Sie sind auf Seite 1von 10

Computers and Structures 77 (2000) 91100

www.elsevier.com/locate/compstruc

Analysis and diagnostic testing of a bridge


Andrzej S. Nowak*, Sangjin Kim, Pawel R. Stankiewicz
University of Michigan, Ann Arbor, MI 48109-2125, USA Received 25 November 1998; accepted 27 June 1999

Abstract The paper deals with analytical and experimental study to determine the cause of transverse crack patterns on two parallel seven-span haunched steel-girder bridges. Major cracks are concentrated between the beginning and the middle of the third span. The crack patterns are symmetric with respect to the fourth span and can be found on both bridges. Diagnostic tests were performed with crawling-speed and high-speed trucks. The measured distribution factors are consistently more uniform than code-specied values and the impact factors are smaller than specied values. Tests on concrete cores indicated that the water/cement ratio is higher than code specied ratio and that the strength of concrete is greater than specied. Three-dimensional nite-element analysis indicated that stresses in the slab due to live load do not match with existing crack patterns. However, the analysis of the concrete pouring sequence showed that maximum concrete stresses are large and the locations correspond to those of the most severe slab cracking. Therefore, it is concluded that the bridge deck slab cracks resulted primarily from the deck pouring sequence, combined with concrete shrinkage. Since the primary causes of cracks are not active anymore, an epoxy injection method has been recommended to seal cracks. 7 2000 Elsevier Science Ltd. All rights reserved.

1. Introduction The objective of this paper is to present the causes of transverse crack patterns found in a seven-span haunched steel-girder bridge built in 1968. Successful repair procedures should take into account the causes of the cracking. If the cracking was primarily due to drying shrinkage, then it is likely that after a period of time the cracks will stabilize. On the other hand, if cracks are primarily due to live load and insucient structural capacity, then simple sealing of cracks will be of no use. Therefore, the rst step for successful repair is to check the structural capacity of the bridge. This study includes a series of diagnostic tests, concrete core tests and nite element analysis to check the

* Corresponding author. Tel.: +1-734-764-9299; fax: +1734-764-4292.

structural capacity of the bridge and to nd the causes of transverse crack patterns. The bridge is on the Huron Parkway over the Huron River in Ann Arbor, Michigan. It carries mainly car trac and only a small number of trucks were observed. It consists of two completely separated bridges with about 25 mm gap between them: east bridge and west bridge (Fig. 1). Both bridges have the same girder section and slab thickness and each bridge carries two-lane trac. The major dierences are girder spacing, presence of a sidewalk on the east bridge and trac lane arrangement. Figure 2 shows the elevation view. The rst and last three spans are continuous and a center span has a suspended span. The suspended center span structurally separates the south and north part of the bridge and makes the bridge symmetric with respect to its center. Shear studs are provided only between splice points in the middle of each span. The major transverse crack patterns are shown in

0045-7949/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 5 - 7 9 4 9 ( 9 9 ) 0 0 1 8 8 - 1

92

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

Fig. 1. Cross sections of the Huron Parkway bridge: (a) West bridge; (b) East bridge.

Fig. 2. They are symmetric with respect to span 4. The cracks are concentrated between the beginning and the middle of the third span. The cracks can also be found in the fth span, symmetrically with respect to the center of the bridge. The crack spacing is approximately 1 to 1.5 m. Some cracks are found in spans 1 and 7. Span 4 has virtually no cracks. Spans 2 and 6 are relatively clean. The same crack patterns were found on both east and west bridges but the cracks are slightly more severe in east bridge. Minor corrosion of steelgirders was found due to water leakage through transverse deck cracking. 2. Field tests of the bridge Field tests were planned so that the bridge's response to static and moving truck loading could be investigated. Specic purposes were to investigate actual stresses under test trucks, load distribution and impact eects. The results of the tests were used to validate the nite-element modeling. 2.1. Tests and instrumentation Strain transducers were placed at selected sections as shown in Fig. 2. Important factors considered in the selection of strain transducer locations are observed crack pattern, maximum stress, maximum bending moment, trac control and ease of approach to steel girders. Strain transducers were installed at both bot-

tom ange and top ange or web of the steel girders. The steel girders were approached with a ladder, a cherry picker type vehicle, or a snooper truck depending on the span. To compare the response between east and west bridges, strain transducers were installed in both east and west bridges for spans 1, 5 and 7. Each bridge is capable of carrying two lanes of trafc. Hence, crawling-speed tests and high-speed tests were performed with a single truck and with side-byside trucks. Each test run was repeated at least twice. Each truck weighing approximately 200 kN maintained the same trac lane during the tests. In addition, crawling-speed tests with two bumper-to-bumper trucks in each lane were performed to observe maximum stresses from test trucks. Tandem-axle dump trucks loaded with gravel or sand were used to apply the test loads. The test trucks were provided by the City of Ann Arbor, Michigan. Typical truck conguration used in the tests is shown in Fig. 3. 2.2. Test results Strains were collected with a sampling rate of 15 samples/s for the crawling-speed tests and with a sampling rate of 200 samples/s for the high-speed tests. Strains from crawling-speed tests were ltered with a low-pass digital lter to remove minor dynamic strains and to obtain static strains. Typical stress records measured at span 5 of the east bridge are shown in Fig. 4 together with approximate locations of trucks. When trucks were on spans 1 to 3,

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

93

Fig. 2. Elevation view of the bridge, major crack pattern and strain transducer locations.

the vibration was transferred to the span 5 across the suspended span. This was expected because the hinge in the suspended span is not a perfect hinge. Maximum stress measured under the combinations of two trucks was 22 MPa at bottom ange, which was caused by the bumper-to-bumper trucks. No appreciable dierence between stresses from the two bridges was observed. 2.3. Evaluation of superposition The rst indication that the bridge behaves satisfactorily would be how well the principle of superposition applies. Using superposition, the sum of the stresses measured under a single truck in left and right lanes should be equal to the stresses measured under sideby-side trucks. Figure 5 shows the maximum bottomange stresses obtained at section C (Fig. 2) of the east bridge under a single truck and under side-by-side trucks. The sum of the stresses under a single truck in left and right lanes is also shown for comparison with measured maximum stresses under side-by-side trucks.

There is good agreement between the values determined under the single- and side-by-side truck tests. 2.4. Measured distribution factors The knowledge of actual distribution and impact factors is important for a rational evaluation of bridges. The factors from the test could be used to check the behavior of the bridge. If necessary, the measured factors could be used to rene rating calculations instead of the factors dened by AASHTO [1]. Strains from the crawling-speed tests were used to calculate wheel-load distribution factors. Fig. 6 shows wheel-load distribution factors obtained at section C (Fig. 2) of the east bridge from side-by-side truck tests. The distribution factors are calculated with two methods: (1) the ratio of the strain at the girder to the sum of all the bottom-ange strains and (2) the weighted ratio considering the section modulus dierence of each girder due to the parapet and sidewalk. When section modulus dierence of each girder is considered, wheel-load distribution factors from the tests

94

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

Fig. 5. Evaluation of superposition of girder stresses.

Fig. 3. Test truck conguration.

show a more uniform distribution. Whether the section modulus dierence is considered or not, the measured distribution factors are well below the AASHTO value [1].

Fig. 4. Typical stress records measured at span 5: (a) Crawling-speed test, (b) High-speed test.

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

95

Fig. 8. Isometric view of discretized east bridge.

3. Material test of concrete slab A wide variety of poor construction practices can result in cracking in concrete structures. Foremost among these is the common practice of adding water to concrete to improve workability [2]. Several concrete cores were taken during strain transducer instrumentation. Petrographic analysis on concrete cores indicates that the water/cement ratio varies from 0.57 to 0.65. The ratio is much higher than AASHTO specied ratio of 0.45 [1]. ACI Committee 201 also recommends the water/cement ratio of 0.45 for bridge decks [3]. The strength of the concrete is greater than 34.5 MPa and most of the concrete is still in a good condition. 4. Finite element analysis of the bridge 4.1. Description of the FEM model Finite element analysis was performed to identify the causes of transverse cracking of concrete deck. The stresses in the concrete deck and steel girders under dead load and design live loads was calculated. The results of the analysis were also used to check the load carrying capacity of the bridge according to the current design standards. Field test measurements indicate a good agreement between the measurement for side-by-side trucks and the superposition of the measurements for left- and right-lane trucks. Therefore, a linear-elastic nite element model was used for the analysis. The discretized bridge consists of a total of 2381 nodes and 1273 nite elements including: . beam elements to model steel girders; . beam elements to include the stiness of the sidewalk and parapets; . eight-node quadrilateral shell elements to model the slab; . rigid links to connect between the centroids of shell and beam elements and to satisfy the compatibility of composite or non-composite behavior.

Fig. 6. Wheel-load distribution factors.

2.5. Measured impact factors The impact factor is an important component of bridge loads. The impact factor (1+I ) is dened as the ratios of stresses recorded in high-speed truck tests to crawling-speed test trucks (Fig. 4): Figure 7 shows impact factors obtained at section C (Fig. 2) of the east bridge from side-by-side truck tests. Measured impact factors are much smaller than AASHTO specied values except for girder 4. However, the large impact factor in girder 4 has no practical signicance since the stress in girder 4 (Fig. 5) is small compared with stresses in other girders. Generally, measured impact factors are well below AASHTO specied values [1] and the bridge is behaving satisfactorily.

Fig. 7. Impact factors.

96

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

Fig. 9. Comparison of nite-element analysis and eld measurement.

Non-composite behavior was assumed at the locations of no shear studs. It was modeled by removing horizontal displacement compatibility of rigid links between girders and slab. Each bearing was assumed to locate at the centroid of the bottom ange of the girder. Figure 8 shows an isometric view of the discretized east bridge. 4.2. Validation of the nite element discretization scheme To test the validity of the adopted discretization scheme, a numerical simulation of the eld test was conducted and compared with the eld measurements. For the purpose of comparison, the horizontal time axis of the measurements was transformed to geometrical axis. Figure 9 compares nite-element analysis and eld measurement for a west exterior girder at section

C (Fig. 2) of the east bridge from the left-lane truck. Remarkably good agreement with the eld measurements has been obtained. It shows that the bridge with even severely cracked concrete decks can be modeled reasonably well by linear-elastic nite elements. 4.3. Results of FEM analysis and causes of transverse cracking of deck For the purpose of identifying the potential causes for cracking of concrete deck and checking the capacity of the structure, the analyses have been performed for the following load cases: . dead load of the structure considering pouring sequence of concrete deck; . design live load according to AASHTO [1]; and . Michigan 11-axle truck (686 kN) [4].

Fig. 10. Concrete pouring sequence.

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

97

Fig. 11. Stresses in concrete deck due to concrete pouring sequence.

Figure 10 shows the concrete pouring sequence according to the built-in drawings. The sequence is symmetrical with respect to span 4. Such sequence is surprising considering that it induces large tensile stresses in the concrete deck in spans 3 and 5 and smaller

but reasonably high stresses in spans 1, 2, 6 and 7. It was later found that a coerdam was built the during concrete pouring period for construction equipment. Concrete was poured without shoring. Because of the almost symmetrical design of the bridge and the sym-

Fig. 12. Tensile stress envelope in concrete deck under HS20 live load.

98

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

Fig. 13. Steel stress envelopes under dead load, HS20 live load and impact load.

metry of the pouring sequence the stresses induced by construction sequence were also symmetrical. The intensity of cracking along the bridge and its symmetry can be well explained by the results of nite element analysis.

Figures 11 and 12 illustrate the stresses in the concrete due to dead load and live load for HS20 [1] and Michigan 11 axle truck respectively. The thin median line represents the level of stress due to dead load, taking the construction sequence into account. Thick

Fig. 14. Steel stress envelopes under dead load, Michigan 11-axle truck and impact load.

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

99

Fig. 15. Steel stress envelopes under dead load and Michigan 11 axle truck for top ange of girders (impact factor is 1.15).

curves show the envelopes of live loads superposed on dead load lines. The construction sequence induced the tensile stresses up to 3.3 MPa in spans 3 and 5 and about 1.2 MPa in spans 1 and 7. The total level of stresses in concrete due dead load and live load (impact factor 1.15) can reach even higher values (5

MPa in spans 3 and 5 and 2.7 MPa in spans 1 and 7). These values exceed the tensile strength of fresh concrete. The cracking was worsened by shrinkage strains due to high water/cement ratio. Additional contribution was probably due to the tensile fatigue of concrete in highly stressed regions.

Fig. 16. Steel stress envelopes under dead load and Michigan 11 axle truck for bottom ange of girders (impact factor is 1.15).

100

A.S. Nowak et al. / Computers and Structures 77 (2000) 91100

Figures 1216 present the stress envelopes in the steel girder under dead load and live load including in impact factor of 1.15. Michigan Department of Transportation uses HS20 live load for inventory rating and Michigan 11-axle truck for operating rating [4]. The calculated stresses are within the allowable limits for dead load combined with either HS20 live load or 11axle truck (686 kN). The bridge has a sucient load carrying capacity for the heaviest trucks acceptable on Michigan roads. 5. Conclusion The measured load-distribution and impact factors are well below AASHTO specied values [1]. Tests on the concrete cores taken indicate higher water/cement ratio than AASHTO specied values. Most of the concrete in the slab is still in a good condition with the strength greater than 34.5 MPa. Finite-element analysis showed that the stresses in the slab due to live load do not match with the observed crack patterns. This suggests that live-load stresses alone played a relatively minor role in transverse cracking. However, the analysis of the concrete pouring sequence showed that maximum concrete stresses are large and that the locations correspond to the those of the most severe slab cracking. Stresses in the steel girders due to dead load and live loads are below allowable values. It is concluded that the bridge deck slab cracks resulted primarily from the deck pouring sequence, combined with concrete shrinkage due to high water/ cement ratio. Since the stresses which cause the cracks have already been relieved by their occurrence, the structural integrity can be restored with some expec-

tation of permanence. Therefore, sealing the transverse cracks with epoxy injection methods has been recommended to repair the transverse cracks and to prevent further deterioration of the bridge as the most cost eective repair method.

Acknowledgements The presented research was performed with McNamee, Porter and Seeley, represented by Sean Kelley. The project was sponsored by the City of Ann Arbor, represented by Steve Rojeck, which is gratefully acknowledged. However, the results presented here are those of the authors and do not necessarily reect the opinions of the sponsor. Thanks are due to Chan-Hee Park and Vijay Saraf for their help in eld measurements and John M. Kulicki for fruitful discussions.

References
[1] Standard Specications for Highway Bridges. Washington, DC: AASHTO, 1992. [2] ACI Committee 224. ACI 224.1R-93, Causes, evaluation and repair of cracks in concrete structures. Manual of Concrete Practice, American Concrete Institute. Detroit, MI, 1994. [3] ACI Committee 201. ACI 201.2R-92, Guide to durable concrete. Manual of Concrete Practice, American Concrete Institute. Detroit, Michigan, 1994. [4] MDOT. Michigan Bridge Analysis Guide, Michigan Department of Transportation. Lansing, MI, December 1983.

Das könnte Ihnen auch gefallen