Loading to failure and 3D nonlinear FE modelling of a strengthened RC bridge

Abstract

A reinforced concrete railway trough bridge in Ö rnsköldsvik, Sweden, was strengthened in bending with rods of carbonfibre-reinforced polymer and loaded to failure. The aim was to test and calibrate methods developed in the European Research Project 'Sustainable Bridges' regarding assessment and strengthening of existing bridges. A steel beam was placed in the middle of one of the two spans and was pulled downwards. Failure was reached at an applied load of 11.7 MN. It was initiated by a bond failure caused by a combined action of shear, torsion as well as bending after yielding in the longitudinal steel reinforcement and the stirrups. The bond failure led to a redistribution of the internal forces from the tensile reinforcement to the stirrups, causing the final failure. The computer models developed to simulate the loading process were improved step by step from linear shell models to more detailed models. The most developed model, a three-dimensional nonlinear finite element model with discrete reinforcement, gave accurate accounts of the response of the bridge.

Nonlinear FE analysis of a full scale in situ test on a RC bridge by means of a 1D layered frame model

“…It can be observed that the continuum simulations are very sensible to the parameters of the model and may present great discrepancy of results, even in terms of capturing the failure load. The higher quantity of parameters involved in these models, comparing with the 1D model, makes it (Puurula et al, 2013) fits well with the experimental results. These very good results were achieved with a great cost of detail and complexity of the model performed in a steadily manner by increasing the nonlinear materials step-by-step due to convergence problems (Puurula et al, 2013).…”

“…The higher quantity of parameters involved in these models, comparing with the 1D model, makes it (Puurula et al, 2013) fits well with the experimental results. These very good results were achieved with a great cost of detail and complexity of the model performed in a steadily manner by increasing the nonlinear materials step-by-step due to convergence problems (Puurula et al, 2013). Even tough, the model does not contemplate interface mechanism between concrete and FRP, so, it is not capable of capturing FRP debonding.…”

“…A good agreement was achieved in terms of load-displacement curve, with an underestimation of the failure load. Regarding the 3D model, the bridge was simulated as a single solid part with nonlinear materials and discrete reinforcement by means of 152,460 elements and 164,003 nodes with a total number of variables of 511,317 (Puurula et al, 2013): shell elements for the steel beam, solid elements for the concrete bridge, 2-node linear 3D truss elements for reinforcement.…”

Test to failure of a steel truss bridge – Calibration of assessment methods

“…Some 40 strain gauges, 10 deflection gauges and 3 temperature sensors were mounted on the bridge during the summer of 2012 in order to check strains and deformations in critical sections, Blanksvärd, 2013. The highest strains observed when ordinary trains passed over the bridge were of the order of 250 x 10 -6 corresponding to a stress of 50 MPa.…”

“…Täljsten et al 1994, 2006, Carolin, 2003, Blanksvärd, 2009, Sas, 2011& Puurula et al, 2012. This question was studied in the EU Project Sustainable Bridges, 2007, and a full scale test to failure was there performed on a reinforced concrete trough railway bridge, SB-7.3, 2008& Puurula, 2012 The bridge remained elastic up to about three times the original design load and the load could then be almost doubled with substantial yielding deformations before a buckling failure appeared in the top girders for a load of 11 MN (1000 short tons) for a midpoint deflection of ca.0,2 m (8 inches). No brittle or fatigue failure in any of the joints appeared and the bridge proved to behave in a ductile way with a substantial hidden capacity.…”