The Birmingham Bridge is located in Pittsburgh, PA and carries SR2085 over the Monongahela River. The main span of the structure is a 620 ft span tied arch designed in 1973 and consists of multi-girder approach spans, which flank each side of the tied arch.
Over the past several years fatigue cracks have been found in nearly all of the transverse floorbeams at the connection to the tie girders.
Specifically, horizontal cracks, some of which have several branches, have been reported at the upper web/flange weld in the web gap between the top flange and the connection angles. Many of these cracks are several inches long.
The fatigue cracking on the Birmingham Bridge has been observed on several other tied arch, truss, and plate girder bridges in the U.S. In these bridges, the floorbeams were also connected to the main r tie girders with a shear connection only. No direct connection existed between the flanges and the tie girders. Because there is no direct connection, longitudinal displacements between the top flange of the floorbeam the face of the tie girder can occur. This relative movement is focused within the horizontal web gap between the connection angles and the bottom of the top flange and results in the development of horizontal fatigue cracks along the web-to-flange weld in the floorbeam.
Although the magnitude of the displacement is very small, it is concentrated within the small web gap. The restraint provided by the top flange and the connection angles on the web force the sections of the web within the small web gap to bend in double curvature.
Longitudinal displacements are comprised of the global deflection of the bridge and local deflections/rotations due to trucks passing in the adjacent floorbeams. Because the displacement is related to the global deflections of the bridge, free vibration of the structure also produces the necessary driving force.
Hence, multiple cycles are accumulated during the passage of a single truck. The crack surfaces "moved" relative to each other in the longitudinal direction. The movement was found to be constant, no matter what longitudinal or transverse positions the vehicles were in.
Prototype retrofit and field instrumentation
To alleviate this problem, a portion of the floorbeam and connection angles were removed at the top of the connection in order to provide sufficient flexibility to allow the required movement.
In order to evaluate the performance of the proposed retrofit, the upstream and downstream connections at one interior floorbeam were retrofitted and instrumented.
The objectives of the instrumentation were as follows:
- Verify the adequacy of the retrofit as subjected to the random variable load spectrum;
- Determine the driving mechanism behind the observed cracking;
- Establish the magnitude of relative displacement between the top flange of the floorbeam and the face of the tie girder; and
- Use the field-measured data to calibrate and confirm the results of the FE model of the retrofit.
Data were collected for a period of almost 40 days as random vehicles crossed the bridge. Triggered time-history data and stress-range histograms were developed during the monitoring period.
While on site, data were viewed in real-time using a laptop computer directly connected to a Campbell Scientific CR9000 data logger. In order to upload monitoring programs, download data and view data in real time remotely from the ATLSS Laboratories, a high-speed wireless Internet connection was used. The connection to the Internet was provided by Carnegie Mellon University from the Pittsburgh Technology Center on the north shore of the river upstream from the bridge. Data were transmitted wirelessly from the north pier to a receiver that was connected to the Internet.
The measurements have confirmed that the primary cause of the observed cracking is due to relative longitudinal displacements between the top flange of the floorbeam and the face of the tie girder.
This relative movement was focused within the horizontal web gap between the connection angles and the bottom of the top flange. This resulted in the development of horizontal fatigue cracks along the web-to-flange weld in the floorbeam. Several cracks were observed to branch and begin to turn downward into the web.
Instrumentation of the prototype retrofits indicated that stress ranges produced by the random variable spectrum are below the CAFL at all locations. The retrofit provides sufficient flexibility at the connection and the remaining floorbeam connections can be retrofitted using the same detail to correct this problem at all locations.
