Stability of precast prestressed concrete bridge girders considering imperfections and thermal effects

An experimental and analytical study was performed to determine the stability behavior of prestressed concrete beams. Two stability phenomenons were investigated: (1) lateral-torsional buckling and (2) rollover. An emphasis was placed on the effects of initial imperfections on the stability behavior the effect elastomeric bearing pads and support rotational stiffness was investigated. The experimental study consisted of testing six 40-in. (1016 mm) deep, 4-in. (102 mm) wide, 32-ft. (9.75 m) long rectangular prestressed concrete beams with varying prestressing force and prestressing strand eccentricity and testing one 100-ft. (30.5 m) long PCI BT-54 bridge girder. Elastic and nonlinear analyses were performed on the seven test specimens, on a hypothetical rectangular beam with a series of varying initial imperfections and a PCI BT-72 with varying imperfections.The first set of experiments was performed on the six rectangular beams. The beams were designed to fail by lateral-torsional buckling. The results showed that the prestressing strands did not restrain the beams from buckling out-of-plane or destabilize the beam like in the case of a beam-column. The beams buckled after flexural cracking had occurred and did so at a load much less than what elastic lateral-torsional buckling theory predicted. Initial imperfections were shown to decrease the inelastic lateral-torsional buckling load due to a rotated neutral axis, additional torsion on the cross-section and progressive rotation that led to a larger component of flexure about the weak-axis (P-delta effect).A material and geometric nonlinear, incremental load analysis was performed on the six rectangular beams. The nonlinear analyses matched the experimental load versus lateral displacement and load versus rotation behavior, and the analysis predicted the experimental maximum load within an error of 2%.The nonlinear analysis was extrapolated to several different initial imperfection conditions to parametrically study the effect of initial lateral displacement and initial rotation on the inelastic lateral-torsional buckling load. A simplified expression for lateral-torsional stability of beams with initial imperfections was developed based on an elastic stability expression (Goodier, 1941 and 1942). The data from the parametric study were used to develop reduction parameters for both initial sweep and initial rotation.The first experiment with the PCI BT-54 was a study on the deformation of the girder due to solar radiation. Solar radiation on the top and side of the girder, wind speed, internal strain, air temperature, internal temperature and surface temperature were recorded to determine additional sweep or rotation in the girder due to non-uniform heating. The research showed that the initial sweep of the 101-ft. (30.8 m) PCI BT-54 girder increased up to 40% due to the effect of solar radiation on the girder, an additional sweep of 0.0515-in. (1.31 mm) per 10-ft. (3.05 m) of girder length. However, only 0.000212 radians of additional rotation was developed due to the non-uniform heating of the girder.The PCI BT-54 was tested under midspan point load to examine its rollover behavior. For the stability experiment, full torsional restraint was not provided at the supports. Instead, torsional restraint was only provided by the couple created by the bottom flange and the elastomeric bearing pads. The load versus lateral displacement and load versus rotation response corresponded well with the prediction from the nonlinear incremental analysis that included a bearing pad model. A rollover failure occurred well before an inelastic lateral-torsional buckling mode was anticipated. In fact, the girder never cracked during the testing. The nonlinear incremental analysis did not predict the rollover failure because of assumptions made in the elastomeric bearing pad model. Imperfect bearing conditions were not modeled and nonlinear bearing stiffness behavior at large rotations was most likely inaccurate. The rollover methodology proposed by Mast (1993) predicted the rollover limit state very well.From the research, it was apparent that rollover is the controlling stability phenomenon for prestressed concrete bridge girders. The nonlinear lateral-torsional stability failure is unlikely because prestressed concrete bridge girders are designed to not crack under self-weight alone. Therefore, the inelastic lateral-torsional buckling simplified equation initial imperfection reduction parameters do not apply to bridge girders. Instead, the elastic lateral-torsional buckling predictions should be used. However, the elastic lateral-torsional buckling loads were found to be greater than the rollover limit for girders with no end support lateral bracing.