Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Contents lists available at ScienceDirect

Journal of Wind Engineering & Industrial Aerodynamics

journal homepage: www.elsevier.com/locate/jweia

Fragility analysis of a transmission tower under combined wind and

rain loads
Xing Fu a, Hong-Nan Li a, b, Gang Li a, *, Zhi-Qian Dong a
a
State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian, 116023, China
b
School of Civil Engineering, Shenyang Jianzhu University, Shenyang, 110168, China

A R T I C L E I N F O A B S T R A C T

Keywords: Frequently, only one parameter is used in fragility evaluations for engineering structures under environmental
Transmission tower loads. However, there are two intensity measures (IMs) for combined wind and rain loads. Consequently, this
Wind load study presents a process for constructing the fragility surface of tower structures under combined wind and rain
Rain load
loads, and an alternative IM based on basic wind speed and rain intensity is proposed to satisfy the requirement of
Uncertainty analysis
Fragility surface
fragility surface with two IMs. Then, a fragility evaluation is performed for an actual transmission tower-line
Performance evaluation system. The uncertain parameters and corresponding probability distributions are provided to establish the un-
certainty numerical models. Intensity measure groups of wind and rain loads are generated, followed by a
nonlinear dynamic analysis. The fragility surfaces are consequently obtained, and meteorological data of typhoon
Hagupit are used to calculate the failure probabilities under wind and rainfall excitations. Wind attack angle is
also found to exert great influence on the fragility surface, with the most unfavorable wind attack angle being 90
.
For lighter rain intensities, the fragility surface changes abruptly. Finally, the concept of critical collapse surface, a
quick and convenient tool for judging the structural state at varying wind speeds, directions, and rain intensities,
is proposed.

1. Introduction mechanism of transmission towers during severe gales and thunder-

storms, the primary task is to present a method for calculating the rain
Transmission lines are one of the most important infrastructures in load. Li et al. (2013) proposed a momentum conservation method to
modern society. They are used for transferring electric energy from calculate the rain load of a transmission tower; then, they performed
power plants to consumers. The main networks of transmission lines are simulation and experimental studies of a transmission tower under wind
supported by steel lattice towers, and the entire energy system will be and rain loads. The results indicated that the influence of rain load should
disrupted when the tower collapses. Therefore, it is of great importance be considered during strong rainstorms. Fu et al. (2015b) validated that
to guarantee the tower safety under severe loading conditions. Due to the horizontal velocity of a raindrop is higher than the corresponding
uncertainties in the structural parameters and external loads, the struc- wind speed; they derived the fitting formula of the velocity ratio, which
tural design is a decision-making process involving risks, constraints of was used to further modify the rain load formula. Fu et al. (2015a)
the national economic conditions, and a balance between the initial developed a hybrid method for calculating the rain load based on a
construction costs and the potential collapse costs in the future. Many single-raindrop impingement experiment and carried out wind tunnel
collapse events of transmission lines have been reported during typhoons tests to verify its effectiveness. Fu and Li (2018a) further investigated the
or hurricanes, and nearly all of these collapse cases are attributed to wind effects of different raindrop size distributions on both the rain pressure
loads (He et al., 2019). Meanwhile, recent studies declare that the rainfall and tower response. The results revealed that the raindrop size distri-
effect has also great influence on tower collapse (Fu et al., 2019b; Li et al., bution has a considerable effect on the rain pressure distribution and its
2013). Thus, it is necessary to perform a fragility analysis of transmission time interval has a large impact on the total rain pressure. Fu and Li
towers under combined wind and rain loads. (2016) presented a method for calculating the wind and rain loads acting
During typhoon or hurricane landfall, strong winds are usually on a transmission conductor and performed a dynamic analysis of a
accompanied by continuous rainfall. To clearly reveal the collapse tower-line system under wind and rain loads. The results demonstrated

* Corresponding author.
E-mail address: [email protected] (G. Li).

https://doi.org/10.1016/j.jweia.2020.104098
Received 8 August 2019; Received in revised form 11 January 2020; Accepted 11 January 2020
Available online xxxx
0167-6105/© 2020 Published by Elsevier Ltd.
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

that the maximum percentage of average displacement induced by the 2. Fragility surface of transmission towers under combined wind
rain loads relative to the wind loads can reach 22.00%, indicating that and rain loads
the rain load has a significant effect on the response of tower-line sys-
tems. Currently, a complete set of rain load models for transmission lines, 2.1. Definition of fragility surface
which can be used in the dynamic analysis of transmission tower-line
systems, has been established preliminarily. Fragility of a system is the probability that the system response to
Previous investigations on the strength capacity estimation of trans- external loads of specified intensities exceeds a critical value (Kafali and
mission towers often ignored the parametric uncertainties. They adopted Grigoriu, 2007). Fragility curves represent a relationship between this
standard values, such as fixed material properties and geometric di- exceedance probability and a scalar measure, and the probability of
mensions (Tian et al., 2016; Wang et al., 2018; Zhang et al., 2013). system failure as a function of two measures.
However, the structural parameters and external loads are variable in The regression fit method based on a probabilistic demand analysis
practice (Dolsek, 2009). Fu et al. (2016) calculated the fragility curves for (PDA), which has the advantage of high calculation efficiency, is used to
transmission towers under a wind load with the uncertainty of wind only. fit the fragility surface in this study. The relationship between median
Park et al. (2016) developed seismic fragility curves of high-voltage engineering demand parameter (EDP) and the selected IM yields a linear
transmission towers using a deterministic structural model. Fu and Li model (Nielson and DesRoches, 2007):
(2018b) presented an uncertainty analysis method for tower structures
subjected to a wind load, and a static nonlinear buckling analysis was lnðEDPÞ ¼ a lnðIMÞ þ ln b (1)
conducted using a finite element method. Then, the fragility curves of a For the fragility curve, there is only one IM, while the fragility surface
transmission tower were obtained considering the uncertainties of has two different IMs, namely, IM1 and IM2. Thus, an alternative IM
structural parameters (Fu et al., 2019a). To improve the computation based on two IMs, e.g., basic wind speed and rain intensity is proposed:
efficiency, Cai et al. (2019) established a Kriging surrogate model for the
tower capacity surface as a function of uncertain structural parameters. lnðEDPÞ ¼ a lnðIM1 Þ þ b lnðIM2 Þ þ c (2)
The model was used to estimate the strength capacity of each sample, and
then a Monte Carlo simulation was carried out to obtain the fragility where a, b, and c are the regression coefficients; IM represents the vector
curves. (IM1, IM2). By assuming that the EDP at a certain IM coincides with the
Wind speed and rain intensity are often used to measure the intensity lognormal probabilistic distribution, the conditional probability for a
of wind and rainfall loads. However, they are unsuitable for fragility transmission tower reaching the damage LS under a certain IM can be
analysis because only one variable is required in the fragility curve to developed as follows:
estimate the strength capacity. When conducting a structural perfor-

lnðEDPÞ  lnðLSÞ
mance evaluation with two intensity measures (IMs), the fragility surface P½EDP  LSjIM ¼ Φ (3)
is the best choice. The concept of fragility surface originated from βEDPjIM
earthquake engineering (Kafali and Grigoriu, 2007), while its application vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
in the field of wind engineering has not been reported so far. Kafali and uP n
u ½ln EDPi  ða lnðIM1i Þ þ b lnðIM2i Þ þ cÞ2
Grigoriu (2007) investigated the fragility surfaces of nonlinear systems t
βEDPjIM ¼ i¼1
(4)
based on two parameters, namely, the earthquake moment magnitude n2
and the distance from the seismic source to the system site. Seyedi et al.
(2009) proposed a complete methodology for building fragility surfaces where Φð⋅Þ denotes the standard normal distribution function, and
based on the damage calculation through nonlinear numerical analysis. βEDPjIM is the lognormal standard deviation of the EDP. By substituting
The results showed that an increase from one to two ground-motion Eqs. (2) and (4) into Eq. (3), the fragility surface can be obtained.
parameters leads to a significant reduction in the scatter in the fragility
analysis and allows the uncertainties related to the effect of the second 2.2. Wind and rain loads
ground-motion parameter to be accounted for within risk assessments.
Guo et al. (2016) obtained the fragility surfaces of a bridge at various For a transmission tower-line system, the rain load acting on a
damage states, which were conditional on both bridge service time and transmission tower is composed of the raindrop impinging force, whereas
seismic intensity. the rain load of conductors mainly reflects the change in aerodynamic
Based on a review, it was found that previous studies about the force. Thus, the rain loads for different components have different acting
fragility analysis of transmission towers under wind loads are limited. mechanisms and calculation methods (Fu et al., 2019b). Notably, in this
Although many transmission towers collapsed during typhoons or hur- study, we are more concerned with the wind speed and rainfall during a
ricanes, the effect of rainfall on the fragility has not been considered until strong typhoon or hurricane. Therefore, the possible ice formation
the present. Thus, it is necessary to conduct a fragility analysis of trans- caused by freezing rain is ignored and conductor galloping is not
mission towers under combined wind and rain loads, where the un- considered.
certainties of both structural parameters and external loads are
considered. Before conducting fragility analysis, the coupling IM based 2.2.1. Rain load of the tower body
on basic wind speed and rain intensity should be proposed firstly. The profile of mean wind speed along the altitude uses the power law
Additionally, the quick estimation of collapse state for transmission as
towers under multi-hazard needs to be solved. In Section 2, the calcu-
α
lation method of wind and rain loads for a tower-line system is intro- H
Va ¼ V10 (5)
duced and the process of constructing the fragility surface of tower 10
structures under wind and rain loads is developed. Section 3 presents the
determination of uncertain parameters and limit state (LS) indices. where V10 is the basic wind speed representing the mean wind speed over
Moreover, the simulation of dynamic nonlinear responses to obtain the 10 min at an altitude of 10 m, H is the altitude, and α is the power law
fragility surfaces is discussed including the influence of wind attack exponent.
angle. In Section 4, the concept of critical collapse surface, which is used The wind loads can be simulated by
to quickly estimate the collapse state of transmission tower-line systems 
Fa ¼ μs Aρa Va 2 2 (6)
under wind and rain loads, is proposed. Section 5 concludes the study.

2
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Table 1 influence on the drag coefficient have been neglected during the dynamic
Attached rainfall mass and its proportion at various water film thicknesses. analysis. To illustrate this issue quantitatively, the rainfall mass and its
Average water film thickness Attached rainfall mass Rainfall mass/Tower proportion to the total mass of the tower were estimated. At first, the
(mm) (kg) mass surface area of the entire tower employed in Section 3 was calculated,
0.05 22.91 0.096% with the value of 916.5 m2, and the windward area is about 916.5/2 ¼
0.1 45.82 0.19% 458.25 m2. The tower mass is 23,984 kg, and then Table 1 lists the
0.2 91.65 0.38% attached rainfall mass and its proportion at various water film thick-
0.5 229.12 0.95% nesses. It can be seen that the proportion of rainfall mass is very small,
which implies that the rainfall mass has limited influence on the dynamic
response of the transmission tower.

2.2.2. Aerodynamic force of the transmission line

In addition to the raindrop impinging force, the rainfall has also an
important effect on the aerodynamic characteristics of the transmission
conductor (ESDU, 1980; Fu and Li, 2016). The total horizontal and ver-
tical loads, Ftotal h and Ftotal v , respectively, acting on a transmission
conductor under wind and rain excitations can be calculated by (Fu and
Li, 2016)
Z ∞
ρa V 2a
Ftotal h ¼ k ρw SðγVa ; RÞnðD; RÞγ 3 Va 3 D3 dD þ CD ðVa ; RÞA (8)
0 2
Z ∞
Ftotal v ¼ kρw SðVterm ðDÞ; RÞnðD; RÞVterm 3 ðDÞD3 AdD (9)
0

Fig. 1. Cross-section of LP-810 (Kikuchi et al., 2002).   

Vterm ¼ 9:40  1  exp 0:557  D1:15 (10)

where ρa is the air density equal to 1.235 kg/m3, while μs and A are the where CD ðVa ; RÞ is the drag coefficient of the transmission conductor
drag coefficient and the windward projected area, respectively. μs can be under both wind and rain excitations, R is the rain intensity in mm/h, and
determined based on standards (ASCE-No.74, 2010; DL/T5154-2012, Vterm ðDÞ denotes the vertical terminal velocity of a raindrop with diam-
2012; IEC60826, 2003). eter D.
Fu et al. (2015a) proposed a hybrid method for calculating rain load So far, the measured data of CD ðVa ; RÞ are limited, and only Kikuchi
based on a single-raindrop impingement experiment, followed by a wind et al. (2003) has measured the aerodynamic force of a newly designed
tunnel test conducted to validate the effectiveness of the method. To conductor LP-810 under combined wind and rainfall excitations. The
avoid repetition, these detailed validations were not included in this cross-section of LP-810, which has a diameter of 37.2 mm and a mass of
paper, and concerned readers can refer to Fu et al. (2015a). The rain load 2.761 kg/m, is plotted in Fig. 1, and the experimental results as shown in
of the tower body for a specified rain intensity and wind speed yields Fig. 2 are used in this study.
Z ∞ Only a limited number of drag coefficients have been measured by
Fr ¼ 2 kρw SðγVa ; RÞnðD; RÞγ 3 Va 3 D3 AdD (7) Kikuchi et al. (2002), and the drag coefficients under other rain in-
0
tensities and wind speeds are still required to perform the fragility
where the detailed descriptions of all the above parameters can be found analysis. To fill this gap, a simple interpolation method is proposed. In
in Fu et al. (2015a). Fig. 2(a), when the wind speed is fixed, the drag coefficients under the
Most tower members are in vertical or inclined position, and it is three rain intensities can be obtained via linear interpolation. Then, a
difficult for rainfall to accumulate on the member surface. The water film quadratic polynomial is used to fit the drag coefficients at various rain
attached on the tower surface is very thin and the water density is much intensities. As can be seen in Fig. 2(b), for a specified wind speed of 40
smaller than that of steel. Therefore, the attached raindrop mass and its m/s, the measured drag coefficients agree well with the fitting curve,
indicating that the proposed interpolation method is feasible.

Fig. 2. Drag coefficients of LP-810 under combined wind and rainfall excitations.

3
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 3. Process of constructing the fragility surface of tower structures under combined wind and rain loads.

2.3. Developed fragility surface for tower structures combined wind and rainfall excitations, with the detailed process illus-
trated in Fig. 3.
The current research indicates that the dynamic responses of trans- The process consists of four main steps, namely, establishing uncer-
mission tower-line systems under wind and rain loads are very remark- tainty finite element models (FEMs), generating wind and rain loads for
able, and the rainfall effect has great contribution as well (Li et al., 2013; each IM group, nonlinear dynamic analysis, and regression analysis.
Wang et al., 2013). Consequently, it is necessary to consider the rainfall There are many uncertain variables for both structural model and envi-
effect when evaluating the collapse state of transmission towers. In ronmental loading. When establishing uncertainty numerical models, all
addition, nearly all the structural parameters are uncertain, conforming the known uncertain parameters should be considered.
to a particular probability distribution. Even for the same batch of steel, The second step is generating wind and rain loads for each IM group,
the yield strength is not exactly consistent. To meet the requirement of which is a little different from the conventional process of calculating
reliability, a 0.05 fractile of the probability distribution for material fragility curves. During this study, the basic wind speed V10 and rainfall
strength is used to determine the standard value during design intensity R are selected as the IMs, and the tower tip displacement d0 is
(GB50068-2001, 2001). In the structural performance evaluation, the selected as the EDP. For the fragility curve under wind loading, the wind
fragility analysis method is commonly used to consider the uncertainties speed is divided into various small ranges to produce a wind speed dis-
of both resistance and action effects. The conventional fragility curve has tribution over a wide and uniform range. For the fragility surface, there
only one IM and cannot satisfy the coupling analysis of wind and rain are two IMs, and the wind speed and rain intensity should be combined
loads (Seyedi et al., 2009). Thus, this study developed a fragility surface totally for each IM group. If the wind speed and rain intensity are divided
construction method for transmission tower-line systems under into n1 and n2 ranges, respectively, the total combinations for each group

Fig. 4. FEM of the employed transmission tower-line system.

4
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

can reach n1  n2, increasing in multiples. No obvious correlations can be Table 2

found between wind speed and rain intensity (Ni et al., 2007), and thus, Probability distributions of the selected uncertain parameters.
this has not been considered in the fragility analysis. Uncertainty Average Coefficient of Distribution Reference
The third step is the nonlinear dynamic analysis, and one uncertainty source value variance type
numerical model corresponds to one IM group. If the number of uncer- Elastic — 0.03 Lognormal Fu and Li
tainty tower models is m, the number of IM groups will also be m, and the modulus (2018b)
nonlinear dynamic analysis should be performed m  n1  n2 times, of Poisson’s 0.3 0.03 Lognormal Fu and Li
which the calculation amount is very huge, taking up most of the time ratio (2018b)
Yield strength — 0.07 Lognormal Fu and Li
among the four steps. Based on the above calculation results and (2018b)
regression analysis, the fragility surface can be obtained. Damping 2% 0.3 Normal Bezabeh et al.
ratio (2018)
3. Case study Drag — 0.05 Normal Minciarelli
coefficient et al. (2001)
Web 0.985  0.032 Normal Chen and Li
A real operational 220 kV transmission line located in Guangdong thickness Standard (1985)
Province, China is used in this study. The exposure category of the value
employed transmission line is type B based on Chinese standard, which Width 1.001  0.008 Normal Chen and Li
has a power-law exponent of 0.15. This category in Chinese standard Standard (1985)
value
includes farms, countryside, jungles, hills, and towns with sparse houses,
which is similar to Exposure C in American standard and category II in Note: “—” indicates that the average values for different materials or shapes are
both European and Japanese standards. The tower height is 68.5 m and different, and thus, no fixed value is assigned.
the tower members consist of steel angles, with the main members of
Q345-type steel and auxiliary members of Q235-type steel. The modulus
of elasticity for steel angles is 206 GPa, and the standard values of yield
strength for the Q345- and Q235-type steel angles are 345 MPa and 235
MPa, respectively. The two adjacent towers are tension supports, and the
span lengths are 843 m and 1061 m, respectively, which are extremely
long. The FEM of the transmission tower-line system is then established
using the ANSYS software, as shown in Fig. 4. The BEAM188 element,
which is a type of fiber element model, supporting plasticity, material
and geometry nonlinearities, is used to simulate the tower member,
whereas the LINK180 element is used to simulate both the transmission
line and insulator. The bilinear isotropic hardening plasticity model is
used to simulate the constitutive model of the steel material. Both the
material and geometric nonlinearities are considered during the simu-
lation. To accurately simulate the member stress and judge the failure
state, the main members and diagonal members were divided into 10 and
3 elements, respectively. Due to that most of the tower failure can be
attributed to member buckling, the joint failure was not considered in
this study. Once the joint distance of a tower member changes dramati-
cally or a cross-section enters the plastic stage mostly, the tower member
is judged to fail.
Fig. 5. Statistical results of yield strength at various sample sizes.
3.1. Uncertain parameters
current research papers, all the uncertain parameters used and corre-
The key issue of fragility analysis is to include the uncertain param- sponding statistical properties are summarized as given in Table 2
eters comprehensively. While many sources of uncertainty exist, they are (Bezabeh et al., 2018; JCSS, 2001).
generally categorized as either aleatory or epistemic (Kiureghian and Random samples are then generated based on the technique of Latin
Ditlevsen, 2009). Aleatory uncertainty is caused by the inherent hypercube sampling (LHS) (Dolsek, 2009). LHS is a selective sampling
randomness of the uncertain parameters, including the uncertainties of technique in which, for a desirable accuracy level, the sample size is
loading, section dimension, material strength, and constitutive relation, significantly smaller than that in the direct Monte Carlo simulation
and this type of uncertainty is objective and inevitable. Epistemic un- (Thomos and Trezos, 2006). Taking the yield strength of Q345-type steel
certainty is the model uncertainty and statistical uncertainty caused by as an example, the mean values and standard deviations at various
the subjective limitations of researchers and the lack of current knowl- sample sizes are calculated as shown in Fig. 5. Considering both the ac-
edge. Such uncertainty can be gradually reduced with the accumulation curacy and time consumption, a sample size of 20 is used.
of knowledge of researchers and the increase in sample size.
In previous uncertainty analyses of transmission towers, only the
uncertainties of material properties and section dimensions were 3.2. Limit state index
considered (Fu and Li, 2018b; Fu et al, 2019a, 2019b). Because only the
static analysis was performed, it is reasonable to ignore the uncertainty of For the PDA method, only a limited discrete values of IMs, corre-
damping ratio. However, in this study, a large number of nonlinear dy- sponding to the IM groups shown in Fig. 3, have been selected to carry
namic analyses should be carried out, and damping ratio is a crucial out the nonlinear dynamic analysis. The results of each selected IM group
factor that must be considered. The drag coefficient is also uncertain and may not match exactly with the LSs. Thus, the nonlinear dynamic anal-
plays a decisive role in the wind loading. The span length of the trans- ysis can not be used to determine the LS indices for the PDA method. To
mission line is usually very long, and the average values of diameter and obtain the LS index, a static nonlinear pushover analysis of the deter-
density vary only slightly; therefore, the uncertainties of diameter and ministic FEM is conducted. The wind loads with various basic wind
density for the transmission line are ignored in this study. Based on speeds are calculated, and the static non-linear responses, with a

5
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 8. Simulated wind speed of the highest point.

Fig. 6. Pushover curve of the deterministic tower model.

transmission tower-line system and the different components have
different wind loads, which is almost impossible to simulate at the cur-
Table 3 rent level of technology. Therefore, the wind load is simplified into a
LS indices dLS for the transmission tower at various wind attack angles (unit: m). concentrated load acting on the nodes of the numerical model (Fu et al.,
2016; Zhang et al., 2013). The simplified points of the generated wind
Wind attack Damage state
angle
speeds are shown in Fig. 7.
Slight (50%  Moderate (75%  Collapse The harmonic superposition method is used to generate fluctuating
dCollapse) dCollapse)
wind speeds with the Davenport spectrum (Zhang et al., 2013). The total
0
0.3645 0.54675 0.729 time for each time series is 600 s and the time interval adopted is 0.1 s.
22.5
0.273 0.4095 0.546
The simulated cutoff frequency is 5 Hz, and the number of divided fre-
45
0.215 0.3225 0.430
67.5
0.223 0.3345 0.446 quencies is 1024. The simulated wind speed of the highest point, which
90
0.227 0.3405 0.454 has a basic wind speed of 10 m/s, is plotted in Fig. 8. The wind spectrum
is also calculated to compare with the target Davenport spectrum, as
shown in Fig. 9, which shows good consistency. The spatial correlation
gradually increasing basic wind speed until the tower collapses, are then coefficients of the simulated fluctuating wind speeds along the left bot-
calculated. Fig. 6 plots the curves between tower tip displacement and tom conductor span are calculated, as shown in Fig. 10, to compare with
basic wind speed with the wind direction perpendicular to the trans- the target formula. It can be observed that the simulated fluctuating wind
mission line, corresponding to a wind attack angle of 90
. It is assumed speeds satisfy the demand of spatial correlation.
that the tower will collapse once the displacement reaches or exceeds the Assuming that wind direction is perpendicular to the transmission
buckling point (Fu and Li, 2018b). In Fig. 6, the displacements at the line, the IM groups of the wind and rain loads are then generated, fol-
buckling points are 0.454 m and 0.450 m, which are very close, with a lowed by a nonlinear dynamic analysis. The observation point for the
relative error of (0.454–0.450)/0.454 ¼ 0.9%. Consequently, only the tower tip displacement during the dynamic analysis is marked in Fig. 11.
result under wind loading is selected as the LS index of the complete With the simulated dynamic responses under different combinations
collapse state. The slight and moderate damage LSs are also determined of wind speed and rain intensity, all the parameters of Eq. (2) can be
as given in Table 3 (Fu et al., 2019a). fitted through a regression analysis as shown in Fig. 12. The goodness of
fit reaches 0.9197, indicating that the dynamic simulation results
conform very well to a binary first-order polynomial. Thus, Eq. (2) can be
3.3. Performance evaluation
written as
3.3.1. Fragility analysis lnðEDPÞ ¼ 1:747 lnðV10 Þ þ 0:03265 lnðRÞ  6:593 (11)
In reality, the approaching wind interacts on the surface of the

Fig. 7. Simplified points of generated wind speeds.

6
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 9. Comparison between simulated and target wind spectra. Fig. 11. Observation point of the tower tip displacement.

Substituting Eq. (11) and the LS index in Table 3 into Eq. (3), the
expressions of fragility surface for different LSs yield


 1:747lnðV10 Þþ0:03265lnðRÞ6:593lnð0:227Þ
P EDPLSSlight jIM ¼Φ
0:07342
(12)


1:747lnðV10 Þþ0:03265lnðRÞ6:593lnð0:3405Þ
P½EDPLSModerate jIM¼Φ
0:07342
(13)


 1:747lnðV10 Þþ0:03265lnðRÞ6:593lnð0:454Þ
P EDPLSCollapse jIM ¼Φ
0:07342
(14)
The fragility surface of Eq. (14) for the collapse LS is then plotted as
shown in Fig. 13. Generally, the wind speed has greater influence on the
Fig. 12. Regression analysis of the tower tip displacement vs. in-
fragility probability than the rain intensity. For lighter rain intensities,
tensity measures.
the fragility surface changes abruptly. The fragility curves at various rain
intensities are then captured to further determine the influence of rain
station, China from September 22, 2008 at 21:00 UTC to September 26,
load, as shown in Fig. 14. Obviously, the fragility curves move to the left
2008 at 20:00 UTC, have been adopted. The sample interval is 1 min, and
with increasing rain intensity. The moving distance is relatively large for
to be consistent with the fragility surface, the measured wind speed data
light rain intensities, while for heavy rain intensities larger than 100
are converted into 10-min average values, as shown in Fig. 15. The
mm/h, the fragility curves are very close. The fragility surfaces for slight
maximum rain intensity reaches 150 mm/h, occurring under a low wind
and moderate LSs are very similar to the results of the collapse LS; thus,
speed. The maximum basic wind speed and maximum rainfall intensity
they are not demonstrated here.
did not occur synchronously.
To further investigate the failure probability of the transmission line
Fig. 16 plots the scatter diagram of the basic wind speed and corre-
under multiple hazards of wind and rainfall excitations, the meteoro-
sponding rain intensity. It can be observed that most rainfalls are located
logical data of typhoon Hagupit recorded by the Dianbai meteorological
in low wind speeds of less than 5 m/s, and no obvious trend can be found.

Fig. 10. Spatial correlation coefficients along the left bottom conductor span.

7
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 13. Fragility surface of the transmission tower under both wind and rainfall excitations.

Fig. 14. Fragility curves of the transmission tower at various rain intensities.

Fig. 15. Meteorological data of typhoon Hagupit.

By combining the measured meteorological data of typhoon Hagupit 5.04e-12, indicating that the transmission line is very safe. Most rainfalls
and simulated fragility surface in Fig. 13, the failure probabilities of the occurred with low wind speeds, which have negligible influence on the
transmission line at each moment are calculated, as shown in Fig. 17. maximum failure probability. Assuming that the rain intensity maintains
Because the recorded wind speeds during typhoon Hagupit are not very a constant value of 100 mm/h during typhoon Hagupit, the failure
strong, the maximum failure probability of the transmission line is only probabilities are then calculated as shown in Fig. 18. The maximum

8
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 16. Correlation between basic wind speed and rain intensity. Fig. 19. Collapse basic wind speeds for various rain intensities.

The basic wind speed corresponding to the probability of 10% is

defined as the collapse basic wind speed (Fu et al., 2016). Fig. 19 pro-
vides the collapse basic wind speeds at various rain intensities. With
increasing rain intensity, it can be observed that the collapse basic wind
speed decreases abruptly at first and then gently. The collapse basic wind
speeds with rain intensities of 0 and 240 mm/h are 27.4 m/s and 23.7
m/s, respectively. The difference is relatively large, indicating that the
rainfall effect has a significant impact on the strength capacity of the
transmission tower-line system. In Guadeloupe, Caribbean, 38.1 mm of
rain fell in 1 min in 1970, which corresponds to 2286 mm/h; it was the
heaviest rainfall ever recorded. In Xi’an, China, 59.1 mm of rain fell in 5
min in 1973, which corresponds to 709.2 mm/h (Fu et al., 2015b). Based
on the two cases, it can be observed that the transient rain intensity can
be very heavy, and the rain intensity of 240 mm/h can occur naturally in
extreme conditions. Thus, the contribution of rain load on the tower
collapse cannot be neglected and should be paid more attention.

3.3.2. Influence of wind attack angle

For a transmission tower, the wind may come from any direction, and
five wind attack angles of 0
, 22.5
, 45
, 67.5
, and 90
are selected due
Fig. 17. Failure probabilities of the transmission line under typhoon Hagupit.
to the biaxial symmetry, as defined in Fig. 20. A detailed description of
calculating the wind loads for each wind attack angle can be found in
relevant standards; thus, it is not given here (ASCE-No.74, 2010;
IEC60826, 2003). The LS indices dLS of the transmission tower at various
wind attack angles are also calculated as given in Table 3. It can be
observed that the ultimate deformation capacities of the employed
transmission tower at different wind attack angles are different, and the
critical collapse displacement at the wind attack angle of 0
is the largest.
Based on the presented process of constructing the fragility surface,
the nonlinear dynamic responses are evaluated, and the regression co-
efficients at various wind attack angles are then derived as listed in
Table 4.
Consequently, the fragility surfaces for different wind attack angles
are obtained as shown in Fig. 21. The shapes of all these fragility surfaces
are very similar but the ranges of basic wind speed vary widely, indi-
cating that the wind attack angle has great influence on the fragility
surface and the most unfavorable wind attack angle is 90
. In addition, it
can be concluded that the wind load depends on not only the wind speed
but also the wind direction.

Fig. 18. Failure probabilities with a constant rain intensity of 100 mm/h. 4. Quick estimation of the collapse state for tower-line systems
under combined wind and rain loads
failure probability reaches 25.41%, demonstrating that the transmission
line is in extreme danger. It can be found that if a heavier rain intensity
There are two coupling IMs of wind and rain loads, namely wind
and a stronger wind speed occur synchronously, the failure probability
speed and rain intensity, and the number of load parameters will be three
will increase significantly.

9
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 20. Definition of wind attack angle for a transmission tower-line system.

excitations. For a severe gale and thunderstorm event attacking the

Table 4 employed transmission line, the coordinate points of basic wind speed,
Regression coefficients of fragility surface at various wind attack angles.
corresponding wind direction, and rain intensity can be plotted in
Wind attack angle Parameter Fig. 23. If all the coordinate points are below the critical collapse surface,
a b c βEDPjIM the transmission tower-line system is safe. However, if one or some of the


coordinate points are above the critical collapse surface, the employed
0 2.360 0.02756 11.04 0.1074
22.5
1.676 0.03901 8.194 0.08149
transmission tower will be in danger and may collapse under this severe
45
1.606 0.04102 6.864 0.07924 meteorological condition. The concept of critical collapse surface pro-
67.5
1.582 0.04147 6.299 0.07797 vides a quick and convenient tool for judging the structural state, and it
90
1.747 0.03265 6.593 0.07342 can also be used to initiate an early warning.

if the wind direction is considered. The related issue is how to quickly 5. Summary and conclusions
estimate the collapse state of a transmission line under combined wind
and rain excitations with three load parameters. To fill this gap, the In this study, a process of constructing the fragility surface of tower
concept of critical collapse surface is proposed in this study, and the structures under combined wind and rain loads is developed, and an
calculation process is described as follows: alternative IM based on basic wind speed and rain intensity is proposed
to satisfy the requirement of fragility analysis with two IMs. Then, a real
(1) Calculate the fragility surface of the transmission line at various operational transmission line is employed to simulate the fragility sur-
wind attack angles. face, and the meteorological data of typhoon Hagupit are used to
(2) Cut the fragility surface with a 10% probability horizontal plane; calculate the failure probabilities under combined wind and rainfall ex-
the intersecting curve is the critical collapse curve. citations. The influence of the wind attack angle is also discussed. Finally,
(3) Obtain the critical collapse surface by plotting the intersecting the concept of critical collapse surface, which provides a quick and
curves at various wind attack angles in one figure. convenient tool for judging the structural state, is proposed. The main
conclusions drawn from the fragility analysis can be summarized as
When the coordinate point of wind speed, wind direction, and rain follows:
intensity is located below the critical collapse surface, the structure is
safe; when the coordinate point is in the critical collapse surface, it is in (1) A sample size of 20 can provide enough accuracy to generate
critical state; when the coordinate point is above the critical collapse random samples for the uncertainty tower model, and the rela-
surface, the structure will collapse. For a real operational transmission tionship between EDP and the selected IMs (wind speed and rain
line or other types of structures, it can be observed that the concept of intensity) yields a linear model.
critical collapse surface is a convenient tool for quickly judging the (2) It can be observed from the fragility surface that the wind speed
structural state with multiple load parameters. has greater influence on the fragility probability than the rain
In Eq. (3), when the probability is 10%, the relationship between intensity. For lighter rain intensities, the fragility surface changes
wind speed and rain intensity, which is defined as the critical collapse abruptly, which should be paid more attention.
curve, yields (3) The fragility curves at various rain intensities move to the left with
increasing rain intensity. The moving distance is relatively large
a lnðV10 Þ þ b lnðRÞ þ c  lnðLSÞ  βEDPjIM Φ1 ð10%Þ ¼ 0 (15) for light rain intensities, while for heavy rain intensities larger
than 100 mm/h, the fragility curves are very close.
Substituting the parameters in Tables 3 and 4 into Eq. (15), the crit-
(4) The maximum failure probability of the transmission line during
ical collapse curves for various wind directions are then obtained as
typhoon Hagupit is only 5.04e-12. Assuming a fixed rain intensity
shown in Fig. 22. The basic wind speeds of all the five curves decrease
of 100 mm/h during typhoon Hagupit, the maximum failure
with increasing rain intensity, and the critical collapse curve moves down
probability can reach 25.41%, indicating that the failure proba-
with increasing wind attack angle.
bility will increase significantly if heavier rain intensities and
Finally, the critical collapse surface of the employed transmission
stronger wind speeds occur synchronously.
tower under combined wind and rain loads is derived as shown in Fig. 23.
(5) With increasing rain intensity, the collapse basic wind speeds
It can be observed that the surface is swept by the critical collapse curve
decrease sharply at first and then gently. The collapse basic wind
in Fig. 22. Because only five wind attack angles are selected, the swept
speeds with the rain intensities of 0 and 240 mm/h are 27.4 m/s
critical collapse surface is a little unsmooth along the dimension of the
and 23.7 m/s, respectively. The difference is relatively large,
wind attack angle. However, it can still be used to judge the collapse state
indicating that the rainfall effect has a significant impact on the
of the employed transmission tower under combined wind and rainfall
strength capacity of transmission tower-line systems.

10
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 21. Fragility surfaces of the employed transmission tower at various wind attack angles.

Fig. 22. Critical collapse curves for various wind directions.

The simulated critical collapse surface presented in Section 4 is not For wind and rain loads with multiple IMs, the fragility curve is no
very smooth. This is because the calculation amount for each wind attack longer applicable. The fragility surface can represent two-dimensional
angle is very great and only the fragility surfaces with five selected wind information, which is superior to the fragility curve. The presented
attack angles had been calculated. In the future, the divided wind attack concept of critical collapse surface can be used not only for combined
angles can be refined to develop a smooth critical collapse surface based wind and rain loads, but also for other coupling excitations. In addition,
on the requirement of practical engineering. only the LS of collapse has been considered in this study, but the critical

11
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Fig. 23. Critical collapse surface of the employed transmission tower under combined wind and rain loads.

collapse surfaces corresponding to the LSs of slight and moderate dam- Fu, X., Li, H.-N., 2018a. Effect of raindrop size distribution on rain load and its mechanism
in analysis of transmission towers. Int. J. Struct. Stab. Dyn. 18, 1850115.
ages can also be calculated for setting the monitoring threshold in the
Fu, X., Li, H.-N., 2018b. Uncertainty analysis of the strength capacity and failure path for
next step. a transmission tower under a wind load. J. Wind Eng. Ind. Aerodyn. 173, 147–155.
Fu, X., Li, H.-N., Li, G., 2016. Fragility analysis and estimation of collapse status for
transmission tower subjected to wind and rain loads. Struct. Saf. 58, 1–10.
Author contributions section
Fu, X., Li, H.-N., Tian, L., Wang, J., Cheng, H., 2019a. Fragility analysis of a transmission
line subjected to wind loading. J. Perform. Constr. Facil. 33, 04019044.
Xing Fu: Investigation, Methodology, Data Curation, Writing - Orig- Fu, X., Li, H.-N., Wang, J., 2019b. Failure analysis of a transmission tower subjected to
inal draft preparation, Hong-Nan Li: Supervision, Writing- Reviewing and combined wind and rainfall excitations. Struct. Des. Tall Special Build. 28, e1615.
Fu, X., Li, H.-N., Yang, Y.B., 2015a. Calculation of rain load based on single raindrop
Editing, Gang Li: Conceptualization, Writing- Reviewing and Editing, impinging experiment and applications. J. Wind Eng. Ind. Aerodyn. 147, 85–94.
Zhi-Qian Dong: Visualization, Formal analysis. Fu, X., Li, H.-N., Yi, T.-H., 2015b. Research on motion of wind-driven rain and rain load
acting on transmission tower. J. Wind Eng. Ind. Aerodyn. 139, 27–36.
GB50068-2001, 2001. Unifide Standard for Reliability Design of Building Structures.
Declaration of competing interest China Architecture & Building Press, Beijing.
Guo, X., Wu, Y., Guo, Y., 2016. Time-dependent seismic fragility analysis of bridge
The authors declare that they have no known competing financial systems under scour hazard and earthquake loads. Eng. Struct. 121, 52–60.
He, B., Zhao, M., Feng, W., Xiu, Y., Wang, Y., Feng, L., Qin, Y., Wang, C., 2019. A method
interests or personal relationships that could have appeared to influence for analyzing stability of tower-line system under strong winds. Adv. Eng. Software
the work reported in this paper. 127, 1–7.
IEC60826, 2003. Design Criteria of Overhead Transmission Lines. International
Electrotechnical Commission, HIS, Switzerland.
Acknowledgements JCSS, 2001. Probabilistic Model Code-Ppart 3 - Material Properties. JCSS.
Kafali, C., Grigoriu, M., 2007. Seismic fragility analysis: application to simple linear and
This research was supported by the National Natural Science Foun- nonlinear systems. Earthq. Eng. Struct. Dyn. 36, 1885–1900.
Kikuchi, N., Matsuzaki, Y., Banse, H., Kaneko, T., Yukino, A., Ishida, H., 2002.
dation of China (grant no. 51708089), the China Postdoctoral Science Development of conductors with reduced wind drag and wind noise for overhead
Foundation (grant nos. 2017M620101 and 2019T120207) and the power transmission lines. Furukawa Rev. 21, 50–55.
Fundamental Research Funds for the Central Universities (grant no. Kikuchi, N., Matsuzaki, Y., Yukino, T., Ishida, H., 2003. Aerodynamic drag of new-design
electric power wire in a heavy rainfall and wind. J. Wind Eng. Ind. Aerodyn. 91,
DUT19RC(4)021). 41–51.
Kiureghian, A.D., Ditlevsen, O., 2009. Aleatory or epistemic? Does it matter? Struct. Saf.
References 31, 105–112.
Li, H.-N., Tang, S.-Y., Yi, T.-H., 2013. Wind-rain-induced vibration test and analytical
method of high-voltage transmission tower. Struct. Eng. Mech. 48, 435–453.
ASCE-No.74, 2010. Guidelines for Electrical Transmission Line Structural Loading. ASCE,
Minciarelli, F., Gioffre, M., Grigoriu, M., Simiu, E., 2001. Estimates of extreme wind
USA.
effects and wind load factors: influence of knowledge uncertainties. Probabilistic Eng.
Bezabeh, M.A., Bitsuamlak, G.T., Popovski, M., Tesfamariam, S., 2018. Probabilistic
Mech. 16, 331–340.
serviceability-performance assessment of tall mass-timber buildings subjected to
Ni, Y.Q., Wang, X.Y., Chen, Z.Q., Ko, J.M., 2007. Field observations of rain-wind-induced
stochastic wind loads: Part II - structural reliability analysis. J. Wind Eng. Ind.
cable vibration in cable-stayed Dongting Lake Bridge. J. Wind Eng. Ind. Aerodyn. 95,
Aerodyn. 181, 112–125.
303–328.
Cai, Y., Xie, Q., Xue, S., Hu, L., Kareem, A., 2019. Fragility modelling framework for
Nielson, B.G., DesRoches, R., 2007. Seismic fragility methodology for highway bridges
transmission line towers under winds. Eng. Struct. 191, 686–697.
using a component level approach. Earthq. Eng. Struct. Dyn. 36, 823–839.
Chen, X.-G., Li, J.-H., 1985. Statistical parameters of material strength and geometric
Park, H.-S., Choi, B.H., Kim, J.J., Lee, T.-H., 2016. Seismic performance evaluation of high
properties of shapes for steel members. J. Chongqing Constr. Eng. Coll. 1, 1–23.
voltage transmission towers in South Korea. KSCE J. Civil Eng. 20, 2499–2505.
DL/T5154-2012, 2012. Technical Code for the Design of Tower and Pole Structures of
Seyedi, D.M., Gehl, P., Douglas, J., Davenne, L., Mezher, N., Ghavamian, S., 2009.
Overhead Transmission Line. China Planning Press, Beijing.
Development of seismic fragility surfaces for reinforced concrete buildings by means
Dolsek, M., 2009. Incremental dynamic analysis with consideration of modeling
of nonlinear time-history analysis. Earthq. Eng. Struct. Dyn. 39, 91–108.
uncertainties. Earthq. Eng. Struct. Dyn. 38, 805–825.
Thomos, G.C., Trezos, C.G., 2006. Examination of the probabilistic response of reinforced
ESDU, 1980. Mean Forces, Pressures and Flow Field Velocities for Circular Cylindrical
concrete structures under static non-linear analysis. Eng. Struct. 28, 120–133.
Structures: Single Cylinder with Two-Dimensional Flow. Engineering Sciences Data
Tian, L., Ma, R.-S., Li, H.-N., Wang, Y., 2016. Progressive collapse of power transmission
Unit, London.
tower-line system under extremely strong earthquake excitations. Int. J. Struct. Stab.
Fu, X., Li, H.-N., 2016. Dynamic analysis of transmission tower-line system subjected to
Dyn. 16, 1550030.
wind and rain loads. J. Wind Eng. Ind. Aerodyn. 157, 95–103.

12
X. Fu et al. Journal of Wind Engineering & Industrial Aerodynamics 199 (2020) 104098

Wang, F.Y., Xu, Y.L., Qu, W.L., 2018. Multi-scale failure analysis of transmission towers Zhang, Z.-Q., Li, H.-N., Li, G., Wang, W.-M., Tian, L., 2013. The numerical analysis of
under downburst loading. Int. J. Struct. Stab. Dyn. 18, 1850029. transmission tower-line system wind-induced collapsed performance. Math. Probl.
Wang, Z., Zhao, Y., Li, F., Jiang, J., 2013. Extreme dynamic responses of MW-level wind Eng. 2013, Article ID 413275.
turbine tower in the strong typhoon considering wind-rain loads. Math. Probl. Eng.
13. Article ID 512530.

13