Static and dynamic behaviour of soil-steel composite bridges obtained by field testing

Long span soil-steel composite bridges, which are generally assembled from corrugated flexible plates and surrounded by compacted granular soil material started to become more frequent in highway and railway networks. They provide quick and more economical solutions compared to traditional bridges. The performance of soil-steel strucutures is governed by soil-structure interaction. The flexible steel part of the structure acquires part of its rigidity from the confining backfill soil. In the past, the design calculations were kept empirical because of the limited usage of soil-steel structures. Today there is a need for deeper understanding the soil-structure interactive behaviour and review of the available design procedures. This thesis presents the evaluation of several soil-steel composite bridges made of deep corrugated steel plates by in-situ measurements. The full-scale field tests comprised measurements during backfilling, under service and ultimate loads. One of the tested structures was an 11 m span single radius arch a railway bridge. The other 4 structures were box culverts with spans of 8 and 14 meters. Two of these four box culverts were reinforced by rib plates at the crown arch. The actual dynamic response of the railway bridge is studied during passages of a locomotive at different speeds. Service response and the ultimate bearing capactities of the structures were studied under different cover depths with a truck load, block weights and a hydraulic jack. Some of the measured results were compared with the theoretical design values. It was also possible to compare part of the results with results from a finite element model developed as a part of a master thesis using Plaxis, under supervision of the author. The field tests showed the effect of the depth of soil cover and soil confinement in the performance of soil-steel bridges. Under live loads, the structures became more vulnerable to applied loads as the soil cover decreased. The difference in moments and thrusts were 4 to 4.5 times when the depth of cover increased 3 times in structure with 8 m span. The moment and thrust ratios were higher for the structure with 14 m span. It was observed that the load bearing capacity increased linearly with increasing depth of cover. Finite element analysis results were conservative when live loading was concerned but the crown displacements and thrusts during backfilling were underestimated. Application of crown stiffening was more effective under shallow soil covers. The maximum crown displacement was reduced to half and the ultimate load capacity was doubled at the depth of cover of 45 cm by using crown stiffeners. Dynamic amplification factor, DAF, values as high as 1.45 were obtained for the moments at the quarter points and 1.25 for the crown displacements. The braking of the locomotive was found to increase dynamic displacements by 5% at the crown. DAF values were quite high compared to the values given in Eurocode and the Swedish bridge design code BV Bro. A revision in BV Bro’s method of the dynamic factor is suggested. The two measurements campaigns done on the 11 m long-span arch railway bridge, which were seven months apart (including the first winter), indicated that the structure has become more flexible in time. The Swedish design model calculates the thrusts due to backfilling with a safety margin of 2 for the arch cluvert bridge. The moments in the structures due to dead loads were found to be sufficiently predicted by the Swedish design method. The moments from the Canadian code, were conservative at zero depth of cover and then did not sufficiently increase with increasing soil cover. The design live load thrusts in the arch structrue calculated according to BV Bro loads was 50% higher than maximum thrust due to locomotive loading. If the real axle loads were used in the calculation the safety margin dropped to 1.25. The live load thrusts in the box culvert, however, were underestimated by the Swedish and Canadian bridge codes. Comparisons showed that the measured ultimate loads were considerably larger than what the design methods provide. The Canadian method, which is only applicable up to spans of 8 meters, was less conservative if the soil stiffness is high.


  • English

Media Info

  • Pagination: pp 55
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 01448340
  • Record Type: Publication
  • Source Agency: Swedish National Road and Transport Research Institute (VTI)
  • Files: ITRD, VTI
  • Created Date: Oct 1 2012 1:14PM