Technical University of Civil Engineering of Bucharest
ABSTRACT: Gradual changes of weight and geometry of vehicles combined with natural damaging process caused by weather and traffic sometimes leads to old existing bridges having to be strengthened and retrofitted. The optimal solution to follow depends on the resistance capacity reserves of each structure, related with the existing traffic level at the moment of retrofitting works and for an estimated future traffic. Almost all existing road bridges in Romania were designed according to national provisions/standards, using the old loading classes I and E respectively. Starting with 2010, following the European Community’s decision, the projects should be designed according to the Eurocodes and the existing bridges, after the strengthening/retrofitting processes must comply with the new European load models, even though these structures do not present
obvious damages which are claiming such kind of works. The process of checking existing bridges for the action of load models described in Eurocodes is a complicated task, involves hard work and consumes a lot of time. However, for the solutions consisting in the use for the bridge superstructure of precast concrete girders, the process for the evaluation of the bearing capacity reserves can be significantly simplified using a probabilistic approach. The approach presented in this paper is based on the effects, in terms of bending moments and vertical displacements, produced by several types of vehicles on the bridge superstructure. For presenting the use of this proposed methodology, typical bridge superstructure with precast concrete girders of different lengths were analyzed. These bridges are frequently used on the road network in Romania and they were designed using the loading models existing in the national standards at that time. The proposed methodology allows to establish the “vulnerability” of existing bridges on the action of load models presented in the Eurocodes in order to establish the most efficient retrofitting solutions.
1 INTRODUCTION
1.1 Strength capacity
The strength capacity of a bridge during service is determined in the design phase by adopting an efficient and functional solution which depends, among others, on using durable, high performance materials and correct estimation of the loads and behavior of the structure. Designing a structure is generally based on the available data and the level of knowledge at the time but, in the case of bridges, extra measures are taken in order to accommodate for the wide variety of actions (especially traffic) foreseen in the future for the entire estimated lifespan. During the service life of a structure, design norms can change several times to such a degree that it may be necessary to recheck the existing structure in accordance with the new provisions in order to make sure the safety levels under service are still at an acceptable level.
1.2 Romanian road network
There are many bridges on the road network of Romania designed using the old national standards that were available at the time of construction. Among these, many are still performing well under service, only requiring regular maintenance works, but there is also a large number of bridges showing significant degradation to different levels due to lack of maintenance or inappropriate strength capacity. In the year 2010, the need to use the European Norms (Eurocodes) became mandatory in both designing and erecting new structures. The new requirements in durability and loads claim reviewing the existing structures, both the ones with signs of degradation and the ones performing well, in order to check whether or not they satisfy the new demands and keep an acceptable level of safety.
1.3 Assessing the safety under service
A convenient method for assessing the level of safety under service for an existing bridge within the
European context, can be estimating strength capacity reserve using a probabilistic approach. Comparing the old norms with the Eurocodes, it can be considered that there are only small changes in the case of permanent loads considered in the design phase. The most significant changes in the internal stress levels are given by the new variable actions. In the case of concrete bridges with the superstructure made up of precast, prestressed concrete girders, staged construction designs were made considering the different phases of erection with permanent loads and variable traffic loads. The methodology proposed in this paper for estimating the vulnerability of existing concrete bridges implies establishing a law of distribution for the maximum stress levels and determining the strength capacity reserves by comparison with the design stress levels. In order to make a realistic study, it would have been necessary to have information regarding the types of road vehicles frequently transiting these bridges. The information should have contained details about the number, distribution and weight of the axles of each type of vehicle. Sadly however, this type of data does not exist at an administrative level, and collecting it requires significant costs and personnel mobilization. Under these conditions, in order to finish the study, several load models considered to be found more often on Romanian roads were selected from both the old and new design norms, as well as a number of convoys previously used for testing bridge structures under service loads. Maximum stress levels for the service limit state were determined by using finite element models and linear static analyses considering phases of construction and traffic loads of the old bridges and the ones required for retrofitting them. These were then associated to a statistic distribution law. Based on probability density functions and cumulative distribution functions, the probability of exceedance was determined for each case (old and retrofitted).
2 CASE STUDY
2.1 Description of the structure
In order to illustrate the proposed method, a typical 30.00m long bridge superstructure widely used in Romania in the 1980’s was chosen. The superstructure is made up of 4 precast prestressed concrete girders placed at 2.83m apart, an 18cm slab and 3 crossbeams (at each bearing and in the middle of the span) linking them. The girders were made of B500 concrete (equivalent C32/40) and the slab of B400 concrete (equivalent C25/30). It supports a 7.80m carriageway and 2 footways of 1.00m each,
Figure 1 – Cross section of existing bridge for the case study
The retrofitting solution chosen for the case study is removing all the carriageway’s and footway elements (asphalt, slope concrete, kerbs, footways, pedestrian guardrails, etc.), adding a concrete slab of C35/45, widening the superstructure so that H4b barriers can be added to the footways and rebuilding all the carriageway and footway elements.
Figure 2 – Cross section of retrofitted bridge for the case study
Several
analyses
were
made considering the
existing bridge and the retrofitted bridge with a
variable minimum thickness of the 10cm, 15cm and 25cm.
new deck
of
2.2 Approach method
The idea of this study was inspired by the vulnerability analyses of the structures at the seismic action Vamvatsikos (2002), Annan (2009). The first phase of the proposed method in the paper is based on a distribution of probability of the set of values for normal stresses or vertical displacements calculated in the lateral beam in the middle of the span, considering loads from the Serviceability Limit State (SLS). The second consists in determining the function of density probability and the one for cumulative distribution. Using these two functions, it is possible to determine the value of reliability in terms of probability of exceedance of the of the maximum admissible stress or displacement at the middle of the span.
The link between all these three functions can be expressed by the following relationships, Lungu
(1982):
P a
X
b
b
f
x
x
a
F
X
x
X
x
f
X
t
dt
1
X
(1)
(2)
(3)
where P=distribution of probability, a,b=interval of values, f _{X} =probability density function, F _{X} =cumulative distribution function, R _{X} =reliability function
2.3 Finite element model
Three finite element models were built up (one for each concrete slab thickness) using beam elements for the precast concrete girders and crossbeams and shell elements for the slab. One of these models in shown in figure 3 bellow.
a) Discrete view of the finite element model
b) Extruded view from the top
c) Extruded view from the bottom
Figure 3 – Finite element model
After determining the internal forces, stresses and displacements in the mid span of the lateral girder due to permanent loads, a number of 51 static moving load analyses were performed considering different load models to simulate traffic conditions
during bridge service life. The permanent loads taken into consideration are:
the dead load of the initial structure’s concrete elements (beams, crossbeams and slab), the dead load of the over slab of the retrofitted structures, the weight of the carriageway’s supporting elements (asphalt layers, kerbs, H4b barriers) and the footways (concrete fillings, pedestrian guardrails, etc.). As for the traffic loads, several design convoys from different regions of the world and real truck loads have been taken into account as follows: the design convoys used in the initial design used in Romania at the time, according to STAS 3221/86 for classes I, II and E, the load models presented in Eurocode SR EN 19912/2004, a few convoys used in the American AASHTO and 4 real convoys used in the testing of the retrofitting solution used for the GiurgiuRuse Bridge over the Danube. The vehicles used to simulate the response were grouped and loaded according with the provisions of their specific norms at the time, in order to get the most unfavorable effects in terms of bending moment in the middle of the span for the lateral beam as well as maximum vertical displacement of the same element. Several arrangements of the convoys are provided in the following figures:
Figure 4 – Examples of traffic load models
In order to obtain a sufficient number of values required for determining a correct variation for the distribution of probability, 51 combinations were made using the vehicles presented above. With the same configurations for the traffic loads, 51 static analyses were made for each of the 4 models, all of them having included the staged construction phases necessary for erection. A comparison between the bending moments was considered irrelevant, as the beams have different sections depending on the construction stage, and thus, a comparison of the normal stress at the bottom fiber was made. The reference for this was
considered the one corresponding to load class E and convoy V80 according to the Romanian standard, as this was the norm initially used to design the bridge. As such, all stress values are compared to this, the base line being set at 22.62 MPa as 0, positive values meaning more tension in the fiber by that specific amount, and negative meaning more compression. The maximum acceptable limit is only in tension, and was calculated at 3.2 MPa (maximum acceptable tensile stress of the old concrete). Regarding the values for displacement, only the mobile loads were considered as they have the more relevant impact. Just as before, all values are compared to the one corresponding to the V80 convoy load case, but in this case, the base line was set at 0. The maximum acceptable limit was the one prescribed by PD 1652000, as no more then L/800 =
36.8mm.
3 PRELIMINARY RESULTS
After running all the analyses presented above, extracting the bending moments in the middle of the span for the marginal beam and overlaying them according with the provisions of their norms, a set of 51 variables was obtained for each model. By considering a sufficient number of intervals and choosing the correct size for them, histograms of the relative frequencies and relative cumulative frequencies were determined. Based on these, the distribution of probability for the 51 values was calculated. A similar approach was done for the 51 displacements values calculated in the middle. The relative frequency and relative cumulative frequency histograms are presented in the following figures:
Figure 5 – Relative frequency histograms
Figure 6 – Cumulative frequency histograms
Figure 7 – Frequency of normal stress in the bottom fiber over intervals
By analyzing the histograms above, it can be considered that the law of distribution of probability is similar to the lognormal distribution for both the bending moments and displacements results. In a lognormal distribution, the density of probability function, respectively the cumulative distribution function are defined as follows:
The maximum acceptable tension in the concrete for class C32/40 is 3.2MPa, according to Romanian national appendix. Most of the values for the stress are situated between 3 and 2 MPa, so the structure checks out for most load models, but there is a significant percent of values above this limit.
e
2 The maximum displacements calculated in the
ln x
2
2 (4)
middle of the span for the marginal girder were only
considered from the actions given by the traffic loads. The extreme values, as well as a partition of the
(5)
values obtained over intervals can be found in table 2 bellow.
1
f
x
x
F x ln
Where x is the point in which the function is calculated, μ represents the average of the ln x values, σ represented the standard deviation of the ln x values and _{} is the cumulative distribution function for the normal standard distribution which is determined by the following relation:
(6)
The maximum stress levels according to the characteristic group of the service limit state in the bottom fiber of the marginal girder were calculated in the middle of the span for each, accounting for staged construction and dynamic effects of the convoys. The extreme values, as well as a partition of the values obtained over intervals can be found in table 1.
Table 1. Extreme stress values in the bottom fiber
Normal stress 
Without 
10cm 
15cm 
25cm 
[MPa] 
retrofit 
slab 
slab 
slab 
Minimum 
3.07 
5.75 
4.78 
2.78 
Maximum 
12.33 
6.52 
6.42 
6.47 
*compression is considered negative, and tension positive
Table 2. Extreme displacement values
Displacement 
Without 
10cm 
15cm 
25cm 
[mm] 
retrofit 
slab 
slab 
slab 
Minimum 
3.00 
2.27 
2.00 
1.66 
Maximum 
34.18 
25.78 
22.56 
18.61 
*positive values indicate lowering of the girder
Figure 8  Frequency of displacement over intervals
4 CONCLUSIONS
The purpose of study presented in this paper was to analyze the vulnerability of existing road bridges in Romania following the action of the actual vehicles/convoys, which are different with respect to those used for the initial design. The study presents for comparison the safety levels of the initial
nonretrofitted structure, but also of the retrofitted structure using different values for the thickness of the concrete slab above the prestressed beams. The vulnerability of the analyzed structures was described through the values of the reliability, by considering as response values the values of the stresses in the bottom extreme fibers of the marginal precast beam and the vertical displacements in the middle of the span. The limit values were chosen those obtained in the design stage using the vehicles A30 and V80 both described in the Romanian norms used at that time. Following the performed analyses on considered bridge it can be concluded that the values of the stresses in the bottom extreme fibers of the marginal precast beam based on the bending moments in the middle of the span and the values of the vertical displacements in the same section follow a distribution function close to the lognormal one. Based on the obtained results one can see that in terms of stresses, presented in table 3, the structure is vulnerable if it is not strengthened and shows an acceptable level of safety after the retrofitting process. The values of the reliability are in the range 4.298.37% (Fig.10) for the retrofitted structure and the most efficient solution appear to be the one with 10cm thick concrete slab above the precast beams. This result can be explained by the fact that this thickness ensure a balance between the permanent and live loads acting on the structure. By overloading the bridge, during the retrofitting process, by a thicker slab, will decrease the solution efficiency with respect to the live load actions.
Figure 9 – Probability density function for normal stress in the bottom fiber
Table 3. Stress comparison of retrofitting solutions
Without 
10cm 
15cm 
25cm 

retrofit 
slab 
slab 
slab 

Probability of 

nonexceedence 
35.73% 
95.71% 
95.01% 
91.63% 
Reliability 
64.27% 
4.29% 
4.99% 
8.37% 
*compression is considered negative, and tension positive
Figure 10 – Cumulative distribution function and reliability for normal stress in the bottom fiber
In terms of vertical displacements, the values of reliability are extremely small for the retrofitted bridge, in the range 0.391.68%, in this case the solution with a 25cm slab showing a better behavior. This fact is normal assuming a higher value of the bending stiffness of the bridge superstructure. Even the nonretrofitted bridge shows a satisfactory behavior in terms of displacements.
Figure 11 – Probability density function displacement
Table 4. Displacement comparison of retrofitting solutions
Without 
10cm 
15cm 
25cm 

retrofit 
slab 
slab 
slab 

Probability of 

nonexceedence 
92.23% 
98.32% 
99.05% 
99.61% 
Reliability 
7.77% 
1.68% 
0.95% 
0.39% 
*positive values indicate lowering of the girder
Figure 12 – Cumulative distribution function and reliability for displacement
For a more realistic study the traffic measurements are necessary and on this base, the values of the reliability for the existing bridges can be established. Following such analyses the necessity of the use of an intervention plan for bringing the bridges at the safety level from the design stage can be established.
REFERENCES
CEN. (2005) EN 1990:2002 Annan, C.D., Youssef, M.A., El Naggar, M.H. (2009) Seismic Vulnerability Assessment of Modular Steel Bridges, Journal of Earthquake Engineering 13:8, Taylor&Francis,
pag.10651088
Aviram, A., Mackie, K.R., Stojadinović, B. (2008) Guidelines for Nonlinear Analysis of Bridges Structures in California, Peer Report 2008/03, Pacific Earthquake Engineering Research Center, University of California, Berkeley CEN. (2005) EN 1990: 2002/A1 Basis of structural Design CEN (2003) EN 19912 Eurocode 1: Actions on structures – Part 2 : Traffic loads on bridges Lungu, D., Ghiocel, D. (1982) Metode probabilistice în calculul construcţiilor, Editura Tehnică (Probabilistic methods in the analysis of structures) Vamvatsikos, D., Cornell, C.A. (2002) Incremental Dynamic Analysis, Earthquake Engineering and Structural Dynamics 31 (3), John Wiley&Sons Ltd., pag.491514