| nisee | National Information Service for Earthquake Engineering University of California, Berkeley |
Laura N. Lowes
University of California at Berkeley
One of the most vivid images of the 1989 Loma Prieta earthquake was that of the collapsed upper deck of the Cypress Street Viaduct in Oakland, California (fig. 1).
Fig. 1. Cypress Street Viaduct
At EERC, researchers have recently completed experimental tests of three 1/3-scale models of interior "T" joints.
Fig. 2. Multicolumn reinforced concrete bridge bent with test subassemblage shown in detail. Cross section of bent cap beam for the as-built and retrofit models are shown.
Caltrans provided the funding for this research program. These models represent an as-built beam-column connection typical of RC highway structures dating from the late 1950s and two retrofit as-built connections. Each model consisted of an interior column, the beam-column joint, and a portion of the bent cap beam on each side of the joint. The models were tested by applying a pseudostatic cyclic load to the base of the column in the plan of the cap beam. A constant axial load of approximately 0.03Agf'c was applied to the column throughout all tests. The tests were conducted by Prof. J. P. Moehle and graduate students Stephen Jirsa and Laura Lowes.
The first model tested was representative of an as-built joint (fig. 2). The design and detailing for this model was determined through review of engineering drawings. The design details that were expected to control the seismic performance of the joint included grade 40 column bars that projected straight in the joint, with development lengths of 20 times the bar diameter, minimal beam positive moment reinforcement continuous through the joint, and no joint transverse reinforcement. Reinforcement ratios for the bent cap beam and the interior column were designed within the observed ranges in the engineering drawings to give a nominal joint shear stress of 8 f'c psi at yield and to provide column flexural capacity less than the sum of the beam flexural capacities. Finally sufficient transverse reinforcement was provided in the cap beam and the column to ensure that neither would fail in shear.
Testing of the as-built joint produced the force-versus-displacement relation seen in figure 3.
Fig. 3. Hysteretic behavior of as-built model
The figure presents the theoretical capacity of the connection based on the nominal flexural capacity of the cap beam and column sections, and a model stiffness computed from gross section properties. The expected yield mechanism for this connection was yielding of the cap beam in positive flexure followed by flexural hinging of the column. The test results indicate that the beam-column connection carried a peak load of only 65 percent of the theoretical capacity, which corresponds to a nominal joint shear stress of 6 f'c psi. The softening of the system prior to reaching the peak load and subsequent failure of the connection were attributed to deterioration of the column bar bond within the joint that resulted in significant column bar slip. This conclusion is based on data from the instrumentation measuring the slip of the embedded ends of the column bars with respect to the top of the beam and is based on the observed crack pattern in the joint that clearly delineated the location of the column bars embedded within the joint.
Following testing of the as-built joint the first retrofit was designed. The retrofit consisted of RC bolsters cast on each side of the as-built beam section (fig. 2). The additional concrete thickness provided by the bolsters was intended to increase the joint area and to improve the bond stress capacity of the column bars. To prevent the flexural yielding of the beam from contributing to the deterioration of the column bar bond strength additional beam flexural reinforcement was added in the bolsters. Prior to casting the retrofit bolsters, the as-built beam section was drilled and dowel bars were expoxied into the holes to ensure an adequate connection between the existing beam section and the bolsters. The dowel bars were sized and spaced to transfer all tensile stress developed in the retrofit flexural steel into the existing concrete cap beam.
The expected yield mechanism for this retrofit connection was flexural yielding of the column. The connection did carry the nominal yield capacity of the column and a maximum nominal joint shear stress of 6 f'c psi. However, the observed cracking in the joint region was typical of what would be expected if the joint were overstressed in shear, and as loading of the joint continued beyond the yield point, significant slip of column bars and inelastic joint shear strain were observed. Because the failure mode of this connection was not the desired flexural yielding of the column, this retrofit was not recommended for implementation.
A second retrofit was designed based on a current Caltrans retrofit for this type of system. This retrofit consisted of adding post-tensioned RC bolsters on each side of the existing beam section (fig. 2). The post-tensioning force and the thickness of the bolsters were designed to reduce the maximum nominal joint tensile stress to 3.5 f'c psi at yield of the column. Dowel bars expoxied into holes in the as-built beam were sized and spaced to provide for even distribution of the post-tensioning force between the bolster and the beam concrete at a distance equal to the beam depth away from the joint face. Elsewhere, dowels were sized and spaced to provide for transfer of the maximum tensile force that could be developed in the bolster flexural steel.
The response of the second retrofit joint is shown in figure 4.
Fig. 4. Hysteretic behavior of retrofit #2 model
The bilinear envelope shown for this retrofit is based on gross section properties and on the nominal yield capacity of the column. The envelope does not account for strain hardening. The maximum load corresponds to a maximum nominal joint tensile stress of 4.3 f'c psi. The failure of this system resulted from fracture and buckling of the column longitudinal reinforcement at displacement levels beyond those shown. Observation of damage to the retrofit system seems to indicate that the joint region remained elastic during the entire test. However, data from strain gages embedded in the joint region concrete indicate that inelastic shear strain occurred. In addition, some slip of the column bars with respect to the top of the beam was recorded.
The results of these tests will enable engineers to better understand the issues critical to the retrofit design of beam-column joints. Further analysis of the data collected during this experimental program will concentrate on developing criteria for evaluating the capacity of as-built and retrofit beam-column connections. Additionally, it is expected that these data will improve engineers' understanding of stress distribution and force transfer in joints.
This article was originally published from EERC News, Vol. 15, No. 4, October 1994. The final report for this project was the Seismic Behavior and Retrofit of Older Reinforced Concrete Bridge T-Joints by Laura N. Lowes and Jack P. Moehle, report no. UCB/EERC-95/09, September 1995.
Updated October 28, 1998.
Send comments or questions to
EERC Library