roof is
approximately 2:1.
Since these two counties are neither in the same region, nor share same weather
conditions, it is reasonable to assume that these two counties are randomly
representative of Florida in general. Therefore the ratio of 2:1 for gable and hip
was extrapolated to all the other counties.
Recall that the ratio of gable and hip roof in the post-hurricane survey in the
HAZUS report is 3:1. But it should be noticed that HAZUS report only shows the
ratio among the buildings, which are damaged. Since gable roofs are less durable
than hip roof, it is logical to see higher ratio of gable and hip in HAZUS report than
tax appraiser’s database which covers most residential buildings.
4.3.3 Exterior Wall
Exterior wall failures are much less commonly cited in post damage reports than
roofing system failures. There are two main major types of exterior wall material:
concrete block and wood frame. The difference of these two types lies more in their
damage mechanism than material properties.
Wood frame wall could fail in several different ways: First of all, it can fail by
losing the structural integrity with the roof system. In the typical structural analysis
of a wood frame house, the wood panel wall can be modeled as a vertical simply-
supported panel with one end attached to the foundation and the other supported by
the header or studs which are linked to roof by metal straps. Once the roof is
damaged or blown out, the wall panel will lose one of its supports. Secondly,
because the installation of wood frame walls requires a large number of nailing and
metal strap connections, each of these connections could be a weak link of the
whole wall system. If designed and constructed strictly under the guidance of
69
wood design codes and specifications, the number of nails and sizes of the
connections can be ensured. But the quality of connections is still largely dependent
upon the installation. A picture in the post-damage survey of Hurricane Andrew
shows nails that were supposed to connect plywood wall panel to a 2x4 stud were
totally out of alignment and none of them penetrated into the stud. The connection
fails when the uplift force or shear force exceeds its capacity. And finally, the wall
could also fail due to excessive shear or bending moment. In some case, the
missile impact can also damage wall panels.
Damage to masonry walls is less prevalent than to wood frame walls. The masonry
walls are also dependent on the integrity of the roof system for their survival.
Damage surveys have shown that un-reinforced masonry might be a weak link in
the structural system. When the roof fails, the un-reinforced concrete blocks will
stand as a vertical cantilever panel, due to the weak bound between individual
concrete blocks, the blocks are very likely to collapse. In other cases, after the
failure of an opening, the increased internal pressure can lead to the collapse of
masonry walls, which triggers the collapse of the whole structure.
Due to large difference in the performance of exterior wall materials, it is critical to
get the statistical information of the exterior wall material. Table 16 shows the
distribution of exterior wall material distribution of single family residential
buildings in Brevard County.
70
Table 16. Brevard County Single Family Buildings Exterior Wall Materials
External material type Percentage C.B. Stucco 40% C.B.Plain 31% Wood Sheathing 9% Wood Frame Stucco 8% Vinyl/aluminum 4% Brick on msnv 3% Brick on wood 2% Exterior plywood 1% Wood Frame no shingle 1%
Although exterior wall materials usually have large versatile distribution, the author
defines two major exterior wall types from the structural point of view: Wood
Frame (Wood), Concrete Block (Masonry). In the classification, the author
encounters a problem of “Brick”. Almost every county’s database has large number
of records of various kinds of “Brick”. As decoration material, brick usually do not
serve any structural function. Field tax appraiser officials obviously record what
they see from the appearance of buildings. Given the fact that there is nearly
impossible for us to tell what is the true structural exterior wall material behind the
brick, the authors did the following: using the information from Figure 12, classify
“Brick” kind into “Wood Frame” when the building is in northern area county,
otherwise, classify them into “Concrete Block”.
4.3.4 Year Built
The year built record of a building can link the building to the building codes in
effect when the structure was built. The design requirements stipulated in the
building code directly affect the wind resistance of a building. For example, the
building code requirements in southern Florida were more lenient before Hurricane
Andrew and severe after it. However, different jurisdictions ranging from single
city to county have adopted different building codes at various historical times. The
71
author had spent a considerable amount of time to do investigations on the
evolution history of building codes and construction practices for each county but
got no substantial information that can be used to set up a cutting date for different
building codes. For this reason, year built information is not included in the loss
projection model. A sample chart of year built is shown in Figure 25.
0%
1%
2%
3%
4%
5%
6%
1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000
year built
Figure 25. Distribution of Year Built for Single Family Homes in Brevard County
Since detail information on year built distribution is collected for each sample
counties, it could possibly be used in the future under certain circumstances. For
example, given known cut off date of implementing different building codes in a
given area, the year built distribution information can be used to divide all
buildings to categories of each building codes, thus link the construction practice
and prevalent design requirement to the buildings. The general idea of building
characteristics can be obtained simply though year built to building codes without
going into details of tax appraiser’s database.
72
4.3.5 Number of Stories
Obviously, two stories family buildings have large differences from one story
buildings in terms of structural characteristics, number of openings, value etc.
Most one-story buildings have either masonry exterior wall or timber frame, in
other words, one type of exterior wall material. However, most second story
buildings have mixed exterior wall material. Typical second story single family
buildings are built with concrete block wall for first story and timber frame for
second story. The majority of single family homes are one story buildings. For
example, Table 17 describes the number of stories distribution of Pinellas County.
Table 17. Percentage of Number of Story Distribution of Pinellas County
Number of story Percentage 1 92% 2 8%
4.3.6 Building Areas
Another characteristic collected from the tax appraisers’ database is building area
information. The results show that buildings with hip roofs usually have larger area
than building with gable roofs and buildings with concrete block exterior wall
generally have larger area than wood frame buildings. The percentile distribution
of home areas of single family homes in Brevard County, regardless of structural
characteristics, is shown in Figure 26 (the range is in ft2, 1ft2 = 0.1 m2). It should be
noted that the area information in the tax appraiser’s database are the building area
under air conditioning. So it does not contain the garage area except the database
for Brevard County.
73
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
0~50
0
500~
1000
1000
~150
0
1500
~200
0
2000
~250
0
2500
~300
0
3000
~350
0
3500
~400
0
4000
~450
0
4500
~500
0
5000
~550
0
5500
~600
0
6000
or a
bove
range
per
cen
tag
e
Figure 26. Brevard County Single Family Area Range Distribution (unit: ft2)
4.4 Counties Statistics
4.4.1 Brevard County
Brevard County’s tax appraiser database contains 146799 single family homes.
74
4.4.1.1. Roof materials
SHINGLE ASP .85%
CEM ENT TILE3%
SHINGLE ASB.2%
CLAY TILE3%
B.U./T.6%
SHEET M ETAL1%
Figure 27. Brevard County Roof Material Detail Distribution
75
Table 18. Brevard County Combined Roof Material Distribution
Classfication Number Percentage MEMBRANE B.U./T. & G./MEMBRANE 8552 ROLL COMP. 451 Subtotal 9003 6.1% SHINGLE/TILE CEMENT TILE 4692 CLAY TILE 4198 SLATE 97 SHINGLE ASB. 3249 SHINGLE ASP. 123906 WOOD SHAKES 164 WOOD SHINGLES 90 Subtotal 136396 92.9% METAL ENAMEL METAL 308 METAL LEAF 74 SHEET METAL 1018 1.0% Subtotal 1400 Total 146799 100.0%
76
4.4.1.2. Roof Type
GABLE65%
HIP32%
WOOD DECK1% FLAT SHED
2%
Figure 28. Brevard County Roof Type Detail Distribution
Table 19. Brevard County Combined Roof Type Distribution
Classification Number Percentage FLAT FLAT SHED Subtotal 3593 2.5% GABLE/HIP GABLE 95098 64.9% HIP 46481 31.7% IRREGULAR 96 0.1% MANSARD 287 0.2% Subtotal 141962 96.6% OTHERS PRE-STRS. CONC. 34 0.0% WOOD TR. WOOD DECK 974 0.7% B.J. & RIGID 16 0.0% LACK OF INFO 220 0.2% Subtotal 1244 0.8% Total 146799 100%
77
4.4.1.3. Exterior Wall
C .B. ST UC CO50%
WOOD FR M . ST UC CO
13%
WOOD F RM . N O SH .1%
BR IC K ON M SN V.2%
WOOD SHEAT HIN G6%
EXT. PLYWOOD3%
CEDA R BOAR D1%
H AR DB OA RD LAP1%
BR IC K ON WOOD3%
VIN YL/A LUM INUM3%
C.B. PLAIN17%
Figure 29. Brevard County Exterior Wall Material Detail Distribution
78
Table 20. Brevard County Combined Exterior Wall Distribution
Classification Number Percentage C.B. C.B. PLAIN 25229 C.B. STUCCO 73128 BRICK ON MSNV. 2626 STYROFOAM STUCCO 168 Subtotal 101151 68.9% WOOD FRAME WITH SIDINGS WALL BOARD 33 WOOD FRM. ASBESTOS 427 WOOD FRM. NO SH. 1324 WOOD FRM. STUCCO 18230 WOOD SHEATHING 8455 WOOD SHINGLES 560 EXT. PLYWOOD 4562 EXT. HDBD. PANEL 667 CEDAR B & B 839 HARDBOARD LAP 1027 BRICK ON WOOD 4087 VINYL/ALUMINUM 5055 Subtotal 45266 31% OTHER COMPOSITION 38 PERMASTONE 242 REINFORCED CONCRETE 9 SHEET METAL 17 ENAMEL STEEL 27 Subtotal 295 0.2% Total 146799 100%
4.4.1.4. Number of Story (Story Height)
Brevard County’s database contains information roof height instead of number of
story. So the author has to assume all buildings in Brevard County are one story.
79
Roof height distribution of single family buildings in Brevard County
9 feet
97%
10 feet2%
11feet1%
Figure 30. Brevard County Roof Height Distribution
80
4.4.1.5 Year Built Distribution
Percentile Distribution of year built of single family residence in Brevard County
0%
1%
2%
3%
4%
5%
6%
1875 1901 1905 1909 1913 1917 1921 1925 1929 1933 1937 1941 1945 1949 1953 1957 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997
year built
per
cen
tag
e
Figure 31. Brevard Year Built Distribution
81
4.4.1.6. Area Distribution
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
0~50
0
500~
1000
1000
~150
0
1500
~200
0
2000
~250
0
2500
~300
0
3000
~350
0
3500
~400
0
4000
~450
0
4500
~500
0
5000
~550
0
5500
~600
0
6000
or ab
ove
Range in square f t
Figure 32. Brevard County Area Distribution
82
4.4.2 Hillsborough County
4.4.2.1 Roof Cover Materials
Asphalt/Comp. Shingl82%
Concrete Tile4%
None1%
M odular M etal3%
Blt.up Tar & Gravel
7%
M inimum2%
Rolled Composition
1%
Figure 33. Hillsborough County Roof Cover Detail Distribution
83
Table 21. Hillsborough County Combined Roof Cover Distribution
Classification Number Percentage SHINGLE ASPHALT/COMP.SHINGL 3 CORRUGATED ASBESTOS 5 ASBESTOS SHINGLE 6 CEDAR SHAKE 9 WOOD SHINGLE 10 282827 89.7% TILE CONCRETE TILE 7 CLAY OR BERMUDA TILE 8 SLATE 11 451 0.1% OTHER ROLLED COMPOSTION 2 NONE 0 MINIMUM 1 BLT. UP TAR & GRAVEL 4 MODULAR METAL 12 EMAMAL METAL 13 COPPER 14 EXCEOTIONAL 15 31952.5 10.1% 315232 100.0%
84
4.4.2.2. Roof Type
Gable o r Hip90%
Steel Frame or Truss
1%
Flat4%
None1%
Rigid F rame/ Barjoist
2%
Wood Truss2%
Figure 34. Hillsborough County Roof Type Detail Distribution
85
Table 22. Hillsborough County Combined Roof Type Distribution
Classification Number Percentage FLAT FLAT SHED Subtotal 12848 4.1% GABLE/HIP GABLE/HIP WOOD TRUSS SAWTOOTH MANSARD GAMBREL IRREGULAR BOW STRING TRUSS Subtotal 309475 90.3% OTHERS NONE RIGID FRAME/BARJOIST STEEL FRAME OR TRUSS REINFORCED CONCRETE PRESTRESS CONCRETE Subtotal 17652 5.6% Total 315232 100.0%
86
4.4.2.3. Exterior Wall
Tile/Frame Stucco4%
C.B. Stucco43%
Single Siding/No Sh1%
Cedar or Redwood1%
Prefab wood panel3%
Concrete Block29%
Wood on Sheathing5%
Asbestos Shingle2%
Face Brick1%
Modular Metal4%
Alum/Vinyl Siding6%
Common Brick1%
Figure 35. Hillsborough County Exterior Wall Detail Distribution
87
Table 23. Hillsborough County Exterior Wall Combined Distribution
Classification Number Percentage C.B CONCRETE BLOCK C.B. STUCCO CEMENT BRICK COMMON BRICK STONE Subtotal 224514 71.2% WOOD FRAME WITH SIDING COMP/WALL BOARD SHINGLE SIDING/ NO. SH AVERAGE FACE BRICK BOARD & BATTEN AVG. ASBESTOS SHINGLE CORRUGATED ASBESTOS WOOD ON SHEATHING ABOVE AVERAGE BOARD & BATTEN CEDAR OR REDWOOD PREFAB WOOD PANEL SWOOD SHINGLE TILE/FRAME STUCCO ALUM/VINYL SIDING BELOW AVERAGE Subtotal 75444 23.9% OTHER NONE MINIMUM REINFORCED CONCRETE CORRUGATED METAL MODULAR METAL PREFINISHED METAL GLASS THERMOPANE UNFINISHED PRECAST PANEL Subtotal 15274 4.8% Total 315232 100.0%
88
4.4.2.4. Number of story
There are 315232 single family residential buildings. 268388 1-story single family
residential buildings and 41686 second story buildings.
4.4.2.5. Year Built
E ffec tiv e Y ear D is trib u tio n C h art o f S ing le F am ily R es id en c e in H ills b oro u gh C o u n ty
0 %
1 %
2 %
3 %
4 %
5 %
6 %
7 %
0 1 94 6 1 9 4 9 1 95 2 1 9 55 19 5 8 1 9 61 19 6 4 1 96 7 1 9 7 0 19 7 3 1 9 76 19 7 9 1 98 2 1 9 8 5 1 98 8 1 9 91 19 9 4 1 99 7 2 0 0 0
Figure 36. Hillsborough County Year Built Distribution
4.4.2.6. Area
Hillsborough County’s area information comes in form of size index. It only
provides discrete number of area. The author used weighed average method to
calculate the area information in the analysis.
89
0%
5%
10%
15%
20%
25%
4001 3201 2401 2001 1801 1601 1401 1201 1001 801 601
Area (square feet)
Figure 37. Hillsborough County Area Distribution
4.4.3 Pinellas County
There are 262685 records in PINELLAS_RES, including 250826 single family
residential buildings
90
4.4.3.1. Roof Cover Materials
COMPOS-SHGLE69%
B.U.-TAR&GRAV-OTHER FLAT
6%
CONC TILE23%
MH ROOF COVER1% MEMBRANE
1%
Figure 38. Pinellas County Roof Cover Detail Distribution
91
Table 24. Pinellas County Roof Cover Combined Distribution
Classification Number Percentage
MEMBRANE MTL-CORR/SHEET COMPOSITION ROLL B.U-TAR & GRAV-OTHER FLAT Subtotal 23656 10.1% TILE CONC TILE/AVG METAL CLAY TILE/GLAZED Subtotal 56311 22.4% SHINGLE SLATE/GOOD METAL COMPOS-SHGLE ASBEST/WD SHGL Subtotal 173885 66.2% METAL/OTHER MH ROOF COVER MH ROOF OVER METAL-SHGLE Subtotal 8550 3.6% Total 262685 100.0%
92
4.4.3.2. Roof Type
GABLE-HIP 92%
M ANSARD/ GAMBREL
0%
Blank1% FLAT-SHED
7%
Figure 39. Pinellas County Roof Type Distribution
Table 25. Pinellas County Roof Type Distribution
Classification Number Percentage FLAT-SHED FLAT-SHED 0 Subtotal 18125 6.9% GABLE-HIP GABLE-HIP MANSARD/GAMBREL Subtotal 241145 91.8% Blank 3152 1.2% Total 100.0%
93
4.4.3.3. Exterior Wall
FRAM E-STUCCO3%
CONC BLK23%
C.B. STUCCO/ CB
RECLAD50%
FRAM E/ RECLAD ALUM / VINYL
5%
FRAM E-M ETAL1%
FRAM E-SIDING17%
FRAM E/ CUSTOM WOOD
1%
Figure 40. Pinellas County Exterior Wall Detail Distribution
Table 26. Pinellas County Exterior Wall Distribution
Classification Number Percentage FRAME-WITH SIDING FRAME-METAL FRAME-SIDING FRAME-STUCCO FRAME/RECLAD ALUM/VINYL FRAME/CUSTOM WOOD Subtotal 72238 27.5% C.B. CONC BLK C.B. STUCCO/CB RECLAD Subtotal 188870 71.9% OTHERS FRAME-COMPOS MASONRY BRICK MASONRY STONE FRAME/BRICK/STONE Subtotal 1576 0.6%
Total 262685 100.0%
94
4.4.3.4. Number of Story
There are 235242 1 story single family residential buildings and 15584 more story
buildings.
4.4.3.5. Year Built
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%
7.0%
019
0319
0919
1419
1919
2419
2919
3419
3919
4419
4919
5419
5919
6419
6919
7419
7919
8419
8919
9419
99
lack of record
Figure 41. Pinellas County Year Built Distribution
95
4.4.3.7. Area
0%
10%
20%
30%
40%
50%
60%
0- 1000 1001- 2000 2001- 3000 3001- 4000
Range (square ft)Range (square ft)Range (square ft)Range (square ft)
Percentage
Percentage
Percentage
Percentage
Figure 42. Pinellas County Area Distribution
96
4.4.4 Escambia County
4.4.4.1. Roof Cover Materials
C OM P S H I N GLE
8 6 %
B U I L T U P M TL / GYP
2%
D I M / A R C H S H I N GLE
4 %
TI LE/ C LA Y/ C EM EN T
1%
B U I LT U P ON WOOD
4 %
C OR R M ETA L
1%
R OLL ED R OOF
1%
A S B ES TOS / WOOD
SH I N GLE
1%
Figure 43. Escambia County Roof Cover Material Detail Distribution
97
Table 27. Escambia County Combined Roof Cover Material Distribution
Classification Number Percentage SHINGLE/TILE ASBESTOS/WOOD SHINGLE 643 COMP SHINGLE 87560 DIM/ARCH SHINGLE 4076 SLATE 28 TILE/CLAY/CEMENT 632 Subtotal 92939 91% METAL/OTHER BUILT UP MTL/GYP 2357 BUILT UP ON WOOD 3672 CORR METAL 1144 ENAMEL METAL 388 ROLLED ROOF 917 Subtotal 8962 9% Total 102289 100%
4.4.4.2. Roof Type
GABLE
18%
HI PITCH GABLE
8 %
HIP
6 %
HI PITCH HIP
6%
GABLE & HIP
COM BO2 %
HI PITCH GABLE &
HIP COM BO3%
WOOD FRAM E
54%
CONCRETE
3 %
Figure 44. Escambia Roof Type Detail Distribution
98
Table 28. Escambia Roof Type Combined Distribution
Classification Number Percentage GABLE/HIP GABLE 18091 HI PITCH GABLE 8222 HIP 6006 HI PITCH HIP 5867 GABLE & HIP COMBO 2091
HI PITCH GABLE & HIP COMBO 3080 IRREGULAR 402 EXTR COMBO DES 8 MANSARD/GAMBREL 493 WOOD FRAME 54116 Subtotal 98376 96% OTHER CONCRETE 2572 RIGID 96 STEEL TRUSS 363 DOME/UNUSUAL 14 Subtotal 3913 4% Total 102289 100%
4.4.4.3. Exterior Wall
Siding Below Avg6%
Siding Sht Avg 13%
Siding Lap Avg 9%
Vinyl Siding 5%
Brick- Face Brick48%
Brick-Common6%
Concrete Block 10%
Concr Blck w Stucco
3%
Figure 45. Escambia County Exterior Wall Material Detail Distribution
99
Table 29. Escambia County Exterior Wall Material Distribution
Classification Number Percentage FRAME WITH SIDING ALUM SIDING 579 ASBESTOS SIDING 1029 COMPOS SIDING 11 GENERAL SIDING MINIMUM 595 SIDING BELOW AVG 6443 SIDING SHT AVG 12711 SIDING LAP AVG 8087 VINYL SIDING 5063 WOOD SHAKE/SHINGLE 60 STUCCO OVER WOOD 891 STUCCO OVER SYN 1118 BRICK FACE BRICK 45801 CLAY TILE 73 STONE 37 Subtotal 82498 81% C.B. BRICK BLK 306 BRICK CEMENT 698 BRICK COMMON 5824 CONCRETE BLOCK 8834 CONCRETE BLOCK STUCCO 2719 Subtotal 18381 18% OTHER GLASS 0 LOG 31 METAL 22 PRECAST PANEL 1333 CORRUG METAL 24 Subtotal 1410 1% Total 100%
4.4.4.4. Number of Stories
Table 30. Escambia Number of Story Distribution
1story 87% 2 story or more 13%
100
Database shows that 99.9% of single family residential buildings have one story.
This is not likely true, however this is will be used as an assumption for present
research.
4.4.4.5.Year Built
Residential Building Construction in Escambia County
0
500
1000
1500
2000
2500
3000
3500
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Year
Res
iden
tial
Bu
ildin
gs
Co
nst
ruct
ed
Figure 46. Escambia County Year Built Distribution
101
4.4.5 Leon County
4.4.5.1. Roof Cover
Shingle95%
Built-Uproof3%
Tile1%
Other1%
Figure 47. Leon County Roof Cover Distribution
102
Table 31. Leon County Roof Cover Distribution
Classification Number Percentage SHINGLE/TILE COMP SHINGLE CORRUGATED ASBESTOS CEMENT OR CLAY TILE ASB SHINGLE SLATE CORR COMP ASB SHINGLE WOOD SHINGLE Subtotal 57279 95% OTHER ROLLED ROOFING METAL BUILT UP Subtotal 3015 5% Total 60294 100%
4.4.5.2. Roof Type
Gable/HipGable/HipGable/HipGable/Hip
90%90%90%90%
Flat/otherFlat/otherFlat/otherFlat/other
10%10%10%10%
Figure 48. Leon County Roof Type Distribution
103
Table 32. Leon County Combined Roof Type Distribution
Classification Number Percentage GABLE/HIP GABLE/HIP MANSARD WOOD FRAME/TRUSS Subtotal 54265 90% OTHER FLAT OTHER BAR JOIST/RIGID FRAME STEEL TRUSS Subtotal 6029 10% Total 60294 100%
4.4.5.3. Exterior Wall
Concrete Block Concrete Block Concrete Block Concrete Block
8%8%8%8%
Timber frameTimber frameTimber frameTimber frame
44%44%44%44%
Stucco over Stucco over Stucco over Stucco over
woodwoodwoodwood
3%3%3%3%
Brick count as Brick count as Brick count as Brick count as
timber frametimber frametimber frametimber frame
42%42%42%42%
cement brickcement brickcement brickcement brick
3%3%3%3%
Figure 49. Leon County Exterior Wall Material Detail Distribution
104
Table 33. Leon County Exterior Wall Material Distribution
Classification Number Percentage C.B. CONCRETE BLOCK CONCRETE BLK/WOOD CONCRETE BLK/STUCCO CONCRETE BLK/BRICK COMMON BRICK CEMENT BRICK Subtotal 9630 16% FRAME WITH SIDING COMMON BRICK/WOOD SIDING MINIMUM SIDING BELOW AVG. SIDING AVG SIDING ABOVE AVG WOOD SIDING FACE BRICK Subtotal 50574 83.8% METAL/OTHER CORRUGATED METAL PREFINISH METAL PRECAST PANEL GLASS PRECAST PANEL FB Subtotal 72 0.1% Total 100%
4.4.5.5. Number of Stories
Database shows that 99.9% of single family residential buildings have one story.
This is not likely true, however this is will be used as an assumption for present
research.
105
4.4.5.6. Year Built
Single Family Home Construction in Leon County
0
500
1000
1500
2000
2500
1900
1903
1906
1909
1912
1915
1918
1921
1924
1927
1930
1933
1936
1939
1942
1945
1948
1951
1954
1957
1960
1963
1966
1969
1972
1975
1978
1981
1984
1987
1990
1993
1996
1999
Year
Nu
mb
er o
f S
ing
le F
amily
Res
iden
ces
Co
nst
ruct
ed
Figure 50. Leon County Year Built Distribution
4.4.6 Walton County
Walton County has a population of 40601. Its database’s format is older version
Access file. It has 93613 single family residential buildings.
106
4.4.6.1. Roof Cover Materials
MetalMetalMetalMetal
5%5%5%5%
Composite Composite Composite Composite
ShingleShingleShingleShingle
68%68%68%68%
TileTileTileTile
5%5%5%5%
Asb.shingleAsb.shingleAsb.shingleAsb.shingle
10%10%10%10%
OtherOtherOtherOther
12%12%12%12%
Figure 51. Walton County Roof Cover Material Detail Distribution
107
Table 34. Walton County Combined Roof Cover Material Distribution
Classification Number Percentage SHINGLE/TILE MINIMUM COMP LIGHT CORG ASB ASB SHINGLE CONCRETE TILE CLAY TILE CEDAR SHAKE WOOD SHINGLE SLATE Subtotal 77885 83.2% OTHER ROLL COMPOSITE BUILT-UP MODULAR METAL MODERN TIN SEAMED TINE COMP HEAVY OUTLINE Subtotal 15728 16.8% Total 100.0%
108
4.4.6.2. Roof Type
Gable/HipGable/HipGable/HipGable/Hip
97%97%97%97%
Flat/otherFlat/otherFlat/otherFlat/other
3%3%3%3%
Figure 52. Walton County Roof Type Distribution
109
Table 35. Walton County Combined Roof Type Distribution
Classification Number Percentage GABLE/HIP GABEL/HIP WOOD TRUSS SAWTOOTH MANSARD GAMBREL IRREGULAR Subtotal 90737 97.0% OTHER FLAT SHED RIDGE FRAME STEEL FRAME BOWSTRING TRUSS REINFORCED CONCRETE PRESTRESSED CONCRETE Subtotal 2853 3.0% Total 100%
110
4.4.6.3. Exterior Wall
CONCRETE BLOCK10%
STUCCO ON CONCRETE BLOCK
5%
BRICK24%
SHINGLE SIDING28%
WOOD EXTERIOR16%
STUCCO ON WOOD FRAME
6%
VINEER SIDING7%
OTHER4%
Figure 53. Walton County Exterior Wall Material Detail Distribution
111
Table 36. Walton County Exterior Wall Material Distribution
Classification Number Percentage C.B. CONCRETE BLOCK STUCCO ON CONCRETE BLOCK BRICK STONE Subtotal 35868 38.3% WOOD FRAME WITH SIDING WALL BOARD BELOW AVG SHINGLE SIDING AVG BD/BATTEN ASB SHINGLE WOOD ON PLY COR. ASB ABOVE AVG CEDAR WOOD SHINGLE STUCCO ON WOOD FRAME VINEER SIDING Subtotal 54135 57.8% OTHER MINIMUM METAL NO DESP REINFORCE CON. PREFAB PANEL Subtotal 3605 3.9% Total 100%
4.4.6.5. Number of Story
There are 80507 one story single family buildings, which counts for 86% of total
single family buildings and 13105 second story single family buildings.
112
4.4.6.6. Year Built
Residential Building Construction in Walton County
0
500
1000
1500
2000
2500
3000
3500
4000
<190
019
0319
0619
0919
1219
1519
1819
2119
2419
2719
3019
3319
3619
3919
4219
4519
4819
5119
5419
5719
6019
6319
6619
6919
7219
7519
7819
8119
8419
8719
9019
9319
9619
99
Year
Res
iden
tial
Bu
ildin
gs
Co
nst
ruct
ed
Figure 54. Walton County Year Built Distribution
4.4.7 Broward County
Broward County has a massive population of 1623018. The county’s database has
many blank records. The single family residential file contains 371745 record and
344508 records are not blank.
113
4.4.7.1. Roof Cover
Shingle/TileShingle/TileShingle/TileShingle/Tile
37%37%37%37%
TileTileTileTile
54%54%54%54%
Built compositeBuilt compositeBuilt compositeBuilt composite
9%9%9%9%
Figure 55. Broward County Roof Cover Material Detail Distribution
114
Table 37. Broward County Roof Cover Material Distribution
Classification Number Percentage SHINGLE/TILE SHINGLE/TILE SHINGLE/ASBESTOS SHINGLE COMPOSITE TILE TILE Subtotal 305675 90.8% OTHER BUILT COMPOSITE Subtotal 31041 9.2% Total 100.0%
4.4.7.2. Roof Type
Hip/GableHip/GableHip/GableHip/Gable
96%96%96%96%
FlatFlatFlatFlat
3%3%3%3%
Prestressed(Flat Prestressed(Flat Prestressed(Flat Prestressed(Flat
shape,concrete shape,concrete shape,concrete shape,concrete
beam or slab)beam or slab)beam or slab)beam or slab)
1%1%1%1%
Figure 56. Broward County Roof Type Detail Distribution
115
Table 38. Broward County Roof Type Distribution
Classification Code Number Percentage GABLE/HIP GABLE/HIP 8 TRUSS WOOD 11 Subtotal 331293 96.3% OTHER FLAT 7 TRUSS STEEL 15 PRESTRESSED 12 STEEL BAR JOIST 9 BAR JOIST 12 Subtotal 12852 3.7% Total 100.0%
4.4.7. 3. Exterior Wall
Frame Stucco(Wood Frame Stucco(Wood Frame Stucco(Wood Frame Stucco(Wood
Studs with Lath Studs with Lath Studs with Lath Studs with Lath
and Stucco)and Stucco)and Stucco)and Stucco)
1%1%1%1%
CB CB CB CB
Stucco/Frame/HollStucco/Frame/HollStucco/Frame/HollStucco/Frame/Holl
ow Tileow Tileow Tileow Tile
99%99%99%99%
Figure 57. Broward County Exterior Wall Distribution
116
Table 39. Broward County Exterior Wall Distribution
Classification Code Number Percentage C.B. C.B.PLAIN 31 C.B.STUCCO 33 BRICK 38 Subtotal 338222 99.3% FRAME WITH SIDING FRAME 25 FRAME (COUNT AS C.B.) 33 FRAME STUCCO 28 ASBESTOS SIDING 28 CORRUGATED METAL 16 WALL BOARD 16 Subtotal 2344 0.7% Total 100.0%
The Broward County’s database maintenance staffs use the same code “33” for
both C.B.Stucco and Frame, which is obviously impropriate. However, there is no
other information for us to get C.B. Stucco and Frame breakdown within all records
with code “33”. At this stage, we consider code “33” stands for “C.B.Stucco” since
Broward County is one of the southern counties where Concrete Block exterior
wall has predominant distribution.
4.4.7.4. Number of Story
There are 232220 (80%) single story residential buildings. And 65090 (20%)
second story residential buildings.
117
4.4.7.5. Year Built
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
1919 1927 1932 1937 1942 1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997
year builtyear builtyear builtyear built
Percentage
Percentage
Percentage
Percentage
Figure 58. Broward County Year Built Distribution
118
4.4.7.6. Area
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
1-5001-5001-5001-500 501-1000501-1000501-1000501-1000 1001-15001001-15001001-15001001-1500 1501-20001501-20001501-20001501-2000 2001-25002001-25002001-25002001-2500 2501-30002501-30002501-30002501-3000 3001-35003001-35003001-35003001-3500
Area Range (square feet)Area Range (square feet)Area Range (square feet)Area Range (square feet)
percentage
percentage
percentage
percentage
Figure 59. Broward County Area Distribution
4.4.8 Palm Beach County
Palm Beach County has a population of 1131184. Its county’s database is very
large and can not be opened using conventional database software. The authors
used C language programming to extract useful data. It has 222918 single family
residential buildings.
119
4.4.8.1. Roof Cover
ASPH/COMP. SHG.ASPH/COMP. SHG.ASPH/COMP. SHG.ASPH/COMP. SHG.
30%30%30%30%
CORR.ASB.CORR.ASB.CORR.ASB.CORR.ASB.
4%4%4%4%
ASB. SHGASB. SHGASB. SHGASB. SHG
4%4%4%4%CONC.TILECONC.TILECONC.TILECONC.TILE
26%26%26%26%
CLAY/BERMUDA CLAY/BERMUDA CLAY/BERMUDA CLAY/BERMUDA
TILETILETILETILE
34%34%34%34%
ROLLED ROLLED ROLLED ROLLED
COMP./MODULAR COMP./MODULAR COMP./MODULAR COMP./MODULAR
MTLMTLMTLMTL
2%2%2%2%
Figure 60. Palm Beach Roof Cover Material Detail Distribution
120
Table 40. Palm Beach Combined Roof Cover Material Distribution
Material Code Number Percentage SHINGLE MINIMUM 1 ASPH/COMP. SHG. 3 66875 30% CORR.ASB. 5 8917 4% ASB. SHG 6 9067 4% CEDAR SHAKES 9 WOOD SHAKE 10 38% TILE CONC.TILE 7 57959 26% CLAY/BERMUDA TILE 8 75792 34% SLATE 11 Subtotal 60% OTHER ROLLED COMP. 2 MODULAR MTL 12 Subtotal 3676 1.8% Total 100%
4.4.8.2. Roof Type
FLAT3%
GABLE/HIP94%
WOOD TRUSS3%
Figure 61. Palm Beach Roof Type Detail Distribution
121
Table 41. Palm Beach Roof Type Distribution
Classfication Code Number Percentage GABLE/HIP GABLE/HIP 3 WOOD TRUSS 4 SAWTOOTH 5 MANSARD 6 GAMBREL 7 IRREGULAR 8 Subtotal 210176 94.3% OTHER FLAT 1 SHED 2 RIGID FRM 9 STEL FRM OR TRUSS 10 REINF. CONC. 12 PRESTRESSED CONC. 13 Subtotal 12712 5.7% Total 100.0%
4.4.8.3. Exterior Wall
CB STUCCO79%
VI NLY/ ALUMI NUM SI DI NG
1%
CONC. BLOCK1%
COMMON BR1%
WOOD FRM STUCCO/ HOLLOW
6%
WOOD SI DI NG9%
ABOVE. AV1% BD & BATTEN
2%
Figure 62. Palm Beach Exterior Wall Material Detail Distribution
122
Table 42. Palm Beach Exterior Wall Material Combined Distribution
Classification Code Number Percentage C.B. CONC. BLOCK 15 CB.STUCCO 17 COMMON BR. 19 STONE 21 CEMENT BR. 18 FACE BR. 20 Subtotal 171300 77.2% SIDING MINIMUM 1 COMP OR WALL BD 2 BELEOW AV. 3 SINGLE SIDING 4 WOOD SIDING 5 BD.& BATTEN 6 ASB. SHG. 7 WOOD SHT/PLY 8 CORR.ASB. 9 ABOVE AV. 10 BD. & BATTEN AV 11 CEDAR/REDWOOD 12 WOOD SIDING 14 WOOD FRM STUCCO 16 VINLY/ALUMINUM SIDING 26 Subtotal 50508 22.8% Total 100.0%
4.4.8.4. Number of Story
There are 205085 (92%) 1 story single family residential buildings and 17833 (8%)
2 story single family houses.
123
4.4.8.5. Year Built
0%
1%
1%
2%
2%
3%
3%
4%
4%
5%
1951 1955 1959 1963 1967 1971 1975 1979 1983 1987 1991 1995 1999
Year built
Per
cen
tag
e
Figure 63. Palm Beach County Year Built Distribution
124
4.4.8.6. Area
0%
5%
10%
15%
20%
25%
30%
0~50
0
501~
1000
1001
~150
0
1501
~200
0
2001
~250
0
2501
~300
0
3001
~350
0
3501
~400
0
4001
~450
0
4500
~
Area Range (square feet)
per
cen
tag
e
Figure 64. Palm Beach Area Range Distribution
4.4.9 Monroe County
The key region contains only one county – Monroe County. Due to its geographical
uniqueness, we distinguish it from other region. It has a population of 79589.
Monroe County’s database can be easily opened by any database software. It has
total 23480 single family residential buildings.
125
4.4.9.1. Roof Cover
Asphal t Shi ngl e 47%
Conc/ cl ay t i l e8%
Met al26%
Mi n/ Pai nt Conc10%
Rol l ed compos1%
Tar & Gr avel7%
Wood Shi ngl e1%
Figure 65. Monroe County Roof Cover Material Detail Distribution
126
Table 43. Monroe County Roof Cover Material Distribution
Classification Number Percentage SHINGLE/TILE ASPHALT SHINGL CONC/CLAY TILE WOOD SHINGLE Subtotal 13204 56.3% OTHER METAL 5570 MIN/PAINT CONC NONE ROLLED COMPOS TAR & GRAVEL Subtotal 10259 43.7% Total 100.0%
4.4.9.2. Roof Type
Fl at or shed9%
Gabl e/ Hi p71%
I RR/ Cust om12%
Rei nf or ced Conc7%
Mansar d1%
Figure 66. Monroe County Roof Type Detail Distribution
127
Table 44. Monroe County Roof Type Combined Distribution
Classification Number Pecentage GABLE/HIP GABLE/HIP IRR/CUSTOM MANSARD Subtotal 19961 85.1% OTHER NONE PRESTRESS CONC REINFORC CONC STEEL TRUSS FLAT OR SHED Subtotal 3502 14.9% Total 100.0%
4.4.9.3. Exterior Wall
Above aver age wood12%
B & B2%
C. B. S.51%
Conc. Bl ock1%
Concr et e St uc.11%
Met al / Al um1%
Rei n Conc1%
Vi nyl si di ng4%
WD FR St ucco1%
WD FRAME16%
Figure 67. Monroe County Exterior Wall Detail Distribution
128
Table 45. Monroe County Exterior Wall Combined Distribution
Classification Number Percentage C.B. C.B.S. CONC BLOCK CUSTOM STONE/BRICK Subtotal 14928 63.6% WOOD W/SIDING B & B AB AVE WOOD SIDING ABOVE AVERAGE WOOD AVE WOOD SIDING MIN WOOD SIDING VINYL SIDING WD FR STUCCO WD FRAME WD WITH CONC BLK Subtotal 7919 33.8% METAL/OTHER METAL SIDING METAL/ALUM REIN CONC Subtotal 608 2.6% Total 100.0%
4.4.9.4. Number of Story
There are 22772 (97%) 1 story single family residential buildings and 665 (3%)
second story buildings in Monroe County’s database.
129
4.4.9.5. Year Built
0%
1%
2%
3%
4%
5%
6%
1876
1900
1904
1910
1918
1922
1926
1930
1934
1938
1942
1946
1950
1954
1958
1962
1966
1970
1974
1978
1982
1986
1990
1994
1998
year built
per
cen
tag
e
Figure 68. Monroe County Year Built Distribution
130
4.4.9.6. Area
0%
2%
4%
6%
8%
10%
12%
14%
16%
0~50
0
500~
1000
1000
~150
0
1500
~200
0
2000
~250
0
2500
~300
0
3000
~350
0
3500
~400
0
4000
~450
0
4500
~500
0
5000
~550
0
5500
~600
0
6000
~650
0
6500
~700
0
Area Range (square feet)
Per
cen
tag
e
Figure 69. Monroe County Area Range Distribution
4.4.10 Define the Most Common Structural Types
Based on the information presented in the preceding sections, the principal
structural characteristics classification is shown in Table 46.
Table 46. Simplified Structural Characteristic Types
Roof Cover Roof Type Exterior Wall Number of Story Shingle Gable Wood frame 1 Tile Hip Masonry 2 Others Others Others more
The main structural types of single family residential building are, result from
combination of these 4 characteristics: number of story (either 1 or 2), roof cover
(shingle/tile), roof type (either gable or hip) and exterior wall material (either
131
concrete blocks or wood). In addition, due to the variety of the building stock in
the Keys, 4 additional types were defined for the Keys (corresponding to buildings
with metal roof cover). Based on the above criteria, Table 47 listed the assumed 16
structural types.
Table 47. Assumed Structural Type Definitions
No of Stories Exterior Wall Roof Materials Roof Type Type 1 1story concrete blocks Shingle/Tile Gable Type 2 1story concrete blocks Shingle/Tile Hip
Type 3 1story Wood Shingle/Tile Gable
Type 4 1story Wood Shingle/Tile Hip
Type 5 2 story 1 story: concrete block; 2 story:
wood Shingle/Tile Gable
Type 6 2 story 1 story: concrete block; 2 story:
wood Shingle/Tile Hip
Type 7 2 story Wood Shingle/Tile Gable
Type 8 2 story Wood Shingle/Tile Hip
Type 9 1story concrete blocks Metal Gable
Type 10 1story concrete blocks Metal Hip
Type 11 1story Wood Metal Gable
Type 12 1story Wood Metal Hip
Type 13 2 story 1story: concrete block; 2 story:
wood Metal Gable
Type 14 2 story 1story: concrete block; 2 story:
wood Metal Hip
Type 15 2 story Wood Metal Gable
Type 16 2 story Wood Metal Hip
Compared with the ISO construction types listed in Table 4 in Section 4.2.1, these
structural types are defined to look at the component resistance of the building parts,
for example, wood siding exterior wall are defined separately with concrete block
exterior wall because they have totally different structural behavior, whereas, ISO
construction types are used to evaluate the policies. It is based on characteristics
such as combustible and non-combustible materials, which does not directly reflect
the structural characteristics of the buildings but rather their fire resistance
132
4.4.11 Distribution of the Most Common Structural Types Per County
Once the structural types were defined, the next step was to find the in statistical
distribution in each county. The author used computer database language to find the
exact numbers of buildings of each assumed structural type, eliminate those types
having negligible percentages and get the final most common structural types for
single family residential buildings for each counties. The following Tables 47-50
present the results county by county. Area information was also collected for each
structural type. The unit of area is square feet. The standard deviation for each area
statistics was computed as followed:
STDEV for each structural type = n
xxn
ii∑
=
−1
2)(, where ix is the area for every
building surveyed, x is the average area of all building survey under certain
category. n is the number of building being surveyed. The counties are grouped
by regions that are defined in the next section.
133
Table 48. Central Region Counties Model Distribution
Central Region
Brevard County Number Percentage Area (ft2) STDEVof area(ft2)
Type 1 57639 39.3% 1802 724 Type 2 28026 19.1% 2290 834 Type 3 26342 15.9% 1954 879 Type 4 12718 8.7% 2190 979
83% Unknown type 13%
Hillsborough County Number Percentage Area (ft2) STDEV of area(ft2)
Type 1 or 2 195499 62% 2390 910 Type 3 or 4 39933 13% 1998 789 Type 5 or 6 23106 7% 5076 3675 Type 7 or 8 15180 5% 3624 1882
87% Unknown type 13%
Pinellas County Number Percentage Area(ft2) STDEV of area(ft2)
Type 1 or 2 169223 72% 2232 377 Type 3 or 4 46691 20% 1753 1022 Type 5 or 6 2017 1% 2128 155 Type 7 or 8 6736 3% 2108 57
96% Unknown type 4%
134
Table 49. Northern Region Counties Model Distribution
Northern Region Leon County Number Percentage Area(ft2) STDEV of area(ft2) Type 1 or 2 3768 6.2% 1361 675 Type 3 or 4 48527 80.4% 1626 705 86.6% Unknown type 13.0% Escambia County Number Percentage Area(ft2) STDEV of area(ft2) Type 1 or 2 12746 12% 1733 810 Type 3 or 4 67445 66% 2058 958 Type 5 or 6 1057 1% 4289 3633 Type 7 or 8 10798 11% 3424 2882 90% Unknown type 10% Walton County Number Percentage Area(ft2) STDEV of area(ft2) Type 1 or 2 30681 22% 2013 286 Type 3 or 4 43833 32% 2040 233 Type 5 or 6 2017 1% 2128 155 Type 7 or 8 6736 5% 2108 57 61% Unknown 39%
135
Table 50. Southern Region Counties Model Distribution
Southern Region Palm Beach County
Number Percentage Area (ft2)
Area STDEV(ft2)
Type 1 or 2 147798 66% 2118 931 Type 3 or 4 31343 14% 1702 814 Type 5 or 6 9736 4% 4302 2185 Type 7 or 8 9553 4% 2964 1289
89% Unknown Type 11%
Broward County Number Percentage Area(ft2) Area STDEV (ft2)
Type 1 or 2 242466 70% 2013 286 Type 3 or 4 1385 0% 2040 233 Type 5 or 6 56732 16% 2128 155 Type 7 or 8 48 0% 2108 57
87% Unknown Type 13%
Table 51. The Key Region – Monroe County Model Distribution
Key Region Monroe County
umber percentage Area (ft2) Area
STDEV(ft2) Type 1 or 2 7683 34% 3289 2069 Type 3 or 4 4159 18% 270 1026 Type 5 or 6 2755 12% 2691 2285 Type 7 or 8 2267 10% 2108 57 Type 9 or 10 64 0% 3680 2596 Type 11 or 12 63 0% 3230 1110 Type 13 or 14 83 0% 3670 1714 Type 15 or 16 409 2% 2325 1305
77% Unknown 23%
136
4.5 Florida Regions
4.5.1 Introduction
The aim of the structural survey is to generate a manageable number of building
models to cover the majority of Florida building stock. To define building models
for each county could be cumbersome and unnecessary because the major building
types are similar over a large area with same geographical and climate conditions.
So the author chooses to divide Florida counties into four regions with different
building mixes based on several reasons:
1. The author reference of the Frame building distribution investigation results
from FCHF, listed in Figure 12. The author aims to divide the region with similar
building characteristics, such as whether the majority homes have exterior wall or
not.
2. Based on geographical vicinity, all regions are defined on adjacent counties.
3. To make sure that each region have at least two sample counties and also
consider the ratio population covered by the sample counties. For example, sample
counties with large building data should be included in the region with more
population. Figure 66 shows the resulting regional division.
137
Northern Region
Central Region
Southern Region
The Keys Region, Monroe County
Escambia, Walton, Leon
Pinellas Hillsborough Brevard
Palm Beach Broward
Figure 70. Four Regions with Sample Counties Highlighted and Name Marked
In each region, there are at least two counties per region which are: Escambia,
Walton, and Leon in the Northern region; Brevard, Pinellas, and Hillsborough in
the Central region; Palm Beach, and Broward in the Southeast region; and Monroe
County fully covers the Keys region. These counties for which we have obtained
database are referred as sample counties. The number of counties and sample
counties in each region and the population associate with them are shown in
Table 51.
138
Table 52. Number of Counties and Population in Each Region
Central region
Northern region
Southeast region
The Keys
Total number of counties 27 34 5 1 Number of sample counties 3 3 2 1 Population of the region 7,690,240 2,885,559 5,326,990 79589 Population in sample counties 2,396,660 57,712 2,754,202 79589 Percentage of population covered 30% 20% 52% 100%
The Central Region covers 27 counties running from west coast to the east coast of
the peninsular. According to the census 2000, central region has a population of
7,690,240 people and 3 counties of which building information have been obtained
cover population of 2,396,660, approximately 30% of the total population of the
region.
The Northern Region comprises 34 northern – panhandle counties and 3 sample
counties which are Escambia, Leon County and Walton County. Northern Region
has a population of 2,885,559, three sample counties’ population is 574,463 which
covers roughly 20% of total population.
The Southern region includes 5 counties with a population of 5,326,990. Two
sample counties are Broward County and Palm Beach County, which covers 54%
of total population (2754202).
The key region contains only one county – Monroe County. Due to its geographical
uniqueness, we distinguish it from other region. It has a population of 79,589.
139
4.5.2 Distribution of Structural Types Per Region
The distributions of structural types for the counties were extrapolated to the
regions that contained these counties. The author uses the following procedure for
the extrapolation.
First, we assume that the counties for which we have data were selected randomly
from all the counties in the region and are referred to as “sample counties”. In
which case, we will have what is called a SIMPLE ONE STAGE cluster sample [5].
We also assume that all counties have a similar building population density in each
region.
For each region we define the following:
M = Total number of counties in the region. For example, the central region has 27
counties.
m = Number of counties in the sample. For example, central region has 3 sample
counties for which we have obtained data. (e.g. Brevard, Pinellas, Hillsborough)
yi = Number of single family homes in sample county i. (obtained directly from
databases) (e.g. 146799 single family homes)
xi = Number of homes of type I in ith sample county (obtained directly from
databases)
x = total number of homes of Type I in the whole sample = ∑=
m
iix
1
y = Total number of homes in the whole sample = ∑=
m
iiy
1
140
N = Total number of homes in region – which is unknown
X = Total number of homes of Model Type I in the region
p = X/N = proportion of Type I in the region. p is unknown, we estimate p usingp̂ .
p̂ = y
x = estimate of p, proportion of Type I in the region. The results of the
extrapolation are listed in Table 53, 54.
Table 53. Probability of Occurrence of Structural Type for 3 Regions.
Structural Type
definition Central Region Northern Region Southern Region
Model (Number of stories,
exterior wall, roof cover, roof type)
p̂ Area (ft2)
STDV (ft2)
p̂
Area (ft2)
STDV (ft2) p̂
Area (ft2)
STDV (ft2)
Type 1 1story, concrete blocks,
Shingle/Tile, Gable 42% 12% 46%
Type 2 1story, concrete blocks,
Shingle/Tile, Hip 22%
2222 550 6%
1702 590 23%
2147
734
Type 3 1story, Wood frame,
Shingle/Tile, Gable 12% 39% 4%
Type 4 1story, Wood frame,
Shingle/Tile, Hip 6%
1941 913 20%
1908 399 2%
2022 662
Type 5
2 stories, 1st story: concrete block; 2nd story:wood frame, Shingle/Tile, Gable
2% 1% 8%
Type 6
2 stories, 1st story: concrete block; 2nd story: wood frame, Shingle/Tile,
Hip
1%
3602 1915
0.4%
3208 1894
4%
3215
1170
Type 7 2 stories, Wood frame,
Shingle/Tile, Gable 1.4% 5% 1%
Type 8 2 stories, Wood frame,
Shingle/Tile, Hip 1%
2866 969 2.3%
2766 1470 1%
2118
673
Total coverage
87% 86% 89%
Unknown type
13% 14% 11%
141
Table 54. Probability of Occurrence of Structural Types for the Key Region
4.5.3 Result Analysis
The information in Table 53 and 54 matches closely the distribution of frame
construction in the FCHF database, shown in Figure 12. Table 53 indicates an
overwhelming number of building Type 3 and Type 4, which are with timber frame
walls, in Northern region, and more Type 1 and Type 2, which have masonry walls,
in the other regions.
The classification of structures is based on the comprehensive study of tax
appraiser’s database. The author first defined assumed structural types according to
the distribution of building characteristics, and then went to the tax appraiser’s
database to find the statistics. Hence, each structural type is backed up with solid
statistics. This time-consuming yet simply-philosophy approach ensures that the
structural types defined here reflect the real building stock in the given area.
Type Description Percentage Area (ft2)
STDEV(ft2)
Type 1 1story, concrete blocks, Shingle/Tile, Gable 23%
Type 2 1story, concrete blocks, Shingle/Tile, Hip 11%
3295 2071
Type 3 1story, Wood frame, Shingle/Tile, Gable 12% Type 4 1story, Wood frame, Shingle/Tile, Hip 6%
2771 1027
Type 9 1story, Concrete blocks, Metal, Gable 8% Type 10 1story, Concrete blocks, Metal, Hip 4%
3295 2070
Type 11 1story, Wood frame, Metal, Gable 7% Type 12 1story, Wood frame, Metal, Hip 3%
2179 1341
2story,all type 3% Total Coverage
77%
Unknown types
23%
142
4.5.4 Error Estimation
The error estimation is proposed by the statistician in our Project, Dr. Sneh Gulati
from Department of Mathematics at Florida International University.
The error of “p” for each structural type in each region was calculated and the
calculation procedure is as followed:
Define xclu = x/m = average number of single family homes of Type I in whole
sample
Define yclu = y/m = average number of single family homes in whole sample
Now SE(xclu) = standard error of xclu = 1M
mMˆ
m
1x −
−σ , and xσ̂ =
M
M
m
xxm
iclui
x
1
1
)(ˆ
2/1
1
2
−
−
−=∑
=σ , where M
M 1− is a finite collection factor.
Also SE(yclu) will be defined similarly, just by substituting yσ̂ in the above
expression for xσ̂ , with M
M
m
yym
iclui
y
1
1
)(ˆ
2/1
1
2
−
−
−=∑
=σ
SE (p̂ ) = standard error of p̂ is calculated as:
143
SE (p̂ ) =
[ ] [ ] ( )2/1
cluclucluclu
2clu
2clu
2clu
2clu yx(Cov
yx
1
M
mM
m
2
y
)y(SE
x
)x(SEp̂
−
−+
where Cov (xclu, yclu) is an estimate of the covariance between xclu and yclu and is
given by:
Cov (xclu, yclu) =
−
−−∑=
1m
)yy)(xx(m
1icluiclui
Finally, a 99% confidence interval for p, the true proportion of homes of model
type I in the region is:
p̂± 3 SE(p̂ )
At this stage, the distribution for p̂ is assumed to be normal, however, further
research is needed for defining a more precise distribution type.
Table 55 is a summary of results for the percentage of each structural type and error
estimation for 3 regions. Because there is only Monroe County in the Key region,
there is no need for extrapolation or error estimates in that particular region.
144
Table 55. p̂ Values and Error for Each Type for 3 Regions
Central Region Northern Region Southeast Region
p̂ SE p̂ 99%cnfdn. Intvl. p̂ SE p̂
99%fnfdn. Intvl. p̂ SE p̂
99%cnfdn. Intvl
Type 1 43% 2% 49%-37% 12% 5% 27%-0% 46% 1% 49%-43%
Type 2 22% 1% 25%%-19% 6% 3% 15%-0% 23% 0.5% 25%-22%
Type 3 13% 2% 19%-7% 39% 8% 63%-15% 4% 3.4% 14%-0% Type 4 7% 1% 10%-4% 20% 4% 32%-8% 2% 1.7% 7.1%-0% Type 5 2% 1.7% 7%-0% 1% 0.4% 2.2%-0% 8% 3% 17%-0% Type 6 1% 1% 4%-0% 0.4% 0.2% 1%-0% 4% 1.5% 8.5%-0% Type 7 1.4% 1.2% 5%-0% 5% 1.7% 10%-5% 1% 1.1% 4.3%-0% Type 8 1% 0.6% 3%-0% 2.3% 0.9% 5%-0% 1% 0.5% 3%-0%
We can test the extrapolation result with individual county structural type
distribution. For example, we chose Leon County from Northern Region (Table 49),
it has Type 3 or 4 has a probability of occurrence 80.4%, use the ratio 2:1 between
gable and hip to divide two types, we have Type 3 with 53% probability of
occurrence. Look up in the Table 55, in Northern Region, Type 3 covers range 63%
- 15%. 53% falls into this range. So we can say that Type 3 covers Northern
Region 39% of building stock with 8% of error.
145
Chapter 5. Statistical Analysis of Florida Manufactured Homes
5.1 Introduction
Manufactured Home can be defined as a structure, transportable in one or more
sections, which, in the traveling mode, is eight body feet or more in width or forty
feet or more in length, or when erected on site is three hundred twenty or more
square feet, and, which is constructed on a permanent chassis and designed to be
used as a "dwelling" with or without a permanent foundation when connected to the
required utilities, and includes the plumbing, heating, air conditioning, and
electrical systems contained therein.
Manufactured home, depending on its affordability, variety of style, sizes and floor
plans and flexibility of installation location, are welcomed by retiree and lower
income population. Florida has a large amount of population of retiree or lower
income families residing in manufactured homes. Although it only counts a
fraction of building stock, usually less than 10% shown from Table 10, it has
brought many most unfortunate effects. According to HUD (US Department of
House and Urban Development), in 1992, 97% of all manufactured homes in
Hurricane Andrew's path in Dade County, were destroyed, compared to 11% of
single-family, non-manufactured homes.
146
For a long time in history, unlike site-built housing, manufactured homes were
usually built in a different jurisdiction than the one in which they will be occupied.
This systemic gap in building regulation had a number of unfortunate effects in the
middle decades of this century: the lack of regulatory oversight made the quality of
"manufactured homes" unreliable at best and prompted many local governments to
ban or severely restrict their use. Since 1976, the Manufactured Home Construction
and Safety Standards have protected consumers and helped ensure the quality and
availability of this important affordable housing resource by providing a system for
regulating the design and construction of manufactured homes nationwide, known
as the “HUD Code”. The “HUD Code” is set of factory–built to meet federal
manufactured home construction and safety standard, which is administrated by the
U.S. Department of Housing and Urban Development (HUD). The Code regulates
the manufactured home design and construction, strength and durability, during
transportability, fire resistance and energy efficiency.
The “HUD Code” is authorized under the National Manufactured Housing
Construction and Safety Standards Act of 1974, Title VI, Public Law 93-383 (42
U.S.C. 5401). After June 15, 1976, all buyers of manufactured homes benefit from
the HUD regulations. Manufacturers who build manufactured homes for sale in the
United States must comply with HUD-mandated design and construction standards.
This implementation of HUD Code provides consumer protection through
enforcement of these standards for manufactured homes, investigation of consumer
complaints, and certifications, testing, in-plant inspections, and review of
manufacturers' designs and quality assurance programs. Either HUD or State
Administrative Agencies in 36 States may enforce these standards. The "HUD
Code" takes into account existing State and local laws but preempts those that
differ from the Federal standards.
147
Another milestone in the manufactured building code happens after Hurricane
Andrew’s devastating impact, new higher standards have been adopted in the
construction of manufactured housing since 1994. Even though these new homes
are considered "safer" than older models, hurricane winds and related dangers
continue to threaten all manufactured/mobile home residents.
For this reason, our Project investigates manufactured homes together with typical
single family houses and to develop loss prediction models for manufactured
homes for Florida insurance companies.
5.2 Sources of Information
5.2.1 Florida Hurricane Catastrophe Fund Exposure Database
Again, from the analyst from NIST, we obtained the statistical distribution for
manufactured homes in the FHCF database. (Note: the mobile home mentioned
in the figures refer the same building category as manufactured home)
Ratio of zip0 - 0.0410.041 - 0.1210.121 - 0.2250.225 - 0.4040.404 - 0.804
Ratio of Manufactured Housing in Zipcode
Figure 71. Ratio of Manufactured Housing
148
Ratio of zip0 - 0.0320.032 - 0.090.09 - 0.1680.168 - 0.2890.289 - 0.725
Ratio of Mobile Home - Fully Tied DownPre 7/13/1994
in Zipcode
Figure 72. Ratio of Manufactured Home – Fully Tied Down Pre 7/13/1994.
Ratio of zip0 - 0.0170.017 - 0.0470.047 - 0.0910.091 - 0.1670.167 - 0.341
Ratio of Mobile Home - Fully Tied DownPost 7/13/1994
in Zipcode
Figure 73. Ratio of Manufactured Home – Fully Tied Down Post 7/13/1994
149
Ratio of zip0 - 0.0020.002 - 0.0050.005 - 0.010.01 - 0.0180.018 - 0.036
Ratio of Mobile Home - Partially Tied Downin Zipcode
Figure 74. Ratio of Manufactured Home – Partially Tied Down
Ratio of zip0 - 00 - 0.0010.001 - 0.0020.002 - 0.0040.004 - 0.007
Ratio of Mobile Home - Not Tied Downin Zipcode
Figure 75. Ratio of Manufactured Home – not Tied Down
150
Ratio of zip0 - 0.0010.001 - 0.0040.004 - 0.0080.008 - 0.0170.017 - 0.046
Ratio of Mobile Home - Unknownin Zipcode
Figure 76. Ratio of Manufactured Home – Unknown
Upon observation of Figure 70 – 76 we can reach the following visual conclusions:
1. From Figure 70 and 75, we can see that manufactured homes have wide
distribution. Generally speaking, there more manufactured homes in north
rather than south; inland than coastal area.
2. From Figure 71-73, the tie down information tells us that 1994 is a cut off
date in tie down requirement as mentioned in the previous section. But the
author didn’t observe sharp contrast of the tie down distribution prior and
post 1994.
The above figures can give us a conceptual knowledge of distribution of
manufactured homes. But, as the same case as the single family building category,
they can not provide detail structural information of manufactured homes.
151
5.2.2 HAZUS Manual
The HAZUS manual provides a rough list of the materials commonly used in
manufactured housing construction.
Table 56. Sample Manufactured Home Materials in HAZUS Manual
There is no information about siding of new manufactured homes in the manual.
The basic manufactured models used in the damage simulation are presented in the
following Table 57:
Table 57. Criteria of Defining Manufactured Homes in HAZUS Manual
A total of 28 damage state curves were presented in the manual. Combinations of
the possibilities listed in the Table 57 were tested for several surface roughness:
0.03m, 0.1m, 0.35m representing respectively a typical open terrain, a relatively
open terrain and a typical suburban terrain.
The manufactured building information in HAZUS manual provides no
comprehensive statistical survey results. Other sources of information should
therefore be considered to obtain a distribution of the different types of roof covers
and materials used for the siding. However, the building material types and the
criteria of defining manufactured home for vulnerability model can be used as
reference to us.
Roof membrane systems Metal skin attached directly to the roof trusses or roof sheathing attached to the roof trusses
Siding of older mobile homes Wood, vinyl or metal panels
Tiedowns Yes or No Construction types Pre-HUD, HUD
152
5.2.3 Tax Appraiser’s Databases
As detailed in Section 4.2.3, there are considerable large amount of information
contains in the tax appraiser’s databases in the 9 Florida Counties, namely: Brevard,
Pinellas, Hillsborough, Broward, Palm Beach, Escambia, Leon, Walton and
Monroe. The building characteristics extracted from the database include: roof
cover material, roof type, exterior wall material, area, year built. However, there is
a big shortcoming of the tax appraiser’s database. Most of the counties databases
don’t record whether the manufactured home is tied down or not. Compared to site-
built homes, manufactured homes are relatively lightweight. They have flat sides
and ends, and they are built on frames rather than foundations. Almost all
manufactured homes are elevated, situated on top of some sort of pier or foundation
system. Wind can get under the homes and lift them up. In addition, the wind
passing over the top of your manufactured home can create an uplift force. Tie
downs distribute the load to the foundation and provide the attachment to ground
anchor so as to resist wind over turning and sliding as imposed by the respective
design code.
There are different types of tie downs in response to different soil conditions
including concrete slab. Auger anchors have been designed for both hard soil and
soft soil. Rock anchors or drive anchors allow attachment to a rock or coral base.
This type of anchor is also pinned to the ground with crossing steel stakes. a
concrete anchor should be installed before pouring a concrete base.
153
Figure 77. Most Common Type of Manufactured Tie Downs
The tie down can be an important topic of future manufactured home research.
5. 3 Information Gathered in the Survey
The five structural characteristics information is extracted from the tax appraiser’s
database. The statistics are also calculated. They include the statistics for roof
cover material, roof type, exterior wall, area and year built. Since almost all
manufactured homes are single-story building, the number of story is not
investigated.
Compared with single family residential buildings, the manufactured buildings
have the following characteristics:
1. Unlike most single family homes have shingle or tile roof cover,
manufactured homes have large portion of metal or aluminum as roof cover
154
materials. This distribution agrees with HAZUS manual results listed in
Table 56.
2. Most exterior wall material for manufactured homes are wood siding or
vinyl aluminum siding, while single family homes have large variety of
different type of exterior materials. Masonry walls (concrete block) are less
likely to be used in manufactured home construction. However, some
county database has concrete block exterior wall materials for manufactured
homes. This could probably rise from the building evaluation officials
mistaking veneer materials as exterior wall materials. In the following
section, the results for each county’s manufactured distribution are listed.
Because it is no way for us to tell what exterior wall materials for those data,
those records with concrete block materials can be simply taken out. But at
this stage, all exterior wall materials are still listed.
5.4 Counties Statistics
The structural characteristics distribution for manufactured home is listed county
by county.
155
5.4.1 Brevard County
Sheet Metal38%
Shingle ASB1%
Shingle ASP47%
Enamel Metal9%
B.U/T.&G /Membrane5%
Figure 78. Brevard County Manufactured Home Roof Material Distribution
Flat Shed28%
Gable71%
Wood TR.Wood Deck1%
Figure 79. Brevard County Manufactured Home Roof Type Distribution
156
VINYL/ALUMINIUM79%
STYRO FOAM STUCCO1%
EXT.HDBD Panel1%
Sheet Metal1%
EXT.PLYWOOD1%
WOOD FRM. ND .SH1%
WOOD FRM.STUCCO1%
WOOD SHEATING10%
Enamel Steel5%
Figure 80. Brevard County Manufactured Home Exterior Wall Material
Distribution
0%
1%
2%
3%
4%
5%
6%
1940
1950
1954
1956
1958
1960
1962
1964
1966
1968
1970
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
year built
per
cen
tile
Figure 81. Brevard County Manufacture Homes Year Built Distribution
157
0~500
500~1000
1000~1500
1500~2000
2000~27000%
5%
10%
15%
20%
25%
30%
35%
40%
45%
0~500 500~1000 1000~1500 1500~2000 2000~2700
Area Range (square feet)
Per
cen
tag
e
Figure 82. Brevard County Manufactured Home Area Range Distribution
5.4.2 Pinellas County
COMPOS-SHGLE2%
MH ROOF COVER54%
MH ROOF OVER44%
Figure 83. Pinellas County Manufactured Home Roof Cover Material Distribution
158
FLAT-SHED83%
GABLE-HIP17%
Figure 84. Pinellas County Manufactured Home Roof Type Distribution
FRAME-METAL65%
FRAME-SIDING15%
FRAME/RECLAD ALUM/VINYL
20%
Figure 85. Pinellas County Manufactured Home Exterior Wall Material
Distribution
159
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
0- 499 500- 999 1000- 1499 1500- 1999 2000- 2499 2500- 2999 3000- 3499 4000- 4499Ar ea Range( squar e f eet )
per
cen
tag
e
Figure 86. Pinellas County Manufactured Home Area Range Distribution
0%
1%
2%
3%
4%
5%
6%
7%
8%
9%
019
4819
5119
5319
5519
5719
5919
6119
6319
6519
6719
6919
7119
7319
7519
7719
7919
8119
8319
8519
8719
8919
9119
9319
9519
9719
99
lack of record
Figure 87. Pinellas County Manufactured Home Year Built Distribution
160
5.4.3 Hillsborough County
Asphalt/Comp. Shingl48%
Minimum9%
Rolled Composition4%
Blt.up Tar & Gravel13%
Concrete Tile1%
Clay or Bermuda Tile1%
Modular Metal24%
Figure 88. Hillsborough County Manufactured Home Roof Cover Material
Distribution
161
Gable or Hip54%
Wood Truss12%
Rigid Frame/Barjoist7%
Steel Frame or Truss7%
Flat19%
Reinforced Concrete1%
Figure 89. Hillsborough County Manufactured Home Roof Type Distribution
0%
1%
2%
3%
4%
5%
6%
1878
1902
1905
1910
1913
1915
1917
1919
1921
1923
1925
1927
1929
1931
1933
1935
1937
1939
1941
1943
1945
1947
1949
1951
1953
1955
1957
Figure 90. Hillsborough County Manufactured Home Year Built Distribution
162
5.4.4 Broward County
Database can not be processed for roof material, roof cover and exterior wall
materials and area for Broward County.
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
1933 1950 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000Year bui l t
Per
cen
tag
e
Figure 91. Broward County Manufactured Home Year Built Distribution
5.4.5 Palm Beach County
163
FLAT 47%
GABLE/HIP51%
STL FRM OR TRUS1%
WOOD TRUSS1%
Figure 92. Palm Beach County Manufactured Home Roof Type Distribution
VINLY/ALUMINUM SIDING
81%
WOOD SHT/PLY2%
WOOD SIDING10%
MINIMUM1%
MOD MET1%
ABOVE AV5%
Figure 93. Palm Beach County Manufactured Home Exterior Wall Material
Distribution
164
0%
2%
4%
6%
8%
10%
12%
14%
1938
1950
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
year built
per
cen
tag
e
Figure 94. Palm Beach County Manufactured Home Year Built Distribution
5.4.6 Monroe County
Asphalt Shingle20%
Metal44%
Min/Paint Conc33%
Tar & Gravel1%
Rolled2%
Figure 95. Monroe County Manufactured Home Roof Cover Material Distribution
165
Flat or shed73%
Gable & Hip27%
Figure 96. Monroe County Manufactured Home Roof Type Distribution
Metal/Alum83%
Vinyl siding13%
C.B.S.1%
WD FRAME3%
Figure 97. Monroe County Manufactured Home Exterior Wall Distribution
166
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
0~500 500~1000 1000~1500 1500~2000 2000~2500 2500~3000 3000~3500
Area Range (square feet)
per
cen
tag
e
Figure 98. Monroe County Manufactured Home Area Range Distribution
5.4.7 Escambia, Leon and Walton
These three northern counties’ databases can not be properly process for mobile
home. For Walton County, the database doesn’t keep separate record for mobile
home. The database for 2109 manufactured homes in Escambia is not accessible for
the author. Leon County has 6398 manufactured homes. But the detail distribution
has also been collected together with single family homes and residential buildings.
The joint distribution shown is also shown in Chapter 4 Leon County section.
Because of lack detail information on Manufactured homes in Northern Region
counties, the most common types are not defined in that Region.
167
5.5. Most Common Manufactured Homes Types in Florida
The similar procedures have been used to process the manufactured home types,
namely 1) find structural statistical information from the database 2) extract
statistics for each possible manufactured types from the tax appraiser’s database for
each county and come up with the most common structures types 3) extrapolate the
results for each county to 4 regions 4) estimate the error.
The following Table 58-60 presents the final results for manufactured homes.
Compared with single family homes, manufactured homes have less variety of
structural types. Since the roof materials recorded in the database lack of detailed
description or simply recorded as “mobile home roof cover”, the roof cover
materials are not included in the model.
168
Table 58. Probability of Occurrence of Manufactured Home Types for 3 Regions.
Structural
Type definition
Central Region Southern Region
Type No.
Roof type, exterior wall
p Area (ft2)
STDV(ft2) p Area (ft2)
STDV(ft2)
M1 Gable/hip,
wood 5% 1200 430 7% 2700 1209
M2 Gable/hip,
metal/vinyl/aluminum
35% 1343 632 32% 2571 890
M3 Flat/shed,
wood 1% 890 430 6% 1903 782
M4 Flat/shed,
metal/vinyl/aluminum
34% 910 540 29% 1876 896
M5 Reinforced concrete,
wood 8% 1690 703 8% 2468 673
Total Coverage
83% 82%
Unknown type
17% 18%
Note: M1,M2..M5 stands for Manufactured homes Type 1, 2..5. The descriptions are in the second column.
Table 59. Probability of Occurrence of Manufactured Home Types in Key Regions.
Type Roof type Exterior Wall Percentage Area (ft2) STDV (ft2) M7 Gable/Hip Metal/Vinyl/Aluminum 26% 1670 980 M8 Flat/shed Metal/Vinyl/Aluminum 51% 1398 1023 77%
169
Table 60. p̂ Values and Error for Each Type for 3 Regions
Central Region Southeast Region
p̂ SE p̂
99%cnfdn. Intvl. p̂
SE p̂
99%cnfdn. Intvl
Type 1 5% 5% 0%-20% 7% 6% 0%-25% Type 2 35% 20% 0%-90% 32% 18% 0%-96% Type 3 1% 1% 0%-4% 6% 3.5% 0%-18% Type 4 34% 21% 0%-90% 29% 15% 0%-74% Type 5 8% 7% 0%-29% 8% 3% 0%-17%
5.6 Result Analysis
We can see in the Table 58, Type M1, M3 and M5 are with wood exterior wall,
they count for relatively small proportion of building stock. When we combine the
statistics for M2 and M4 with metal or Vinyl aluminum siding, we see similarity
with M7 and M8 in the Key region. The above observation shows that unlike site-
built homes, the distribution of manufactured homes structural types does not
depend on geographical locations largely. Due to this reason, we can either
combine all the manufactured homes from all counties as a whole database to find
the most common structural type, or we may use the common types found for
Central Region to extrapolate those unknown region such as Northern Region.
The year built distribution is not counted as one of the criteria to define structural
type. Actually, since there are two clear cut-off date in the manufactured home
construction regulation: the year of 1974 when the HUD code was newly enforced
and the year of 1994 when the code was reinforced after Hurricane Andrew, the
year built can be a very good indicator of manufactured homes’ construction
quality and wind resistance, thus should be incorporated into manufactured home
modeling in the future research. The HAZUS manual has included the 1994 as one
of the criteria.
170
Based on the above analysis, the future research on manufactured building can be
furthered in several paths:
1) Obtained tie down information from possible sources, assign different resistance
capacity to various anchor types in the vulnerability simulation study.
2) Include year built information in defining the most common structural types. For
manufacture buildings built prior 1974, poor resistance is imposed. Those built
after 1994 should be assign higher capacity than those built in between 1974 to
1994.
3) Combined available manufacture database to define the most common structural
types for the whole Florida State.
4) Interview with Florida manufacture home main provider, to get the statistics
from them the most popular structural types in different area in the Florida State.
171
Chapter 6. Cost Estimation
6.1 Introduction
In Chapter 4, the author showed how the Florida Building population of each
structural type is classified and how the probability of occurrence. Each structural
type defined in the statistical survey will be modeled with the Monte Carlo
simulation. The output of the Monte Carlo simulation is a damage matrix that for
each interval of wind speed gives the probability of occurrence of all the possible
damage states for that particular type of structure. This chapter shows how these
probabilities are used in the damage cost estimation.
6.2 Average Annual Damage
As mentioned before, each of the structural types defined in the statistical survey is
being modeled in a Monte Carlo simulation. The output of the Monte Carlo
simulation is a damage matrix that for each interval of wind speed gives the
probability of occurrence of all the possible damage states for that particular type of
structure. If damage is expressed as a percentage of replacement cost of the home,
the resulting damage, in a given area, for a certain type of home m subject to a
certain wind speed vj becomes:
172
Damage type m (vj)= damage_ state i
∑ P(damage_statei|vj) x c(damagei) (1)
Where P(damage_statei|vj) is the probability of occurrence of each damage state
and c(damagei) is the associated cost percentage for that damage state. In modeling
the repair cost, the procedure needs to incorporate the fact that, the combined repair
cost of components cannot exceed the replacement cost of the facility. In practice,
the combined repair costs taper off to reach the replacement cost. Moreover, if the
repair cost of the combined structure exceeds 50% to 60% of the replacement value
of the building; it is considered economical to demolish the building. For this case
the cost of demolishing and removal of debris must be used in the estimates.
It was stated at the end of Chapter 3 that our 5-mode model is decomposed into 217
combined damage states. It is not reasonable to expect that a distinct cost can be
assigned to each of these states. Rather, the many combinations will be associated
with a handful of classes of damage. For example, 128 states, say, may all lead to
20% of replacement cost, 56 states may lead to 40 % replacement cost, and so forth.
The simplification inherent in this observation is to be incorporated into the
estimation procedure, and will require the input of experienced insurance adjusters.
The damages are then added over all the possible wind speeds to yield the probable
annual damage for type m, as follows:
Annual_Damage type m= ∑jwindspeed Damage type m (vj)*P(vj-∆v/2<vj<vj+ ∆v/2) (2)
Where P(vj-v/2<vj<vj+ v/2) is the annual probability of occurrence of wind
speed vj in the specified interval (Note: these probabilities are estimated with a
wind model described in a companion paper [3]). The process is repeated for each
building type in the area under consideration, and the average annual damage for a
generic home in that area will be:
173
Average_Annual_Damage = ∑typem
Annual_Damage type m* P(typem) (3)
Where P(typem) is the probability of occurrence of the different type of buildings
as listed in Tables 52 and 53.
The total estimated expected damage to buildings for a particular zone is the
damage calculated by using Eq. 3 times the total number n of houses in the zone.
Multiplication of this latter result by the average value of a home in that area yields
the monetary damage. Alternatively, if dealing with a portfolio where the values of
each house in the portfolio is known, the average annual damage (Eq. 3) for one
house can be multiplied by the sum of the values of all the houses insured in that
area. The process is repeated for each zone, and the results for each zone are added
to obtain the estimated expected hurricane-induced annual damage to buildings for
the entire state.
6.3 Numerical Example
Suppose there is a certain area with 1000 buildings of 3 major structural types, with
the distribution of 30% type A, 40% of type B, and 30% of type C. If the average
house value is $100,000, then the total value of houses in this area will be $ 100
million. And for each structural type, the Monte Carlo simulation engine is run and
generates the damage matrix shown in the Table 61. For illustration purposes, we
assume there are only 3 damage states (DS), DS1, DS2 and DS3. This damage
states are preventatives of any combined damage like a combination of moderate
opening damage, heavy roof cover damage, moderate sheathing damage, and light
roof-to-wall connection damage.
174
Table 61. Damage Matrix for Three Structural Types
v (m/s)
DS
50 60 70
DS 1 70% 30% 10%
DS 2 20% 50% 40%
DS 3 10% 20% 50%
v (m/s)
DS
50 60 70
DS 1 80% 50% 10%
DS 2 10% 30% 20%
DS 3 10% 20% 70%
v (m/s)
DS
50 60 70
DS 1 40% 20% 0%
DS 2 40.% 20% 10%
DS 3 20% 60% 90%
Damage Matrix Type A Damage Matrix Type B Damage Matrix Type C
For the sake of simplicity we also assume only 3 possible wind speeds with the
probability density distribution shown in Table 62.
Table 62. Sample Wind Field Model Data
V(m/s) 50 60 70 Probability density 60% 30% 10%
The damage cost of each of the three damage states A,B,C is expressed as a
percentage of replacement cost of homes. Assume the cost of three damage states
for 3 Types is listed in the following Table:
Table 63. Replacement Cost of Three Damage States
Type A Type B Type C DS1 10% 20% 10% DS2 20% 30% 30% DS3 60% 70% 50%
Step 1: Applying equation (1) we have
For Type A:
175
Damage (50m/s, type A) = P (DS1/50m/s) x C(DS1) + P(DS2/50m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.7 x0.1+0.2x0.2+0.1x0.6= 0.17
Damage (60m/s, type A) = P (DS1/60m/s) x C(DS1) + P(DS2/50m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.3 x0.1+0.5x0.2+0.2x0.6= 0.25
Damage (70m/s, type A) = P (DS1/70m/s) x C(DS1) + P(DS2/70m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.1 x0.1+0.4x0.2+0.5x0.6= 0.39
For Type B:
Damage (50m/s, type B) = P (DS1/50m/s) x C(DS1) + P(DS2/50m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.8 x0.2+0.1x0.3+0.1x0.7= 0.26
Damage (60m/s, type B) = P (DS1/60m/s) x C(DS1) + P(DS2/60m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.5 x0.2+0.3x0.3+0.2x0.7= 0.33
Damage (70m/s, type B) = P (DS1/70m/s) x C(DS1) + P(DS2/70m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.1 x0.2+0.2x0.3+0.7x0.7= 0.57
For Type C
Damage (50m/s, type C) = P (DS1/50m/s) x C(DS1) + P(DS2/50m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.4 x0.1+0.4x0.3+0.2x0.5= 0.26
Damage (60m/s, type C) = P (DS1/60m/s) x C(DS1) + P(DS2/60m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.2 x0.1+0.2x0.3+0.6x0.5= 0.38
Damage (70m/s, type C) = P (DS1/70m/s) x C(DS1) + P(DS2/70m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0 x0.1+0.1x0.3+0.9x0.5= 0.48
Step 2. Applying the equation (2), we get
176
Annual _Damage type A = Damage type A(50m/s) x P(50m/s) + Damage type A
(60m/s) x P(60m/s) + Damage type A(70m/s) x P(70m/s) = 0.17x0.4+0.25x0.5+
0.39x0.1=0.232
Annual _Damage type B = Damage type B(50m/s) x P(50m/s) + Damage type B
(60m/s) x P(60m/s) + Damage type B(70m/s) x P(70m/s) = 0.26x0.4+0.33x0.5+
0.57x0.1=0.326
Annual _Damage type C = Damage type C(50m/s) x P(50m/s) + Damage type C
(60m/s) x P(60m/s) + Damage type C(70m/s) x P(70m/s) = 0.26x0.4+0.38x0.5+
0.48x0.1=0.342
Step 3: Applying equation (3) we have
Average_Annual_Damage = Annual_Damage type A x P (type A) +
Annual_Damage type B x P( type B) + Annual_Damage type C x P (type C) =
0.232x 0.3 + 0.326x 0.4 + 0.342x 0.3 = 0.3026
Step 4: The total estimated expected damage to buildings in a certain area is the
damage calculated by using Eq. 3 times the total replacement cost of the building.
For example, for the 1000 buildings with a replacement average about $ 100million,
if assume the replacement cost is approximately equals the total value, then
Annual monetary damage prediction = Average_Annual_Damage x Total
replacement cost = 0.3026 x $ 100million = $ 30.26 million
In other words, given the conditions and assumptions, the annual Hurricane damage
prediction in this certain area with 1000 houses will be $30.26 million dollars.
Since the particular damage is expressed as the percentage of the replacement value,
the estimation model is flexible and also applicable in other cases prediction for
177
total loss for a certain area. For example, intermediate results for each individual
type can be used to calculate the damage for different structural type. Given the
special wind field data of a certain wind storm, the model could also be applied to
estimate for event-based wind storm or other user defined scenarios and historical
events. It can be also used to estimate the damage for a particular event.
For example, we assume for a particular hurricane, say Hurricane Liang. We know
that the peak 3-second gust wind speed 10 m above the ground is 70 m/s. and we
consider the same area in the previous example.
We use the results from step 1 in previous example,
Damage (70m/s, type A) = P (DS1/70m/s) x C(DS1) + P(DS2/70m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.1x 0.1+0.4x0.2+0.5x0.6= 0.39
Damage (70m/s, type B) = P (DS1/70m/s) x C(DS1) + P(DS2/70m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0.1 x0.2+0.2x0.3+0.7x0.7= 0.57
Damage (70m/s, type C) = P (DS1/70m/s) x C(DS1) + P(DS2/70m/s) x C(DS2) +
P(DS3/50m/s) x C(DS3) = 0 x0.1+0.1x0.3+0.9x0.5= 0.48
Since the wind speed has already be disclosed, we can omit step 2 and jump to
step3.
Hurricane Liang_Damage = Hurricane Liang_Damage type A x P (type A) +
Hurricane Liang_Damage type B x P( type B) + Hurricane Liang_Damage type C x
P (type C) = 0.39x 0.3 + 0.57x 0.4 + 0.48x 0.3 = 0.489
And the actual cost will be easily calculated by multiplying Hurricane
Liang_Damage ratio to the total replacement cost of the area, which is $100 million,
the result will be 0.489 x $ 100 million = $ 48.9 million.
178
Chapter 7. Conclusions and Recommendations
7.1 Summary and Conclusion
This report presents a component approach hurricane damage prediction model.
The component approach explicitly accounts for both the resistance capacity of the
various building components and the load effects produced by wind events to
predict damage at various wind speeds. In this approach the resistance capacity of
a building can be broken down into the resistance capacity of its components and
the connections between them. Damage to the structure occurs when the load
effects from wind or flying debris are greater than the component’s capacity to
resist them. Once the strength capacities, load demands, and load path(s) are
identified and modeled, the vulnerability of a structure at various wind speeds can
be estimated based on past laboratory tests and building material capacity obtained
from manufacturers. Estimates are affected by uncertainties regarding on one hand
the behavior and strength of the various components and, on the other, the load
effects produced by hurricane winds.
The probabilistic damage state modeling incorporates the building components
vulnerability into the damage prediction model. It is assured that no type of damage
is counted more than once, no type of possible damage is omitted from the
calculations, and interactions between various types of damage are accounted for.
179
The costs are calculated by correctly accounting for the dependence between
various damage modes (e.g., window breakage and roof uplift). The damage is
appropriately modeled as a stepwise process, such as damage to openings gives
sudden rise to increases of internal pressures, and sudden collapse of the roof
results in immediate damage to walls.
The framework developed in this report is illustrated for the case of five basic
damage modes, i.e.
(1) Breakage of openings (O);
(2) Loss of shingles (T);
(3) Loss of roof or gable end sheathing (S);
(4) Roof to wall connection damage (C); and
(5) Masonry wall damage (W).
The report also introduces the use of damage matrices for the estimation of
expected damage due to a windstorm event.
Another important part of this report presented the results of a statistical survey and
analysis of the building population in Florida. Based on the results of the survey,
the most common structural types are defined for each of four regions and their
probabilities of occurrence are estimated. The goal is to predefine the structural
make up of the building population and its corresponding wind vulnerability in any
given area of the State. Thanks to these results, even in the case of portfolio files
with no or incomplete information regarding the structural strength of the insured
properties, insured losses due to hurricane winds will be predicted based on
180
location. The method can be used to compute expected annual losses as illustrated
in the report, and expected losses induced by a specified hurricane event.
Finally, the procedures of cost estimation are discussed and a numerical example is
included for illustration purposes. The cost estimation explicitly shows the logical
relationships between wind field modeling, vulnerability modeling and building
statistics study. It illustrates how to incorporate the results of probabilistic damage
states modeling (damage matrices), building statistics information, wind field data
and damage matrices filled by Monte Carlo simulation to calculate the annual
hurricane loss for a certain return period.
Instead of predict the annual hurricane loss, the cost calculation is also flexible to
estimate the loss for a historical hurricane event or a certain residential building
stock.
7.2 Uncertainties
The uncertainties existing in this prediction model lies in each of its four
components: wind field modeling, vulnerability modeling, building statistics and
cost calculation. Since wind field modeling is beyond the scope of this report, its
uncertainties will not be discussed here.
Main types of uncertainties in the probabilistic component vulnerability modeling
part include:
1) Selection of the building components and their connections, the appropriateness
of current five components will be improved with the development of more robust
computer-aid probabilistic and simulation and a more detail component breakdown
can be realized.
181
2) Limitation of properties or parameter inputs into the Monte Carlo simulations
and the translation from wind speed to resultant force on roof. Current the inputs
for Monte Carlo simulation come from our thorough literature research. Some
properties of building component lack study and numerical information of their
capacity. Under these circumstances, best knowledge has been used including
engineering judgments.
Main sources of uncertainties lies for the building inventory statistics study include
1) lack of completeness and errors in the county tax appraiser’s databases. Despite
of huge effort of research group, only 9 Florida counties tax appraisers’ databases
have been obtained and processed. Needless to say, the more accurate the building
statistics used in the model reflects actual building inventory, the less error in the
loss estimation for the target area.
2) Error in the process of extrapolation and interpretation of the data. The fact that
not all county databases were available for analysis made us to use extrapolation to
find the building statistics in those areas that buildings information is not available.
This type of error has been calculated in Chapter 4.
3) System error in the tax appraisers’ database and error generating when process
the database. Due to the limit of structural knowledge of property tax officials,
some structural information recorded in the tax appraisers’ database are not 100%
correct. For example, some wood exterior walls with brick decoration are recorded
as brick exterior wall. And most counties’ database has blank or corrupted records.
Although author has resorted computer-aided database language to process the
database, due to the large records of data and limited time, the error in the process
of database is almost inevitable.
182
Another source of uncertainty in the cost estimate is the attribution of a cost
percentage to the different damage combination. The uncertainties of cost
estimation can be expressed as a whole percentage of the total cost or can be
expressed in the intermediate steps.
7.3 Research Underway and Recommendations
The research currently underway includes wind field modeling, Monte Carlo
simulations for more structural types, quantifying the uncertainties in estimation for
probability matrices, probabilistic modeling and Monte Carlo simulations.
The future research could be conducted on the basis of the framework of the current
component approach modeling. Since the lack of statistical information of the
building stock is one of the main uncertainties in the exposure component of this
model, the future refinement of the model could benefit from obtaining more
counties’ databases with various geographical locations. The Florida counties
should be more supportive in providing more building stock information. The
future research path for manufactured homes is proposed in Chapter 5. The
probabilistic and Monte Carlo simulation procedures are all highly logical and
programmable. A computer flat form will greatly facilitate the procedure of
revising damage states. So a close collaboration between engineering group and
computer group in the program the modeling is an important task in the future.
The estimation of damage states will need more coordination with actuaries and
financial experts to compute the economic losses for a given insurance exposure,
taking into account specific coverage terms like limits and deductibles. More
laboratory test data for different building components are expected to increase the
accuracy of the Monte Carlo Simulations. And the selection of new building
components can be considered for the typical building types in the target area.
183
Besides, detail and reliable insurance claim data could be used as a validation of the
resulting cost estimation.
184
References
1. Cope, A., Gurley, K., Pinelli, J.-P., and Hamid, S., “A Simulation Model For
Wind Damage Predictions In Florida,” Proceedings, 11th ICWE, Lubbock,
Texas, June 2003.
2. Simiu, E., Pinelli, J-P., Subramanian, C., Zhang, L., Cope, A., Gurley, K.,
Filliben, J., and Hamid, S., “Hurricane Damage Prediction Model for
Residential Structures,” Proceedings, 11thICWE, Lubbock, Texas, June 2003
3. Powell,M., Soukup, G., Morisseau-Leroy,N., Landsea, C., Cocke,S., Axe, L.,
and Gulati, S., “State of Florida Hurricane Loss Projection Model: Atmospheric
Science Component,” 11th ICWE, Lubbock, Texas, June 2003.
4. Khanduri A.C., Morrow G.C., “Vulnerability of buildings to windstorms and
insurance loss estimation,” Journal of Wind Engineering and Industrial
Aerodynamics 91 (2003) 455-467
5. Paul S. Levy, Stanley Lemeshow, Sampling of Population: Method and
Applications, book Publisher: Wiley-Interscience; 3rd edition April 1999 ISBN:
0471155756
6. American Association for Wind Engineering, “Wind Engineering: New
Opportunities to Reduce Wind Losses and Improve the Quality of Life in the
USA”, August 1997.
185
7. Federal Emergency Management Agency, “ Building Performance: Hurricane
Andrew in Florida,” report FIA-22, 1993.
8. Huang, Z., Rosowsky, D. V., and Sparks, P. R., (2001), “Long-term hurricane
risk assessment and expected damage to residential structures,” Reliability
Engineering and System Safety, 74, 239-249.
9. Landsea, C. W., Pielke, Jr. R. A., Mestas-Nunez, A. M. and Knaff, J.A. (1999),
“Atlantic Basin Hurricanes: Indices of Climatic Changes”, Climatic Change, 42,
89-129.
10. Minor, J. E. and P. J. Schneider (2001), “Hurricane Loss Estimation – The
HAZUS Preview Model,” America’s Conference on Wind Engineering,
Clemson, SC June 4-6 2001.
11. Mitsuta, Y., T. Fujii and I. Nagashima (1996), “A Predicting Method of
Typhoon Wind Damages”, Probablistic Mechanics and Structural Reliability:
Proceedings of the 7th Specialty Conference, 970-973.
12. Pielke, Jr., R. A., and Landsea, C. W., "Normalized Atlantic Hurricane Damage,
1925-1995," Weather Forecasting 13, 621-631, 1998.
13. Bhinderwala. S. Insurance Loss analysis of Single Family Dwellings Damaged
in Hurricane Andrew, Master thesis, 1995, May
14. Amirkhanian, Serij; Sparks, Peter & Watford, Sandy (1994)“Statistical analysis
of wind damage to single family dwellings due to Hurricane Hugo”Proceedings
of the Structures Congress ’94.
15. Cook R.A., editor (1993)Hurricanes of 1992: Lessons Learned and Implications
for the Future, Cook, Ronald A. (Editor), ASCE
16. FEMA (1992)“Building Performance: Hurricane Andrew in Florida
Observations, Recommendations, and Technical Guidance”, FEMA, Federal
Insurance Administration Coastal Archives call # G 409.52 F4 B8
186
17. Kareem, Ahsan (1985) “Structural Performance and Wind Speed-Damage
Correlation in Hurricane Alicia,” Journal of Structural Engineering, vol 111, n
12, 2596-2610.
18. Kareem, Ahsan (1986) “Performance of Cladding in Hurricane Alicia”, Journal
of Structural Engineering, vol 112, n 12, 2679-2693.
19. NAHB Research Center (1993). “Assessment of Damage to Single-Family
Homes Caused by Hurricanes Andrew and Iniki,” Prepared for U.S.
Department of Housing and Urban Development, Office of Policy
Development and Research
20. NAHB Research Center (1996).“Assessment of Damage to Homes caused by
Hurricane Opal”, Prepared for Florida State Home Builders Association.
21. NAHB Research Center (1999).“Reliability of Conventional Residential
Construction: An assessment of Roof Component Performance in Hurricane
Andrew and Typical Wind Regions of the United States”, Prepared for U.S.
Department of Housing and Urban Development and National Association of
Home Builders
22. Sparks, Peter R (1991)“Damages and lessons learned from hurricane Hugo”,
Proceedings of the 23rd Joint Meeting of the US-Japan Cooperative Program
in Natural Resources Panel on Wind and Seismic Effects.
23. Phang, M. K. (1999) “Wind Damage Investigation of Low Rise Buildings,”
Proceedings of the 1999 Structures Congress, sponsored by ASCE
24. Sill, B. L. and P. R. Sparks, editors (1990)Hurricane Hugo One Year Later,
Proceedings of the Symposium and Public Forum Sep 13-15 1990 ASCE
25. Boswell, M.R., R.E. Deyle, R.A. Smith, and E.J. Baker “Quantitative Method
for Estimating Probable Public Costs of Hurricanes,” Environmental
Management, v23 n3.
187
26. Holmes, J. (1996) “Vulnerability Curves for Buildings in Tropical Cyclone
Regions”, Probablistic Mechanics and Structural Reliability: Proceedings of the
7th Specialty Conference, 78-81.
27. Minor, J. E. and P. J. Schneider (2001)“Hurricane Loss Estimation – The
HAZUS Preview Model,” America’s Conference on Wind Engineering,
Clemson, SC June 4-6 2001.
28. Mitsuta, Y., T. Fujii and I. Nagashima (1996) “A Predicting Method of
Typhoon Wind Damages”, Probablistic Mechanics and Structural Reliability:
Proceedings of the 7th Specialty Conference, 970-973.
29. Sill, B.L. and R.T. Kozlowski (1997)“Analysis of Storm Damage Factors for
Low-Rise Structures”, Journal of Performance of Constructed Facilities, vol 11,
n 4, 168-176.
30. Baskaran, A. (1999)“Discussion on ‘Wind Uplift Model for Asphalt Shingles",
Journal of Architectural Engineering, vol 5, n 2, 67-69.
31. Cunningham, Jr., Ph.D., Thomas P. (1993) “Roof sheathing fastening schedules
for wind uplift”, APA Report T92-28.
32. Devlin, Paul A. (1996) “Wind Resistance of Roof Coverings,” Natural Hazard
Mitigation Insights, Insurance Institute for Property Loss Reduction, n 4 July
1996
33. Hosoya, N. Cermak, J.E., and Dodge, S.M. (1999)Area-averaged pressure
fluctuations on surfaces at roof corners and gable peaks, Wind Engineering in
the 21st Century,Larsen, Larose, and Livesey (eds), Bakkema, Rotterdam.
34. Mahendran, Mahen (1995) “Wind Resistant Low-rise Buildings in the Tropics,”
Journal of Performance of Constructed Facilities ASCE, Nov 1995.
35. Mizzell, Dana P. (1994) Wind Resistance of Sheathing for Residential Roofs,
Master’s Thesis, Clemson University Department of Civil Engineering
188
36. Peterka, J.A., J.E. Cermak, L.S. Cochran, B.C. Cochran, N. Hosoya, R.G.
Derickson, C. Harper, J. Jones and B. Metz (1997) “Wind Uplift Model for
Asphalt Shingles”, Journal of Architectural Engineering, vol 3, n 4, 147-155.
37. Reed, Thomas D., David V. Rosowsky and Scott D. Schiff (1996) “Structural
Analysis of Light-Framed Wood Roof Construction,” A Wind Load Test
Facility Report for Blue Sky, PBS-9606-02.
38. Rosowsky, David, Scott Schiff, Tim Reinhold, Peter Sparks and Ben Sill
(1998)“Performance of Light-Frame Wood Structures Under High Wind Loads:
Experimental and Analytical Program”, FPS Specialty Publication Wind
Performance and Safety of Wood Buildings (1998)
39. Rosowsky, David V. and Scott D. Schiff (1999)“Combined Loads on Sheathing
to Framing Fasteners in Wood Construction”, Journal of Architectural
Engineering, vol 5, n 2, 37-43.
40. Rosowsky, David V. and Timothy A. Reinhold (1999) “Rate-of-Load and
Duration-of-Load Effects for Wood Fasteners”, Journal of Structural
Engineering, vol 125, n 7, 719-724.
41. Rosowsky D., Cheng N., (July 1998)“Reliability of a Light Frame Roof
Systems subjected to Wind Uplift”, Clemson University, Prepared for NAHB
Research Center and the National Association of Home Builders
42. Canfield, Laurence R., Stanley H. Niu and Henry Liu (1991)“Uplift Resistance
of Various Rafter-Wall Connections,” Forest Products Journal, vol 41, n 7/8,
27-34.
43. Conner, H.W.; Gromala D.S. & Burgess D.W. (1987)“Roof connections in
house; Key to wind resistance”, Journal of structural engineering, 113(12),
2459-2474.
44. Reed, Rosowksy and Schiff (1997)“Uplift Capacity of Top Plate Connections”,
Journal of Architectual Engineering, vol 3, n 4,
189
45. Dawe, J.L. and G.G. Aridru (1993) “Prestressed concrete masonry walls
subjected to uniform out-of-plane loading”, Canadian Journal of Civil
Engineering, vol 20, 969-979.
46. Foliente G., Kasal B., Paevere P., Macindoe L., Banks R., Mike S., Seath C.,
Collins M. (2000).“Whole Structure Testing and Analysis of a Light Frame
Wood Building, Phase 1 – Test House Details and Preliminary Results”,
Prepared for NAHB Research Center.
47. Pinelli, J.-P., and O’Neill, S. (2000)“Effect of Tornadoes on Residential
Masonry Structures,” Wind and Structures Journal, Vol. 3, No 1, March 2000.
48. Stricklin Jr., David Lee (1996) Investigation of Light-Framed Wood Wall
Systems Under Wind Uplift Loads, Master’s Thesis, Clemson University
Department of Civil Engineering
49. Yancey, C.W., Cheok, G.S. Sadek, F. and Mohraz B., 1988"A summary of the
structural performance of single-family, wood-frame housing", NISTIR 6224,
National Institute of Standards and Technology, Sept. 1988
50. IBHS (2000)“Industry Perspective: Impact Resistance Standards,” Natural
Hazard Mitigation Insights, Institute for Business and Home Safety, n 12
February 2000
51. Simiu, E. and Cordes, M.E. 1980 Probabilistic Assessment of Tornado-Borne
Missile Speeds, NBS Interagency Report 80-2117; NBS; Washington, DC
(09/80)
52. Simiu, E. and Cordes, M. R. 1983"Tornado-borne Missile Probabilities,"
Journal of Structural Engineering; pp. 157-168 ASCE; New York, NY (01/83)
53. Cope, Anne D. (1997)Load Duration Effects on Peak Minimum Pressure
Coefficients, Master’s Thesis, Clemson University, Department of Civil
Engineering, August 1997.
190
54. Cope, Anne D. and Kurtis R. Gurley (2001)“Spatial Characteristics of Pressure
Coefficients on Low Rise Gable Roof Structures,” America’s Conference on
Wind Engineering, June 2001.
55. Ginger, J.D. and C.W. Letchford (1995) “Pressure Factors for Edge Regions on
Low Rise Building Roofs,” Journal of Wind Engineering and Industrial
Aerodynamics, vol 54/55, 337-344.
56. Gioffre, M., V. Gusella and M. Grigoriu (2000) “Simulation of Non-Gaussian
Field Applied to Wind Pressure Fluctuations”, Probabilistic Engineering
Mechanics, vol 15, 339-345.
57. Hajj, M.R., D.A. Jordan and H.W. Tieleman (1998) “Analysis of Atmospheric
Wind and Pressures on a Low-Rise Building”, Journal of Fluids and Structures,
vol 12, 537-547.
58. Hosoya, Noriaki, Jack E. Cermak and Jon A. Peterka (unk.)“Probabilistic
Analysis of Area-Averaged Pressures on Model TTU Building”, Location being
determined
59. Ho, T.C.E., A.G. Davenport and D. Surry (1995)“Characteristic Pressure
Distribution Shapes and Load Repetitions for the Wind Loading of Low
Building Roof Panels”, Journal of Wind Engineering and Industrial
Aerodynamics, vol 57, 261-279.
60. Jordan, D.A., M.R. Hajj, R.W. Miksad and H.W. Tieleman (1999)“Analysis of
the Velocity-Pressure Peak Relation for Wind Loads in Structures”, Wind
Engineering into the 21st Century, Proceedings of the Tenth International
Conference on Wind Engineering, Balkema, Rotterdam, 443-448.
61. Kasperski, Michael (1996) Design Wind Loads for Low-Rise Buildings: A
critical review of wind load specifications for industrial buildings”, Journal of
Wind Engineering and Industrial Aerodynamics, vol 61, 169-179.
191
62. Kumar, K. Suresh and T. Stathopoulos (1998)“Power Spectra of Wind
Pressures on Low Building Roofs”, Journal of Wind Engineering and Industrial
Aerodynamics, vol 74-46, 665-674.
63. Maruta, E., M. Kanda and J. Sato (1998)“Effects on Surface Roughness for
Wind Pressure on Glass and Cladding of Buildings”, Journal of Wind
Engineering and Industrial Aerodynamics, vol 74-76, 651-663.
64. Marwood, R. and C.J. Wood (1997)“Conical Vortex Movement and its Effect
on Roof Pressures”, Journal of Wind Engineering and Industrial Aerodynamics,
vol 69-71, 589-595.
65. Peterka, J.A., N. Hosoya, S. Dodge, L. Cochran and J.E. Cermak (1998)“Area
Average Peak Pressures in a GableRroof Vortex Region”, Journal of Wind
Engineering and Industrial Aerodynamics, vol 77-78, 205-215.
66. Pettit, C.L., N.P. Jones and R. Ghanem (1999)“Detection, Analysis and
Simulation of Roof-Corner Pressure Transients”, Wind Engineering into the
21st Century, Proceedings of the Tenth International Conference on Wind
Engineering, Balkema, Rotterdam, 1831-1838.
67. Reinhold, T.A. Y. Sun and H.W. Tieleman (unk.)“Boundary Layer Wind
Tunnel Simulation of Area Averaged Wind Loads”, Location being determined
68. Robertson, A.P. (1992) “The Wind-Induced Response of a Full-Scale Portal
Framed Building,” Journal of Wind Engineering and Industrial Aerodynamics,
vol 41-44, 1677-1688.
69. Uematsu, Yasushi and Nicholas Isyumov (1999)“Wind Pressures Acting on
Low-Rise Buildings”, Journal of Wind Engineering and Industrial
Aerodynamics, vol 82, 1-25.
70. Young, Michael A. (1997)Effect of Open Fields on Low Building Wind Loads
in a Suburban Environment, Master’s Thesis, University of Western Ontario,
Department of Civil Engineering, June 1997.
192
71. Beste, Frank and Jack E. Cermak (1997)“Correlation of Internal and Area-
Averaged External Wind Pressures on Low-Rise Buildings”, Journal of Wind
Engineering and Industrial Aerodynamics, vol 69-71, 557-566.
72. Ginger, J.D. and C.W. Letchford (1999)“Net Pressures on a Low-Rise Full-
Scale Building”, Journal of Wind Engineering and Industrial Aerodynamics,
vol 83, 239-250.
73. Harris, R.I. (1990)“The Propagation of Internal Pressures in Buildings”, Journal
of Wind Engineering and Industrial Aerodynamics, vol 34, 169-184.
74. Sharma, R.N. and P.J. Richards (1997)“The Effect of Roof Flexibility on
Internal Pressure Fluctuations”, Journal of Wind Engineering and Industrial
Aerodynamics, vol 72, 175-186.
75. Vickery, B.J. (1994) “Internal Pressures and Interactions with the Building
Envelope”, Journal of Wind Engineering and Industrial Aerodynamics, vol 53,
125-144.
76. Vickery, B.J. and P.N. Georgiou (1991) “A Simplified Approach to the
Determination of the Influence of Internal Pressures on the Dynamics of Large
Span Roofs”, Journal of Wind Engineering and Industrial Aerodynamics, vol
38, 357-369.
77. Vickery, B.J. (1986) “Gust-Factors for Internal-Pressures in Low Rise
Buildings”, Journal of Wind Engineering and Industrial Aerodynamics, vol 23,
259-271
78. R.D. Marshall, (1977).The Measurement of Wind Loads On A Full-Scale
Mobile Home, National Bureau of Standards, Washington, D.C., 1977
79. R. D. Marshall, (1993)Wind Load Provisions of the Manufactured Home
Construction and Safety Standards -- A Review and Recommendations for
Improvement, NIST IR 5189, National Inst. of Standards and Technology, 1993.
193
80. R.D. Marshall, (1994)Manufactured Homes -- Probability of Failure and The
Need for Better Windstorm Protection Through Improved Anchoring Systems,
NIST IR 5370, 1994
81. R.D. Marshall and F.Y. Yokel, (1995)Recommended Performance-Based
Criteria for the Design of Manufactured Home Foundation Systems to Resist
Wind and Seismic Loads, NISTIR 5664, NIST 1995
82. Pearson, John; Longinow, Anatol & Meinheit, Donald F. (1996)Wind
protection tie- downs for manufactured homes, Practice Periodical on Structural
Design and Construction, V1 n4, Nov 1996
