Abstract:

There are a significant number of older on-system metal truss bridges still in vehicular service in Texas, USA.  A number of these are of significant historical interest due to their age and other unique features, and are either listed or eligible for the National Register of Historic Places.  Considerable interest exists in maintaining historic metal truss bridges in continued vehicular service.  However, achieving this goal is often problematic because of structural and functional deficiencies found in these bridges.  The structural load rating can often be low due to the initial low design loads used for the bridge combined with damage and deterioration that has occurred over the service life of the bridge.

Based on commonly used evaluation procedures, many of these older truss bridges may show deficient load ratings based on current standards such as HS20.  This paper examines if such low load ratings accurately reflect the true load carrying capacity of these bridges, and whether more realistic load ratings can be achieved through the use of more accurate structural analysis methods and field load testing.  A case study bridge located in Llano, Texas, USA, was selected for evaluation. This bridge, typical of the circa 1930 on-system truss bridges, featured non-composite slab on steel girder bridge decks and Parker trusses.  The bridge was studied through the use of conventional AASHTO load rating techniques, the use of more advanced structural analysis models, and extensive field load testing. 

Initial evaluation of the bridge indicated that the truss members showed inventory load ratings well in excess of HS20 using standard AASHTO load rating techniques.  Consequently, the truss members themselves did not pose a problem with respect to inadequate load rating.  However, the steel beams and stringers in the bridge floor system showed inventory ratings well below HS20, using standard AASHTO based load-rating techniques.  Based on initial evaluation of the Llano Bridge, it was clear that the floor system was controlling the load rating, and was the primary structural issue of concern for the bridge. 

An elastic finite element model was developed for the bridge floor system to determine if a higher load rating could be justified by using analysis methods that are more advanced and more exact than used in conventional load rating.  The model was constructed using the commercially available finite element analysis software package SAP2000.  The bridge floor model used standard beam elements to represent the beams and stringers, and shell elements to represent the concrete slab.  The model was constructed to represent a floor system with no composite action between the steel members and the concrete slab.  AASHTO HS20 truck loads were applied in positions on the model to produce maximum flexural response for both stringers and beams.  The maximum moments due to live load effects were computed and used to load rate the members.

The finite element analysis showed significantly lower moments in the stringers and beams than conventional AASHTO calculations.  This reduction in moment was attributed to two factors.  One factor was that the finite element analysis predicts that the reinforced concrete slab resists substantial moment, thereby reducing the moment that must be carried by the steel beams and stringers.  Thus, even without composite action, the finite element analysis shows that the slab provides a significant contribution to the load-carrying capacity of the bridge deck.  Conventional AASHTO load rating procedures do not regard a non-composite slab as a load-carrying part of the deck system.  A second factor was that the finite element analysis predicted a different and more advantageous distribution of moments among the steel members than obtained from the AASHTO calculations.

Field load tests were subsequently conducted on the Llano Bridge to obtain the most accurate assessment of live load effects on the floor system, and to assess the accuracy of the finite element model.  In these tests, selected portions of the floor system were instrumented with strain gages.  Trucks of known weight and geometry were then driven slowly over the bridge, and the response of the instrumented members was measured.

The field test data showed live load stresses in the floor beams and stringers that were significantly lower than predicted by the standard AASHTO load rating.   In many cases, the field test data showed stresses that were less than half of those predicted by standard AASHTO calculations.  This confirmed that the bridge floor system was significantly stronger than indicated by the standard load rating, and that an increased load rating for the bridge floor members is justified.

The field test data also showed live load stresses smaller than predicted by the finite element model of the bridge floor system.  However, the difference between the field test data and the finite element model predictions were much smaller than the difference with the standard load rating calculations.  The field test data showed stresses that were 5 percent smaller than predicted by the finite element analysis for the stringers, and 25 percent smaller for the transverse beams.  Thus, the finite element model, although predicting much smaller stresses than standard load rating calculations, still provided conservative predictions compared to field test data. 

Figure 1 below shows the stress comparisons of AASHTO, FEA, and field load testing of the bridge deck members.  Shown are longitudinal members S2 and S5, members S3 and S4, and interior transverse member B2. 

Figure 1. Stress comparisons between AASHTO, finite element analysis, and field load testing.