Advances in Steel Structures - part 33 ppt

300
S.P. Chiew and C.W. Dai
external ring. The connection that used re-bar as the stiffener (UCN-8) performed very
well in terms of initial stiffness and ductility. Compared with UCN-1, its initial stiffness
and ductility are 2.09 and 2.13 times respectively.
Table 5 Initial Stiffness and Ductility
Specime UCN- 1 UCN-2 UCN-3 UCN-4 UCN-5 UCN-6 UCN-7 UCN-8
Stiffness 1.0 1.30 0.73 1.52 1.31 1.11 0.78 2.09
Ductility 1.0 1.31 1.38 0.95 0.99 2.23 1.63 2.13
3.4 Failure Modes
Figure 6 shows the failure modes of some specimens. For all simple connections,
the failure began with the tube tearing at the beam flange attachment point on the tension
zone and then, with the increment in load, buckling appeared on the beam flange near
joint. For rigid connections, the failure modes of specimen UCN-5 and UCN-6 were
similar to those of the simple connections, but no buckling can be observed on
compression zone. The failure of the specimen UCN-7 began with the external ring
rupture at the beam flange attachment point, with the increment in load, the crack
developed through the external ring and finally the tube tore at the web attachment
position. Buckling can also be observed on the flange near the external ring (Fig.6 (c)).
The failure of the specimen UCN-8 began with the buckling near the end of the re-bar
(Fig.6 (d)). The twisting of the beam can be observed on the later stage (Fig.6 (e)). In
order to check the inside condition of the concrete core, specimen UCN-4 and UCN-8' s
steel tube skin around the joint were cut away after test. It was found that the concrete
core behind the connection exhibited no signs of crushing or distress even though it has a
lower compressive strength (Fig.6 (f)).
4. CONCLUSIONS
The following conclusions can be obtained from the tests:
1) Simple connections have weaker stiffness, ductility and strength. In this experiment,
their yield moment had only 40%-48% of the beam cross-section plastic moment
except UCN-2. The change of the thickness of the steel tube obviously influence the
Experimental Study of Steel I-Beam to CFT Column Connections

301
Fig.6 Failure Modes
302
S.P. Chiew and C. HI. Dai
composite behavior of the connection.
2) With the same cross-section area, the selection of deeper and wider, but thinner steel
1-beam can improve the connection moment transfer capacity, but the ratio of the
yield moment to beam' s plastic moment capacity is almost the same.
3) Stiffeners have some effect on improving the connection' s stiffness, ductility and
strength. The degree of improvement on strength varies from 26% to 46% for shear
plate, cover plate and external ring.
4) Re-bar can greatly improve the composite behavior of the connection. The ultimate
strength of the connection with the re-bar stiffener is about 2.4 times of the
connection without stiffener. The main reasons are, firstly, the re-bar can improve the
boundary condition, eliminate the stress concentration point that appeared on the
column wall at the connections without stiffener; secondly, the re-bar can improve the
beam' s cross-section bending stiffness; and finally, the re-bar can move the plastic
hinge away from the column face.
5) Test results proved that the finite element models built up for the composite
connections are feasible to be used for the strength predictions.
ACKNOWLEDGEMENTS
The authors wish to Nanyang Technology University, Singapore for their
financial support in this project. The authors would also like to thank the technical staff
of the CSE Construction Lab and CSE Heavy Structures Lab for their assistance in
fabricating and testing all the specimens.
REFERENCES
1. Shakir-Khalil. H. (1992), "Full Scale Tests on Composite Connection" ASCE
Proceedings, Composite Construction of Steel and Concrete II, pp. 539-554.
2. Kato, B., Kimura, M., Ohta, H. and Mizutani, N. (1992),
"Connection

of Beam
Flange to Concrete-filled Tubular Columns" ASCE Proceedings, Composite
Construction of Steel and Concrete II, pp. 528-538.
Experimental Study of Steel I-Beam to CFT Column Connections
303
3. Gang, H.G., Chung, I.Y., and Hong, S.G. (1998), "Performance of Concrete Filled
RHS Column-to-Beam Connections with Exterior Plate Diaphragm" Structural
Steel PSSC' 98 Vol.2, pp. 729-734.
4. 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, 1998.
5. Oh, Y.S., Shin, K.J. and Moon, T.S. (1998), "Test of Concrete-filled Box Column to
H-Beam Connections" . Structural Steel PSSC' 98 Vol.2, pp. 881-886.
6. Oh, Y.S., Shin, K.J., Lee, M.J. and Moon T.S. (1995),
"A
Study on the Bending
Behaviour of Connections for Empty and Concrete-Filled Box Steel Column and H-
Beam by Stiffened Triangular Plates" , Proceedings of 4 th pacific Structural Steel
Conference, V.2, Singapore, pp. 57-64.
7. MARC analysis research corporation, MARC manual Vol. A, 1995.
This Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left BlankThis Page Intentionally Left Blank
BEHAVIOUR OF T-END PLATE CONNECTION TO RHS
PART I: EXPERIMENTAL INVESTIGATION
M. Saidani, M. R. Omair, and J. N. Karadelis
School of the Built Environment
Coventry University, Priory Street, Coventry CV 1 5FB
ABSTRACT
This paper is concerned with investigating the behaviour of welded T-end plate connections to
rectangular hollow sections when subjected to tension. A series of static tests were conducted to
failure with varying parameters for the tube wall thickness and the cap plate thickness. The cleat

plate thickness was kept constant for all tests. Stresses, strains and deflections at different locations
in the connection were recorded and plotted against the applied load. Numerical modelling of the
connection was undertaken using the finite element suite ANSYS as discussed in the companion
paper.
KEYWORDS
Hollow sections, T-end plate connection, tests, tension, stresses, deformations, design.
INTRODUCTION
The excellent properties of structural hollow sections have long been known Cran (1977), CIDECT
(1984), and Packer et al (1992). Connections made with hollow sections are often said to be
complex and expensive. In reality they can be made simple and cost-effective. This is to add to their
e,:cellent aesthetic appearance making them the ideal choice in many elegant structures.
Rectangular hollow section (RHS) members are often used as compressive members due to their
good buckling stiffness. Often such members are also required, and indeed should be designed, to
take also tensile forces. One of the simplest ways to connect tubular members is by cutting the ends
and welding together. However, depending on joint configuration and number of members
connected, this may result in complex and expensive connections. The alternative would be to
connect the members together through some other means. Figure 1 shows types of end connection
details for hollow tubes and which are used in practice. One of the most economic solutions is to
weld a cap plate to the tube (CHS or RHS) and then weld on to it a cleat plate (Figure 2). The
connection could be made entirely in the workshop, thus reducing labour work on site.
305
306 M. Saidani et al.
Figure 1: Type of end-connections
In the UK there is very little guidance on the design of welded T-end connections. Elsewhere,
research work was mainly carded out by Kitipornchai and Traves (1986), Stevens and Kitipornchai
(1990), and Granstrom (1979). The absence of design recommendation very often leads designers to
specify uneconomical solutions.
Figure 2: Welded T-end plate connection
Research has shown that welded T-end connections subjected to uniform tension may fail in
different ways. The failure mode is dictated by parameters such as:

9 Tube wall thickness tw;
9 Cap plate thickness te;
9 Cleat plate thickness to;
9 Weld size and quality.
The possible resulting modes of failure are as follows:
9 Tube yielding;
9 Local fracture in tube (in the region adjacent to weld);
9 Fracture of the weld;
9 Yielding of the cap plate;
9 Shear failure of the cap plate;
9 Yielding of the cleat plate.
Combination of more than one mode of failure is possible. In a truss environment (when the
connection forms part of the truss assembly), there is also the possibility of the bolts failing.
Behaviour of T-End Plate Connections to RHS Part I
307
The problem being investigated in this research programme is to answer the fundamental question:
how does the cap plate thickness influence the mode(s) of failure of the connection? In order to
answer this question, an experimental programme was conducted on a series of specimens with
varying parameters. In the companion paper, Karadelis et al (1999), a finite element model is
developed, analysed and results are compared with the test results.
EXPERIMENTAL PROGRAMME
The experimental programme followed a similar procedure adopted by Stevens and Kitipornchai
(1990) so that useful comparison could be made. The testing work included 8 specimens with
varying tube wall and cap plate thickness. Each specimen was loaded in tension, taking all
precautions to avoid any accidental eccentricity. Strains and deformations were also measured. The
test arrangement for the strain gauges and LVDT's is shown in Figure 3. The programme of tests is
summarised in Table 1. In order to keep the investigation manageable, only one cleat plate
thickness was used and kept equal to 15mm for all specimens.
The test programme was devised to concentrate on the yielding of the tube wall and the deformation
of the cap plate as these were found to be the main causes of failure. Strain gauges were located on

the tube wall (four faces), the cap plate, and the cleat plate with the aim of closely monitoring strain
(and stress) variations across the specimen. The LVDT's will give readings of the deformations and
an indication of any in-plane and/or out-of-plane movements.
TABLE 1
TESTING PROGRAMME
Test
No.
1
2
3
4
5
6
7
8
Cap plate thickness Cleat plate thickness Tube size Steel grade Comments
(mm) (mm)
10 15 60x60x4.0 $275 test No.1 was
10 15 60x60x4.0 $275 repeated due
15 15 60x60x4.0 $275 to premature
20 15 60x60x4.0 $275 failure of the
30 15 60x60x4.0 $275 weld
10 15 80x80x4.0 $275
15 15 80x80x4.0 $275
20 15 80x80x4.0 $275
It is important that the strain gauges are kept far enough from welds in order to avoid any influence
from the residual stresses on readings. The total length of the tube is 500mm again for the same
raison. Strain gauges were placed on opposite sides so that in-plane and out-plane bending moments
could be monitored and calculated. The general arrangement for the testing is shown in Figure 4. In
total 12 stain gauges and 5 LVDT devises were used to monitor the joint behaviour and obtain the

necessary information. In some locations (at some distance from the welds) rosette gauges were
used with the aim of obtaining strains at different angles at a point. This was decided after the first
two tests when it became evident that strains (stresses) were not uniaxial, but were in fact
developing at an angle to the longitudinal axis of the tube.
British Steel was used for all the tests. Samples were cut out from each specimen and were tested in
accordance with British Standards for testing in order to check the material properties (Young's
modulus, yield strength, and ultimate tensile strength). Accurate material properties are important to
308
M. Saidani et al.
obtain since these are needed for accurate numerical modelling of the specimens as described in the
companion paper, Karadelis et al (1999).
Finally, although two tube sizes were tested, the tube thickness was kept constant at 4mm, again
with the aim of keeping the investigation manageable.
Figure 3: Specimen dimensions (all in mm)
Figure 4: Joint testing arrangement
THE SPECIMENS TESTING AND RESULTS
The DENISON machine with a capacity of 500kN was used for the testing of the joints. A tensile
load is applied in increments of 10kN up to failure. The strains and deformations are recorded for
Behaviour of T-End Plate Connections to RHS Part I
309
each load increment into a computer logged to the testing machine. Using a simple spreadsheet
program, stresses are calculated and various graphs are plotted.
Table 2 summarises the results from the specimen testing. For test No. 1, the failure was due to weld
fracture. On close examination of the specimen it was discovered that weld penetration was not
adequate. As a result, it was decided that welding should be done very carefully making sure it is
evenly spread with sufficient material penetration. In the subsequent specimens, failure was mainly
due to tube yielding. Yielding was also noticeable in the cap plate.
TABLE 2
SUMMARY OF TESTING RESULTS
Test

No.
Cap plate
thickness
Tube size
Failure mode
Full tension
capacity Pc
(kN) [theo]
First yield
Py (kN)
[exp]
Ultimate
load Pu
(kN) [exp]
10 60x60x4.0 Weld fracture 409.6 190 260
10 60x60x4.0 Tube yielding 409.6 230 313
15 60x60x4.0 Tube yielding 409.6 280 350
20 60x60x4.0 Tube yielding 409.6 280 385
30 60x60x4.0 Tube yielding 409.6 240 350
10 80x80x4.0 Local fracture 559.2 230 329
in tube
15 80x80x4.0 Tube yielding 559.2 320 450
20 80x80x4.0 Tube yielding 559.2 370 490
Pu/Py
1.37
1.36
1.25
1.38
1.46
1.43

1.40
1.32
In all the tests, the determination of the first yield point was proven very difficult to do accurately.
The values shown in Table 2 were obtained by considering the load-axial deflection curve and
taking the point where departure from the initial elastic path became measurable. The load-stress
curves were also used to help with the determination of the first yield point. Figures 5 and 6 show
typical load-deformation and load-stress curves for specimen No.2.
Using the strain gauge readings, in-plane and out-of-plane bending moments were calculated in
order to check their relative significance. Again, typically, results for specimen No.2 are shown in
Figures 7 and 8 respectively. For each specimen, the following graphs were plotted:
i. Strain vs. stress
ii. Load vs. deformation
iii. Load vs. in-plane bending moment
iv. Load vs. out-of-plane bending moment
v. Load vs. axial force in the tube
Only samples of these graphs are shown for specimen 2.
As can be seen from Table 1, at the exception of specimen No.5, both the ultimate loads and
observed first yield loads increase with the cap plate thickness. Similarly, with increased tube size,
both the ultimate loads and observed first yield loads increased. Specimen No.5 was quite
interesting. This particular test was repeated three times since there were some doubts about the
results, and every time the results were almost the same. This seems to suggest that, as the cap plate
thickness increases beyond 20mm (or may be 25mm), the ultimate capacity of the connection
decreases. This could be explained by the fact that, as the cap plate becomes very thick, the tube
would yield earlier resulting in a reduction in the joint capacity. Three particular modes of failure