Cyclic behavior of stiffened joints between concrete-filled steel tubular column and steel beam with narrow outer diaphragm and partial joint penetration welds

Abstract

This paper presented an investigation on a stiffened joint in practical engineering which was between concrete-filled steel tubular column and steel beam with narrow outer diaphragm and partial joint penetration welds. Through the low-frequency cyclic loading test, the cyclic behavior and failure mode of the specimen were investigated. The results of the test indicated the failure mode and bearing capacity of the specimen which were influenced by the axial compression ratio of the concrete-filled tubular column. On the contrary, the inner diaphragm and inner stiffeners had limited impacts on the hysteretic behavior of the joint. There was no hysteresis damage fracture on the narrow outer diaphragm connected to the concrete-filled steel tubular column with partial joint penetration welds. Due to the excellent ductility and energy dissipating capacity, the proposed joint could be applied to the seismic design of high-rise buildings in highly intensive seismic region, but axial compression ratio should be controlled to avoid unfavorable failure modes.

Keywords

narrow outer diaphragm / concrete-filled tubular column / joint / inner and outer stiffening / cyclic behavior

Introduction

Researches on steel-concrete composite structure are prevalent in China and abroad, mainly because that steel-concrete composite structures exhibit excellent seismic performance. It is more suitable for application in long-span bridges or high-rise buildings. For example, the US and some European countries, such as Germany and UK, mainly study concrete-filled circular or square tube and composite structure with encased steel. However, the study on stiffened joints between concrete-filled steel tubular column and steel beam with narrow outer diaphragm is scarce.

This kind of joints is generally designed with outer diaphragm or interior diaphragm. Experimental results [ 1] show that the connection with outer diaphragm is more feasible, because the transmission of force is clear, stress distribution of joint region is homogeneous and it is easy to be constructed. It has relatively large stiffness, good plasticity, and high bearing capacity [ 1]. Technical specification for structures with concrete-filled rectangular steel tube (CECS154: 2004) and Technical specification for structures with concrete-filled steel tube (DBJ13-51-2003) specify design formulas and structural measures for steel beam to concrete-filled steel tube with regular sized outer diaphragm. According to the formulas, combined with the axial force and moment, the control width b and thickness t₁ can be calculated. But sometimes in practical engineering, due to the limitation of wall thickness and edge distance of concrete-filled steel tubular column, narrow outer diaphragm is usually adopted, and the weld between the outer diaphragm and column face cannot be fully penetrated. This kind of connection is unconventional. There is no further study about its mechanical properties and failure modes. Shim et al. [ 2], Fujimoto et al. [ 3], Li et al. [ 4], and Shin et al. [ 5] investigated seismic performance of regular sized outer diaphragm, but there has been a lack of research on the influence on mechanical properties of joint with narrow outer diaphragm.

To understand the mechanical properties of each components in the structural system of this special kind of joints, experimental study and theoretical analysis are necessary. This paper presented an investigation on a joint between concrete-filled steel tubular column and steel beam with narrow outer diaphragm and partial joint penetration welds in a practical project. A scaled model experiment was tested. Combined with the results from finite element analysis (FEA), cyclic performance of this joint under cycle loads was investigated.

Joint construction

Narrow outer diaphragm

A 1:2.5 reduced-scale model was designed in this experiment. Dimensions of prototype and specimen are shown in Table 1. Concrete in the tube was C80, and steel was Q345C. d_c, t_c are diameter and thickness of column; h_b, b_b, t_w, t_f are height, width, thickness of web, and thickness of flange of steel beam respectively; t_of is thickness of outer diaphragm; t_if is thickness of interior diaphragm; and t_stt is thickness of vertical stiffening rib. The specimen was a cross-shaped joint, which was composed of 3.47-meter-high column and 4.5-meter-span deep beam. According to the relevant provisions in the Code for design of concrete-filled steel tubular structures (GB50936-2014), the thickness of diaphragm should not be less than the thickness of beam flange, and the width of the diaphragm should not be less than 70% of the width of beam flange. In the prototype, the width of beam flange is 600 mm, but the width of the diaphragm is only 200 mm, which does not comply with the above specification. As a result, this joint was classified as a special joint with narrow diaphragm. GB50936-2014 specifies if the dimensions of the diaphragm are not compliant with the above provisions, the diaphragm can be designed with the following method, which is listed in the appendix of the code.

The thickness of the diaphragm should be determined by the axial force of the beam flange N.

(1) t_{\mathrm{1}}={\displaystyle \frac{N}{b_{\mathrm{s}}f},}

where t₁ is the thickness of the diaphragm; b_s is the width of the beam flange.

The control width of the diaphragm should be calculated by the following formulas:

(2) b\ge {\displaystyle \frac{F_{\mathrm{1}}(\alpha )N}{t_{\mathrm{1}}{f}_{\mathrm{1}}}}-{\displaystyle \frac{F_{\mathrm{2}}(\alpha )b_{\mathrm{e}}tf}{t_{\mathrm{1}}{f}_{\mathrm{1}}}},

where

(3) F_{\mathrm{1}}(\alpha )={\displaystyle \frac{\mathrm{0.93}}{\sqrt{\mathrm{2}{\sin }^{\mathrm{2}}\alpha +\mathrm{1}}},}

(4) F_{\mathrm{2}}(\alpha )={\displaystyle \frac{\mathrm{1.74}\sin \alpha }{\sqrt{\mathrm{2}{\sin }^{\mathrm{2}}\alpha +\mathrm{1}}},}

(5) b_{\mathrm{e}}=(\mathrm{0.63}+{\displaystyle \frac{\mathrm{0.88}{b}_{\mathrm{s}}}{d}})\sqrt{dt}+t_{\mathrm{1}},

where α is the angle between the tensile force N and the section for calculation; t is the thickness of column wall; d is the external diameter of the steel column; f is the design tensile strength of steel column; f₁ is the design tensile strength of the diaphragm; b_e is the effective width of steel column wall participating in force with diaphragm.

Apart from the above requirements, the dimension of the diaphragm must also meet the following structural demands:

(6) \mathrm{0.25}\le {\displaystyle \frac{b_{\mathrm{s}}}{d}}\le \mathrm{0.75},

(7) \mathrm{0.10}\le {\displaystyle \frac{b}{d}}\le \mathrm{0.35}

(8) {\displaystyle \frac{b}{t_{\mathrm{1}}}}\le \mathrm{10.}

Obtained by calculation, the dimensions of the prototype structure described in this article met the above requirements written in the appendix of the code. However, these formulas in the appendix are generally introduced from abroad without related detailing [ 6], such as Architectural Institute of Japan [ 7], so it is still necessary to investigate the hysteretic behavior of this joint in the way of experiment and FEA.

Partial joint penetration welds

According to the relevant provisions Code for design of concrete-filled steel tubular structures (GB50936-2014), full penetration welds should be adopted between the diaphragm and the column wall. However, the thickness of the joint described in this article is only 30 mm, and the thickness of the diaphragm is 80 mm, so full penetration welds cannot be used.

Inner stiffeners

As shown in Fig. 1, in addition to the outer diaphragm and vertical stiffeners, there are also inner diaphragm and vertical stiffeners. The influence of these inner stiffeners was also investigated.

Experimental programme

Material test

The results of steel material test are shown in Table 2. Tested steel specimens have reached the design requirement of yield strength, and the elongations are greater than 20%. Moreover, their strength-yield ratios are all greater than 1.30. Concrete samples were standard cubes, and their average strength was about 64 MPa.

Test setup and loading

An overall view of typical test setup is shown in Fig. 2. Boundary conditions of top and bottom of the column and beam end were hinge joints. In order to prevent the lateral buckling, two out-of-plane braces should be set. At the top of the column, there was a vertical actuator to provide constant load of 160 kN. The compressive bearing capacity of the concrete-filled steel tubular column can be calculated by the following formulas, (GB50936-2014). The axial compression ratio was about 0.01.

(9) N_{\mathrm{u}}={\phi }_{\mathrm{l}}{\phi }_{\mathrm{e}}{N}_{\mathrm{o}},

if {\theta }_{\mathrm{s}\mathrm{c}}\le [{\theta }_{\mathrm{s}\mathrm{c}}], ,

(10) N_{\mathrm{o}}=\mathrm{0.9}{A}_{\mathrm{c}}{f}_{\mathrm{c}}(\mathrm{1}+\alpha {\theta }_{\mathrm{s}\mathrm{c}}),

if {\theta }_{\mathrm{s}\mathrm{c}}>[{\theta }_{\mathrm{s}\mathrm{c}}], ,

(11) N_{\mathrm{o}}=\mathrm{0.9}{A}_{\mathrm{c}}{f}_{\mathrm{c}}(\mathrm{1}+\sqrt{{\theta }_{\mathrm{s}\mathrm{c}}}+{\theta }_{\mathrm{s}\mathrm{c}}),

where N_o is the design value of axial compression bearing capacity of concrete-filled steel tubular short column; {\phi }_l is the bearing capacity reduction coefficient influenced by slenderness ratio; {\phi }_e is the bearing capacity reduction coefficient influenced by eccentricity; A_c is the cross sectional area of core concrete in the steel tubular column; f_c is the design value of compressive strength of core concrete; {\theta }_{\mathrm{s}\mathrm{c}} is the confinement index of concrete-filled steel tube; α is the coefficient related to the strength grade of concrete; [{\theta }_{\mathrm{s}\mathrm{c}}] is the limit value of confinement index related to the strength grade of concrete.

The specimen was loaded by synchronous controlling antisymmetric displacement of beam end. The specimen was tested pseudo statically according to the AISC cyclic loading protocol [ 8]. The amplitudes of relative drift angle were 0.375% (3 circles), 0.5% (3 circles), 0.75% (3 circles), 1% (3 circles), 1.5% (2 circles), 2% (2 circles), 3% (2 circles) and so on. The specimen was equipped with displacement transducers and strain gauges to measure deformation contributions and strain distributions at critical points.

Experimental phenomena

After applied the axial pressure 160 kN, the specimen did not appear obvious deformation. In each cycle, the east actuator went up first, then the sign was plus, and in the opposite direction, the sign was minus. Below the bottom outer diaphragm, in the first cycle at 0.75% rad, east of tube column wall yielded, and in the first cycle at 1% rad, west of tube column wall yielded. Above the top outer diaphragm, in the first cycle at ‒1% rad, east and west of tube column walls yielded, and at the same time, west of top outer diaphragm near the column yielded. In the first cycle at 2% rad, west of top outer diaphragm near the beam yielded, and at ‒2% rad, east of top outer diaphragm near the beam and west of bottom outer diaphragm near the beam yielded. In the first cycle at ‒3% rad, east of bottom outer diaphragm near the beam yielded. In the first cycle at 5% rad, south of tube column wall above the outer diaphragm yielded. In the first cycle at ‒6% rad, total cross-section of steel tube column yielded, and the tube column wall failed in a brittle manner, with a crack at west of column wall below the bottom outer diaphragm, and then the load of west actuator decreased. In the second cycle at ‒6% rad, half of the cross-section of the tube column wall below the outer diaphragm cracked, meantime east of the tube column wall above the top outer diaphragm cracked, and then the test was stopped.

The failure mode of test specimen is shown in Fig. 3. The tube column wall near the outer diaphragm cracked, and the local buckling of vertical stiffening rib also can be seen. During the test, the tube column wall near the outer diaphragm yielded first, and two ranks later outer diaphragms near the beam yielded. Besides, outer diaphragms near the beam also entered into plastic early and dissipated energy together. Because the axial compression ratio was too small (about 0.01), the tube column wall failed in yielding by tension, and this was confirmed by the results of finite element analysis as shown in the subsequent sections. By means of the finite element analysis, when the axial compression ratio increased (to about 0.5), the failure mode changed into column wall buckling by compression and overall instability.

Results

Cyclic behavior

As shown in Fig. 4, the hysteretic curve appeared antisymmetric till the end. The time when the specimen entered into plastic energy dissipation can be seen from this curve. M_p (1267 kN•m) means full plastic moment of beam end calculated with results from material test. When the hysteretic curve does not maintain straight, closed area is the dissipated energy through the plastic deformation of joint region. In the cycle at±1% rad, the curve of east beam end moment-rotation was no longer a straight line, and formed a closed hysteretic loop. It indicates that at this loading rank, the joint region began to dissipate energy through the plastic deformation. The maximum story drift was 6% rad, while the maximum bending moment did not reach M_p.

Shear deformation of joint region

The curve of joint region shearing force-shear deformation was plumper. In Fig. 5, V_y means theoretical value of shear capacity of this kind of joint calculated with results from material test and by formulas from the article [ 9]. As seen in Fig. 5, the negative amplitude of shear angle of joint region was close to the negative amplitude of story drift ‒6% rad. This illustrates that in the composition of story drift, the shear angle of joint region took up a big percentage. The positive amplitude of this curve was less than the negative amplitude. And they differed bigger for the following reasons. As seen in Fig. 6, the measured beam-end displacement is comprised of the following four parts: (1) the deformation caused by the bending of beam δ_b; (2) the deformation caused by the bending of column δ_c; (3) the deformation caused by shear of joint region δ_s; (4) the deformation caused by rigid body rotation δ_r. As seen in Fig. 2, there was a horizontal brace connected to the reaction frame at the east side of the top of column. When the east actuator moved up, namely the sign of the story drift plus, the horizontal brace at top of the column was in tension, so the displacement of the top of column was larger than in the negative direction of loading. That led to the increase of beam-end displacement caused by rigid body rotation δ_r, in other words, the proportion of the shear angle of joint region in total story drift decreased, and also the residual shear plastic deformation after unloading. With the increase of loading level, the contribution of residual shear deformation to the asymmetry of total shear deformation sent up. With the loading loops constantly accumulated, ultimately the measured shear angle presented obvious asymmetric. In addition, the cracking of west of tube column wall below the outer diaphragm also can affect the asymmetry of curve.

Bearing capacity

In Fig. 4, P_uc (593 kN) is the theoretical value of the actuator loading when the column became full plastic, and P_ub (760 kN) is the theoretical value of the actuator loading when the beam became full plastic. When the tube column wall below the bottom outer diaphragm cracked, the bearing capacity of this joint dropped a little and the crack developed slowly. After some time, until the crack went through most of the tube column wall, the loading of actuator dropped significantly. At that time, the crack was unstable, and it meant the failure of the specimen. During the process, the maximum loading of actuator was 682.5 kN, more than P_uc, the theoretical value of actuator loading when the column became full plastic.

Ductility

Among the seismic performances of structures, ductility is a very important property. The ductility of structural elements or joints is often presented by ductility coefficient. The higher the ductility coefficient is, the better the ductility is. As for the calculation of displacement ductility coefficient, the following formula can be used, where D_u is the ultimate displacement of specimen, and D_y is the yield displacement.

(12) \mu ={\displaystyle \frac{{\mathrm{\Delta }}_{\mathrm{u}}}{{\mathrm{\Delta }}_{\mathrm{y}}}},

At present, there are no universal standards for determining the ultimate displacement and yield displacement. More commonly used methods are graphic method, energy method, two-tangent method and so on. The principle of the method used in this paper was shown in Fig. 7. Take the secant stiffness of a line from the origin of the curve to the point corresponding to one third of the ultimate bearing capacity as the elastic stiffness. Draw a horizontal line through the point corresponding to one third of the ultimate bearing capacity and name the intersection of this line and the loading-displacement curve point A. Draw a slash OA through the origin, and extend it to intersect with the horizontal line, which passes the point of ultimate bearing capacity, then get the intersection point C. Draw a vertical line through the point C, and the intersection of it and curve is point B, namely equivalent yield point. Take the displacement corresponding to the loading decreased to 85% of ultimate loading in the curve as the ultimate displacement D_u. While if the loading reduces, but not decreases to 85% of ultimate loading, take the displacement of termination point of test as the ultimate displacement.

The curve of beam-end loading and story drift was used to analyze the ductility of the specimen in this paper. Skeleton curve is the ligature through peak point in the first of all cycles at every loading level. Skeleton curve was shown in Fig. 8, and ductility coefficient was shown in Table 3. From the skeleton curve, the ductility coefficient of specimen was close to 3, and the maximum of story drift was 5.85% rad. In the AISC seismic provisions [ 8], there are limitations of structural story drift against the seismic force. When the moment of beam end decreases to 80% of full plastic moment of beam end, the story drift must be no less than 40% rad. As seen in Fig. 4, the ductility of specimen satisfied the requirements of the code.

Energy dissipation

As seen in the macro experimental phenomena, the deformation of joint region made up the large proportion of story drift. This indicates that the joint region exhibited strengths of the capacity of shear plastic dissipation. Besides, the joint released energy by cracking on the surface of column, and the intrinsic plasticity of material was brought into fully realized. During the analysis of seismic, equivalent viscous damping coefficient h_e and energy dissipation coefficient C_e are often used to judge the capacity of energy dissipation. Equivalent viscous damping coefficient and energy dissipation coefficient was shown in Fig. 9. The calculation method is shown as follows:

(13) h_{\mathrm{e}}={\displaystyle \frac{\mathrm{1}}{\mathrm{2}\pi }}{\displaystyle \frac{S_{(\mathrm{ABC}+\mathrm{CDA})}}{S_{(\mathrm{OBE}+\mathrm{ODF})}}},

(14) C_{\mathrm{e}}={\displaystyle \frac{S_{(\mathrm{ABC}+\mathrm{CDA})}}{S_{(\mathrm{OBE}+\mathrm{ODF})}}.}

The hysteresis loop through ultimate bearing capacity among the beam-end loading and displacement curves was chosen to calculate the equivalent viscous damping coefficient and energy dissipation coefficient. The hysteresis loop was shown in Fig. 10, and the values calculated were shown in Table 4. Equivalent viscous damping coefficient was 0.352, and energy dissipation coefficient was 2.21.

Degradation of strength and stiffness

By reference to Specification of Testing Methods for Earthquake Resistant Building, the strength degradation coefficient on same loading level l_i or global strength degradation coefficient l_j are required to present the strength degradation. l_i equals to the ratio of peak loading in every cycle on the same loading level to the peak loading in first cycle of corresponding loading level. l_j reflects the decreasing degree of global strength of the specimen with increasing displacement, and it equals to the ratio of peak loading on every loading level P_i to the ultimate loading of the specimen P_max. The change of strength degradation coefficient on same loading level l_i and global strength degradation coefficient l_j with changing displacement (D/D_y) was shown in Fig. 11. It indicated that after the specimen yielded, the strength degradation on same loading level was not obvious, even increased slightly sometimes. In the global strength degradation curve, after the specimen yielded, the curve still maintained longer horizontal interval length, and not lost bearing capacity quickly. Even if the specimen reached the failure load, it can still bear loads. The global strength degradation trend of the specimen was flat, and it shows that the ductility of the specimen in yielding and failure stage was good.

Stiffness can reflect the capacity of resisting deformation. In elastic stage, the deformation is recoverable without any residual deformation. In plastic stage, the specimen presents nonlinear deformation, and the stiffness will decrease with increasing displacement. This phenomenon is called stiffness degradation. As for the quantitative analysis of stiffness degradation, there are many methods. Because when the specimen failed, it produced obvious plastic deformation, and cannot recover completely, it was reasonable to choose secant stiffness to reflect the stiffness degradation in this paper [ 10, 11]. The calculation of stiffness degradation of the specimen is as follows.

(15) {{\displaystyle K}}_i={\displaystyle \frac{\left|{{\displaystyle P}}_i\right|+\left|{{\displaystyle -P}}_i\right|}{\left|{{\displaystyle \mathrm{\Delta }}}_i\right|+\left|{{\displaystyle -\mathrm{\Delta }}}_i\right|}},

where P_i and ‒P_i are the peak loading in the ith time of cyclic loading respectively, D_i and ‒D_i are corresponding displacements. As seen in Fig. 12, take the ratio of stiffness value in every cycle K_i to the stiffness value in the first cycle K₁ as y axis, and take D/D_y as x axis, and finish the curve. It can be seen that the stiffness of the specimen went downward with increasing displacement. The reason was the development of plastic deformation after the joint yielded can lead to cumulative damage [ 12], but the degradation was slow, which means good seismic capacity.

Numerical modeling of connection

FE modelling methodology

On the basis of the experimental results, the computer code ABAQUS was utilized for numerical analysis. Fig. 13 shows the boundary conditions of FE model. To increase computational efficiency, hybrid models with both solid elements and shell elements were adopted. To model all components but concrete the element specification was Shell S4R, and the concrete was modelled using Solid C3D8R elements. A combined kinematic/isotropic model capturing the cyclic inelastic behavior of metals was used to represent the steel material properties from tensile coupon tests. The plastic-damage constitutive model of concrete was adopted. The material grade of concrete was C65. The elastic modulus and Poisson’s ratio was 38.7 GPa and 0.2 respectively. Between steel tube and concrete, “Hard” contact was used for normal behavior, and for tangential behavior the friction coefficient was 0.25. First load step was applying axial force at column end. Secondly, the cyclic load was applied onto the loading point at the beam end using a displacement control mechanism (referring to AISC Seismic Provisions [ 8]), which was same as experiment.

Validation of FE model

BF32-0.01 indicates the thickness of beam flange was 32 mm, and the axial force ratio was 0.01 (axial force: 160 kN). The dimension of other components was same as test specimen. The stress of outer diaphragm and panel zone was higher to plastic stage. Because the numerical simulation of fracture was not imported, the crack could not be observed at the stage of later loading. However, the slight buckling of panel zone, vertical stiffeners, and the column wall under the outer diaphragm could be seen, in accordance with experiment, as shown in Fig. 14. Good correlation was observed between the global hysteretic response of the FE model and the test specimen, as shown in Fig. 4. The hysteretic curve of this joint had slight reduced pinch phenomenon, and kept anti symmetry in the process. The initial stiffness and unloading stiffness match well with the experimental results. The failure mode of the FE model BF32-0.01 was also the yielding of total cross-section of the column wall under the diaphragm as same as experiment. After the column wall became plastic, the plasticity developed slowly. The specimen reached the damage state over a period of time, and the ultimate bearing capacity was very close to P_uc. The bearing capacity calculated by FE was lower than the result from experiment, so this shows a tendency to safety if use FE to design.

Parametric study

Effect of thickness of beam flange

As shown in Table 1, the thickness of beam flange of prototype t_f was 35 mm, and the thickness would be 14 mm in the experiment according to the reduced scale factor 2.5. However, in order to research the behavior of diaphragm and joint clearly, so the failure mode of damage occurring to beam end should be avoided. As a result, the thickness of beam flange was increased to 32 mm in the experiment, as same as the thickness of diaphragm. Now different hysteretic response and failure mode of joints with different t_f would be analyzed through FE.

BF14-0.01 means the thickness of beam flange of FE model was 14 mm, and the axial force ratio at column end was 0.01. As shown in Fig. 15, when the FE model BF14-0.01 was uploaded to ‒0.06 rad, the stress of beam web near the diaphragm was little higher on account of transferring most shear. The beam flange buckled severely to form plastic hinge. The column wall of panel zone became plastic and buckled slightly. Most part of the diaphragm still kept elastic, and only a little part near the beam end entered into plastic. As shown in Fig. 16, the hysteretic curve of model BF14-0.01 was plumper relatively and its initial stiffness was lower than model BF32-0.01. When BF14-0.01 was loaded to the drift level of 0.05 rad, P_E reached peak value 465 kN, which was smaller than BF32-0.01, 578 kN. When BF14-0.01 was loaded to 0.06 rad, the counter force at beam end was 83% of peak value, which showed the bearing capacity declined slowly. M_p, which means the theoretical value of total cross-section yielding moment of beam, was 654 kN. When BF14-0.01 was loaded to 0.06 rad, the moment of beam end reduced to 95.8% of M_p. This satisfied the demands of AISC Seismic Provisions on ductility [ 1]. So, if the axial force ratio was still 0.01, but the thickness of beam flange decreased, the bearing capacity and initial stiffness of the joint decreased, and the failure mode was changed to plastic hinge produced at beam end.

Effect of axial force ratiot

BF14-0.2 indicates the thickness of beam flange was 14 mm, and the axial force ratio was 0.2 (4000 kN). BF14-0.5 represents the thickness of beam flange was 14 mm, and the axial force ratio was 0.5 (10000 kN). As shown in Fig. 17 (a), if the axial force ratio was 0.2, the failure mode was same as BF14-0.01, when the model BF14-0.2 was loaded to ‒0.06 rad, the beam flange buckled severely and formed a plastic hinge. Besides, the panel zone buckled and entered into plastic state. As shown in Fig. 17 (b), if the axial force ratio was 0.5, when loaded to ‒0.04 rad, the model failed in overall instability. Besides, the east of column wall and vertical stiffness near the diaphragm buckled. The stress of panel zone, diaphragm and nearby column wall was higher into plastic. As shown in Fig. 18, both of BF14-0.01 and BF14-0.2 failed in plastic hinge at beam end, and the axial force ratio had little effect on the bearing capacity of them. However, BF14-0.5 failed in overall instability, so higher axial force was not good for joint with lower bearing capacity. The curves of first two were fuller, and presented the degradation of stiffness and strength, which could not be seen in the curve of BF14-0.5. When the beam end moment of these three models decreased to 80% of M_p, the story drift of them were all no less than 0.04rad, which met the provision about ductility, but the ductility of BF14-0.5 was worst.

BF32-0.2 means the thickness of beam flange was 32 mm, and the axial force ratio was 0.2 (4000 kN). Similarly, the axial force ratio of BF32-0.5 was 0.5 (10000 kN). As shown in Fig. 19, the column wall under the diaphragm and panel zone of BF32-0.2 buckled slightly before cracked. Finally, BF32-0.2 failed in tensile failure of the column wall under the diaphragm, as same as the results of BF32-0.01 and experiment, so the increase of axial force was beneficial to the joint. However, BF32-0.5 failed in overall instability, which was similar to BF14-0.5, so the axial force was not good for the bearing capacity of the joint. The ultimate loads of beam end of these three models were 578 kN, 593 kN, and 493 kN respectively, as shown in Fig. 20. Besides, it also can be seen the ductility of BF32-0.5 was worst. In a word, if the axial force increased, the ductility would be worse, and the bearing capacity changed with failure mode.

Effect of inner stiffeners

BF14-0.5 means the thickness of beam flange was 14 mm, and the axial force ratio was 0.5, without inner diaphragm and vertical stiffeners. As shown in Fig. 21, a part of inner diaphragm of model BF14-0.5 entered into plastic. Whether there was inner stiffeners or not, the failure modes were same. As shown in Fig. 22, the bearing capacity of BF14-0.5noin was 440 kN, which was slightly lower than BF14-0.5, 458 kN. So, the existence of inner diaphragm and stiffeners had little effect on the stiffness, bearing capacity and ductility of joint.

Conclusions

In this paper, the results of test on stiffened joint between concrete-filled steel tubular column and steel beam with narrow outer diaphragm and partial joint penetration welds were reported. The main conclusions are as follows.

1) The main deformation features and failure modes were: a) shear plastic deformation of joint region; b) crack at heat-affected zone between the tube column wall and the outer diaphragm; c) buckling of vertical stiffening rib; d) buckling of the column wall beyond the welds between the tube column and outer diaphragm.

2) There was no hysteresis damage fracture on the narrow outer diaphragm connected to the concrete-filled steel tubular column with partial joint penetration welds.

3) If the thickness of beam flange was thinner to some extent, the failure mode would be changed, with lower stiffness and bearing capacity.

4) When the axial force ratio increased, the ductility would decrease. With different failure mode of the specimen, the bearing capacity of joint would be different.

5) Whether there was inner diaphragm and inner stiffeners or not, the hysteretic behavior of joint basically unchanged.

6) Sometimes in practical engineering, due to the limitation of wall thickness and edge distance of concrete-filled steel tubular column, narrow outer diaphragm and partial joint penetration welds can be used. This kind of joint can be applied to the seismic design of high-rise buildings in high intensive seismic region, but high axial compression ratio should be controlled to avoid unfavorable failure mode.

References

[1] Choi S M, Shin I B, Eom C H, Kim D K, Kim D J. Elasto一Plastic behavior of the beam to concrete filled circular steel column connections with external stiffener rings. In: Building for The 21st Century. Australia: Griffith University Gold Coast Campus, 1995

[2] Shim J S, Han D J, Kim K S. An experimental study on the structural behaviors of H-shaped steel beam-to-concrete filled steel square tubular column connections. In: Building for The 21st Century. Australia: Griffith University Gold Coast Campus, 1995

[3] Fujimoto T, Inai E, Kai M, Mori K, Mori O. Behavior of beam-to-column connection of CFT column system In the 12th WCEE 2000: the 12th World Conference on Earthquake Engineering. New Zealand: New Zealand National Society for Earthquake Engineering, 2000

[4] Li X, Xiao Y, Wu Y T. Seismic behavior of exterior connections with steel beams bolted to CFT columns. Journal of Constructional Steel Research, 2009, (65): 1438–1446

[5] Shin K J, Kim Y J, Oh Y S. Seismic behaviour of composite concrete-filled tube column-to-beam moment connections. Journal of Constructional Steel Research, 2008, (64): 118–127

[6] Cheng-Yu L, Liang-Li X, Lie L. Design Formula Indicated for Connection with Exterior Diaphragms to Concrete Steel Tubular Based Yield-line Method. Steel structure, 2010, 363–368

[7] AIJ. AIJ standard for structural calculation of steel reinforced concrete structures. Architectural Institute of Japan, 2001

[8] ANSI/AISC 341-10. Seismic Provisions for Structural Steel Buildings

[9] Qin G. Research on Behavior and Design of Steel Beam to CFST Column Joints with External Stiffening Ring. Lanzhou: Lanzhou University of Technology, 2009

[10] Wen Y, Zeng Q Y. A novel approach to elasto-plastic finite element analysis of beam structures using the concept of incremental secant stiffness. Finite Elements in Analysis and Design, 2010, 46(11): 982–991

[11] Ichinohe Y, Matsutani T, Nakajima M. Elasto-plastic behavior of concrete filled steel circular columns. In: Proceedings of the Third International Conference on Steel-Concrete Composite Structures: Special Volume. Japan: Association for International Cooperation and Research in Steel-Concrete Composite Structures, 1991

[12] Tang K K, Li Z X, He D D, Zhang Z H. Evolution of plastic damage in welded joint of steel truss with pre-existing defects. Theoretical and Applied Fracture Mechanics, 2010, 54(2): 177–181

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg