Engineering Structures 33 (2011) 37553761

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Damage due to heavy traffic on three RC road bridges

Martin Pircher a, , Bernhard Lechner b,1 , Oliver Mariani b,1 , Andreas Kammersberger c,2
a b c

ABES Consulting International GmbH, Bergmanngasse 7, 8010 Graz, Austria Kompetenzzentrum - Das Virtuelle Fahrzeug Forschungsgesellschaft mbH, Inffeldgasse 21a, 8010 Graz, sterreich, Austria Amt der Steiermrkischen Landesregierung, Landhausgasse 7, 8010 Graz, Austria

article

info

abstract
An investigation of the damage caused by road traffic on three reinforced concrete bridges was performed. This study was performed with a focus on the determination of the damage induced by heavy vehicles in relation to the damage caused by average everyday traffic. A damage model based on fatigue of reinforcement bars was employed. The stress cycles in the reinforcement bars were determined using measurements of the crack widths under traffic loading. Stress cycles were analyzed using the Rainflow Method and Miners Rule was employed to gain a measurement for the damage. Monitoring systems were installed on each of the three bridges and operated for several weeks continuously to collect input data for the damage model. Computer software was developed to process the monitoring data in the sense of the damage model. The described methodology was developed and used for the first time in the described project. Heavy traffic was shown to cause a disproportionate high portion of the overall damage on all three bridges. The obtained results indicated that the damage caused by singular events, such as the passage of a heavy vehicle, in relation to the damaging effects of every-day traffic differed significantly between the considered bridges. For one of the bridges the greatest damage from singular traffic events was computed for passages of special transport vehicles. The same bridge was also found to display the highest damage by a single passage in relation to average everyday traffic. Passages of short and heavy trucks with four and five axles were identified as the most detrimental traffic events on the other two bridges. 2011 Elsevier Ltd. All rights reserved.

Article history: Received 23 September 2010 Received in revised form 6 July 2011 Accepted 1 August 2011 Available online 1 September 2011 Keywords: Heavy vehicle loading Bridge monitoring Accumulated damage Reliability Failure probability Bridge damage model

1. Introduction Valid bridge designs according to most modern design codes should ensure that the probability of failure remains under a given critical limit throughout the design life time. The reliability of a structure is the inverse of the probability of failure and can be computed by comparing the loading with the resistance of a structure [1,2]. Any structure is exposed to environmental influences, service loading and permanent loads which can all be seen to cause small portions of damage accumulating over time. In general terms, this accumulated damage can be interpreted as reducing the structural resistance, and therefore the structural reliability of a structure. On the other hand, code-specific loading also changes, and in the case of traffic loading on bridges these loads have increased significantly over time in most design codes.

Corresponding author. Tel.: +43 0 316 670619 0; fax: +43 0 316 670619 99. E-mail address: martin@abes-australia.com (M. Pircher).

1 Tel.: +43 316 873 9001; fax: +43 316 873 9002. 2 Tel.: +43 0 316 877 2466. 0141-0296/$ see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2011.08.012

Additionally, traffic volume and therefore the number of loading events on bridges also tend to increase accelerating the rate at which damage accumulates. Fig. 1 illustrates the described changes of loading S and resistance R over time. Discrete values for S and R are usually not available and therefore they are assumed to be distributed according to some probability functions. The area under the intersecting parts of these two probability functions is a measurement for the probability of failure. The initial position of these curves at time t = 0 illustrates the situation immediately after completion of a structure, and the dotted lines in Fig. 1 show the situation at some later point in time during the design life. In the present paper, the term damage is used in the sense of a reduction in structural resistance. Accumulated damage as a result of many loading events throughout the life time of a structure causes a steady decrease of resistance. It can be assumed, that the probability of failure remains below the defined critical level throughout the life time of a structure as long as these damage loading events remain within the regulations of the applied design code indicated by the dotted line labeled Design loading in Fig. 2. However, singular overloading events beyond the service load levels prescribed in the design code, such as illustrated

3756

M. Pircher et al. / Engineering Structures 33 (2011) 37553761

the reinforcement bars as a measure. Using this damage model singular loading events were evaluated, the accumulated damage of daily traffic was measured and the effects of overloading by special transport vehicles were studied for three distinctly different RC bridges. 2. Considered bridge structures Three reinforced concrete bridges, namely the Weizbach Bridge, the Arndorfbach Bridge and the Salmbach Bridge, along a continuous stretch of country road B64 in the province of Styria, Austria, were evaluated for the present paper with a focus on determining the damage caused by traffic on these bridges. Road B64 links an area with industrial and agricultural activity with the major highway network. Around 11,000 vehicles travel across the Weizbach Bridge and the Arndorfbach Bridge on an average work day, and around 13,000 vehicles travel across the Salmbach Bridge. 12% of these vehicles is heavy traffic consisting of trucks and occasionally of special transport vehicles (Fig. 7). A large proportion of this heavy traffic serves the industrial fabrication yards along this roads, and also delivers agricultural goods from and to farms in that particular area. The special transport vehicles using road B64 typically move heavy industry products. To serve as an example, the axle layout for such a typical special transport vehicle is given in Fig. 7. The Weizbach Bridge (Fig. 3) is a 3-span bridge with a doubleTee deck cross-section and a 10 + 20 + 10 m span arrangement. The bridge is slightly curved in plan view and has skew supports. There are elastomeric bearings at the abutments and hinged connections at the piers. The bridge was built in 1969. In the second (middle) span there are distinct vertical bending cracks at mid-span with widths from 0.1 to 0.2 mm. Near the piers typical shear cracks running at a 45 angle can be observed in the side view. The crack pattern for this bridge was deemed fully developed. The Salmbach Bridge (Fig. 4) is a slightly skew simply-supported slab bridge with a width of 13 m spanning 7.5 m. The bridge was built in 1993. Some cracks in the transversal direction could be observed on the underside of the deck slab, however, none of the cracks could be observed to run the full width of the bridge. The irregular distances between cracks indicated that the cracking pattern was only partially developed. The Arndorfbach Bridge (Fig. 5) is a 3-span slab bridge with a span arrangement of 6.3 + 12.6 + 6.3 m. There are elastomeric bearings at the abutments and hinged connections between piers and the deck. The bridge was built in 1967. Some cracks in the transversal direction could be observed on the underside in the middle of bridge. Again, the crack pattern was only partially developed. During close inspection of this bridge it could be observed that there are a few more cracks under the lane in the WeizGleisdorf (WG) direction.

Fig. 1. Changes in loading (L) and structural resistance (R) over time.

Fig. 2. Schematic representation of the relationship between overloading and design life time reduction.

schematically by the peak labeled Overload in Fig. 2, can reduce the structural resistance suddenly by shifting the accumulated damage to a higher level. If the accumulated damage remains below the given critical value the structure can remain in service after such an overloading event. Whether the line for accumulated damage progresses parallel to the initial line (line (a) in Fig. 2), or whether the increase in damage is accelerated (line (b) Fig. 2) after an overloading event depends on the nature of the damage and on the specific structure. A reduction in remaining life time must be assumed in any case because the critical value for accumulated damage will be reached earlier. The present paper focuses on damage caused by road traffic on reinforced concrete bridges. A damage model was developed that used the fatigue damage caused by stress oscillations in

Fig. 3. Weizbach Bridge.

M. Pircher et al. / Engineering Structures 33 (2011) 37553761

3757

Fig. 4. Salmbach Bridge.

Fig. 5. Arndorfbach Bridge.

3. Damage model Most certainly, the described structures have accumulated, and continue to accumulate, damage through a multitude of compounding effects, one of which is fatigue in the reinforcement steel bars. Reinforcement fatigue was assumed to be representative of the damage accumulated in these structures for the purpose of this study. In order to measure fatigue damage, suitable bending cracks were identified on the bridge undersides in mid-span areas of predominant bending action. Displacement sensors were mounted to observe the cracks opening and closing under traffic loading. Using these crack width measurements the associated stresses in the reinforcement bars crossing these cracks could be computed using the considerations described in Gilbert [3]. These considerations assume constant bond shear stress between reinforcement bars and surrounding concrete. Fatigue effects on this bond shear are not considered in this model. The first natural frequencies were estimated for all three bridges using numerical models. It was found, that in order to gain a sufficient number of readings per vibration cycle in the first vertical mode a sampling rate of 200 Hz was adequate. This rate was used throughout the monitoring program to gain timestress diagrams. Loading cycle information could be filtered from these diagrams using the Rainflow method [4,5]. After deducting the load

spectrum from this data, Miners rule [6,7] could be applied to determine the fatigue damage of various measurement data subsets by taking the specific fatigue properties and the SN curve (k1 = 5, k2 = 9, N = 106, Rsk = 195 N/mm2 ) of the present reinforcement steel [810] into account. 4. Monitoring system LVDT displacement sensors with a resolution of 10 nm were used to measure the changes in crack width due to traffic loading, not including the effects of permanent loading. The sensors were mounted on one side of each selected crack using bolted and glued connections with the deck. The distance to a fixed point across the crack was then measured as seen in the inset of Fig. 6. The deck temperature was also measured in two places and the measurements were used for plausibility checks. All sensors were calibrated before they were mounted. DAQ USB 6212 data loggers by National Instruments were used. Power supply was organized mostly with solar panels and batteries and partly with a temporary connection with the power grid for the Weizbach Bridge. More details can be found in Pircher and Lechner [1113]. A view of the underside of the Weizbach Bridge with the monitoring system in place is given in Fig. 6. Additionally, a weight-in-motion system was implemented that allowed the calculation of velocity, axle

3758

M. Pircher et al. / Engineering Structures 33 (2011) 37553761

Weiz Gleisdorf

Gleisdorf Weiz

Displ. sensor Mahr 1303 Hirth Displ. sensors HBMWeta 2 Mahr 1300 box

70mm 30mm

crack Temperature sensor

Fig. 6. Monitoring system: displacement sensors, temperature sensors, box for data logger and power supply management.

Fig. 7. Axle loads and distances for a typical special transport vehicle.

arrangement and axle loads for individual vehicles driving across either one of the three bridges. This weight-in-motion system utilized the same input data generated by the same monitoring system as for the damage model [14]. All three bridges carried one lane for traffic traveling in the WG (WeizGleisdorf) direction and a second lane for traffic traveling in the opposite direction, GW. The Salmbach Bridge has one additional lane in the middle for left-turning traffic. Sensors were mounted underneath each lane in order to distinguish between the effects of traffic in both directions. The monitoring system and the weigh-in-motion system were tested and calibrated during a series of calibration runs using a car with a weight of 2 tons, a 2-axle truck with a weight of 18 tons and a 3-axle truck with a weight of 25 tons driving across the bridges in both directions at various velocities in both directions. During these calibration runs the bridges were blocked to all other traffic. Axle loads and distances between axles were accurately measured prior to these calibration runs. Using this calibration data the relationship between traffic loading by discrete vehicles and changes in crack width could be established. After these calibration runs the monitoring system was in operation for a minimum of 8 weeks for each bridge, first at the Weizbach Bridge during the summer of 2008, and then at the two other bridges simultaneously during summer in 2009. Measurements were taken continuously during these monitoring periods. Several passages of special transport vehicles were recorded during the monitoring periods. The monitoring data for these passages was identified and analyzed separately in order to gain an understanding of the level of damage caused by these special transports in relation to everyday traffic.

Fig. 8. Damage caused in one minute intervals on Weizbach Bridge in the WG direction.

5. Results 5.1. Contribution of heavy traffic The monitoring data was split into data sets of one minute duration. The damage according to the described damage model was computed for each individual one-minute data set. The average damage caused by one passage of the 2-axle truck during the calibration runs was computed for each direction, on each one of the three considered bridges, and was used as a unit for the damage computation for the one minute intervals. Fig. 8 shows a graphical representation of the damage accumulated in these one

M. Pircher et al. / Engineering Structures 33 (2011) 37553761

3759

Fig. 11. Accumulated damage Arndorfbach Bridge WG. Fig. 9. Damage caused in one minute intervals on Arndorfbach Bridge in the GW direction.

Fig. 12. Accumulated damage Weizbach Bridge WG. Table 1 Damage contribution of traffic percentiles. Percentile (%) Arndorfbach Bridge GW 100 99 95 90 80 100 95.09 81.44 68.3 47.89 WG 100 92.95 76.12 60.95 38.74 Salmbach Bridge GW 100.00 93.71 77.53 62.51 40.21 WG 100.00 91.72 72.84 56.94 35.69 Weizbach Bridge GW 100.00 69.4 31.8 9.5 1.0 WG 100.00 48.2 18.7 7.3 2.5

Fig. 10. Accumulated damage Salmbach Bridge WG.

minute intervals for a period of 17 days for the Weizbach Bridge direction WG, and Fig. 9 shows similar results for the Arndorfbach Bridge direction GW. There is a driving ban for trucks during nights and on week ends which is reflected very clearly in Figs. 8 and 9. The damage computed for the one minute intervals was summed up resulting in diagrams showing the accumulation of damage over time for each lane on each bridge. Figs. 1012 show these diagrams for vehicles traveling in the WG direction for each of the considered three bridges. Furthermore, the one minute data sets were sorted according to the computed damage. For the lines labeled 99% in Figs. 1012 the most damaging 1% of the data sets was omitted. For the lines labeled 95% the most damaging 5% were omitted and so forth for the lines labeled 90% and 80%. The damage contributions for these percentiles are tabulated for both directions in Table 1. These values can be interpreted as a measure for the contribution of each respective percentile to the overall damage. On all three bridges, the contribution of heavy traffic is disproportionate; however, for the Weizbach Bridge this contribution was especially high. On the Weizbach Bridge the contribution of the top one percent of data sets accounted for more than half of the overall damage on this bridge in the WG direction, compared to just under 10% for the other two bridges. 5.2. Damage caused by special transport vehicles The most damaging vehicle passages for each bridge were determined and the data for these loading events was analyzed with a weigh-in-motion system giving the number of axles, individual axle loads and distances between axles for these passages. Table 2 lists key data for the two most detrimental passages on each bridge in the two directions. The damage of these

loading events is measured again in equivalent damage caused by an 18 ton 2-axle truck. It should be noted here, that loaded special transport vehicles typically travel on this road with finished pieces of industrial equipment from the fabrication yards toward the highway system, which is in the WG direction. Therefore, for the sake of comparing damage caused by special transport vehicles with the damage caused by ordinary trucks and everyday traffic only the data for this direction should be considered. The Weizbach Bridge has a 20 m middle span which can accommodate most axle loads on a typical special transport vehicle, thus minimizing the favorable effects of loading on adjacent spans. The most detrimental passages on this bridge during the monitoring period in the WG direction were all caused by special transport vehicles. The Arndorfbach Bridge on the other hand has a much shorter middle span of 12.6 m resulting in less detrimental loading and less damage generated by the passage of special transport vehicles. The most detrimental passages in this case were recorded for relatively short and heavy trucks with five (and in one case 4) axles. The number of axles of a special transport vehicle fitting on the Salmbach Bridge is limited by the relatively short simply supported span of 7.5 m resulting in a limitation of the damage that can be caused by these special transport vehicles on this bridge. It was observed, that short and heavy trucks with four or five axles caused

3760 Table 2 Most detrimental passages. Bridge WG

M. Pircher et al. / Engineering Structures 33 (2011) 37553761

GW Comment Special transport vehicle Special transport vehicle 5-axle truck 5-axle truck Special transport vehicle 5-axle truck Equivalent 2-axle trucks 1386 1170 11 11 Comment 4-axle truck 4-axle truck 5-axle truck 5-axle truck (n/aa ) (n/aa )

Equivalent 2-axle trucks Weizbach Bridge Arndorfbach Bridge Salmbach Bridge

a

1935 978 9 7 21 15

The signal for this lane was not clear enough to be used as input for the weigh-in-motion system.

damage at a similar level to the damage caused by special transport vehicles on this bridge during the monitoring period. Looking at the data for damage caused by passages in the other direction GW all bridges showed the highest damage values for relatively short and heavy 4-axle trucks and 5-axle trucks. The difference in damage caused by heavy traffic between the two directional lanes is further illustrated by Table 1 where it can be seen, that for all percentiles of the one minute data sets the damage was consistently higher in the WG direction than it was in the GW direction. This trend indicates that the damage by heavy transportation from fabrication yards toward the highway system causes proportionally more damage than the traffic traveling the other way. It is especially interesting to compare the damage values for the top percent of the one minute data sets for the two directions. The top percent for the WG direction contains the special transport vehicles, along with other heavy traffic. On the other hand, the top percent of the one minute intervals in the GW direction only contains heavy traffic other than special transport vehicles. The difference between the two directions is around 2% for the Salmbach Bridge and the Arndorfbach Bridge, but more than 20% for the Weizbach Bridge, indicating that special transport vehicles have an especially damaging effect on the Weizbach Bridge. For the Weizbach Bridge the data indicated that the damage of one day of average traffic caused about as much damage as one passage of a special transport vehicle. For the Arndorfbach Bridge the damage of one average day is matched by the passage of 330 heavy 5-axle trucks as listed in Table 2 and for the Salmbach Bridge it takes 75 of the most detrimental passages to match one day of average traffic. 5.3. Comparison of results for the three bridges Heavy traffic causes disproportionate damage on all three considered bridges. However, on the Weizbach Bridge the damaging effects of heavy traffic, and especially of special transport vehicles, are much higher than on the other two. Close inspection confirmed, that the Weizbach Bridge can be considered fully cracked with bending cracks at regular distances clearly visible around the mid-span on the underside of the bridge and obvious shear cracks near the piers. The other two bridges displayed some bending cracks perpendicular to the travel direction near mid-span, but at irregular and rather long distances. From the inspection data, it could be assumed that the Salmbach Bridge and the Arndorfbach Bridge were nowhere near a fully cracked state. For the present study, crack widths were measured and stresses in the reinforcement across these cracks were computed from the measurement data. It can be assumed that cracks open wider, and stresses in the reinforcement across these cracks are higher, in a bridge that has developed a full crack pattern as is the case for the Weizbach Bridge. In this context, it should be noted that typical SN curves for reinforcement steel are given in a double logarithmic scale. Small increases in stress cause comparatively

high increases in damage when using these curves as a basis for applying Miners rule to determine a measurement for damage. In the opinion of the authors of this study, the fact that the Weizbach Bridge is in a fully cracked state is the main reason for this bridge to display higher damage values for heavy traffic than the other two bridges. Additionally, it should be noted that the structural system of the Weizbach Bridge deck is that of a double-Tee beam while the other two bridges are slab bridges. The two-dimensional load distribution of slab bridges also yields favorable results in the context of the present analysis when compared to the strictly onedimensional global load bearing system of the Weizbach Bridge. Discounting passages of special transport vehicles it could be shown, that short but heavy trucks have a highly detrimental effect on all three bridges. 4-axle vehicles and 5-axle vehicles with short distances between axles were identified as the most common source of passages with high damage values. These passages were found to be especially damaging when axle loads were unevenly distributed. 6. Conclusions A damage model for the determination of damage caused by traffic was developed and applied on three reinforced concrete bridges of various length, span arrangement and structural system. Monitoring systems were mounted and operated for several weeks on each of these bridges in order to gather input data for the developed damage model. The three bridges considered for this study are situated along a rural highway that connects an industrial area with the major highway system. Special transport vehicles with high loading travel along this rural highway and over the three considered bridges. The applied damage model used the stress oscillations in the reinforcement steel under traffic loading as a basis for fatigue considerations. The stresses were computed from measurements of bending crack width changes under traffic loading. Using this damage model it could be shown that heavy traffic contributes disproportionately to the overall damage caused by traffic in general. The extent of this contribution varied considerably between the Weizbach Bridge and the other two bridges. It was concluded, that the main reason for this variability was the fact that the Weizbach Bridge displayed a fully developed crack pattern while for the other two bridges only partial cracking had occurred. It could also be shown, that special transport vehicles had a detrimental effect in the sense of the applied damage model. The disproportional damage caused by special transport vehicles was very small for the Arndorfbach Bridge where the load distributing properties of these vehicles could work to the best advantage. The damage by special transport vehicles was the highest for the Weizbach Bridge where the mid-span was long enough to avoid favorable loading of adjacent spans. It could also be shown for all three bridges that relatively short, but heavy trucks can cause high damage, especially in the presence of high single axle loads.

M. Pircher et al. / Engineering Structures 33 (2011) 37553761

3761

Acknowledgments The authors wish to thank Professor Emeritus Russell Bridge for his generous support. The research project described in this paper was supported by funding from the Styrian Research Foundation (SFG) and the Local Styrian Government. References
[1] Moses F. Bridge reliability concepts and methods. In: Frangopol DM, editor. Bridge safety and reliability. Reston VA: ASCE; 1999. [2] Eichinger M. 2003 Beurteilung der Zuverlssigkeit bestehender Massivbrcken mit Hilfe probabilistischer Methoden, Dissertation Technische Universitt Wien. [3] Gilbert RI. Control of flexural cracking in reinforced concrete. ACI Struct J 2008; 105(3):3017. [4] Matsuiski M, Endo T. Fatigue of metals subjected to varying stress. Japan Soc Mech Eng 1969. [5] Downing SD, Socie DF. Simple rainflow counting algorithms. Int J Fatigue 1982; 4(1):3140.

[6] Palmgren A. Die lebensdauer von kugellagern. Zeitschrift des VDI 1924;58: 33941 [in German]. [7] Miner MA. Cumulative damage in fatigue. J Appl Mech 1945;12:15964. [8] DIN 2003 DIN Fachbericht 102 Betonbrcken, Beuth Verlag, Berlin. [9] ON 2007 Eurocode 2 Bemessung und Konstruktion von Stahlbeton- und Spannbetontragwerken, Teil2: Betonbrcken Bemessungs- und Konstruktionsregeln, ON sterreichisches Normungsinstitut [in German]. [10] Zilch K, Zehetmaier G, Glser C. Ermdungsnachweis bei Massivbrcken. In: Betonbaukalender 2004. Berlin: Ernst & Sohn; 2004. p. 309406 [in German]. [11] Pircher M, Lechner B. 2008 Schadenspotential von Sondertransporten auf der Weizbachbrcke, Report to Steiermrkische Landesregierung, FA 18B, Graz [in German]. [12] Pircher M, Lechner B. 2010 Schadenspotential von Sondertransporten auf der Salmbachbrcke und der Arndorfbachbrcke, Report to Steiermrkische Landesregierung, FA 18B, Graz [in German]. [13] Pircher M, Lechner B, Mariani O, Kammersberger A. Schdigung einer schlaff bewehrten Betonbrcke durch Verkehrsbelastung. Beton und Stahlbetonbau 2009;104(3):15463 [in German]. [14] Lechner B, Lieschnegg M, Mariani O, Pircher M, Fuchs A. A wavelet-based bridge weigh-in-motion system. Int J Smart Sensing and Intell Syst 2010;3(4): 57391.