Schneider and Alostaz 1998

From Composite Systems
Jump to: navigation, search

An analytical and a subsequent experimental study were conducted on steel girder to circular CFT beam-column exterior moment connections (Alostaz and Schneider, 1996; Schneider and Alostaz, 1998). Three-dimensional finite element analysis of various connection details was performed for monotonic flexural loading. According to the analysis results, six large-scale specimens were designed and then tested under cyclic loading. The authors compared the performance of each detail and highlighted the differences and similarities observed in their inelastic cyclic responses.

Analytical Study

A total number of 6 connection types (I, II, III, IV, V, and VI) were modeled in ABAQUS. The D/t ratios of the columns were 40, 53.5 and 80. The axial load over nominal axial load strength ratio was ranging between 0% and 36%. The third parameter, which was moment over shear ratio, varied from 0.0197 to 0.299. In addition, some connections were also modeled and analyzed as interior connections. The length of the columns were 144 in.. The yield strength of the steel was 45.98 ksi and 35.97 ksi for the steel tube and wide flange girder, respectively. The compressive strength of concrete was 5 ksi. The columns were pinned at the bottom and had a roller support at the top. The girders were subjected to monotonic shear force while a constant axial load was applied on to the columns. The steel tube and girders were modeled with shell elements and brick elements were used to model the concrete core. For steel, a bilinear constitutive stress-strain relationship with 2% strain hardening was selected both in tension and compression. Von Mises’ yield criterion and Prandtl-Reuss flow rule were utilized for the yield surface and plastic deformations, respectively. The 3D concrete model in ABAQUS, which was derived for low confinement, was used. The bond between steel and concrete were simulated with a thin layer of brick elements having low strength and stiffness properties.

Connection type I consisted of a flared connection stub welded to the steel tube and bolted to the girder. High local deformation of the steel tube was observed close to the tension flange of the girder. This caused the stiffness of the connection to decrease. The plastic moment strength of the girder was not attained. Decreasing the wall thickness from 53.5 to 40 did not improve the initial stiffness and it resulted little gain in strength. When the connection was analyzed as an interior joint, the initial stiffness remained almost the same and little reduction in strength was obtained. Despite higher panel shear force in the interior connection, no concrete crushing was observed. The experimental test of the same connection type had been completed at the time of the analytical study and it was found that the experimental and analytical results showed good correlation.

Connection type II had the exterior diaphragm detail. Two different kinds of this connection were analyzed. The first one had the minimum amount of diaphragm plate and the diaphragm plate of the second one had a larger angle at the flange tips moving the girder away from the connection. The first connection had similar response to connection type I. However, the second connection exhibited superior strength and stiffness properties. Its strength reached the plastic strength of the girder and the deformation of the diaphragm plate was very little close to the steel tube face.

Connection type III consisted of weldable deformed bars inserted into the CFT column and welded to the girder flanges. Although the stiffness was about half of the ideal rigid connection, the strength was close to the plastic strength of the girder. The steel wall distortions due to flange forces were not significant. Two modified connection details were also analyzed for this connection type. The first one was the internal connection and the second one had a shear tab between girder and steel tube to eliminate the need for flange weld. All the connections had almost the same strength and the unmodified type had the largest stiffness. Instability was observed for the connection with no flange weld and buckling of the deformed bars occurred at the gap between the flanges and the column. Thus this connection had the lowest stiffness.

Connection type IV had shear studs inside the steel tube at the same locations with the girder flanges. The stiffness of this connection was half of the stiffness of the ideal rigid connection and it behaved linearly up to 93% of the plastic strength of the girder. The shear studs farthest from the girder web found to have insignificant contribution to the tensile capacity of the flanges. As an alternative connection detail, the same connection was analyzed with less number of shear studs and the results showed that the inelastic response remained almost the same. Another parameter investigated for this connection was the moment-to-shear ratio and the analysis made for different values of this ratio. It was found that the behavior was more ductile in the case of low moment-to-shear ratio.

In connection type V, the web plate of the girder was extended into the CFT column with shear studs attached on it. This connection exhibited the least strength of the connections with a value of about 72% of the plastic strength of the girder. Its stiffness was about 28% of the ideal rigid connection. For small moment-to-shear ratio even lower strengths were obtained. The connection was also analyzed without shear studs on the web plate and became identical to the connection type I excluding the continuing web plate. In this case better strength and stiffness properties were obtained than connection type I.

For connection type VI, the steel girder was extended into the steel CFT column. The strength and stiffness of this connection type was close to those of the ideal rigid connection. There was no significant demand on the steel tube and slight increase in stiffness was obtained when D/t ratio decreased.

Experimental Study, Results, and Discussions

The main parameter of the experiments was the connection detail. The labels for the tested connections similar to the analytical study were I, IA, II, III, VI, and VII. The test setup consisted of a W14X38 shape girder having 108 in. length and a circular CFT column having 92 in. length. The girder was connected to the mid-height of the column. For the CFT columns, the D/t and L/D ratios were 56.6 and 6.6, respectively. The nominal yield stress of the steel tube was 42.1 ksi and nominal compressive strength of concrete was 5 ksi. The supports of the columns were pinned at the bottom and roller at the top. A cyclic shear force was applied at the tip of the girder and displacement controlled loading was utilized. The axial load acting on the columns was kept constant throughout the tests. The degradation of the elastic stiffness of the connection was also monitored by means of elastic cycles in the tests.

In connection type I, the flange tip of the connection-stub fracture was followed by the fracturing of the connection-stub weld. The failure of the specimen, at which the flaring connection-stub flange fractured completely, took place at 3.75% rotation. Only one flange fractured and the other flange remained undamaged. Thus, the hysteresis loop was not stable in the positive loading direction. However, it was stable in the negative loading direction while the connection was able to reach to the plastic moment strength of the girder. The concrete in the connection region was examined and observed to be uncrushed. Among the connection details, type I had the smallest shear and bending strength. The shear capacity deteriorated rapidly after the flexural resistance was lost due to fracture of the connection-stub flange. In addition, the intensive damage occurred on the tube wall increased the flexibility. Thus, the authors concluded that it was not appropriate to use type I connection in moment-resisting frames located in seismic regions.

Connection type IA was similar to type I. However, in type IA, the web of the connection-stub was extended into the steel tube. This resulted a better cyclic response. Strain hardening response was observed until 1.25% rotation. After this level of deformation, degradation in strength was observed due to weld fracture and tearing of the steel tube. Steel tube fractured at the tip of the flanges when the flanges were subjected to tension. The fracture grew as the loading continued and this resulted a 20% decrease in strength. Local buckling occurred in the connection stub flanges at 2.25% rotation and then a web fracture was observed in the girder causing a 16% decrease in flexural strength. At the end of the test, the free side of the steel tube remained undamaged. Rapid decay in moment strength was also observed in connection type IA. Identical to the type I, there was high flexural deformation demand on the web after the flange was fractured. Therefore, for fully restraint connections, it was required to prevent fracture of the flange. The continued web might have some rotational restraint in pinned-ended connections. Yet, it should exist for the required shear strength.

In connection type II, stain hardening behavior was observed up to 1.0% rotation. Due to high stress concentrations, fracture took place at the diaphragm plate acting in tension and the moment strength decreased. At the later stages of loading, fractures in the girder flange propagated into the steel tube. This caused the flexural strength to loose half of the plastic moment strength of the girder. At high deformation amplitudes, the groove weld between the girder flange and the diaphragm plate fractured. This was followed by the fracture of the steel wall along the depth of the connection-stub web. At 2.25% rotation, local buckling of one of the diaphragm plate was observed when the girder flange was in tension. The failure of the connection took place with excessive damage at 4.5% rotation. No concrete crushing was observed in the connection region. The external diaphragm used in connection type II improved the cyclic response. However, the degradation of moment strength was still rapid. Although the elastic stiffness was equivalent to that of an ideal rigid connection at initial stages of loading, rapid decay in stiffness was apparent at high amplitude displacement levels. In addition, extensive tearing at the re-entrant corners between the diaphragm and the girder flange imposed high deformation demand on the steel tube.

Connection type III had four 0.63 in. deformed bars welded to the both flanges of the girder and this connection showed a very stable hysteresis loop. At 3.75% rotation, steel tube tearing was observed between the holes opened to insert the deformed bars. However, it did not affect the response of the connection. At the same deformation level, local flange buckling took place beyond the region covered by the steel bars. The connection maintained its stable response up to 5.0% rotation and the connection failed by the rupture of the deformed bars. The deformed bars welded to the flanges resulted a desirable cyclic behavior for connection type III. It was appropriate to use this connection type in moderately seismic regions. The strength was the largest among the other type of connection details with a maximum value of 1.5 times the plastic moment strength of the girder. The elastic stiffness varied between 98% and 106% of an ideal rigid connection. No decay in stiffness was observed throughout the test. The detailing of the deformed bars was critical and the location of first weld on the bars should be adjusted to assure proper inelastic response. The use of small diameter bars was more practical due to the requirements for weld quality and spacing. Connection type VI was different from the connection type with the same label in the analytical study. The flanges of the connection-stub were extended into the CFT column and welded to the steel tube. During the test, the bond between the concrete and the continuity plate was not strong and this deteriorated the hysteresis loop. At 0.5% rotation fracture of the weld between the continuity plate and the steel tube took place. Initially, this did not cause any reduction in flexural strength due to interlocking between the welded parts. However, the strength was not maintained in the following cycles. At later stages of loading, steel tube tearing also occurred. This resulted insignificant reduction in strength but rapid stiffness degradation was observed. For connection type VI, the continuing flange improved the behavior compared to type I. The elastic stiffness underwent continuous degradation during the test. The slip of the connection-stub flange was critical as it imposed high deformation levels on the steel tube wall. For this connection to be used in seismic regions, this slip should be prevented. For this purpose, it was possible to use deformed bars placed on the continuing flanges and transferring the flange force to the concrete core. In addition, plates could be attached to each flange at both sides of the column and transferring the flange force directly to the CFT column.

In connection type VII, the girder continued into the CFT column entirely and welded to the steel tube. The specimen maintained its strength up to a rotation level of 4%. Local flange buckling of the connection-stub took place at about 7.5% rotation. At the same deformation amplitude, the weld between the steel tube and the connection-stub fractured. This resulted higher shear force to act on to the weld at the other side of the column and the weld in that region fractured. Web buckling also occurred and then deterioration in cyclic response was observed. At later stages of loading, fracture in the connection-stub flange also occurred. The flexural strength of this connection was 1.25 times the plastic bending strength of the girder. The concrete in the connection region was examined and there was no crushing. Type VII connection had identical response to an ideal rigid connection. Both the flexural strength and elastic stiffness remained stable throughout the test. This connection detail could be used in highly seismic regions.


  1. Alostaz, Y. M. and Schneider, S. P. (1996). “Analytical Behavior of Connections to Concrete-Filled Steel Tubes,” Journal of Constructional Steel Research, Vol. 40, No. 2, pp. 95-127. doi:10.1016/S0143-974X(96)00047-8
  2. Schneider, S. P. and Alostaz, Y. M. (1998). “Experimental Behavior of Connections to Concrete-Filled Steel Tubes,” Journal of Constructional Steel Research, Vol. 45, No. 3, pp. 321-352. doi:10.1016/S0143-974X(97)00071-0