13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13

Seoul, South Korea, May 26-30, 2019

Wind Reliability of Transmission Line Models using Kriging-Based

Methods
Yousef Mohammadi Darestani
PhD Candidate, Dept. of Civil, Environmental, and Geodetic Engineering, The Ohio State University,
Columbus, Ohio, USA
Zeyu Wang
PhD Candidate, Dept. of Civil, Environmental, and Geodetic Engineering, The Ohio State University,
Columbus, Ohio, USA
Abdollah Shafieezadeh
Associate Professor, Dept. of Civil, Environmental, and Geodetic Engineering, The Ohio State
University, Columbus, Ohio, USA

ABSTRACT: Risk assessment of power transmission systems against strong winds requires models that
can accurately represent the realistic performance of the physical infrastructure. Capturing material
nonlinearity, p-delta effects in towers, buckling of lattice elements, joint slippage, and joint failure
requires nonlinear models. For this purpose, this study investigates the reliability of transmission line
systems by utilizing a nonlinear model of steel lattice towers, generated in OpenSEES platform. This
model is capable of considering various geometric and material nonlinearities mentioned earlier. In order
to efficiently estimate the probability of failure of transmission lines, the current study adopts an
advanced reliability method through Error rate-based Adaptive Kriging (REAK) proposed by the authors.
This method is capable of significantly reducing the number of simulations compared to conventional
Monte Carlo methods such that reliability analysis can be done within a reasonable time. Results indicate
that REAK efficiently estimates the reliability of transmission lines with a maximum of 150 Finite
Element simulations for various wind intensities.

areas. For example, transmission line failures in

1. INTRODUCTION Hurricane Irene (2011) and Hurricane Sandy
(2012) resulted in 6.69 million and 8.66 customers
Electric power is transferred from power plants to to lose their power, respectively (Hoffman and
distant areas through transmission systems, Bryan, 2013).
commonly supported by lattice towers. Although A large percentage of current studies on
lattice towers are usually designed to withstand performance assessment of transmission lines,
high intensity wind hazards, experience from past assumed deterministic models for investigating
weather related events such as hurricanes, failure in lattice towers (Jiang et al., 2011; Jiang
tornados, and downbursts has highlighted their et al., 2017; Ahmed et al., 2009). However, for
vulnerability (Campbell, 2012; Hoffman and risk management purposes, there is a significant
Bryan, 2013; Elawady et al., 2017). For example, need to estimate the reliability of transmission
Hurricane Irene (2011) caused 280 transmission towers through probabilistic models that account
line failures and hurricane Sandy (2012) led to for uncertainty in demand and capacity of lattice
200 transmission line failures (Hoffman and towers as well as various failure modes. There are
Bryan, 2013). Moreover, as transmission lines a few studies that investigated the performance of
perform with minimum redundancy, any failure lattice towers through probabilistic models
can result in power outages for large geographical

1
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

(Rezaei, 2016; Fu et al., 2016). However, as these of power networks (Bhat et al., 2018; Darestani et
studies use simplified models that do not account al., 2016a; Darestani et al. 2016b; Darestani et al.
for post buckling and post yielding behavior of 2017; Darestani and Shafieezadeh, 2017).
tower elements as well as joint slippage and joint
failure behaviors, they are not necessarily 2. FINITE ELEMENT MODELING OF
representative of the true performance of lattice TRANSMISSION TOWERS
towers during high intensity wind hazards such as Due to various complexities such as post yielding
hurricanes. In these studies, it is assumed that any and post buckling nonlinearities, joint slippage,
failure in the tower results in its total failure. and joint failure along with various modes of
However, as lattice towers are significantly failure, high fidelity nonlinear Finite Element
indeterminate structures, they may not fail under models are essential for a reliable estimation of
a single failure unless there is a failure mechanism the performance of transmission lines during high
developed in the tower. To estimate the intensity wind hazards. For this purpose, a
probability of failure of lattice towers, first order nonlinear static pushover analysis is employed in
reliability methods have been used in the literature OpenSEES platform and elaborated in the
(Rezaei, 2016). However, first order reliability following subsections.
methods are not accurate when nonlinear behavior 2.1 Modeling Steel Lattice Elements in
of steel elements (such as post buckling and post OpenSEES
yielding behavior) is taken into account. Monte In order to account for post yield elasticity,
Carlo simulation methods can also be used to Steel01 material model is considered in
estimate the probability of failure of lattice OpenSEES, which assumes a bilinear relationship
towers. However, these methods require a large for stress-strain behavior. Nonlinear
number of realizations to yield a reliable estimate displacement-based beam column elements are
of probability of failure of the system especially defined through five integration points with 10
for small failure probabilities (Zamanian, 2016). fiber sections along the height and three fiber
In order to address the aforementioned sections along the width of angle elements at each
limitations, this study investigates the probability integration point. In addition, p-delta effects and
of failure of lattice transmission towers by geometric nonlinearities are accounted for
developing a high fidelity Finite Element model through a co-rotational geometric transformation.
that accounts for post buckling and post yielding In order to consider buckling accurately,
behavior of steel elements. In addition, joint according to Uriz et al. (2008), each element is
slippage and joint failure are modeled through a divided in half and a camber displacement equal
nonlinear connection model developed in to 1/2000 to 1/1000 of the length of the element is
OpenSEES (McKenna, 2000) Finite Element applied to the middle node.
platform based on the model suggested by
Ungkurapinan (2000). Subsequently, the 2.2 Modeling Connections
probability of failure of a lattice tower is estimated Under strong wind loads such as hurricanes, there
through a set of pushover analyses along with an is a significant level of joint slippage in the
Error rate-based Adaptive Kriging (REAK) model connections. Joint slippage considerably increases
developed by the authors (Wang and the lateral displacement of the tower, which can
Shafieezadeh, 2018). This model can efficiently result in additional p-delta effects and structural
estimate the probability of failure with much less couplings between adjacent towers.
number of Finite Element simulations compared Ungkurapinan (2000) suggested a nonlinear
to ordinary Monte Carlo simulation methods. model for joint slippage behavior based on a set
Such reliability models can be used in risk of experiments he performed for steel angle
assessment procedures to enhance the resilience members. The proposed model follows a
backbone curve similar to Fig.1. This study adopts

2
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

this model to consider joint slippage behavior. For where 𝐾𝑧 is the velocity pressure exposure
this purpose, joint slippage is modeled by coefficient, 𝐾𝑑 is the wind directionality factor,
assigning zero-length elements in OpenSEES at 𝐾𝑧𝑡 is the wind topographic factor, 𝐾𝑒 is the
the connections and applying the joint slippage elevation factor, and V is the 3-second gust wind
behavior to the zero-length elements as a material velocity at 10 m above the ground line. Since the
model (Fig. 2) assumed lattice tower is located in a flat area,
Force
D therefore, 𝐾𝑧𝑡 is equal to 1. 𝐾𝑧 is is obtained from
2/𝛼
C max⁡(4.75, 𝑧)
𝐾𝑧 = 2.01 ( ) (3)
𝑧𝑔
A B where z is the height from the ground. Since the
lattice tower is located in an open terrain area,
Displacement
exposure category is C, and 𝛼 and 𝑧𝑔 are 9.5 and
Phase 1

Phase 2

Phase 4
Phase 3

274.32 m, respectively. The wind directionality

factor, 𝐾𝑑 , is equal to 1. The gust-effect factor, G
Figure 1. Backbone curve of slippage behavior is equal to 0.85. ASCE07 (2016) defines the force
(Ungkurapinan, 2000) coefficient,⁡𝐶𝑓 , for squared trussed towers as
𝐶𝑓 = 4 ∈2 − 5.9 ∈ +4 (4)
where ∈ is the ratio of solid area to gross area of
the tower face under consideration. 𝐶𝑓 is equal to
1, for the conductors (ASCE 74, 2009).

4. RELIABILITY ANALYSIS USING REAK

METHOD
As it was mentioned previously, reliability
analysis of lattice towers requires estimation of
the limit state function for a large number of
realizations of uncertain parameters to perform a
Monte Carlo simulation. However, as nonlinear
Figure 2. Modeling joint slippage behavior in Finite Element analysis is computationally
OpenSEES using zero-length elements. expensive and considerably time consuming,
direct estimation of limit state function for the
3. WIND LOAD ON LATTICE TOWERS entire set of realizations is practically impossible
To estimate the wind induced load on lattice (Ebad-Sichani et al. 2019; Ebad-Sichani et al.
towers, the static gust wind load suggested by 2018; Fereshtehnejad et al., 2016). In order to
ASCE07 (2016) is employed in this study. address this limitation, various reliability analysis
ASCE07 proposes the following equation for methods based on Kriging have been developed in
calculating wind induced load on non-building the literature (Echard et al. 2011; Jones et al.,
structures. 1998; Wen et al., 2016; Wang and Shafieezadeh,
𝑓𝑤 = 𝑞𝑧 𝐺𝐶𝑓 𝐷 (1) 2018). In Kriging-based reliability analysis,
where 𝑞𝑧 is the velocity pressure at height z on the estimation of the limit state function using
tower, 𝐺 is the gust-effect factor, 𝐶𝑓 is the force computationally expensive Finite Element
coefficient, and D is the width of the element method is limited to a small number of candidate
realizations, in which the limit state function
perpendicular to the wind direction. 𝑞𝑧 is
estimated from: (𝐺(𝑥)) is close to zero. Subsequently, a Kriging
model is used to estimate the limit state function
𝑞𝑧 = 0.613𝐾𝑧 𝐾𝑑 𝐾𝑧𝑡 𝐾𝑒 𝑉 2 (2)

3
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

for the entire set of realizations of uncertain Step 6: Among the samples that satisfy Eq.
parameters to efficiently perform a Monte Carlo (4), choose the one with the maximum EFF from
simulation. Therefore, using Kriging-based the following Eq.:
reliability analysis, a large number of Finite 𝐸𝐹𝐹(𝒙) = (𝜇𝐾 (𝒙) − 𝑎) ×
𝑎 − 𝜇𝐾 (𝒙) 𝑎− − 𝜇𝐾 (𝒙) 𝑎+ − 𝜇𝐾 (𝒙)
Element simulations are avoided and [2Φ (
𝜎𝐾 (𝒙)
) − Φ(
𝜎𝐾 (𝒙)
) − Φ(
𝜎𝐾 (𝒙)
)]
subsequently, the probability of failure of lattice −𝜎𝐾 (𝒙) ×
towers are efficiently estimated. Further 𝑎 − 𝜇𝐾 (𝒙) 𝑎− − 𝜇𝐾 (𝒙) 𝑎+ − 𝜇𝐾 (𝒙) (5)
[2ϕ ( )−ϕ( )−ϕ( )]
discussion on limitations of different Kriging- 𝜎𝐾 (𝒙) 𝜎𝐾 (𝒙) 𝜎𝐾 (𝒙)
𝑎+ − 𝜇𝐾 (𝒙) 𝑎− − 𝜇𝐾 (𝒙)
based reliability analyses can be found in the +2𝜎𝐾 (𝒙) [Φ (
𝜎𝐾 (𝒙)
)− Φ(
𝜎𝐾 (𝒙)
)]
paper by Wang and Shafieezadeh (2018). where ϕ(∙) denotes the standard normal
In this study, in order to obtain the probability density function and ⁡Φ(∙) is the
probability of failure of lattice towers, the Error standard normal cumulative density function, 𝑎 =
rate-based Adaptive Kriging (REAK) proposed
0, 𝑎+ = 2𝜎𝐾 (𝒙), and 𝑎− = −2𝜎𝐾 (𝒙). 𝜇𝐾 (𝒙) and
by the authors (Wang and Shafieezadeh, 2018) is
𝜎𝐾 (𝒙) are the mean and standard deviation of
adopted. This method has shown two advantages
Kriging prediction for point x, respectively.
over the existing adaptive Kriging reliability
Step 7: If 𝑚𝑎𝑥(𝐸𝐹𝐹) <0.001 go to step 8
methods. First, this method introduces an
otherwise go to step 4.
effective adaptive sampling region, in which the
Step 8: Check if the upper bound of error rate
points with low joint probability density function
(𝜖̂𝑚𝑎𝑥 ) is less than 0.05
are removed from candidate samples. Second, an 𝑁 𝑓̂ −𝐼𝛺2
upper bound for the rate of error is introduced 𝜖̂𝑚𝑎𝑥 = 𝑚𝑎𝑥𝑅 (|
𝛺2
|)<0.05 (5)
based on the Lindeberg’s condition for the Central 𝑁
𝐼𝛺2 ⁡∈⁡𝑁 𝑓 ̂ +𝐼𝛺2
𝑓
𝛺1
𝛺 2
limit Theorem (CLT). Using this upper bound, a
where 𝛺1 and 𝛺2 are the regions inside and
faster convergence can be obtained for the
outside of the effective sampling region denoted
reliability analysis. In the following, REAK
algorithm is summarized: by Eq. (4), respectively. 𝑁⁡denotes the size of the
Step 1: Generate N realizations of uncertain set, and 𝐼 is an indicator that takes one when the
parameters using Latin Hypercube Sampling sign of the limit state function is estimated
method. wrongly and takes zero when the sign of the limit
Step 2: Define an effective sampling region state function is estimated correctly.
as Step 9: If 𝜖̂𝑚𝑎𝑥 <0.05 is not satisfied, increase
the size of the effective sampling region by
𝑃{𝜌(𝑥)>𝜌𝑡ℎ𝑟} = 𝛼𝑃̂𝑓𝑛−1 . (4)
reducing the value of 𝛼.
where 𝜌(𝑥) is the joint probability density of Step 10: Estimate the coefficient of variation
candidate design samples, 𝛼 is a constant of failure probability
coefficient and 𝑃̂𝑓𝑛−1 is the probability of failure
achieved by Kriging model. Points outside this 1 − 𝑃̂𝑓
𝐶𝑂𝑉𝑃𝑓 = √ (5)
region will be later removed from training 𝑃̂𝑓 𝑁𝑀𝐶𝑆
samples. Step 11: If 𝐶𝑂𝑉𝑃𝑓 < 0.05, stop the process,
Step 3: Randomly select a small number of
otherwise, increase the number of LHS samples.
initial points from Step 1. These points will be
A flowchart of REAK algorithm is provided
used for constructing the initial Kriging model.
in Fig. 3. As it was mentioned, REAK algorithm
Step 4: Construct a Kriging model for
has a better efficiency compared to the existing
estimating the limit state function using Finite
adaptive Kriging reliability methods as it provides
Element analysis of lattice tower.
an adaptive effective sampling region denoted in
Step 5: Update the efficient sampling region
Step 2 and an upper bound for the maximum error
using Eq. (4).
rate presented in Step 6.

4
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

REAK Algorithm various modes of failure. Due to these

complexities, Finite Element analysis methods are
Generate large samples (𝑥𝑠 ) necessary to estimate the performance of lattice
using LHS towers. However, estimation of reliability of
lattice towers through conventional Monte Carlo
Select a few number of initial points simulation methods requires a large number of
(𝑥𝑖 ) and estimate the limit state
function for the tower using FEM in time consuming evaluations of limit state
OpenSEES functions, which makes the process of reliability
analysis practically impossible. For this purpose,
Construct a Kriging model and in this study, a reliability analysis through Error
estimate G(x) for 𝑥𝑠
rate-based Adaptive Kriging is adopted to
efficiently generate a fragility model for a double
Update the sampling region
for points with higher joint circuit vertical steel lattice tower. This tower is
PDF 27.4 m tall, located in a hurricane prone coastal
area in south of the United States. It carries 6 lines
Estimate Expected Learning of Drake ACSR (Aluminum Reinforce Steel
Function (EFF) for the
updated sampling region
Conductors) and two line of Optical Ground wires
(OPGW). The span length is 258 m and it is
Choose a point with Update assumed that multiple spans with similar
max(EFF) as the next sampling properties exist in the system. Therefore, the
sampling point region
impact of structural couplings is negligible
Yes
(Darestani et al., 2016a; Darestani et al. 2016b;
Update 𝑥𝑠 by No max(𝐸𝐹𝐹)
Darestani et al. 2017). A sketch of the modeled
generating
extra samples ≤ 10−3 tower is provided in Fig. 4.

Yes

Estimate maximum
error rate 𝜖̂𝑚𝑎𝑥

No
𝜖̂𝑚𝑎𝑥 ≤ 0.05

Yes

𝐶𝑂𝑉𝑃𝑓 Yes Figure 4. The assumed double circuit vertical steel

No Exit
≤ 0.05 lattice tower

The tower is modeled in OpenSEES platform

Figure 3. REAK algorithm as it was discussed in sections 2 and 3. In order to
perform the reliability analysis, uncertainty in
5. NUMERICAL ANALYSIS material and demand should be considered. Cha et
As it was noted previously, lattice towers al. (2018), performed a sensitivity analysis on
experience complex nonlinear behaviors at the various uncertainties in modeling of the current
prior to and at the verge of failure especially under lattice tower and they found the parameters
strong winds. These complexities stem from post provided in Table 1 as significant uncertain
yielding and post buckling nonlinear behavior of parameters that can affect the performance of the
steel elements, joint slippage and joint failure, lattice tower. Therefore, in this study, the
uncertainties in demand and capacity as well as parameters shown in Table 1 are assumed

5
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

uncertain. The other parameters are set to their probability of failure of lattice transmission
mean value. towers can be efficiently estimated through
In order to perform the reliability analysis, a REAK. For the assumed lattice tower, the number
limit state function should be defined for the of calls to estimate the limit state function through
tower. In this study, a pushover analysis is carried Finite Element analysis is less than 150.
out to obtain the maximum load bearing capacity Comparing this value with conventional Monte
of the tower. The load bearing capacity is defined Carlo simulations which require tens of thousands
as a factor of design wind speed. For the current of simulations highlights the efficiency and
tower, the design wind speed is equal to 130 mph. importance of adaptive Kriging reliability
Subsequently, the limit state function for the methods such as REAK to generate fragility
tower is defined as: models for lattice towers. The fragility analysis
𝐺(𝑥) = 𝐹𝐿 − 1 (5) explained in this paper can be integrated with a
where 𝐹𝐿 is the failure load factor presented in Generalized Linear Model (GLM) to provide
Fig. 5. simple and accurate fragility models for
transmission towers with different configurations
Failure Load =0.95 × 𝑊𝑖𝑛𝑑
including, type, height, span length, and number
and diameter of conductors, among others. A
similar approach was used by the authors in
(Darestani and Shafieezadeh, 2019) to generate
multi-dimensional wind fragility functions for
wood utility poles.

Figure 5. Definition of limit State function through

pushover analysis
For the analysis performed in this figure, the
failure load equals 0.95, which shows that any
wind load greater than 0.95 times the design wind
load of the tower results in the failure of the tower.
Using REAK method a fragility model is
developed for the double circuit vertical lattice
tower (Fig. 5). The results highlight that the Figure 6. Fragility model developed for the double
circuit vertical lattice tower using REAK method

Table 1: Uncertain parameters assumed for reliability analysis

Properties Notation Type of Reference
Distribution Mean COV
Steel material Modulus of elasticity 𝑬 LogNormal 2.0e11 0.06 ASCE07 (2010) and ASCE 74
(N/m2) (2009)
Yield stress of main leg 𝒇𝒚𝒎 LogNormal 4.02e8(N/m2) 0.1
Yield stress of bracing 𝒇𝒚𝒃 LogNormal 2.9e8 (N/m2) 0.1
members
Post yield elasticity 𝑬𝒔𝒕 LogNormal 0.02E(N/m2) 0.25
Wind load Gust effect factor 𝑮 Normal Section 3 0.11 Ellingwood and Tekie (1999)
Force coefficient 𝑪𝒇 Normal Section 3 0.12
Velocity pressure exposure 𝑲𝒛 Normal Section 3 0.16
coefficient
Wind directionality factor 𝑲𝒅 Normal Section 3 0.08
Connection Type B Slippage length Phase 2 Uniform 0.45(mm) 0.15 Ungkurapinan (2000)

6
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

estimate the limit state function through Finite

3. SUMMARY AND CONCLUSIONS Element analysis is less than 150, which
Complex behavior of lattice towers during strong highlights the efficiency of the method,
wind necessitates use of Finite Element analysis considering the high accuracy of the results.
methods. The complexities stem from post
yielding and post buckling nonlinear behavior of 4. REFERENCES
towers, joint slippage and joint failure, and Ahmed, K. I. E., R. K. N. D. Rajapakse, and M. S.
various modes of failure that can occur in a lattice Gadala. (2009) "Influence of bolted-joint
tower, among others. Given that conventional slippage on the response of transmission towers
Monte Carlo simulation methods require a large subjected to frost-heave." Advances in
Structural Engineering 12, no. 1: 1-17. Canada
number of time consuming Finite Element
ASCE (American Society of Civil Engineers) (2010),
simulations to estimate the probability of failure “Minimum design loads for buildings and other
accurately, Monte Carlo simulation methods structures”. Standard ASCE/SEI 7-10.Reston,
cannot be directly used. VA, USA;
In order to address this limitations, this study, ASCE No. 74 (2009) “Guidelines for Elctrical
considers the impact of post yielding and post Transmission Line Structural Loading”
buckling through a bilinear material model American Society of Civil Engineers.
integrated with a displacement beam column Bhat R., Y. Mohammadi Darestani, A. Shafieezadeh,
element modeled in OpenSEES Finite Element A.P. Meliopoulos, R. DesRoches. “Resilience
platform, in which each element is divided in half Assessment of Distribution Systems
and a camber displacement is applied to the mid- Considering the Effect of Hurricanes”, IEEE
PES Transmission and Distribution Conference
node to capture the out of plane displacement of
& Exposition. Denver, Co. USA. April 16-19,
steel elements under buckling effects. In order to 2018
consider p-delta effects and instability caused by Campbell R. J. (2012), “Weather-related power
buckling effects large deformations are accounted outages and electric system resiliency,”
for through a co-rotational geometric Congressional Research Service, Library of
transformation in OpenSEES. In addition, joint Congress, R42696.
slippage and joint failure are accounted for Cha K., Mohammadi Darestani Y., Zamanian S., and
through a nonlinear experimentally validated Shafieezadeh A., “Sensitivity Analysis of
model applied to each connection through Structural Uncertainties for the Hurricane
zerolength elements in OpenSEES. Second, to Performance of Transmission Towers using
estimate the probability of failure of the lattice BART”, Engineering Mechanics Institute
tower, a reliability analysis through Error rate- (EMI) conference, Massachusetts Institute of
Technology (MIT), Cambridge, MA, May 29-
based Adaptive Kriging (REAK) is employed in June 1, 2018
this study. REAK has two advantages compared Darestani, Y. M., and A. Shafieezadeh. "Modeling the
to previous adaptive Kriging reliability analysis Impact of Adjacent Spans in Overhead
methods. First, it defines an adaptive effective Distribution Lines on the Wind Response of
sampling region that neglect realizations with low Utility Poles." In Geotechnical and Structural
joint probability density function, and second, it Engineering Congress 2016a, pp. 1067-1077.
defines an upper bound for the rate of error, Darestani, Y.M , A. Shafieezadeh, and R. DesRoches.
through which the convergence is obtained much "An equivalent boundary model for effects of
faster. Using REAK a fragility model is adjacent spans on wind reliability of wood
developed for a double circuit vertical lattice utility poles in overhead distribution lines."
Engineering Structures 128 2016b: 441-452.
tower. The result indicate that for various wind
Darestani, Y. M., A. Shafieezadeh, and R. DesRoches.
speeds, especially for those with low probability "Effects of Adjacent Spans and Correlated
of failure, REAK can efficiently estimate the Failure Events on System-Level Hurricane
probability of failure. The number of calls to

7
 13th International Conference on Applications of Statistics and Probability in Civil Engineering, ICASP13
Seoul, South Korea, May 26-30, 2019

Reliability of Power Distribution Lines." IEEE functions." Journal of Global optimization 13,
Transactions on Power Delivery 2017. no. 4 (1998): 455-492.
Darestani, Y. M., and A. Shafieezadeh, “Hurricane McKenna F, Fenves GL, Scott MH. Open system for
Performance Assessment of Power Distribution earthquake engineering simulation
Lines Using Multi-scale Matrix-based System (OpenSEES). Berkeley, CA: University of
Reliability Analysis Method”, 13th Americas California; 2000.
Conference on Wind Engineering, Gainesville, Rezaei, Seyedeh Nasim. (2016) "Fragility assessment
Florida, USA, May 21 - 24, 2017 and reliability analysis of transmission lines
Darestani, Y.M. and Shafieezadeh, A., 2019. Multi- subjected to climatic hazards." PhD diss.,
dimensional wind fragility functions for wood McGill University.
utility poles. Engineering Structures, 183, Sichani, M. E., Padgett, J. E., & Bisadi, V. (2018).
pp.937-948. Probabilistic seismic analysis of concrete dry
Echard, B., N. Gayton, and M. Lemaire. "AK-MCS: cask structures. Structural Safety, 73, 87-98.
an active learning reliability method combining Sichani, M. E., M. Hanifehzadeh, J. E. Padgett, and B.
Kriging and Monte Carlo simulation." Gencturk. "Probabilistic analysis of vertical
Structural Safety 33, no. 2 (2011): 145-154. concrete dry casks subjected to tip-over and
Elawady, A., & El Damatty, A. (2016). Longitudinal aging effects." Nuclear Engineering and Design
force on transmission towers due to non- 343 (2019): 232-247.
symmetric downburst conductor loads. Uriz, P., Filippou, F. C., & Mahin, S. A. (2008).
Engineering Structures, 127, 206-226. “Model for cyclic inelastic buckling of steel
Ellingwood B. R., and P. B. Tekie. (1999) "Wind load braces”. Journal of structural engineering,
statistics for probability-based structural 134(4), 619-628.
design." Journal of Structural Engineering 125, Ungkurapinan, N. (2000). “A Study of Joint Slip in
no. 4: 453-463. Galvanized Bolted Angle Connections”, MSc.
Fereshtehnejad, E., M. Banazadeh, and A. Thesis, University of Manitoba, Winnipeg,
Shafieezadeh. (2016):"System reliability-based Canada.
seismic collapse assessment of steel moment Wang, Z., and A. Shafieezadeh. "REAK: Reliability
frames using incremental dynamic analysis and analysis through Error rate-based Adaptive
Bayesian probability network." Engineering Kriging." Reliability Engineering & System
Structures 118 :274-286. Safety (2018).
Fu, X., Li, H. N., & Li, G. (2016). “Fragility analysis Wen Z, Pei H, Liu H, Yue Z. A Sequential Kriging
and estimation of collapse status for reliability analysis method with characteristics
transmission tower subjected to wind and rain of adaptive sampling regions and
loads”. Structural Safety, 58, 1-10. parallelizability. Reliab Eng Syst Saf
Hoffman P. and W. Bryan, (2013) “Comparing the 2016;153:170–9.
impacts of Northeast hurricanes on energy Zamanian, S. (2016). Probabilistic Performance
infrastructure,” Office of Electricity Delivery Assessment of Deteriorating Buried Concrete
and Energy Reliability, U.S. Department of Sewer Pipes (Master thesis, The Ohio State
Energy. University).
Jiang, W. Q., Z. Q. Wang, G. McClure, G. L. Wang,
and J. D. Geng. (2011) "Accurate modeling of
joint effects in lattice transmission towers."
Engineering Structures 33, no. 5: 1817-1827.
Jiang, Wen-Qiang, Yao-Peng Liu, Siu-Lai Chan, and
Zhang-Qi Wang. (2017) "Direct Analysis of an
Ultrahigh-Voltage Lattice Transmission Tower
Considering Joint Effects." Journal of Structural
Engineering 143, no. 5: 04017009.
Jones, D. R., M. Schonlau, and W. J. Welch. "Efficient
global optimization of expensive black-box