You are on page 1of 10

PR COMPOSITE JOINTS UNDER CYCLIC AND DYNAMIC LOADING CONDITIONS: A COMPONENT MODELING APPROACH

G.A. Rassati and S. No Department of Civil Engineering, University of Trieste Trieste, Italy Roberto T. Leon School of Civil and Environmental Engineering, Georgia Tech Atlanta, Georgia, USA

ABSTRACT
This paper presents a mechanical model suitable for the simulation of partially restrained (PR) connections subjected to monotonic and cyclic loads. The model is capable of simulating the behavior of a PR composite joint subjected to a generic load history, taking in account the influence of all the main components, including slip of the bolts, shear deformation of the panel zone and cracking in the slab. The model is also capable of determining the amount of energy dissipated by local plasticification in the various components, and thus allows an evaluation of their relative importance in the post-elastic behavior of the connection. For validation of the model, comparisons to the results of several experiments are discussed. These results show that the model is capable of tracking the main behavioral aspects observed in the tests, and provides very good quantitative fit to the experimental results.

INTRODUCTION Steel-concrete composite PR frames can significantly enhance both the structural and the economical efficiency of office and commercial multi-story buildings, when compared to the more traditional steel or concrete solutions. Furthermore, it is possible to achieve an engineering optimization by incorporating the actual joint response into the analysis and design procedures (1, 2, 3), even in seismic zones. Partially restrained composite frames are in fact suitable for buildings in moderate seismic zones, and their use is explicitly mentioned in recent design codes (4, 5). Existing research (6, 7, 8, 9) demonstrates that PR, or semi-rigid, steel and/or composite frames can provide equal or better seismic performance when compared to their fully rigid counterparts. This is mainly due to the decrease of the natural frequency of the structure, which causes a corresponding decrease of the seismic forces. In addition, the PR connections, if correctly designed, can provide both ductility and a non-degrading hysteretic behavior under cyclic reversal loading. However, particular attention should be paid to the effects of the higher modes in the design of PR frames, which are usually neglected in the design of conventional frames.

213

In order to make allowance for PR connection behavior, it is necessary to incorporate the actual joint response in the analysis and the design. This means that for the partially restrained frame design, a different modeling approach than that used for rigid frames is necessary. Three main categories of connection models can be identified: mathematical models based on curve-fitting to test results, 2D and 3D finite element models, and simplified mechanical models. Each of the different techniques of modeling the actual response of the connections has its advantages and drawbacks. Models based on curve-fitting are generally not recommended because they require extrapolation of numerous parameters and the database for any particular type of joint is rather small (10). The use of analytical models implies the acceptance of semi-empirical coefficients and formulations, in order to understand and explain such complex phenomena, thus losing in overall generality (11, 12, 13). Obviously, the more accurate way to model the actual behavior of a joint is by means of a thorough finite element idealization. This is very expensive in terms of time and effort, and is therefore not suitable for a design procedure (14, 15, 16). Moreover, a finite element analysis provides a large amount of local behavior data, from which it is difficult to identify trends in a global sense. Somewhere between the mathematical models and the finite elements methods lies the mechanical modeling approach. It inherits the characteristics of simplicity from the analytical methods, without directly depending on the experimental assessment of the connection characteristics. At the same time, the mechanical modeling approach works in an environment similar to the classic finite elements analyses, but provides a more focused set of results. The increase in calculation capacity in the modern personal computers has made this latter method the most applied in the recent years (3, 17, 18, 19). This paper presents the development of a generic mechanical model suitable for the simulation of PR connections subjected to monotonic and cyclic loads. The model is capable of simulating the behavior of a partially restrained composite joint subjected to a generic load history, taking in account the influence of all the main components, including slip of the bolts, shear deformation of the panel zone and cracking in the slab. The model will be described first, and will be followed by a series of illustrative comparisons between its performance and published experimental data. THE TRS1 MODEL A robust mechanical model, shown in Figure 1, and known as the IBK model was developed by a team led by Tschemmernegg (17, 20, 21, 22). This model has been validated by means of a thorough experimental program. This model is suitable for the simulation of cruciform joints under unbalanced negative (hogging) moments, but no provisions were made to allow for positive (sagging) or in general cyclically variable moments. Based on the IBK model, and on the provisions of the Annexes J of both Eurocodes 3 and 4, a similar joint model that features full allowance for cyclic reversal loading has been implemented recently (19). The model, shown in Fig. 2, has been implemented

Figure 1 The IBK model.

214

as a sort of super-element in a commercial finite element program, ABAQUS (23), by merging some user-defined elements. Overall, every cruciform beam-to-column joint consists of 25 nodes, 14 axial spring elements and 16 kinematic constraints.

4 5 1 2 3

6 5 2

Figure 2 The TRS1 model.

F
Fy 2Fy Kel Kh

F
F+ y
+ Kel

K+ h


Kel


K
h

Rule 1: springs Rule 2: springs Rule 3: springs Rule 4: springs

3 4 6 2 1 5

F-y

1. Symmetric bi-linear law F


Fy Kel Kh

2. Unsymmetric bi-linear law F


F+ y Ksoft


Kel

F-y

3. Symmetric bi-linear law with gap element action

4. Concrete slab

Figure 3 Constitutive laws for the TRS1 model.

215

The constitutive laws for every spring have been derived basing on Eurocodes 3 and 4, as shown in Table 1. They are basically bi-linear elasto-plastic relationships with 5% hardening branches. Every rule has been derived and adapted to the cyclic case based on Tschemmerneggs monotonic rules, with two additions that were required to extend the model into the cyclic range: (a) the constitutive laws for the top and bottom steel connections do not allow a shortening with respect to the original length of the components; and (2) the deformation rules for the concrete slab that take into account the tension stiffening effect after the cracking has occurred in tension by means of a softening branch in the constitutive laws. The length of the spring to simulate slab effects has been chosen by taking into account the dimension of the portion of the composite beam in which the plane section hypothesis is not valid. A schematic representation of the constitutive laws for the elements is shown in Fig. 3. The most interesting component in the model is the so-called redirection spring: this component accounts for the behavior of the concrete slab immediately outside and between the column flanges under unbalanced moments. In this case, the slab tends to bear on the outside of one flange and on the inside of the opposite one, while it tends to shift away from the column on the other side of the joint. This mechanism is quite complex, as the behavior of the redirection spring requires that account be taken of: the effective width of the slab directly interacting with the column; the behavior under compression of the column web the flexural behavior of the column flanges the variation of the bearing surface of the slab on the column flanges at the increase of the bending deformation of these.

Stiffness Column web panel compression Column web panel shear Column web panel tension Steel connection - compression Steel connection - tension Slab rebars Redirection Strength Column web panel compression Column web panel shear Column web panel tension Steel connection - compression Steel connection - tension Slab rebars Redirection

EC3 ANNEX JJ

EC4 ANNEX J

J.4.4 (1) (b) J.4.4 (1) J.4.4 (1) J.4.4 (3) J.4.4 (1)

J.4.4 (2) -

J.4.4 (2) J.4.4 (2)

EC3 ANNEX JJ

EC4 ANNEX J

J.3.5.3 J.3.5.2 J.3.5.6 J.3.5.4 J.3.5.5,7,8,9

J.3.5.3 -

J.3.5.5 J.6.1, J.6.2

Table 1 EC3 and EC4 clauses used for strength and stiffness evaluation. For the evaluation of this components characteristics, the proposed model has used the work referenced in the Eurocodes (2, 3), Tschemmerneggs work (22), and particularly the work of Bernuzzi and Menapace (24). The only important mechanism that has been neglected in the implementation of this model is the slip between the concrete slab and the steel girder. It will be shown later that in the cases in

216

which that mechanism does not significantly affect the overall behavior of the joint, the model is capable to predict behavior with satisfactory accuracy. Work is underway to add this feature to the model, but as it stands, the model is unsuitable to simulate the behavior of joints and frames in which the shear slip is important (i.e., in presence of joints of columns to short girders). An interesting feature that has been added later in the numerical implementation of the model is the energy evaluation capability of every spring representing a deformation component of the connection. This helps identify which of the connection components needs more careful detailing, keeping in mind the energy dissipation requirements. In the next section, some validation simulations will be presented, as well as some of the analyses conducted in the past using the TRS1 model. ANALYSES USING THE TRS1 MODEL As a first step for the validation of the TRS1 model, simulations have been carried out in order to reproduce a series of experimental tests on cruciform specimens conducted at the University of Trieste (Benussi et al., 1995). The specimens were steel-concrete PR, isolated joints in a cruciform setup. For the sake of brevity, reference will be made only to the CT1C specimen (Fig. 4); simulations have been carried out for other specimens as well, with comparable results. Global results for specimen CT1C are shown in Figs. 5. The values of stiffness and strength for each component were derived according to the guidelines in Table 1 and following the constitutive laws shown in Fig. 3. The specimens were subjected to the ECCS short procedure (ECCS, 1992), implying a symmetrical load history of beam end displacements. Note that due to the symmetry of the loading, the influence of the redirection spring and of the spring simulating the shear deformability of the column web panel is negligible. The comparisons between experimental and numerical moment-rotation show reasonably good agreement. From the moment-rotation diagram (Fig. 5), it is noticeable that the model overestimated both the positive and negative moments.

CT1C
55 65 HEB 260 IPE 330 L200x90x15 M22-10.9

Figure 4 The CT1C joint specimen.

217

M [kNm]

400.0

300.0

200.0

100.0

0.0 -0.03 -0.02 -0.01 0.00 0.01 0.02

[rad]
0.03

-100.0

Experimental Theoretical

-200.0

-300.0

-400.0

Figure 5 Moment-rotation comparison for the left beam of CT1C. In the positive moment region, the overestimation is slight, and the agreement between the experimental and calculated ultimate forces, and the elastic loading and unloading stiffness can be considered satisfactory. The simulation bounds well the pinching behavior, slightly overestimating what could be considered the yielding force. However, in the negative moment zone, the predicted ultimate load is definitely unsatisfactory. This is due primarily to the overestimation of the yield force and the underestimation of the hardening stiffness of some of the key components. As seen from the initial comparisons to specimen CT1C, the model still needed some further calibration of the individual springs, but the model showed great potential to provide a closer look at the behavior of the single components of the joint during the cyclic loading. Further details about the CTxC simulations are given in (Rassati, 1997).

1800

1800

1250

1000 3500

1250

1750

3500 7000

1750

Figure 6 Details of the UCS frame.

218

The next step in the model validation was the simulation of a cyclic, non-reversal loading test on two different sub-frames (Benussi et al., 1995; No, 1995). Both sub-frames were one-story, two-bays setups. One (specimen SCS) had a symmetric configuration (two equal span length of 5 m), while the other (specimen UCS ) had a shorter left span (3.5 m) than the right one (7 m). Figs. 6 and 7 show details of the frames. The same materials and steel sections were used for both specimens. The load history was a monotonic ramp with two unloading cycles to check for the possible stiffness degradation. For the model, the TRS1 model was used, along with two-node cubic beam HEB 260 elements for the composite beams and 810 Feb44k columns. In addition, plastic hinge 65 elements were inserted in 110 45 correspondence to the load introduction 60 thickness sections on the beams. The latter was 120 10mm 60 necessary to take into account that collapse mechanism, which was observed in the experiments. It is IPE 240 27 66 27 important to point out again the absence BOLTS M20 10.9 1000 of a slip spring in the TRS1 model. In cases like the UCS sub-frame, with its Figure 7 Details of the UCS connection. short span, the approximation could not be acceptable anymore.
600

[kN]
500

Left beam

600

[kN]
500

Right beam

400

400

theoretical
300 300

theoretical

200

200

experimental
100 100

experimental

0 0 10 20 30 40 50 60 70

[mm]
80

0 0 10 20 30 40 50 60 70 80 90

[mm]
100

Figures 8a,b SCS frame load-midspan displacements for left and right beams
Right internal joint
600 600

[kN]
500

Right external joint

[kN]
500

theoretical

400

400

theoretical
300 300

experimental

200

experimental

200

100

100

[mrad]
-80 -70 -60 -50 -40 -30 -20 -10

0 0

[mrad]
-40 -35 -30 -25 -20 -15 -10 -5

0 0

Figures 9a,b SCS frame moment-rotation curves for internal and external joints For the SCS frame, Figs. 8(a) and (b) show the load-midspan displacement for the left and right beam, respectively. It should be pointed out that some of the discrepancies shown in these figures between the predicted and measured values may be the result of how the loads are applied to the model. The comparisons of the ultimate rotations and deflections are particularly sensitive as to whether a load or displacement history is used. For these analyses, the loading

219

600

[kN]

Left beam

400

[kN]
350 300 250 200

Right beam

500

theoretical
400

theoretical

300

experimental
200 150 100 100

experimental

[mm]

50 0

[mm]

0 0 20 40 60 80 100 120

20

40

60

80

100

120

Figures 10a,b UCS frame load-midspan displacements for left and right beams

Right internal joint

400

400

[kN]
350 300

Right external joint

[kN]
350 300

theoretical
250 200

theoretical
250 200 150 100 50

experimental

150 100 50

experimental

[mrad]
-70 -60 -50 -40 -30 -20 -10

0 0

[mrad]
-45 -40 -35 -30 -25 -20 -15 -10 -5

0 0

Figures 11a,b UCS frame moment-rotation curves for internal and external joints measured on the real structure was applied on the model. When large inelastic effects occur and relative low hardening stiffness are used, a small discrepancy in the loading values may result in big differences in the rotations or displacements. Irrespective of this, the initial stiffness was predicted with good accuracy, as well as yielding and ultimate loads. For the SCS frame, the overall results of the simulation are fairly good also for the moment-rotation curves (Figs. 9(a) and (b), especially when looking at the extreme simplicity of the constitutive laws used. The results of the UCS frame, on the other hand, were not as good, as shown in Figs. 10 and 11. While the simulation of the behavior of the long (right) beam is as good as the for the SCS frame, the prediction of the behavior for the left (short) beam is unsatisfactory because the model neglects the shear slip between the beam and slab. Further details on these simulations are found in (19). A further step in the validation of the TRS1 model was the modeling of a more complete specimen, subjected to cyclic loading. For this purpose the sub-frame tested by Leon et al. (1987), was modeled. The results are shown in Fig. 12 for a model with identical characteristics as that used for the simulations of specimen CT1C. Fig. 12(a) shows the moment-rotation diagram for the external left joint of the SRCF2C, and indicates that the calculated positive and negative ultimate moments are close to the experimental ones. In fact, the negative moment is predicted almost exactly, while the positive one is slightly underestimated. The elastic initial stiffness prediction is reasonably good, but not so the unloading stiffness. As for the rotations, the simulation results are fairly close to the experimental values. Even better results were obtained for the overall force-displacement diagram (Fig. 12(b)). Except for a slight underestimation of the positive force, probably due to an excessive flexibility of some steel component the calculated diagram appear to follow quite closely the experimental one.In comparison to the CT1C simulation discussed above, the SRCF2C frame simulation shows the influence of the redirection and of the panel zone in shear springs due to the asymmetry of the loading. This is an early evidence on how the slab behavior in the vicinity of the column can influence the overall behavior of the joint.

220

200 150

250

[kNm]
200 150 100 50 0 -0.06 -0.04 -0.02 -50 0

[kN]

theoretical

100 50 0

experimental

[rad]
0 0.005 0.01

-0.015

-0.01

-0.005 -50 -100 -150 -200 -250

[m] 0.02 0.04 0.06

experimental

theoretical
-100 -150 -200

Figures 12a,b SRCF2C frame moment-rotation and load-displacement sample curves CONCLUSIONS Even if the TRS1 model showed some drawbacks, some of them due to uncertainties in the springs characteristics (e.g. redirection spring, concrete spring), some due to the neglecting of determinant behaviors in the connections, it demonstrated capable of simulating in a reasonably accurate fashion the behavior not only of the single connection but also of the whole structures involved. At this point, a new feature has been added to the model, i.e. the energy evaluation algorithms. In this way, the informations provided to the designer for each single component increase in detail, and it becomes possible to find out which component is the main source of energy dissipation in the connection, i.e. which one needs particular attention in detailing due to its importance in the overall behavior. REFERENCES 1. ECCS (European Convention for Constructional Steelwork), 1992, Recommended Testing procedure for assessing the behaviour of structural steel under cyclic loads 2. CEN (European Committee for Standardization), 1994a, ENV 1993-1-1 Eurocode 3: Design of Steel Structures Part 1: General Rules and Rules for Buildings 3. CEN (European Committee for Standardization), 1994b, ENV 1994-1-1 Eurocode 4 - Design of Composite Steel and Concrete Structures: Part 1.1: General Rules and Rules for Buildings 4. NEHRP (National Earthquake Hazard Reduction Program), 1998, Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, FEMA 302, February 5. AISC (American Institute for Steel Construction), 1997, LRFD Manuals, Vol. 1 and 2 6. Mazzolani F.M., Piluso V., 1996, Theory and Design of Seismic Resistant Steel Frames, E.&F.N. SPON, Chapman & Hall 7. Leon R.T., Hoffman J.J., Staeger P.E., 1998, Partially Restrained Composite Connections A design guide, Steel Design Guide Series, No. 8, AISC 8. ASCE (American Society of Civil Engineers), 1998, Design Guide for Partially Restrained Composite Connections, Journal of Structural Engineering, Vol. 124, No. 10, pp. 1099-1114 9. Chen W.F., Kishi N., 1987, Moment-Rotation relation of Top and Seat Angle Connections, Dept. Of Structural Engineering, Purdue University, Report CE-STR-87-4

221

10. Tschemmernegg F., 1988, On the nonlinear behaviour of joints in steel frames, in Connections in Steel Structures: Behaviour, Strength and Design (ed. R. Bjorhovde et al.), Elsevier Applied Science Publishers, London, pp. 158-165 11. No S., Spanghero F., 1993, Un modello del comportamento ciclico di giunti semi-rigidi composti (in Italian), Giornate italiane della costruzione in acciaio, Viareggio (Italy) 12. White D.W., Chen W., 1995, Generalized models for semi-rigid connections, 3rd International Workshop on Connections in Steel Structures, Trento (Italy) 13. Amadio C., Puhali R., Zandonini R., 1989, Leffetto della continuit parziale nei telai composti (in Italian), Giornate italiane della costruzione in acciaio, Capri (Italy) 14. Aribert J.M., 1995, Proposed clause J.4.5. in Annex J for EN1994-1-1, COST meeting, Trento (Italy) 15. Rex C.O., Easterling W.S., 1995, Finite element modeling of partially restrained beam-togirder connections, 3rd International Workshop on Connections in Steel Structures, Trento (Italy) 16. Tschemmernegg F., Brugger R., Hittenberger R., Wiesholzer J., Huter M., Schaur B.C., Badran M.Z., 1994, Zur Nachgiebigkeit der Verbundknoten (in German), Stahlbau, Vol. 63, No. 12, pp. 380-387 17. Tschemmernegg F., Queiroz G., 1995, Mechanical modeling of semi-rigid joints for the analysis of framed steel and composite structures, 3rd International Workshop on Connections in Steel Structures, Trento (Italy) 18. Benussi F., No S., Rassati G.A., 1997, Components modeling of semi-rigid joints in composite frames under monotonic and cyclic loadings (in Italian, English abstract), Giornate italiane della costruzione in acciaio, Abano Terme (Italy) 19. Tschemmernegg F., Brugger R., Hittenberger R., Wiesholzer J., Huter M., Schaur B.C., Badran M.Z., 1995a, Semi-rigid composite joints, Technical paper T1, COST meeting, Trento (Italy) 20. Tschemmernegg F., Frenkel V., Pavlov A.B., 1995b, Comparison between test results and proposal for EC4-Annex J (panel zone of a composite joint), COST meeting, Trento (Italy) 21. Tschemmernegg F., Huber G., 1995a, Compression region in the panel zone of a composite joint, Technical paper T2, COST meeting, Trento (Italy) 22. Tschemmernegg F., Huber G., 1995b, Shear region in the panel zone of a composite joint, Technical paper T3, COST meeting, Trento (Italy) 23. Huber G., Tschemmernegg F., 1998, Modelling of beam-to-column joints, Journal of Constructional Steel Research, Vol. 45, No. 2, pp. 199-216 24. Hibbitt, Karlsson & Sorensen, 1998, ABAQUS 5.8 Standard Users Manual, Vol. 1,2,3 25. Bernuzzi C., Menapace R., 1997, Modellazione in campo ciclico di giunti trave-colonna per strutture composte in acciaio e calcestruzzo (in Italian), Tesi di laurea, Universit di Trento

222

You might also like