M. Sigit Darmawan et al.

277

The 2nd International Conference on Rehabilitation and Maintenance in Civil Engineering

Time Dependent Reliability Analysis of Steel I Bridge Girder Designed

Based on SNI T-02-2005 and SNI T-3-2005 Subjected to Corrosion

M. Sigit Darmawana*, A. N. Refania, M. Irmawana, R. Bayuajia, and

R. B. Anugrahaa
a
Institut Teknologi Sepuluh Nopember, Indonesia

Abstract
Steel bridge girder located in the vicinity of the sea cannot escape from corrosion. Over time the
effect of corrosion becomes more significant, especially when there is no interference
performed by bridge authority. This study will present the reliability analysis of steel I bridge
girder designed using SNI T-02-2005 and SNI T-3-2005 subjected to corrosion as a results of
chloride attacks. The corrosion model utilized in this study is taken from previous research
available in the literature. The study will consider 15m and 20m spans of bridge girder located
close to the sea, with a designed service life of 50 years. To consider the uncertainty of all
parameters affecting the performance of steel I bridge girder in corrosive environment,
probability approach using Monte Carlo simulation will be employed. Three performance
indicators of steel bridge girder under dead and live loads will be considered (i.e. bending
strength, shear strength and deflection). The framework developed will allow more accurate
service life prediction of steel bridge structures in a chloride environment.
Keywords: corrosion; reliability analysis; steel structure; chloride; service.

1. Introduction
Steel bridges in Indonesia constitute a significant number of bridge populations. In
some cases, steel bridges are preferred than concrete bridges mainly due to their lighter
self-weight and easy of construction. However, in contrast to concrete bridges, steel
bridges are more vulnerable to corrosion, especially when they are built close to marine
environment and there is no additional protection applied to the steel structures.
Especially in Indonesia, where most of its region is surrounded by the sea, structural
framework to predict the performance of steel bridges subjected to corrosion due to
chloride attacks is demanded. This study will present the reliability analysis of steel I
bridge girder subjected to corrosion. The steel bridges are first designed based on SNI
T-02-2005 and SNI T-3-2005, which are the Indonesian bridge loading code and steel
design code respectively. The study will consider 15m and 20m spans of bridge girder,
with a designed service life of 50 years. The strength and serviceability limit states of

* Corresponding author.
E-mail address: msdarmawan@ce.its.ac.id.
278 M. Sigit Darmawan et al.

the girder under the loads will be considered (i.e. bending strength, shear strength and
deflection).

2. Structural Bridge Configuration

The bridge considered in this study is a typical simple span two lane (one way) slab
on steel I girder bridge, which has a span of 15 m and 20 m and a clear roadway width
of 8.4 m. The bridge consists of five steel I girders (see Figure 1) with equal spacing of
1.2 m and a 250 mm thick cast-in-place concrete deck. An-50 mm thick asphalt is used
for the wearing surface.
L
BALOK INTERIOR
PELAT LANTAI
JEMBATAN
JEMBATAN

BALOK INTERIOR
JEMBATAN

Lb Lb Lb Lb

Figure 1. Steel Bridge Configuration

The girder was designed according to SNI T-02-2005 and SNI T-3-2005, assuming
unshored construction and considering composite action between the girder and the
cast-in-place slab. The specified steel yield strength (fy) of the girder is 290 MPa and the
specified concrete strength fc’ is 30 MPa.

3. Bridge Load Modeling

Three types of loads for highway bridges are considered in this study: dead load
(DL) and live load (LL). Dead load is the gravity load due to the self-weight of the
structure and the non-structural element permanently attached to the structure, whereas
live load includes a variety of forces as a result of vehicles moving on the bridge.
Due to the different degrees of variability of structural and non-structural elements and
permanent loads, dead load is divided into three components:
D1 = dead load due to factory made-made element (steel girder)
D2 = dead load due to cast-in-place materials (concrete road deck)
D3 = dead load due to asphalt overlay
The statistical parameters for dead load are given in Table 1 (Nowak et al. 2001). Dn
in Table 1 is the nominal dead load.
Live load is the weight of vehicles, mainly trucks, which has two components: static
and dynamic. The effect of live load on the bridge are influenced by many factors, such
as the span length, truck weight, axle load and configuration, position of the vehicle on
 M. Sigit Darmawan et al. 279

the bridge (transverse and longitudinal), number of vehicles on the bridge (multiple
presence), girder spacing, and the stiffness of structural members Nowak, A.S., Park, C.
And Casas, J.R (2001). Further, live and dynamic loads are time-dependent random
variables. For this study, axle spacing and distribution of axle loads are based on the
standard SNI T-02-2005 and these will be used to determine the peak flexural actions.
Table 1. Statistical parameters for dead load (Nowak et al. 2001).
Type of dead load Mean COV Distribution
D1 1.03 Dn 0.08 Normal
D2 1.05 Dn 0.10 Normal
D3 80 mm 0.30 Normal
According to SNI T-02-2005, live load is determined using either lane load (D) or
truck load (T). The largest effect of these loads on the bridge will be used in the
structural analysis. The lane load comprises of Uniformly Distributed Load (UDL) and
Knife Edge Load (KEL) per notional lane. The KEL is taken as 49 KN/m, whereas the
UDL in KN/m2 is dependent on loaded length (L) and can be determined from
w = 9 KPa, L ≤ 30 m (1)
w = 9 (0.15+15/L) KPa, L > 30 m (2)
See Figure 2 for the description of lane load.

Figure 2. Lane Load Configuration

The truck load is caused by three-axle design truck as shown in Figure 3. Only one
truck is considered in the designed lane along the bridge. The truck is positioned in a
such way to cause the largest effect on the bridge girder.
As there is no published statistical data for bridge live loads in Indonesia, the
statistical data from US is utilized.
280 M. Sigit Darmawan et al.

Figure 3. Truck Load Configuration

4. Deterministic Analysis
To determine steel section to support the bridge loads, a trial and error approach is
used. The final steel section used should satisfy both the strength and serviceability
limit states. The strength limit states considered are flexural and shear strength whereas
serviceability limit states considered is the deflection. The maximum allowable
deflection due to live load is L/800. From the analysis it is found that the serviceability
limit states and not strength limit state determined the final steel section properties.
Table 2 shows the steel section properties of the bridge girders.
Table 2. Steel Section Properties of the Bridge Girders
d b tw tf As Zx Ix L
Steel Section
(cm) (cm) (cm) (cm) (cm2) Cm3 Cm4 (m)
I-700x300x15x28 70 30 1.5 2.8 273.6 7344 237000 15
I-800x400x16x30 80 40 1.6 3 358.4 11430 409950 20

5. Corrosion Model
Corrosion model utilized in this study is initially proposed by Komp (1987). The
model is formulated as
C  At B (3)
where
C = average corrosion penetration in µm
t = number of years
A and B in equation (3) are the parameters determined from the analysis of
experimental data. The values of A and B generally depend on the environment
surrounding the bridge. Albrecht and Naeemi (1984) have summarized corrosion test
results for three different environments: rural, urban, or marine. Average values for A
and B are listed in Table 3 for two types of steel: unprotected carbon steel and
weathering steel. These values were determined from tests of relatively small metal
specimens. Hence, the use of these data in predicting corrosion effect of full-size
 M. Sigit Darmawan et al. 281

bridges need to be done cautiously as some error may be introduced as a result of

component size variation, orientation, and local environment.
Table 3. Corrosion Model Parameters
Unprotected carbon steel Weathering steel
Environment
A B A B
Rural 34.0 0.65 33.3 0.50
Urban 80.2 0.59 50.7 0.57
Marine 70.6 0.79 40.2 0.56
Indonesia in general has a relatively higher rainfall, higher humidity and higher
average temperature than US. In addition, workmanship quality in this country is not as
good as in developed countries such as the US. Thus, due to these combining factors, it
is expected that steel bridges in Indonesia will corrode faster than steel bridges in the
US. As there is no well documented corrosion data available in this country, the
corrosion parameters proposed by Albrecht and Naeemi (1984) will be used in this
study. However, as more relevant corrosion data for Indonesia become available in the
future, the parameters A and B can be modified accordingly.
The effect of corrosion on steel cross section is described in Figure 4. It is assumed
that corrosion occurs on the whole web surface and on the top surface of the bottom
flange over the entire span of the bridge. This corrosion damage was obtained from field
survey in the US and may not be used directly for corrosion condition in Indonesia.

Figure 4. Typical Steel Cross Section Due To Corrosion (Kayser 1988)

6. Limit State Formulation

As mentioned earlier, the limit states considered in this study are flexural and shear
strength and serviceability. The limit state can be represented as
G ui (X)  0 safe (4)

G ui (X)  0 fail (5)

If it is assumed that k load events occur within the time interval (0,TL) at times ti
(i=1,2,..…,k), the cumulative probability of structural failure of service proven
structures anytime during the time interval (0,TL) is
282 M. Sigit Darmawan et al.


p f TL   1  Pr G U  0  G U  .....  G U  0
1 2 k

t 1  t 2  .  t k  TL (6)
where G u (X) is the limit state function at times ti. Hence, the cumulative probability of
i

structural failure is dependent on the prior and updated load S(t) and resistance histories
R(t), see Figure 5. For steel bridge structural members subjected to corrosion, the
corrosion will reduce the structural resistance and so structural resistance is time-
dependent. This represents a first-passage probability that the strength limit state will
take place once during the service life of the structure during the time interval (0,TL). As
shown in Figure 5, the limit state G u (X)  0 when structural failure occurs. Clearly,
i

G ui (X) is a function of load S(t) and resistance R(t).

Figure 5. First Passage Probability

For flexural strength, the limit state can be formulated as follows
G M (X)  Mp(t i )  Mu(t i )
i
(7)
where Mp(ti) is the nominal flexural moment capacity of steel section at time ti and
Mu(ti) is the flexural moment due to ith load effect at time ti, updated on survival of the
previous load events. Both Mp(ti) and Mu(ti) is considered in the middle of the span.
Mp(ti) for composite steel can be determined according to Figure 6, depends on neutral
axis position of the steel section considered (i.e. at concrete slab, steel top flange and
steel web). However, for the steel section considered herein, the neutral axis falls in the
top flange. Thus, only this case will be considered here in detail.
 M. Sigit Darmawan et al. 283

B
0,85 f'c

C
ts Pelat Lantai Beton ts

C'
G.N. d'
t1
fy
e
A1

e' y
t3

T
d

A2

t2
b fy

Keterangan :
A1 = Luas penampang sayap atas profil baja yang mengalami tekan
Figure 6. Neutral axis at steel top flange
A2 = Luas penampang sebagian sayap atas, badan dan sayap bawah profil yang
mengalami tarik
Mp(ti) can be determined from following equation
Mp(t i )  C  e  C'  e' (8)
C  B  t s 0.85f c' (9)
C'  As  fy  C (10)
ts
e y (11)
2
d d'
As  A1
y 2 2 (12)
A2
C'
A1  b  d '  (13)
fy
A2  As  A1 (14)
C'
d'  (15)
fy  b

d'
e  y
'
(16)
2
As = (b . t1) + (b. t2) + ((d-t1-t2).t3) (17)
Asc = (b . t1) + (b. t2c + (t3c. C)) +
((d-t1-t2c).t3c) (18)
t2c = t2 – C (19)
284 M. Sigit Darmawan et al.

t3c = t3 – 2C (20)
where As is area of non-corroded steel section, Asc is the corroded steel section area, t1
and t2 is top and bottom flange thickness respectively, t3 is web thickness, C is the
depth of corrosion penetration determined based on Equation (3), t2c and t3c is bottom
flange and web thickness after corrosion, fc’ is the concrete compressive strength and fy
is the steel yield strength. For corrosion case, replace As with Asc to determine Mp(ti).
For shear strength, the limit state can be formulated as follows
G Vi (X)  Vn (t i )  Vu (t i ) (21)
where Vn(ti) is the nominal shear capacity of steel section at time ti and Vu(ti) is the
shear force due to ith load effect at time ti, updated on survival of the previous load
events. Both Vn(ti) and Vu(ti) is considered in the support region. Vn(ti) can be
determined as
Vn (ti) = 0.6 fy Aw (22)
= 0.6 fy d t3c (23)
where Aw is steel web section area.
For the serviceability limit state, the maximum deflection is of interest (i.e. mid-
section deflection) and this limit state can be formulated as
G Si (X)   allow   mid (t i ) (24)
where Δallow is the allowable deflection and Δmid (ti) is the deflection in the middle of the
span due to ith load effect at time ti updated on survival of the previous load events.
Based on SNI T-02-2005, the maximum deflection of a bridge under live load should
not exceed 1/800 of its span. The cumulative probability of serviceability failure of
service proven structures anytime during the time interval (0,TL) is therefore
 1 2 k
 t 1  t 2  .  t k  TL
p s TL   1  Pr G S  0  G S  .....  G S  0 (25)
 M. Sigit Darmawan et al. 285

Table 4. Statistical Parameters Used in the Probability Analysis

Variables Mean COV Distribution Reference
fy 290 MPa 0.1 Lognormal [22]
f’c 30 MPa 0.15 Lognormal [22]
d (steel section depth) d 0.05 Normal [11]
b (steel section width) b 0.05 Normal [11]
t1 (top flange thickness) t1 0.05 Normal [11]
t2 (bottom flange thickness) t2 0.05 Normal [11]
t3 (web thickness) t3 0.05 Normal [11]
ts (concrete slab thickness) ts 0.067 Normal [3]
Corrosion Parameter A 70,6 μm 0.66 LogNormal [12]
Corrosion Parameter B 0,789 0.02 LogNormal [12]
Model Error for flexure 1.01 0.046 Normal [11]
Model Error for shear 1.15 0.125 Normal [11]
qD, qsw , qUDL (distributed load) 1.05 q 0.1 Normal [16]
PKEL (Knife Edge Load) 1.05 PKEL 0.18 Normal [16]
Asphalt, ta (Asphalt thickness) ta 0.25 Normal [16]

7. Statistical Parameters
Probabilistic analysis is used to take into account the uncertainty of parameters
influencing the performance of steel bridge girder subjected to corrosion. The statistical
parameters of steel beam used in the probabilistic analysis are given in Table 4. This
table shows that the parameters influencing flexural, shear strength and deflection of
steel beam have some uncertainty (random variables). These parameters have
coefficient of variations between 0.05 to 0.25. For corrosion model parameter (A), the
coefficient of variation is also on the high side (e.g. 0.66). Therefore, flexural, shear
strength and deflection determination of steel beams based on deterministic analysis is
inaccurate. Monte Carlo simulation will be used to determine the effect of corrosion on
flexural, shear strength and deflection of steel I beams.

8. Results and Discussions

The results of probabilistic analysis of corrosion effect on steel I bridge girder is
given in Figure 7 to 10 for bridge span of 15 m. Figure 7 shows that corrosion has
reduced the flexural moment capacity (Mn). However, after 50 years of corrosion the
reduction of flexural moment is not significant as the mean flexural moment capacity
only decreases slightly from 212 ton-m to 198 ton-m (7% reduction). Compare with the
moment due to dead and live load (Mu), this mean flexural moment capacity is greater.
Thus, after 50 years of service life no failure due to bending action will be expected as
shown in Figure 8.
286 M. Sigit Darmawan et al.

0.03
L=15 m

Probability Density Function (PDF)

Mn(10)

0.02

Mn(0)
0.01 Mn(50)
Mn(30)

0.00
120 170 220 270
Mn (T) ton-m

Figure 7. Corrosion effect on flexural moment capacity Mn (T) for L=15 m

0.08
L=15 m
0.07
Probability Density Function (PDF)

Mu (due to load)
0.06

0.05

0.04

Mn(10)
0.03
Mn(30)
0.02 Mn(0)
Mn(50)

0.01

0.00
70 120 170 220 270
Mn (T) ton-m

Figure 8. Mu vs. Mn (T)

Figure 9 shows the corrosion effect on shear capacity (Vn). It shows that corrosion
has reduced the shear capacity more significant than flexural moment capacity
reduction. After 50 years of corrosion the mean shear capacity decreases from 190 ton
to 146 ton (25% reduction). Compare with the shear force due to dead and live load
(Vu), this flexural moment capacity is greater. Thus, after 50 years of service life no
failure due to shear action will be expected as shown in Figure 10.
 M. Sigit Darmawan et al. 287

0.03
L=15 m

0.03

Probability Density Function (PDF)

0.02

Vn(30)
0.02
Vn(10)
Vn(50)
0.01

Vn(0)
0.01

0.00
25 75 125 175 225 275 325
Vn (T) ton

Figure 9. Corrosion effect on shear capacity Vn(T) for L=15 m

0.30
L=15 m

0.25
Probability Density Function (PDF)

Vu (due to load)

0.20

0.15

0.10
Vn(30)
Vn(10) Vn(0)
Vn(50)
0.05

0.00
20 70 120 170 220 270
Vn (T) ton

Figure 10. Vu vs. Vn(T)

Figure 11 shows the corrosion effect on central safety factor against flexural
moment and shear. It shows that corrosion has reduced the safety factor against shear
force more significant than against flexural moment. After 50 years of corrosion, the
safety factor decreases from 8.2 to 6 (25% reduction). For the same period of time, the
safety factor against flexural moment only reduce from 2.43 to 2.26 (7% reduction).
288 M. Sigit Darmawan et al.

9.00

L=15 m
8.00
Flexural moment
Shear

Central Safety Factor

7.00

6.00

5.00

4.00

3.00

2.00
0 10 20 30 40 50
Time (Years)

Figure 11. Safety factor reduction

Figure 12 to 16 show the corrosion effect on flexural moment capacity (Mn), shear
capacity (Vn) and central safety factor, respectively for bridge span of 20 m. These
show that corrosion has reduced steel I girder shear capacity greater than its flexural
moment capacity. However, after 50 years of service life no failure will be expected due
to either shear or flexural.
0.15
L=20 m
Probability Density Function (PDF)

Mn(30)
0.10
Mn(50)

0.05 Mn(10)

Mn(0)

0.00
200 250 300 350 400 450 500
Mn(T) ton-m

Figure 12. Corrosion effect on flexural moment capacity Mn (T) for L=20 m
 M. Sigit Darmawan et al. 289

0.40
L=20 m
0.35 Mu (due to load)

Probability Density Function (PDF)

0.30

0.25

0.20
Mn(50)
0.15
Mn(30)
Mn(10)
0.10

Mn(0)
0.05

0.00
100 150 200 250 300 350 400 450 500
Mn(T) ton-m

Figure 13. Mu vs. Mn (T)

0.15

L=20 m
Probability Density Function (PDF)

Vn(30)

0.10

Vn(10)
Vn(50)

0.05
Vn(0)

0.00
50 100 150 200 250 300 350
Vn (T) ton

Figure 14. Corrosion effect on shear capacity Vn (T) for L=20 m

2.00

L=20 m
Probability Density Function (PDF)

1.50 Vu (due to load)

1.00

0.50 Vn(50) Vn(30)

Vn(10)
Vn(0)

0.00
0 50 100 150 200 250 300 350
Vn (T) ton

Figure 15. Vu vs. Vn(T)

290 M. Sigit Darmawan et al.

9.00

L=15 m
8.00
Flexural moment
Shear

Central Safety Factor

7.00

6.00

5.00

4.00

3.00

2.00
0 10 20 30 40 50
Time (Years)

Figure 16. Safety factor reduction

Figure 17 and 18 show the corrosion effect on central deflection for bridge span of
15 m and 20 m, respectively. These figures show that corrosion has more significant
effect on deflection than on strength of steel I girder. After 50 years of service life, limit
state for deflection is violated, especially for 20 m span. The allowable deflection used
herein is L/800 due to live load (i.e. 1.875 for 15 m span and 2.5 for 20 m span).
3.00
L=15 m
Probability Density Function (PDF)

T=0 years
T=10 years
T=30 years
2.00 T=50 years

L/800

1.00

0.00
0.50 1.0 1.5 2.0 2.5 3.0
Deflection (Cm)

Figure 17. Bridge deflection due to corrosion for L=15 m

 M. Sigit Darmawan et al. 291

1.60

L=20 m

Probability Density Function (PDF)

1.20

T=0 years
T=10 years
L/800 T=30 years
0.80 T=50 years

0.40

0.00
1.0 1.5 2.0 2.5 3.0 3.5 4.0
Deflection (Cm)

Figure 18. Bridge deflection due to corrosion for L=20 m

9. Conclusion
This paper has described the structural framework developed to study the effect
corrosion has on steel I bridge designed based on SNI T-02-2005 and SNI T-3-2005.
Probabilistic analysis using Monte Carlo simulation is utilized to take into account the
uncertainty of parameters affecting strength and deflection of steel girder and corrosion
process. From the analysis it can be concluded that up to 50 years of service designed
life, corrosion has more significant effect on deflection than on flexural and shear
strength of steel I bridge girder. No failure is observed for corroded steel bridge girder
due to flexural moment and shear force caused by dead and live load. Also, as the
bridge span increases from 15 to 20 m, the probability of deflection of corroded steel I
girder exceeds the allowable deflection also rises.

References
Albrecht P and Naeemi AH., 1984, Performance of Weathering Steel in Bridge, National Cooperative
Highway Research Program, Report 272. Washington DC.
Ang and Tang, 1984, Probability Concepts in Engineering Planning and Design Vol. II – Decision, Risk
and Reliability, John Willey & Sons, Toronto.
Ayyub, B.M. and McCuen, R.H., 1995, Simulation-based Reliability Method. Probabilistic Structural
Mechanics Handbook. Edited by C. Sundararajan. Chapman & Hall, New York. Pp. 53-69.
Badan Standardisasi Nasional, 2005, Perencanaan Pembebanan untuk Jembatan, SNI T- 02-2005.
Badan Standardisasi Nasional, 2005, Perencanaan Struktur Baja untuk Jembatan, SNI T- 03-2005.
Beedle L.S., 1958, Plastic Design of Steel Frames, John Wiley and Sons, New York.
Cheung MS. Dan Li WC., 2001, Serviceability Reliability of Corroded Steel Bridges, Canadian Journal
of Civil Engineering, Vol. 28 pg.419 – 424.
Cheung MS. Dan Li WC., 2002, Reliability Assessment in Highway Bridge Design, Canadian Journal of
Civil Engineering, Vol. 29 pg.799 – 805.
292 M. Sigit Darmawan et al.

Darmawan, M.S. and Stewart, M.G., 2006, Effect of spatially variable pitting corrosion on structural
reliability of prestressed concrete bridge girders, Australian Journal of Structural Engineering, Vol.
6, No. 2, pp. 147-158.
Darmawan, M.S. and Stewart, M.G., 2007, Spatial Time-dependent Reliability Analysis of Corroding
Pretensioned Prestressed Concrete Bridge Girders, Structural Safety, Vol. 29, pp. 16-31.
Ellingwood, B., Galambos, T.V., MacGregor, J.G. and Cornell C. A. (1980), Development of a
Probability Based Load Criterion for American National Standard A58, National Bureau of
Standards Special Publication 577, US Government Printing Office, Washington DC.
Kayser, JR., 1988, The Effect of Corrosion on the Reliability of Steel Girder Bridges, PhD Thesis,
University of Michigan, Ann Arbor, Michigan.
Kayser JR and Nowak AS, 1989, Capacity Loss due to Corrosion in Steel Girder Bridges, ASCE
Structural Journal, 115(6):1525-37.
Kayser JR and Nowak AS, 1989, Reliability of Corroded Steel Girder Bridges, Structural Safety, Vol.6,
pp. 53-63.
Komp ME., 1987, Atmospheric Corrosion Ratings of Weathering Steel – Calculation and Significance,
Material Performance, 26(7) : 42-44.
Nowak, A.S. 1994. Load model for bridge design code, Canadian Journal of Civil Engineering, 21: 36–
49.
Nowak, A.S. and Collins, K.R. (2000), Reliability of Structures, McGraw-Hill Higher Education, New
York.
Nowak A.S. and Czarnecki A.A, 2008, Time-variant Reliability Profiles for Steel Girder Bridges,
Structural Safety, Vol. 309, pg. 49-64.
Nowak, A. S., Park, C. and Casas, J. R. (2001), Reliability analysis of prestressed concrete bridge girders:
comparison of Eurocode, Spanish Norma IAP and AASHTO LRFD, Structural Safety, Vol. 23, pp.
331-344.
Ranganathan R., 1990, Reliability Analysis and Design of Structures, Tata McGraw-Hill, New Delhi.
Schneider, J. 1997, Introduction to Safety and Reliability of Structures, Structural Engineering
Documents, IABSE, Vol. 5.
Tabsh, S.W. and Nowak. A.S., 1991, Reliability of highway girder bridges, ASCE Journal of Structural
Engineering, 117(8): 2372-2388.