Anda di halaman 1dari 8

STEEL TO CONCRETE

MOMENT CONNECTIONS

Dalibor Gregor František Wald


Excon a.s. Czech Technical University in Prague
Czech Republic Czech Republic
E-mail: Gregor@excon.cz E-mail: Wald@fsv.cvut.cz

ABSTRACT
The paper describes the work on the analytical prediction model of behaviour of end plate
joints connecting the steel beams to concrete structure. Three sets of tests simulating the
beam–to–column joint’s components under cycling loading were carried out in the laboratory
of Czech Technical University to observe experimentally the components: the bolt in tension
and the end plate in bending, the anchor bolt in tension, and the end plate in bending and the
concrete in compression. An analytical prediction model for cyclic loading was developed
based on component method taking into account the test’s results as well as the existing
knowledge of the beam to column and base behaviour. The prediction of the model was
checked on published results with a good accuracy.

1 INTRODUCTION
Structural joints are designed assuming to be exposed to forces resulting from the quasistatic
loading. In the structures loaded cyclically by live loads or thermal and seismic actions,
connections are checked against fatigue separately. In many structures, however, the number
of cycles depends on load spectrum may reach 8 · 103÷105. In the joint the local yielding is
present. The low-cycle fatigue needs to be taken into account within the design procedure.
Information about the behaviour of joints subjected to repeated loading is important namely
for connections of elements from different material but is appropriate for all structures, even
for these where quasistatic approach have been approved in practice, as a good prediction
tool. The modified ECCS recommendation and classification for the cyclic loading procedure
[1] is commonly applied to compare the results of tests of the structural joints and to analyse
the results.
A component method was proved to be an effective analytical tool for determination of joint’s
behaviour. The method is based on the analytical modelling of components (individual parts)
of the joint separately [2]. The behaviour of each component is expressed by a spring.
A mechanical model of the joint assembles these springs and infinitely rigid plates. The
description of behaviour of each component by three basic design characteristics: an initial
(elastic) stiffness, strength (design value) and a deformation capacity enables an application of
the method into the practice. The resulting force-displacement relationship for the joint’s
design is thus bi-linear. The analytical description of components offers to designer
a freedom of geometrical variants. Extrapolation of the method to combination of both bolted
and welded parts of joints under loading by combination of the internal forces (bending, shear
and normal force) proved to be sufficiently accurate.
The cyclic overall behaviour of joints is traditionally predicted by curve-fitting models. For
fitting the model behaviour onto the tests, it is important to select main parameters of the
model properly. The parameters are characterised by constants which are to be defined on a
basis of test results or a sensitivity study by FE analysis. These models are reaching, within a
limited range of applications, the required accuracy. Any extrapolation out of the
experimentally proved geometrical set-up is not possible. In the last decade, the sophisticated
mathematical models for the steel beam-to-column joints were derived for an accurate
prediction of frame behaviour under seismic actions [3] and [4] and for column bases, see [5].

Fig. 1: Test of component concrete in compression and Fig. 2: End plate in bending and anchor bolts
end plate in bending; deformed shape of the plate in tension; failure mode of the end plate

2 COMPONENT TESTS
Three set of tests with components were performed. The end plate in bending and the
concrete in compression were tested under repeated loading, see Fig. 1. The tests with
threaded bar cast in the concrete block was designed to prepare the test of the component the
end plate in bending and anchor bolts in tension.
The test set-up of experiment concrete in compression and end plate in bending is shown
in Fig. 2 [6]. The tests were carried in two configurations. Three specimens were attached to
upper horizontal surface of the concrete block (representing concrete slab) and three
specimens were tested on vertical side of the block (representing concrete wall). The surface
of the concrete was cleaned and levelled by a thin (less than 1 mm thick) layer of high
strength grout to achieve smooth surface. The steel plate with nominal dimensions
200 × 100 × 10 mm was placed on the fresh grout layer. The steel bar, nominally
10 × 10 × 220 mm, was centred on the plate. The concrete block was positioned under the
head of the hydraulic actuator and a layer of plaster was made under the block to ensure the
level of the top surface and good support on the laboratory floor. An example of the result of
test C1/1 is presented in Fig. 3 [7].
Threaded bar cast in the concrete block loaded in tension was tested separately to learn a local
behaviour of the tension part of the connection, see [7]. Threaded bar M20 was concreted in
concrete block 500 × 500 × 500 mm. Two hydraulic jacks were placed on the plaster layer to
ensure the vertical position and good transfer of the reactions into the concrete block. The
beam made from two UPN 140 profiles with web stiffeners transfers the forces to the bar,
which was fixed to the beam by washer plate with thickness 20 mm and hole diameter 22 mm.
For the component end plate in bending and anchor bolts in tension two threaded bars M20,
540 mm long were cast into the concrete block. The T-stub was positioned at the concrete
surface four hours after the casting to ensure proper contact with the concrete surface,
no grout was used. The nuts were tightened before the test, after 120 days from concreting,
by torque of 40 Nm to simulate the hand tightening. Two hydraulic jacks were placed on the
block on grout layer. The beam of two UPN 140 profiles transfers the tensile force. Three set
of specimens were tested. An example of the result of a set of tests is presented in Fig. 4.
Force, kN
600

500

400 Experiment C1/1

300
Prediction
200

100

0
0 -1 -2 Deformation, mm

Fig. 3: Comparison of the predicted and measured displacement at centre of the plate for
the component the concrete in compression and end plate in bending, test C1/1

Force, kN
160
Failure load for TC1
140 Experiment TC1

120
Prediction
100

80

60

40

20
Deformation, mm
0
0 1 2 3 4 5 6 7 8 9 10 11

Fig. 4: Comparison of the predicted and measured displacement of the top of the T-stub,
the component end plate in bending and anchor bolts in tension, test TC1

3 MODELLING
For the component model under cyclic loading, the degradation of material and the assembly
of components including the change of the contact area in the given cycle and the history of
yielding in the precedent cycle are taken into account, see [8]. The degradation phenomena
may be described in the models of each component, see [9] and [10]. Due to hysteretic
character of behaviour it is not possible to determine a given point of the force-displacement
or moment-rotation curve without information about the load history. The step-by-step
procedure is used for establishing the force-displacement or moment-rotation curves, allowing
in each step for the cumulative degradation of the material and the deterioration of stiffness.
The curves are often simplified by use of linear approach.
The tests with components were published in [6], and [7]. The knowledge developed for base
plates and applied in European structural steel practice may be used for connections between
steel frame and reinforced concrete part of the structure, columns and walls. Two approaches
may be distinguished for the component in compression (the plate in bending and the concrete
in compression): concrete in compression under rigid plate and effective area of flexible plate.
Stiffness and resistance of the concrete in compression is limited by crushing of the concrete
surface. The behaviour is influenced by the concrete quality, the thickness and area of the
plate, the grout quality and thickness, the location of the plate on the concrete structure, the
size of the concrete structure, and its reinforcement [11] as well. The component plate in
bending and anchor bolt in tension is solved by T-stub analogy. The stiffness and resistance
are predisposed by elongation of the anchor bolts, which prevents development of prying
forces and guided failure modes different to the steel beam-to-column connections. The
proposed analytical prediction models of the major components are summarised in [11]. The
prediction of component behaviour in Fig. 3 and 4 is based on the measured values of yield
strength fy and ultimate strength fu of the steel and average value of concrete strength. The
simple step by step procedure with constant increment of 1/1000 of deformation was applied
to achieve the description of the working diagram of each component as well as of the whole
assembly. The unloading part of working diagram of each deformable component was
studied separately and models were developed according to [13] using a weakening factor.
A simplification adapting the initial stiffness seems to be acceptable even for the advanced
prediction.
SP1
c
54
d e
4x53
d e

54 c
16 300 16 35 80 35

Fig. 5: Test setup of the joint assembly [14]

4 COMPARISON TO TEST
The prediction model was developed from the tests of components, sees [6] and [7], was
compared to the available test of the whole assembly. The Fig. 5 exhibits the test performed
by Dunai at all [14], the bars transferred the tensile forces and the studs the shear as well as
tensile forces. Fig. 6 shows the comparison of the calculation to the test on moment rotational
diagram to the test. The significant influence of the shear studs by creating round 20% of the
bending stiffness of the joint was found. The springs c represent the compressed part
behaviour, springs d the bars, and e the header studs. The model was loaded by cyclic
actions based on the test records. The measured values of the material were applied in
presented simulation. The first, second and sixths cycle shows a good agreement of the
prediction by model to the test.
Moment, kNm Moment, kNm
120 120

80 Test 80 Test
Model Model
40 40

0 -20 -15 -10 -5 0 -20 -15 -10 -5


0 5 10 15 20 0 5 10 15 20
-40 -40

-80 -80
Rotation, mrad Rotation, mrad
-120 -120

Fig. 6a: Comparison of the model to test, Fig. 6b: Comparison of the model to test,
first cycle second cycle

Moment, kNm
120
Test
80
Model

40
-25 -20 -15 -10 -5
0
0 5 10 15 20 25
-40

-80
Rotation, mrad
-120

Fig. 6c: Comparison of the model to test, sixth cycle

Fig. 7: The parameters for the comparison of the model to test by the resistance ratio
The ECCS classification, see [1], of the joint characteristics under cyclic loading may help to
compare the prediction to the test results. The geometrical variables observed by each cycle
of the loading are summarised in Fig. 7. The cycle parameters may be represented by a partial
ductility, which is defined as
ei+ ei−
µoi+ = ; µ −
oi = (1)
ey+ ey−

a full ductility
∆ei+ ∆ei−
µi+ = ; µ i

= (2)
ey+ ey−

a resistance ratio
Fi + Fi −
ε i+ = ; ε i

= (3)
Fy+ Fy−

a rigidity ratio
tg (α i+ ) tg (α i− )
ξ = +
;ξ = −
(4)
tg (α y+ ) tg (α y− )
i i

and an absorbed energy ratio


Ai+ Ai−
ηi+ = ; η −
= (5)
Fy+ ⋅ ( ei+ − ey+ + ei− − ey− ) Fy− ⋅ ( ei+ − ey+ + ei− − ey− )
i

The functions derived from each cycle parameters depends on the partial ductility as full
ductility function ψ ( µo ) , see Fig. 8a, relative rigidity function ξ ( µo ) , see Fig. 8b, and
relative absorbed energy function η ( µo ) , see Fig. 8c. On the functions may be seen a good
description of tendencies in compressed part, on the right side of the diagrams, and on tensile
part, on the left one. The particular accuracy of the prediction is limited compare to the curve
fitting prediction.
Resistance ration
Experiment SP-1
Experiment SP-1
2,6

2,2

Prediction with f u 1,8 Prediction with fu

Prediction with f y 1,4 Prediction with fy

1,0

0,6

0,2

-12 -10 -8 -6 -4 -2 0 2 4 6 8 Partial ductility

Fig. 8a: Comparison of the model to test by the resistance ratio


Ductility ration
1,0
0,9
0,8
Prediction with fu 0,7 Prediction with fy
0,6
Prediction with fy Experiment SP-1
0,5
Experiment SP-1 Prediction with fu
0,4
0,3
0,2
0,1

-12 -10 -8 -6 -4 -2 0 2 4 6 Partial ductility

Fig. 8b: Comparison of the model to test by the rigidity ratio


Absorbed energy ratio
1,0
0,9
Experiment SP-1 Experiment SP-1
0,8

Prediction with fu 0,7


0,6
0,5

Prediction with f y 0,4 Prediction with f y


0,3
0,2 Prediction with fu
0,1

-12 -10 -8 -6 -4 -2 0 2 4 6 Partial ductility

Fig. 8c: Comparison of the model to test by the absorbed energy ratio

5 SUMMARY
The behaviour of connections loaded by cyclic forces may be predicted by component model.
Component prediction brings higher understanding of each connection parts influence. The
method exhibits good quality of prediction based on input data description as well as chosen
accuracy in detailing of model. The calculation is handicapped by the step by step procedure
for the each component as well as whole joint assembly, which is on the other hand natural in
today analyses supported by informatics. Based on analytical nature the method enables to go
to prediction of new developments.
The description of steel to concrete connection is based on set of tests with components. The
prediction of experiment from literature shows a good quality of prediction. The important
influence of shear forces is expected to be incorporated in the nest step of the model.
ACKNOWLEDGMENT
The work has been supported by grant COST C12.10 of Czech Ministry of Education, Youth
and Sport.

REFERENCES
[1] ECCS TC1 TWG 1.3., Recommended testing procedure for assessing the behaviour of
structural steel elements under cyclic loading. European convention for constructional
steelwork, Brussels 1986
[2] Zoetemeijer P., Proposal for Standardisation of Extended End Plate Connection based
on Test results and Analysis, Rep. No. 6-83-23, Steven Laboratory, Delft 1983.
[3] Mazzolani, F.M., Mathematical model for semi-rigid joints under cyclic loads.
Connections in Steel Structures: Behaviour, Strength and Design, Elsevier Applied
Science Publisher, London, 1988, pp. 112-120
[4] Della Corte G., De Matteis, G. Landolfo, R., A mathematical model interpreting the
cyclic behaviour of steel beam-to-column joints. Proc. 17. kongres C.T.A. - Settimana
della construzione in acciaio, Napoli, 1999, pp. 115-126
[5] Ermopoulos J., Stamatopoulos G., Wald, F., Sokol, Z., Mathematical modelling of
semirigid connections in steel column-bases cyclic loading, Report of bilateral
cooperation in civil engineering research (MŠMT ČR and Greek Ministry of Industry,
Energy and Technology), Athens 2000, p. 32
[6] Gregor D., Wald F., Eliášová M., Jírovský I., Joints for mixed building technology with
view to experiments of component steel plate in bending and concrete in compression, in
Eurosteel 2002, Coimbra 2002, pp. 977-986, ISBN 972-98376-3-5
[7] Gregor D., Wald F., Sokol Z., Experiments with End Plate Joints for Mixed Building
Technology, in Experimental Investigation of Building Materials and Technologies, ed.
Konvalinka P., Luxemburg F., ČVUT, Praha 2003, pp. 65-82, ISBN 80-01-02835-6
[8] Rassati G.A., Noè S., Leon, R.T., PR Composite joints under cyclic and dynamic loading
conditions: The component model approach, in Proc. 4th AISC International Workshop
on Connections in Steel Structures, Roanoke, 2000, pp. 213-222
[9] Bernuzzi C., Balado, L., Castiglioni, C.A., Steel beem to column joints: Failure criteria
and cumulative damage models. Proc. STESSA ed. F.M. Mazzolani and H. Akiyama,
Kyoto, 1997, pp. 538-545
[10] Bursi O.S., Ballerini M., Nemati N., Zandonini, R., Quasi-static monotonic and low-cycle
behaviour of steel isolated tee stub connections, Proc. STESSA, ed. Mazzolani F.M. and
Akiyama H., Kyoto, 1997, pp. 554-565
[11] Wald F., Sokol Z., Connection design, ČVUT, Praha 1999, p. 145, ISBN 80-01-0273-8
[12] Gregor D., Wald F., Component model for steel to concrete joints, in COST C12
Conference Innsbruck 2005, paper No. 77, in printing
[13] Penserini P., Caracterisation et modelisation du comportement des liaisons structure
metallique-fondation, theses l'Université Pierre et Marie Curie, Paris 6, 1991
[14] Dunai L., Ohtani Y., Fukumoto Y., Experimental Study of Steel-to-Concrete End-Plate
Connections under Combined Thrust and Bending, Technology Reports of Osaka
University, Vol. 44, No. 2197, Osaka 1994

KEYWORDS
Mixed building technology, Steel to concrete connections, Structural joints, Experimental
observations, Prediction model, Component method, Quasistatic loading, Cyclic loading.

Anda mungkin juga menyukai