Introduction:

The Philippines is a typical pathway for typhoons. It experiences 20 or more

typhoons annually with varying winds speeds. Residential buildings are the most
commonly constructed structures here in the Philippines. It is frequently observed
that low rise residential buildings suffer largely during tropical cyclones. (Li, 2005).
The most vulnerable part of a residential building is its envelope. It includes the
roof, walls, doors and windows. Failure of one of these components may lead to the
decrease of the structural integrity of the building (Veron, 2012).
The common failures that roofing experience are pullout of the connections on roofto-wall, tearing of the roofing material, and pullout of the fasteners connecting the
roof to the purlin. (Bisa, 2013)
The millions of dollars spent in rebuilding low-rise wood structures in the aftermath
of past hurricanes has clearly identified the need for performance based design
(PBD) of light framed wood residential structures (LFWRS). ,(Shanmugam,

2011)
Light frame buildings consist of many elements including the roof, floor, walls, and
intercomponent connections. When LFBs are exposed to hurricane forces, roofs are
most susceptible to damage, followed by walls in particular those with openings,
and then foundations (Ayscue, 1996). Research by Cook (1991) estimated that roof
failures accounted for 80% of structural losses.
During high wind events, structural integrity of low-rise buildings are highly affecter
after roof sheathing panels are damaged. This is due to the progressive damage
caused by water intrusion, breach of the building envelope, and exposure to interior
pressure which ultimately lead to destruction of more structural members such as
the walls. Amini, M. (2010)
The annual economic loss induced by hurricane hits around $5 billiion (Pinelli et al,
2004), which has far outweighed the loss caused by earthquakes and other natural
hazards. Besides the complexity in the highly turbulent wind loads generated in the
lowest part of the atmospheric boundary layer, the limited knowledge on the
structural responses for those non-full engineered structures is another factor
weakening their capabilities to resist wind loads.
Both reliable damage predictions and efficient mitigation measures for residential
buildings need a better understanding of the structural responses, including system
responses and component responses, under realistic hurricane loads.
The wind and structure interaction is by nature a multidiscipline subject that
involves probability and statistics, meteorology, fluid mechanics of bluff bodies, and
structural dynamics (Holmes, 2001).

Wind Storm/ Cyclone:

-roof is the most vulnerable element of the house in high winds or cyclones
-sheet roofing, popular in recent times, is quite prone to flying off during high speed
winds. This may be fatal.
ROOF SYSTEMS:
The roof framing systems of non-engineered building structures (nebs) utilized in
the region, are primarily timber framed truss systems and infrequently metal truss
systems.
WALL:
External walls in the type of buildings studied are most often constructed of loadbearing concrete masonry, or a combination of reinforced concrete frames with
concrete masonry infill panels. In rural areas, masonry walls were typically found to
be un-reinforced, or minimally reinforced near door and window openings. Vertical
re-bars were found to be more common in buildings constructed in urban settings
where there is often more exposure to proper construction techniques.
Roofs
These observations revealed that in general, most of the wind-based failures in nebs
were initiated in the roofs. The most prevalent mode of failure was the loss of roof
sheeting. This was primarily initiated by the pull out of improperly embedded
roofing nails or screws, and less frequently the loss of nail heads. In no instance was
punching shear failure of the fastener heads through metal roofing was observed.
Roof breakdown due to direct failure of truss members appeared to be very
uncommon. Failure of truss connections, however, was found to be very common.
Walls
Wind-induced failures in masonry observed were as a result of excessive spans, and
as a result of the structural breakdown of other supporting systems such as the roof.
The figures below illustrate a few wind-induced masonry failure observed.
Existing Models:
The vulnerability assessment is determined using a database of fragility curves,
developed originally for the FEMAs HAZUS-MH program, and adapted for this use.
The analysis yields the top three recommended retrofits for each house as-is, and
its expected hurricane-induced economic losses compared against the predicted
loss if all the retrofits were conducted.
The use of performance-based design (PBD) is recognized in the engineering
community as the most rational means of assessing and reducing the risks of
engineered buildings subject to natural disasters (Ciampoli et al 2011).
The objective of PBD is to assess the adequacy of a structural system based on a
set of decision variable (DV). Each DV is typically a (quantitative) measure of the
structural performance that can be defined in terms of interest to the stakeholder.
The personalized risk and vulnerability assessment of ResRe requires an
engineering-based loss model to establish the relationship between wind speed and
expected monetary losses for a given residential structure type. This is publicly
available in the HAZUS-HM hurricane catastrophe model, a peer-reviewed, multi-

hazard catastrophe model distributed by Federal Emergency Management Agency

(FEMA) (Vickery et al 2006). A primary component of HAZUS-HM is a library of loss
functions that relate hurricane wind speed and expected building losses (normalized
to the value of the home) for various building types. The loss functions are
extensive, including 1,024 different configurations of residential buildings.
It is understood that HAZUS-HM loss functions represent generalized structures and
therefore have limited accuracy at house-level resolution. However, these curves
can be used to provide a rational basis for comparative analysis between windresistant building features. Their use in this application also enables meaningful risk
communication to homeowners through dollar figure loss estimations.

The FPHLM was initiated by the state of Florida in the early 2000s (Pinelli et al.
2008). It predicts exterior physical damage by engineering principles using a
probabilistic component approach and interior building damage by a model based
on engineering judgement (Pita el al. 2013). The existing types by roof shape and
primary construction material (Cope 2004).
The component based engineering approach is more scientific than econometric
models that regress over historical claim data on the following aspects
1. An engineering model is able to estimate building vulnerabilities for both a
building stock including a variety of construction types and a specific building

type. Whereas, the regressed vulnerability can only represent an aggregated

average building type, since the claim data do not differentiate building
features.
2. The accuracy of a regressed econometric model is affected by the bias and
incompleteness of insurance claim data. This is because insurance companies
usually sell coverage in preference to certain types of building, thus, claims
above the limits or below the deductibles might be truncated or not recorded.
3. An engineering model is flexible to incorporate the updates in either building
materials/construction practices or dynamic change in building inventory. In
comparison the historic claim data is a snapshot of partial damage caused to
a fixed building stock and thus may not well represent the future building
stock.
4. An engineering model is able to quantitatively evaluate the effectiveness of
mitigation measures to the overall building damage.
Numerical Assesment:
Numerical Assessment of Roof Panel Uplift Capacity under Wind Load
(Weixian He, 2010)
A nonlinear finite element model is developed in this study to incorporate the nail
withdrawal uncertainty in terms of maximum withdrawal force, initial stiffness,
proportional limit, and the displacement at maximum force of the nail withdrawal
behaviour. This model is used to investigate the statistical characteristics of the
panel uplift capacity.
Based on the analysis for roof panel subjecting to time-varying uniform wind
pressure, the conclusions are list below
1. The nonlinear static pushover (NSP) analysis and nonlinear incremental
dynamic analysis were adopted for assessing the roof panel uplift
capacity. It was observed that the use of the NSP analysis is adequate for
estimating the uplift capacity of the panel.
2. The consideration of statistical correlation of nail withdrawal behaviour for
the nails used to fastening the roof panel affects the mean or uplift
capacity about 5%, and it reduces the coefficient of variation (cov) of the
uplift capacity significantly as the degree of correlation between the nail
behaviour decreases. The uplift capacity of roof panel can be modelled
adequately as a lognormal value.
3. In general, the use of the simple tributary area approach underestimates
the mean of the uplift capacity by 10% 22% as compared to that
estimated using the NSP analysis. The difference is due to the fact that
the tributary area method does not consider the load sharing among nails
with different stiffness.
The main objective of the present study is to assess the statistical characteristics
and to assign probability distribution of the uplift capacity of the roof panel under
stochastic wind pressure and considering several factors: the uncertainty in nail
withdrawal behaviour, the spatial varying effects of wind load, and the human error
effects.

Based on the analysis results for roof panel subjecting to spatially and temporally
varying wind pressure, the conclusions are below
1. The statistics of the panel uplift capacity in terms of the (ultimate) total
reacting force are not sensitive to the spatially varying wind, but are
significantly influence by the adopted probabilistic model and correlation
of nail withdrawal behaviour.
2. The use of equivalent uniformly distributed wind loading, with the
pressure coefficient equal to the weighted average wind pressure
coefficient, provides sufficiently accurate estimates of the statistics of the
uplift capacity of panel. This approximate approach is therefore
recommended for assessing the panel uplift capacity as it simplifies the
analysis.
3. The spatial correlation coefficient of the instantaneous pressure
coefficients can be modelled using an exponential model with correlation
length within 1.5m to 3.0m
4. The ultimate total reacting force could be modelled as lognormal variate.
The investigation of the effect of missing nail on the uplift capacity
indicates that missing a single critical nail could reduce the mean of the
panel uplift capacity by about 10%.
5. The panel uplift capacity can also be characterized by the critical
reference mean wind speed. As expected, the statistics of the speed
depend on the location of the roof panel because the magnitude of and
uncertainty in the pressure coefficients are location dependent. In all
cases, the statistics represent the uplift capacity of the panel associated
with the instantaneous pressure coefficients, and they do not incorporate
the gust effect or exposure factor.

Probabilistic Assessment of Roof Uplift Capacities in Low-Rise

Residential Construction ,(Shanmugam, 2011)
Vulnerability of light frame roof structures in low rise buildings to extreme wind
events is often a result of insufficient connection strength, for example, roof to wall
connection and sheathing fasteners, in the wind uplift load path. The susceptibility
of a roof component or system can be expressed using a fragility plot which
conveys the probability of failure of the component or a system when subjected to a
particular wind load.
That includes not only a substantial increase in the design wind load in the past two
decades but also a strict enforcement of tighter nailing schedules and stronger RTW
connections (metal straps and hurricane ties). However a significant number of
older buildings constructed with toenailed RTW connections exist and their safety
and reliability needs to be investigated. Hence there is an apparent need to
statistically understand the behaviour of toenailed RTW connections in existing
buildings.
This research study evaluated the in-situ capacity of roof-to-wall connections and
sheathing to rafter fasteners in light-framed wood construction.

A finite element model of a light frame gable roof system was created using the
developed analytical model of the RTW toenail connection and sheathing fasteners.
Assessment of the overall impact of RTW and sheathing connector behaviour on the
wind-uplift fragility curves for the roof system was achieved using a Latin-hypercube
based simulation strategy. It was found that the treatment of post ultimate
connection behaviour had a significant influence on the fragility assessment of the
roof system.

Probabilistic finite element analysis using ANSYS, Reh, S.

(2005)
Deterministic analysis
Is the transformation function representing the relationship between the input
variables influencing the behaviour of a product and the result parameters
characterizing the product behaviour. In simple cases the result parameters can be
expressed as an analytical function, but in realistic cases the input-output
relationship is only given algorithmically for example using finite element program.
Uncertainty analysis
If the input variables influencing the behaviour o a product are uncertain, i.e are
subjected to scatter, then the primary task of an uncertainty analysis is to quantify
how much the result parameters characterizing the product behaviour are affected
by those uncertainties.
Reliability analysis
In order to quantify the reliability of a product it is useful to calculate the failure
probability or non-conformance probability denoted with P f. The reliability Ps is the
probability that the product will survive or conforms to certain requirements, with P s
= 1 Pf.
Reliability based optimization
As the name implies, reliability based optimization tries to optimize reliability or
failure probability. It should be noted, that improving the reliability often conflicts
with the other technical and financial goals. Hence, the optimization process should
try to achieve a reasonable and quantifiable balance between all goals.
ANSYS Probabilistic Design System (PDS) is made to address probabilistic problems.
As such, it can be used for an uncertainty analysis or reliability analysis. It is tightly
integrated into ANSYS, using the same graphical user interface, hence having the
same look and feel as ANSYS itself. The target audience is advanced engineers who
are comfortable with performing finite element analysis using ANSYS on a regular
basis. The PDS is based on the ANSYS Parametric Design Language (APDL), which
allows users to parametrically build a finite element model, solve it, obtain results
and extract characteristic results parameters such as the maximum stress for
example. In addition, the APDL provides the possibility to use input variables as well
as result parameters in arithmetic expressions, do loops and if-then-else constructs.
As such the PDS is completely independent of the physics captured in a finite

element analysis and for example covers the range from a typical linear elastic
stress calculation to a more advanced coupled multi-physics analysis.

Finite element analysis and field observation of a

residential roof subjected to hurricane winds (Jungmann,
M. 2007)
A detailed 3D finite element model for a LFB that was tested by Phillips (1990) was
analyzed by Kasal et al (1994). Kasal et al (1994) research focused on taking
individual substructures and components of Phillips (1990) experimental structure
and incorporated them in a full-structure FE model, which was the first time this had
ever been done.
The model showed that the FE model was significantly higher in wind loading than
the field data because all sheathing elements in the FE model had wind pressures
applied normal to the test roof. Also, the linear interpolation used for the applied
wind pressures to the FE model could have been different from what was measured
in the field over the test roof.

Reliability Assessment of Roof Sheathing Performance in

Light Wood Frame Structures Subjected to Wind Pressure,
Amini, M. (2010)
This paper presents a methodology for reliability assessment of roof sheathing
panels in low-rise structures for predicted wind pressure distributions. Low-rise
buildings encompass the majority of the residential structures in the United States.
Stochastic modelling is employed to investigate the effect of random variables such
as nail withdrawal capacity, and missing nails on sheathing panels due to
construction malpractice. This study combines analytical and stochastic modelling
to develop a comprehensive understanding of wind and structure interaction. The
developed methodology an be used as a tool to evaluate roof sheathing
performance subjected to wind load.
This study employs stochastic modelling of a typical North American roof to
examine the effects of variable parameters such as nail withdrawal capacity and
number of missing nails on sheathing panels. Pressure measurements from a full
scale structure are used to develop a series of generalized functions that represent
pressure variation on the roof of classified low rise buildings. Finally, the developed
pressure distributions and the stochastic analysis were used to develop an
analytical model.

Refined modelling of a typical low-rise building under

hurricane loads using finite element models (Pan, F.,2012)
The current study focuses on developing a validated finite element model that is
able to more rigorously represent load distributions or redistributions when the

building behaves as a unit or any portion is overloaded. The extensive modelling of

structural components, especially nail connections, effectively improves the
accuracy of the building envelope performance evaluation. For the roof sheathing
panels, the analytically predicted damage patterns that are based on the in-plane
shear strength agreed closely with the post disaster observations, which verified
that the proposed FE model could characterize the structural responses more
realistically
Martin et al. (2011) extended the use of the FEM to investigate the load paths in a
complete 3D building. In their study, the wall nail connections were simplified by the
equivalent shear modulus corresponding to specific nailing schedules to account for
the fact that the shear wall stiffness increases with the edge nail density.
Thampi et al. (2011) obtained the von-Mises stress contours by a detailed FEM to
make a comparison with a damaged low-rise building hit by a tornado event. In this
study, nail connections and sheathing panels were modelled in-depth but limited
details of the modelling were released.
The rationale of this investigation is to develop a refined analytical methodology
built upon the previous research to more accurately model the structural responses
of a typical low-rise residential building under hurricane loads.
Validation Steps:
Model validation is necessarily before any conclusion drawn based on the analytical
outputs and there are basically three philosophies to validate FEM used by the
previous study:
1. To verify the FEM directly with experiments conducted on the same prototype
through the comparisons between reaction forces, displacements or stresses
2. To validate certain specific modelling methods, e.g., truss member
connections, for another structure or substructure with different
configurations but accompanied with full scale test counterparts and then to
apply the same modelling methods to the objective structure of particular
interest
3. To validate the FEM modelling with the published analytical work or code
specifications on the structure with the same or close configurations

Damage Prediction of Low-Rise Buildings Under Hurricane

Winds, (Fang Pan,2014)
The empirically prescribed tributary area, load path and load sharing may be
reasonable for design on the conservative side, but not suitable for damage
prediction that demands accurate instead of conservative load distribution among
the entire system.
The most current standards are developed by obtaining equivalent pressure
coefficients that envelope the peak responses calculated from the wind tunnel data

for a range of assumed structural wind resisting system without appropriate

attention on building envelope.
The major objectives of this study includes to
1. Initiate a database-assisted damage prediction framework for both main wind
force resistance system and components and claddings
2. Collect aerodynamic datasets on scaled models by wind tunnel testing
3. Develop a comprehensive and in depth 3D finite analytical model under
realistic wind pressures with limited available post disaster reports
4. Numerically predict the detailed structural responses for ongoing quantitative
validation against the full scale static tests conducted by Florida International
University
5. Develop the vulnerability curves for a selected roof corner sheathing panel by
using a database assisted stochastic finite element modelling approach.
For the design of a new construction, the spatial and temporal fluctuating wind
loads are reasonably estimated in a few surface zones on the conservative side and
the load shares are prescribed as constants among structural components since the
entire building sustains wind loads as a unit. However, the scenario changes for
damage predictions. During the real wind and structure interactions, the spatial and
temporal variations in wind loads, the dependency of wind loads on building
damage status, and the load redistributions due to the malfunctions of overloaded
structure portions must be accounted rigorously.
The high spatial correlation among building surface pressures may produce
significant structural responses but not directly incorporated in ASCE standards for
the sake of simplicity. This may be compensated somehow by enveloping the worst
loading cases in the codification for component and claddings. However, the unclear
principles of how to combine those worst positive and negative pressure coefficients
to the entire building surface as specified in design standards may fail to capture
the worst loading condition at the system level.
While neglecting wall partitions may lead to over predict the shear wall forces by
more than 100%, the structural responses in the higher portion of the building
remain limited since the shear responses in the roof to wall connections are far
below their peak resistances as reported by Asiz et al (2008).
Therefore, the internal wall partitions are not included here to simplify the
modelling. This is also to be consistent with the model used in the wind tunnel test
(Pan et al. 2013) that provides external and internal wind pressures for the present
and future FEM analysis.

This study addresses the damage prediction issues from another viewpoint by
using a database assisted stochastic finite element modeling approach. The
proposed approach can incorporate the surface wind pressure information from
wind tunnel tests as statistics or time histories, account load distributions
rigorously according to the stiffness of all primary and secondary structural
components, and consider the uncertainties in wind loads as well as material
properties. The vulnerability of a selected roof sheathing is assessed in this
chapter to illustrate the procedures regarding how to utilize the up-to-date

engineering techniques and testing facilities for damage predictions. The

engineering based vulnerability curves are then developed for the targeted
structure performance criteria according to the damage tallies output by a
comprehensive 3D finite element model subjected to the wind tunnel simulated
pressures. The resulted relationship between the mean damage ratios and wind
speeds could be combined with local meteorological data to predict the damage
ratio for each concerned performance. This proposed methodology is applicable
to building envelope components, connections and frame systems. Therefore,
fully engineering based vulnerability curves can be developed in both the
component scale and system scale, which advances the current empirical
engineering based damage approach used by the two public hurricane loss
models towards a more realistic prediction.
Stochastic Finite Element Method (SFEM)
Stochastic finite element method is an extension of the classical deterministic FE
approach to the stochastic framework, i.e., to the solution of stochastic (static
and dynamic) problems involving finite elements whose properties are random
(Stefanou 2009). The ANSYS Probabilistic Design System (PDS) is used to
analyze the selected roof corner sheathing involving uncertain input parameters
here. In this study, the uncertain input parameters are assumed to be limited to
the oncoming wind speed , nail stiffness , and the displacements of the nailing
points on the connected frame system . These input parameters are defined in a
deterministic model by using ANSYS Parametric Design Language (APDL).
The developed APDL file describes the deterministic model and is assigned as an analysis file for
the later probabilistic analysis. After the deterministic model is solved, the variation of the input
parameters, i.e., U , nail K , and frame , are defined as random input variables and are characterized
by their distribution type (Gaussian, lognormal, etc) and by their distribution parameters (mean
values, standard deviation, etc). Any interdependencies between random input variables can also
be defined as correlation coefficients. In this study, all random input variables are assumed to be
independent. The important results are defined as random output parameters.
Future Works and Recommendations:
Numerical Assessment of Roof Panel Uplift Capacity under Wind Load
(Weixian He, 2010)
It is suggested that the analyses presented in the present study is to be repeated
for the whole roof system. The obtained results can be used to aid the development
of simple approach to estimating the reliability of roof system. They can also be
used to aid the calibration of design code.
Also, the analysis can be extended to include the roof trusses and impact of the
construction error associated with the toe nails used to fastening the trusses on the
reliability of the roof system.
This study employs stochastic modelling of a typical North American roof to
examine the effects of variable parameters such as nail withdrawal capacity and

number of missing nails on sheathing panels. Pressure measurements from a full

scale structure are used to develop a series of generalized functions that represent
pressure variation on the roof of classified low rise buildings. Finally, the developed
pressure distributions and the stochastic analysis were used to develop an
analytical model.

Probabilistic Assessment of Roof Uplift Capacities in Low-Rise

Residential Construction ,(Shanmugam, 2011)
Recommendations for chapter 2 and 3
a.) Development of resistance statistics and analytical model for roof to wall
metal connecters subjected to wind uplift loads
b.) Estimation of roof to wall metal connector system fragility
Recommendations for chapter 4 and 5
a.) Evaluation of realistic roof system fragility estimation using actual material
properties and realistic wind loading
b.) Estimation of the influence of wind direction, building dimension and roof
configuration on the pressure coefficient correlation matrix and thereby to
identify their influence on roof system fragility estimation
c.) Numerical evaluation of realistic roof panel capacity for dynamic wind loads
d.) Extension the fragility study to include lateral wind loads

Finite element analysis and field observation of a

residential roof subjected to hurricane winds (Jungmann,
M. 2007)
A FE approach to modeling the entire light frame building (LFB) was provided
in this chapter. Modeling the entire structure was important because much of
the response and intercomponent connections (Collins et al. 2005).
Therefore, the connections utilized in the construction of the LFB must be
well defined in a finite element model. Also, proper estimates of the material
properties used in the LFB to ensure a more accurate FE model of the test
roof. Kasal et al. (1994) analyzed the most detailed FE model for a LFB, which
was tested by Phillips (1990).
Another area of interest when modeling a roof system is to incorporate some
of the finite element modeling assumptions that were used by Zhong et al.
(1998) such as joint eccentricities and partial composite action. If the test
structure was tested in the field under static point loads, the results could be
obtained for more accurate representation of the loadsharing effects and
more accurately model the connections used in construction of the rafters
and trusses. Also, with the field test the researchers could utilize the data to
incorporate spring elements at the connections of the rafters and

trusses.performance of a light-frame building (LFB) is dictated by the

structures diaphragms and

Damage Prediction of Low-Rise Buildings Under Hurricane

Winds, (Fang Pan,2014)
Specifically, the following future works are envisioned:
(1) the vulnerability curves of all roof and wall sheathing panels can be studied. The most vulnerable
piece of sheathing could be analyzed by the methodology developed in the present study and by the
conventional procedure that directly compares the wind pressures instead of structural responses.
Through a future comparison of these two approaches, the differences of the results due to the
inclusion of nail connection details and the sheathing-frame interactions can be identified, which may
provide useful information to the future revisions of ASCE7;
(2) while the building envelope is the primary reason for the economic loss, the collapse of the frame
system may result in casualty that is anther unacceptable aspect of hurricane events. The proposed
methodology developed in the present study is applicable to evaluate the vulnerability of the frame
system. A future combination of the vulnerability curves from the building envelope and the frame
system is the key step to obtain the vulnerability curve for the entire system;
(3) the wind-structure interactions, i.e., the wind pressures change with the loss building envelope,
can be traced to obtain the accumulated structural damage that tends to magnify the structural
damages. A future investigation of the progressive failure process is deserved to thoroughly
understand the performance of low-rise buildings in an extreme hurricane event;
(4) wind-born debris and rain water intrusion may be included to quantify the magnification of
structural damages due to the interior damages;
(5) multi-hazards may be included by changing the load input to the developed model, e.g., flood or
earthquake loads

Objectives: