Page 1 of 6
EUROSTEEL 2008, 3-5 September 2008, Graz, Austria
EXPERIMENTAL STUDY ON FLEXIBLE END PLATE CONNECTIONS IN
FIRE
Ying Hua
, Buick Davisona
, Ian Burgessa
, Roger Plankb
a Department of Civil and Structural Engineering, The University of Sheffield, Sheffield S1 3JD, UK
b School of Architecture, The University of Sheffield, Sheffield S10 2TN, UK
INTRODUCTION
Over the last decade research has shown that steel and steel-composite structures can have a
significantly greater fire resistance than is suggested by conventional tests on isolated components.
Burgess [1] explains that this is largely due to the interaction between the beams and floor slabs in
the fire compartment, and the restraint afforded by the surrounding structure. In the design of real
projects, it is implicitly assumed that the connections have sufficient fire resistance because they are
heated more slowly than the connecting members in fire situations. However, the evidence from the
collapse of the WTC buildings suggests that the progressive collapse may have been triggered by
the failure of steel connections [2], [3]. From full-scale fire tests, it has been found that steel
connections may be the weakest components in fire conditions [4]. In research on the performance
of large substructures in fire, non-linear three- dimensional analysis shows that the axial forces
generated in beams are seen to reach very high values [5]. Typically theses forces can vary from
compression in the early stages of a fire, when thermal expansion is restrained by the surrounding
structure, to tension in the later stages due to the heated members hanging essentially in catenary.
Consequently, the connections at the ends of these members are subjected in turn to these axial
forces whilst also undergoing large rotations.
In 1987 a joint SCI/BCSA connection group was established to produce a series of publications
which would standardise detailed design methods for commonly used steel connections. Owens and
Moore [6] carried out a series of tests to investigate the ability of simple steel connections to resist
tying forces, as specified in UK design codes to ensure a minimum level of robustness and prevent
progressive collapse. The test programme comprised 11 tests for web cleat steel connections and 10
tests for flexible endplate beam-to-column connections. It was found that conventional steel
connections have inherent robustness and may provide tensile ties to resist progressive collapse. It
should be noted that Owens and Moore tested tying forces applied to simple steel connections as
horizontal tie effects, as indicated by UK codes. But in a fire situation the tying force is most likely
to be inclined due to the connections experiencing large rotations/deformations. Hence, the
connection will be subjected to both horizontal and vertical force components. Furthermore, the
strength of steel connections exposed to a fire will result in a reduction of resistance, thus an
understanding of the tying resistance of simple steel connections in fire is essential in the
investigation of robustness of steel structures.
The research group at the University of Sheffield developed a series of tests for steel connections,
including three commonly used simple connections and one moment connection. This paper reports
on the experiments conducted on flexible end plate connections and details of the other connections
may be found in Yu et al. [7].
1 TEST PROGRAM
In this connection programme, twelve tests were separated into three classes according to initial
loading angles (α): 35o
, 45o
and 55o
, which implicitly defined the horizontal and vertical
components of tying forces applied to these connections. As a result of the loading mechanism, the
angle of inclination of the applied tensile force varied during the test. The magnitude and angle of
the force were monitored and recorded throughout the experiments. Flexible end plate (Fep)
Page 2 of 6
connections were tested at both ambient and elevated temperatures. The test schedule, test numbers
and temperatures are as shown in Table 1.
Table 1. Test schedule
Initial α
Temps 35o
45o
55o
20o
C Fep35-20 Fep45-20 Fep55-20
450 oC Fep35-450 Fep45-450 Fep55-450
550 oC Fep35-550 Fep45-550 Fep55-550
650 oC Fep35-650 Fep45-650 Fep55-650
1.1 Test Arrangement and Instrumentation
Connection tests were performed in an electric furnace with the experimental set-up illustrated in
Fig. 1. Fig. 1(a) shows a connection in the furnace and the load applied though three linked φ26.5
mm Macalloy bars. Fig. 1(b) is a photo of the electric furnace; note the small rectangular glass
viewing panel in the door and the large circular hole through which the furnace bar was positioned.
(a) (b)
Fig. 1. (a) Experimental set up & (b) Electric furnace
The essential requirement in this test program was to obtain the resistances of steel connections
against tying force at both normal and high temperatures. Measurement of the forces was achieved
by using strain gauges on the loading system (three Macalloy bars: furnace bar, link bar and jack
bar) at ambient temperatures. At high temperatures, the strain gauges close to or inside the furnace
would be damaged. So in order to obtain the force in the furnace bar, strain gauges were fixed to the
link and jack bars and the applied force was then calculated by resolving the system of forces. Any
changes in inclination of the bars were recorded using three angular transducers.
To avoid the problems associated with cooling instrumentation in a furnace, Spyrou and Davison
[8] developed an image-acquisition technique to measure the displacements and rotations for testing
specimens at high temperatures. This approach was adopted and included image acquisition and
processing software, and targets made of ceramic rods, embedded into the specimens before testing.
A digital camera was fixed on the main door of the furnace to record the movements of the
specimen through a 200 mm x100 mm viewpoint. The recorded photos were processed by software
to obtain the rotations and displacements of the steel connection.
1.2 Fabrication and Assembly
Steel sections (beams and columns) were supplied by Corus and fabricated by Billington Structures
Ltd. The holes in the end plates were punched according to current UK practice. Normal
engineering procedures and standard tolerances were adopted during fabrication and no special
effort was taken. All the specimens were assembled in the test rig and standard grade 8.8 M20 bolts
were used for all tests.
Page 3 of 6
1.3 Specimen Details
In this series of connection tests, a 254UC89 was used for the column and a 305x165UB40 for the
beam. The thickness for the end plate was 10 mm and all the bolts were used in 22 mm clearance
holes. The steel used was nominally S275 for universal beams and the column was S355. The
design details and sizes of the specimens have been included in Fig. 2. and the numbers of c-h
show the location of thermocouples used in testing. α is the initial or ending angle of external
loading.
Fig. 2. Design details for flexible end plates
2 EXPERIMENTAL OBSERVATIONS
2.1 Failure Modes of Endplate Connections
Load against Rotation for Flexible End Plates (35o
)
0
40
80
120
160
200
0 1 2 3 4 5 6 7 8 9 10
Rotation (o
)
Load (kN)
Fep35-20 Fep35-450
Fep35-550 Fep35-650
Load against Rotation for Flexible End Plates (45o
)
0
40
80
120
160
200
0 1 2 3 4 5 6 7 8 9 10
Rotation(o
)
Load (kN)
Fep45-20 Fep45-450
Fep45-550 Fep45-650
(a) (b)
Load against Rotation for flexible end plates (55o
)
0
40
80
120
160
200
0 1 2 3 4 5 6 7 8 9 10 11 12
Rotation (o
)
Load (kN)
Fep55-20 Fep55-450
Fep55-650 Fep55-550
(c) (d)
Fig. 3. Test results for flexible endplate connections (a) 35o
(b) 45o
(c) 55o (d) modes of failure
In the series of flexible endplate tests performed by Owens and Moore [6], two different modes of
failure were reported: (a) bearing failure of the endplate and (b) fracture of the endplate close to the