You are on page 1of 8

Evaluation of Damage Equivalent Factor for Roadway Bridges

Summary
The Eurocodes provide a simplified fatigue design control through the use of fatigue load model 3
(FLM3) and damage equivalent factor, , which is now widely used. In applying the Eurocodes
method, the first problem arises when designers try to determine the corresponding fatigue span
length, especially for bridge static systems not listed. Second problem is that in case of long bridges,
the probability of having several trucks simultaneously over bridge exists (in single lane traffic as
well as in multi-lane traffic); it results in larger cycles and, accordingly, higher contributions to total
damage. This paper first evaluates the damage equivalent factors and then provides an all-inclusive
method for determining the fatigue critical length for any bridge static system and the
corresponding damage equivalent factor. It proposes modifications for improving the fatigue design
method using damage equivalent factors.
Keywords: Fatigue Design; Damage Equivalent Factor; Fatigue Critical Length; Multi-Lane
Traffic; Weigh-In-Motion (WIM); Eurocodes; Fatigue Load Model
1. Introduction
Bridges play an important role in the roadway infrastructures which are under a continuous
increasing of traffic. Fatigue is an important consideration in the design of bridges especially those
made of steel. Indeed, the multiple passage of heavy vehicles, can eventually lead to crack and
failure. Cycles (in number and amplitude) resulted in the passage of a truck over a bridge depend on
bridge type, detail location, span length, vehicles weight and axles configuration. For example,
considering negative moment at mid support of a two-span continuous bridge and with assumption
of long spans for simplification, two cycles will occur due to passage of one truck. Also, assuming
presence of more than one truck over bridge makes the problem more complicated. The number of
cycles and stress ranges in a detail thus are function of many parameters, some are known while
others are not studied or poorly studied [1-5]. This is the case for bridges with trucks on several
lanes with bidirectional or unidirectional traffic. The concept of the equivalent factor for fatigue
load model is an effective method to express the traffic actions with equivalent stress range at two
million cycles and to compare the resistance of detail. Development of this concept expands its
range of validity, and improves its accuracy; it also improves the knowledge about the effects of
multi-lane traffic, as demonstrated in this article.
2. Concept of Damage Equivalent Factors based on Eurocodes
The concept of the fatigue correction factor was proposed to eliminate the tedious calculation
procedure of damage accumulation. Thanks to this method which is presented in Fig. 1, the
computation of the usual cases is performed once for all during development of the code. The left
side of Fig. 1 illustrates different elements involved with fatigue verification using damage
Nariman MADDAH
PhD student
EPFL
Lausanne, Switzerland
nariman.maddah@epfl.ch

Nariman Maddah received his
civil engineering degree in 2004,
and his MS degree from Iran
University of Science and
Technology (IUST) in 2007. He is
PhD student in steel structures lab
(ICOM) focusing on fatigue
design of bridges since 2008.


Alain NUSSBAUMER
Professor
EPFL
Lausanne, Switzerland
alain.nussbaumer@epfl.ch

Alain Nussbaumer, born 1964,
received his MS in civil
engineering from EPFL in 1988,
and his Ph.D. from Lehigh Univ.,
PA, USA in 1993. He is Professor
for steel structures at EPFL since
2008 and specialist in fatigue and
fracture.

accumulation which includes:
Simplified model of real traffic
The corresponding stress history
including dynamic effect
The extracted stress cycles and
calculation of equivalent stress range.
The right side of diagram shows the
application of fatigue load model to obtain
maximum stress,
FLM,max
, and minimum
stress
FLM,min
, with placing it on the most
severe positions. To obtain the same value as
the equivalent stress range,
E2
, which takes
into account the damage accumulation, the
engineers will then correct the value of
FLM

with damage equivalent factor . The fatigue
assessment then can be carried out as follows:
( )
Mf
c
FLM Ff




2
(1)
where
Ff
is the partial safety factor for
fatigue loading,
2
is the impact factor
(which is included in the fatigue load model 3
for good pavement quality,
FLM
is stress
range due to FLM3),
c
is the reference
value of fatigue strength (at 2 million cycles)
and
Mf
is the partial safety factor for fatigue
strength. Damage equivalent factor, , can be
obtained from:
max 4 3 2 1
= (2)
where
1
is the factor for damage effect of traffic and it depends on critical length of influence line
and fatigue load model,
2
is for modification of traffic volume,
3
is for modification of the bridge
design life,
4
is the factor which adds up the effect of traffic on the other lanes, and
max
is the
maximum value which takes into account the fatigue limit.
The geometry of FLM3 based on EN1991-2 [6], which is in accordance with damage equivalent
factors, is shown in Fig. 2. The weight of each axle is 120 kN. Where relevant, a second set of axles
in the same lane should be taken into account. The geometry of the second set is similar, but the
weight of each axle is equal to 36 kN (instead of 120 kN). The minimum distance between two
vehicles measured from centre to centre of vehicles is at least 40 m. This model tries to take into
account the effect of multiple trucks on the bridge.
It is important to mention the critical span length is defined in EN1993-2 [7] case by case;
consequently, the critical span length is unknown for undefined cases. Some of defined critical
lengths are:
for a simple span, equal to span length,
for continuous span, in support sections, the mean of the two spans adjacent to that support,
for continuous span, for intermediate supports reaction, the sum of two adjacent spans.
In addition to
1
, the multi-lane traffic effect can be calculated using the factor
4
. This factor takes
into account the effect of having more than one lane by using following formula:
5 / 1
5
1 1
3 3
1
3
5
1 1
2 2
1
2
4
1
(
(

+
|
|

\
|
+
|
|

\
|
+ = L
m
m
m
m
Q
Q
N
N
Q
Q
N
N

(3)

Fig. 1: Determination Procedure of Damage Equivalent
Factor


Fig. 2: Geometry of Fatigue Load Model 3 (FLM3)
where N
i
is the number of trucks per year in the
corresponding lane, Q
mi
is the average gross weight of
the trucks in the corresponding lane and
i
is the value
of the influence line for the internal force that
produces the stress range in the middle of the
corresponding lane (always positive).
Aforementioned method for calculation of damage
equivalent factors is accompanying with some
shortcomings. For example, continuous flow of traffic
on bridge is not considered in calculation of damage
equivalent factors, and it is neglected to have more
than one truck in each lane. In addition, in multi-lane
traffic, several trucks might be on bridge
simultaneously, which means occurrence of one big
cycle instead of two smaller ones.
3. Methodology
3.1 Hypotheses and Scenarios
For modelling traffic over bridge accurately and accomplishing damage accumulation, software
based on Microsoft C# has been developed [8]. Statistical parameters of actual traffic are based on
Weigh-In-Motion (WIM) measurements from station of Mattstetten on the highway of Bern-
Solothurn in 2003. Further studies will be done using more recent WIM measurements. The
program randomly chooses each truck properties from a database which is in accordance with real
traffic. Such a simulation is able to model fluid traffic as well as congested traffic close to reality. It
also allows have several trucks over bridge simultaneously. Traffic distribution can be defined in
different scenarios. In Fig 3, the daily traffic distribution of Mattstetten in 2003 is shown (which is
adapted to highway traffic with 2000000 trucks per year per slow lane). In this study, two
unfavourable traffic scenarios compared to the real traffic distribution were considered as shown in
Fig. 3: 1) flat distribution of traffic over 10 hours of each day 2) two traffic pick which takes long
for 3 hours per day. Traffic is always fluid and trucks circulate on working days (255 days/year).
Then the program calculates damage sum using the Miner linear damage sum rule and equivalent
damage factor using defined fatigue load model and S-N curve parameters. Whereas simulation of
traffic is done with the Monte-Carlo method, traffic simulation is performed for 1 year and 10 years
to verify the coherence of the results. The number of simulated trucks is 2000000 per year per
direction, which corresponds to highway traffic, and then the number of cycles multiplied by
500000 100 / (2000000 Year) to obtain the similar number of trucks given in the Eurocodes.
Damage equivalent factors for following static systems are analysed:
simple span bridge, mid moment (SS-Mid M.),
two equal spans continuous bridge, negative moment (2CS-Mid Sup. M.) and reaction at
mid support (2CS-Mid Sup. R.),
three equal spans continuous bridge, mid moment of second span (3CS-Mid M.), second
support moment (3CS-2nd Sup. M.), first and second supports reaction (3CS-1st Sup. R. and
3CS-2nd Sup. R.).
The fatigue resistance curve of steel is considered , as defined in EN1993-1-9 [9],with slope of 3 for
cycles with stress range higher than constant amplitude fatigue limit (CAFL), and slop of 5 for
cycles lower than CAFL, also cycles lower than cut-off limit are dismissed.
3.2 Damage Equivalent Factor for Single Lane Traffic
First simulations are performed for the mid moment of simple span bridge and the mid support
moment of two-span continuous bridge to compare damage equivalent factors resulted from two
scenarios (S1, S2) as well as simulation period (1 year and 10 years). Fig. 4 shows the results of the
first simulation; the fatigue load model (FLM3) is also plotted for clarity. The corresponding
damage equivalent factors of EN1993-2 [7] are also shown in Fig. 4. Since the average gross weight
of trucks on station of Mattstetten (2003) is 290 kN, partial equivalent factor,
2
= 290 / 480, is used
Fig. 3: Daily Traffic Distribution of
Mattstetten (2003) and Two Considered
Scenarios
to modify the damage equivalent factors of EN1993-2 [7]. For comparison purposes traffic
simulation is also done assuming one-by-one passage of trucks over bridge with the same heavy
vehicle geometry and weight (VbV-10Y curve in Fig. 4). It represents that the damage equivalent
factors obtained for scenario 2 are slightly higher than scenario 1. Whereas, it clears up that effect
of continuous traffic, which allows having multiple heavy vehicle simultaneously over bridge, is
noticeable and it should be taken into account. This effect though should be lower for roadways
with lower traffic. The damage equivalent factors obtained from simulation of traffic in 1 year and
10 years were similar, thus it is not shown in Fig. 4.
In Fig. 4, the damage equivalent factor for mid support moment of two-span continuous bridge is
higher than the curve of EN1993-2 [7] which is adapted to traffic of Switzerland. Thus for existing
bridges, fatigue life of details located at mid support might be lower than expected life assumed in
the current code. However, the maximum traffic volume based on EN1991-2 [6] is used here. The
traffic volume of different stations in Switzerland is presented in Table 1. The stations are all
located in the highway. The heavy vehicle traffic per year per slow lane, N
obs
, is estimated assuming
10 per cent of traffic circulates on fast lanes and traffic volume is similar on both directions. The
corresponding N
obs
value in the code is 2000000 which is quite higher than the measured stations
in Switzerland. Moreover, the average gross weight of trucks in Switzerland ranges between 278
and 312 which is much lower than average gross weight given in the EN1993-2 [7] with Q
m
of
480 kN.
The same simulation has been done to find the damage equivalent factor for various bridge static
systems. Since the Scenario 2 with 10 years traffic simulation gives slightly more conservative
results, thus further simulations are done only with this hypothesis (S2-10Y). Fig. 5 shows the
damage equivalent factor for different bridge static systems in compare with EN1993-2 [7]. The
partial equivalent factor,
2
= 290 / 480, is similarly used to modify the damage equivalent factors
of EN1993-2 [7]. Fig. 5 shows that the values of the damage equivalent factor obtained in the
support area are higher than the curve of Eurocode, so EN1993-2 [7] is non-conservative. The
safety margin of Eurocode for curve (mid span and support) depends on the span length and bridge
static system, which is not desirable.
3.3 Damage Equivalent Factor for
Double Lane Traffic
For double lane traffic, the same bridge
static systems are chosen. Traffic simulation
parameters as well as calculation of damage
equivalent factors are similar to single lane
traffic as described in previous section. Two
separated cases are studied: bidirectional and
unidirectional. In the case of bidirectional,
Table 1: WIM Station Measurements in Switzerland, 2003
Station AADT ADTT Nobs Qm (kN)
Mattstetten 71552 4818 799422 290
Trbbach 30267 1505 349707 285
Ceneri 36747 3620 600510 312
Oberbren 47353 2450 406444 278
AADT: Annual Average Daily Traffic (on 365 days basis)
ADTT: Average Daily Truck (100 kN) Traffic
Nobs: Number of Truck (100 kN) per year per slow lane (Estimated)
Qm: Average Gross Weight of Trucks (100 kN)


Fig. 4: Effect of Traffic Simulation Condition on Damage Equivalent Factors for Single Lane Traffic
the traffic distributed with the ratio of 50-50 per cent on each direction, and traffic of lanes is
opposite. For second case, unidirectional, traffic is distributed on two lanes with 80 per cent of
trucks traffic in slow lane and 20 per cent on the fast lane, also traffic direction is same for both
lanes.
The main objective of double lane simulation is to study on the effect of trucks which are crossing
(or overtaking) on bridge. The determination of actual frequencies of these situations is not part of
this study. The transverse distribution of load has the key effect on this part. In the case of box
section, the most unfavourable case, there is no transverse distribution, and it is acceptable that the
box section can be uniformly charged regardless of axle position. This assumption allows us
consequently neglect the dimension of section.
The damage equivalent factors are shown in Fig. 6 for bidirectional (BD) and unidirectional (UD)
traffic as well as single lane (SL) traffic for two cases of mid span moment of simple span bridge
and mid support reaction of two-span continuous bridge. The traffic volume is similar on the three
cases. The fatigue load model 3 (FLM3), as shown in Fig. 2, is applied. Whereas, damage
equivalent factors were very similar in case of 1 year and 10 years simulation, the results for just 10
years are presented. Fig. 6 shows the probability of having several trucks simultaneously on bridge
(crossing or overtaking) logically increases, like
1
, for bridges with longer span. Also, it is verified
that the damage equivalent factors are higher in the case of bidirectional traffic in compare with
unidirectional because the probability of simultaneity increases with 50-50 per cent distribution of
traffic on each lane. There is also a slight difference between two simulated scenarios. The values

Fig. 6: Effect of Traffic Simulation Condition on Damage Equivalent Factors for Double Lane Traffic

Fig. 5: Single Lane Traffic, Damage Equivalent Factors for Different Static Systems
obtained for Scenario 2 are higher than the
Scenario 1. It can be described with the fact that
Scenario 2 has two traffic peaks which cause more
crossing on bridge. It is important to mention when
two trucks crossing on bridge, instead of having two
cycles due to passage of each vehicle on each lane,
one cycle but equal to sum of two smaller cycles
occur.
4. Proposed Damage Equivalent Factor
4.1 Single Lane
Several attempts have been done to define
appropriate damage equivalent factor as well as
fatigue load model and influence line length. It has
been seen that using fatigue load model with just
one axle would lead to less scattering in the damage
equivalent factors for various bridge static systems.
Therefore, it is proposed to use a single load model
with the weight of 480 kN same as total weight of
fatigue load model in EN1991-2 [6]. In addition, the
fatigue critical length is denominated fatigue
equivalent length, L

, and can be determined as


follows:
inf
inf

=
A
L

(4)
where A
inf
is absolute sum of area under influence
line, as shown in Fig. 7, and
inf
is difference
between maximum and minimum values of
influence line. Fig. 8 gives the fatigue equivalent
length for some types of influence lines which are
determined with Eq. (4). Fig. 9 illustrates the
damage equivalent factor obtained with this
hypothesis for different static systems; the single
axle fatigue load model is also plotted for clarity.
The simulation parameters are similar to the ones
described in Section 3.1. The proposed curved is
based on the 500000 passage of trucks per slow
lane with average gross weight of 290 kN and
design life of 100 years. The curves obtained for
different static systems show a clear trend with a narrow dispersion band, well represented by the
proposed curve.
4.2 Double Lane
The factor
4
allows adding up the damage due to other lanes with the first lane, thus it is always an
increasing factor. In EN1993-2 [7], this factor is calculated in a simplified way accepting the
damage due to the lanes can be accumulated with constant slope fatigue resistance curve equal to 5.
The current definition of the factor
4
does not embrace the effect of having several trucks
simultaneously on bridge. It is possible to redefine the factor
4
to improve the shortcoming. Using
the Miner summation rule and assuming the stress response spectrum due to passage of trucks
simultaneously on both lanes is proportional of stress response spectrum due to passage of traffic on
slow lane, we can write similarly to Eq. (3):
5 / 1
5
1
2
1
5
1
2
1
2
1
4
1 1
(
(

|
|

\
|

+ +
|
|

\
|


+
|
|

\
|
=

N
N
N
N N
N
N
c c c
(5)

Fig. 7: Schematic Influence Line to Determine
Parameters of Fatigue Equivalent Length


Fig. 8: Fatigue Equivalent Length for Some
Types of Influence Line
where N
c
is the number of trucks crossing (or
overtaking) on bridge, and
1
and
2
are
stress ranges due to passage of fatigue load
model on lane 1 (slow lane) and lane 2
respectively.
The factor
4
for each bridge static system
based on the hypothesis given for factor
1
, as
described in the section 4.1, is calculated and
illustrated in Fig. 10 for unidirectional and
bidirectional traffic. Whereas the traffic on
lane 1(slow lane) is less than single lane traffic
(80% for unidirectional and 50% for
bidirectional), the factor
2
is applied to modify
the volume of traffic on lane 1. In Fig. 10, the
factor
4
obtained from Eq. (5) is also presented.
The given factor
4
, in fact, adds up the damage
due to traffic on lane 2 as well as damage due
to crossing or overtaking.
The factor
4
given in EN1993-2 [7] only adds
up the damage due to volume of traffic of
additional lanes (other than slow lane); however, it does not consider the effect of having several
trucks simultaneously on bridge. This effect in the case of unidirectional and bidirectional traffic, as
show in Fig. 10, is about 8 per cent and 20 per cent respectively, which is more pronounced than the
effect of traffic volume for the simulated cases.
By trial and error, it is found that the simulation results can be conservatively represented assuming:
12% crossing ratio (N
c
/ N
1
) for bridges with bidirectional traffic, and 2% overtaking ratio for
bridges with unidirectional traffic. For comparison purpose, the factor
4
with effect of crossing
given by SETRA [10] for highways (only bidirectional traffic) is also calculated and presented in
Fig. 10. It is apparent that the proposed ratio underestimates the crossing effect. The assumed length
in SETRA [10] is the bridge length, which is roughly assumed to be equal to twice L

. The proposed
crossing and overtaking ratios are preliminary results and the study is still in progress. The principal
parameters that have an influence on it are: the traffic volume on each lane, direction of traffic, and
fatigue equivalent length. Future study will clarify the crossing and overtaking ratio based on the
aforementioned parameters.
5. Conclusions
The goal of this paper is to propos the modification to damage equivalent factors () for fatigue

Fig. 9: Comparison of Fatigue Equivalent Factor for
Different Bridge Static Systems and Proposed Curve

Fig. 10: Double Lane Traffic, Damage Equivalent Factors for Different Static Systems
= 0.45(L

)
0.3

design of road bridges according to concept of Eurocodes. The factor can be decomposed into
several partial factors, i.e. =
1

2

3

4
. This study especially focuses on
1
and
4
.
For the factor
1
, our study was conducted with simulation of real continuous traffic, in a manner as
realistic as possible, on different bridge static systems. The results are thus more accurate than the
former simulations which have been done within writing of the Eurocode. Based on these
simulations, a fatigue load model with a single axle is proposed because it leads to a significant
decrease in the dispersion of the values of the factor
1
obtained for different static systems. In
addition, a new method for determining the fatigue equivalent length is also proposed; see Eq. (4).
This expression provides more simple method in determining uniquely the fatigue length for all
influence lines.
For the factor
4
, similar simulations have been done to study the effect of double lane traffic. The
two main cases are studied: 1) unidirectional traffic assuming 80% of the heavy traffic flow on the
slow lane and 20% in the fast lane, 2) bidirectional traffic with an equal distribution of 50% on each
lane. The obtained results show that the effect of overtaking or crossing is important and its effect
cannot be neglected on the factor
4
; however, the factor
4
in Eurocode only considers the damage
accumulation due to volume of traffic in the additional lanes. Consequently, the new factor
4
is
proposed, Eq. (5), which adds up damage due to crossing (or overtaking) as well as damage due to
traffic volume on the second lane. In addition, based on the traffic simulation for highways, the
crossing and the overtaking ratios can be proposed as 10% and 3% respectively.
6. Acknowledgements
The authors acknowledge ASTRA for the support of this project (AGB 2007/004) as well as the
accompanying group BK-C AGB for their monitoring and advice. We appreciate also the advice of
Mr Th. Meystre, office DIC SA Ing. Councils, as partner of this project.
7. References
[1] MIKI C., GOTO Y., YOSHIDA H., and MORI T., Computer Simulation Studies on the
Fatigue Load and Fatigue Design of Highway Bridges, Proceedings of JSCE Structural
Engineering / Earthquake Engineering, Vol. 2, pp. 37-46, 1985.
[2] LAMAN J.A., and NOWAK A.S., Fatigue Load Models for Girder Bridges, ASCE
Journal of Structural Engineering, Vol. 122, pp. 726-733, Jul 1996.
[3] MORI T., LEE H.H., and KYUNG K.S., Fatigue Life Estimation Parameter for Short and
Medium Span Steel Highway Girder Bridges, Engineering Structures, Vol. 29, pp. 2762-
2774, 2007.
[4] NOWAK A.S., Live Load Model for Highway Bridges, Structural Safety, Vol. 13, pp. 53-
66, 1993.
[5] SCHILLING C.G., Stress Cycles for Fatigue Design of Steel Bridges, ASCE Journal of
Structural Engineering, Vol. 110, pp. 1222-1234, 1984.
[6] EN 1991-2, Actions on Structures, Part 2: Traffic Loads on Bridges, European Committee
for Standardization, 2002.
[7] EN 1993-2, Design of Steel Structures, Part 2: Steel Bridges, European Committee for
Standardization, 2006.
[8] MEYSTRE T., and HIRT M. A., valuation de Ponts Routiers Existants avec un Modle de
Charges de Trafic Actualis VSS 594, Laboratoire de la construction mtallique ICOM,
EPFL, Lausanne, 2006.
[9] EN 1993-1-9, Design of Steel Structures, Part 1-9: Fatigue, European Committee for
Standardization, 2005.
[10] SETRA, Ponts Mtalliques et Mixtes : Rsistance la Fatigue - Guide de Conception et de
Justifications, Service dtudes Techniques des Routes et Autoroutes (SETRA), France,
1996, p.84

You might also like