An experimental test program to evaluate the behavior of a 0.4-Scale model of a two-span continuous plate-girder bridge with modularprecast prestressed concrete deck panels has recently been completed. The bridge, designed according to alternate load factor design (alfd), or autostress design, procedures, utilizes noncompact plate girders with slender webs that fall beyond the present limits of the alfd guide specification. A comprehensive plan was followed to subject the model to a series of tests to evaluate specific responses at simulated aashto service load, overload, and maximum load levels. At elastic service-load stress levels, live-load lateral-distribution factors were computed from experimentally developed influence surfaces. These factors were compared with factors computed from a finite-element model, from current aashto procedures, and from proposed empirical formulas. The factors computed from the experimental and finite-element model data were generally in close agreement. The factors computed from the proposed empirical formulas for the interior girder also agreed closely with the experimental data. The factors computed using aashto procedures were quite conservative for the interior girder and less so for the exterior girders. Neither the proposed nor the current aashto procedures were found to account for the observed slight variation of the distribution factor along the span. The data would seem to indicate that finite-element analysis and the proposed empirical formulas are both plausible methods for computing elastic girder wheel-load distribution factors. For similar tests conducted after the formation of automoments along with subsequent shakedown at overload, the computed distribution factors varied less than 10%. Thus it appears that elastic distribution factors may still be used at overload, even though controlled local yielding is allowed in alfd procedures. In addition, for similar tests conducted with selected cross-frames removed, changes in the computed distribution factors were less than 10% in the positive-moment region for both interior and exterior girders. For the interior girder in the negative-moment region, distribution factors varied up to an average of 15% forthe tests conducted with cross-frames in place and with selected cross-frames removed. This suggests that the load was distributed primarily through the concrete deck. This paper appears in transportation research record no. 1275, Bridge research 1990.
Abstract