Beruflich Dokumente
Kultur Dokumente
www.elsevier.com/locate/jcsr
Abstract
∗ Corresponding author. Tel.: +44 (0)1372 371122; fax: +44 (0)1372 371123.
E-mail address: mxie@kwltd.com (M. Xie).
0143-974X/$ - see front matter © 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2004.10.005
626 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
Notation
B width of plate
d bar diameter (d = 25 mm)
E elastic modulus of steel (E = 2.05 × 105 N/mm2 )
Ec elastic modulus of concrete steel
f ck cylinder strength of concrete
f cu cube strength of concrete
fy yield stress
f yB , f yP yield stress of bar and plate, respectively
fu ultimate strength
f uB , f uP ultimate strength of bar and plate, respectively
F shear force applied to a bar–plate connection
Fu maximum shear force applied to a bar–plate connection
hc plate spacing, or length of bar connector
Ks shear stiffness of bar connector
m modular ratio between steel and concrete, m = E/E c
t plate thickness
TB axial force in bar connector, TB = σB ∗ (πd 2 /4)
δ slip between steel plate and concrete
δu shear slip at failure
ε direct strain
ε1 , ε2 axial strains measured at both ends of a bar connector (Fig. 5)
µ friction coefficient
σ direct stress
σB axial stress at centre of bar connector section
σmax maximum principal stress
σP plate stress in the direction of bar shear force
τBu average bar shear stress at failure
1. Introduction
The Bi-steel system of sandwich construction has shear connectors that are friction
welded at both ends (Fig. 1). Whereas shear connectors in conventional composite
construction are embedded in reinforced concrete, the usual practice is that Bi-steel
members do not have embedded reinforcement. Also, the diameter of conventional shear
studs is required to be no more than 1.5 or 2.0 times the plate thickness t depending on
whether the plate is in tension or compression respectively, whereas in Bi-steel the bar
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 627
diameter d (25 mm, Fig. 1) can be well over 2.0t. A further complication is that the bars in
Bi-steel also resist transverse shear and stabilise plates in compression. This paper aims to
determine the shear strength and stiffness as determined by push tests, and the tensile bar
stress caused by applied shear force. The tests are supplemented by finite element analysis
(FE) which enables the interactions to be studied and understood.
2. Existing information
Tests at Southampton University examined the effect of varying plate thickness and
distance between plates [2,3]. The results of those tests are plotted in Fig. 2. The test
arrangement was such that the maximum bar curvature occurred at mid-length.
According to the design guide [4], the shear capacity of a bar connector Fu is the lesser
of
Fus = 0.8 kL f uB π d 2 /4 (1)
Fuc = 0.29 d 2( f ck E c )1/2 (2)
where f uB is the ultimate tensile strength of the bar connector, d is the bar diameter, f ck is
the cylinder strength of concrete, E c is the elastic modulus of concrete and kL is a reduction
factor given by
f yP
kL = (0.024 t + 0.76) and kL ≤ 1 (3)
355
where fyP is the plate yield stress ( f yP ≤ 355 N/mm2 ), and t is the plate thickness
(t ≤ 10 mm).
On the basis of the values tabulated in BS5400 Part 5 [5] for conventional shear
connectors, the nominal static shear strength for a headed 25 mm diameter stud with an
overall height not less than 100 mm can be calculated from
Fu = 1.46 · f cu + 109.9 kN (4)
628 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
Fig. 2. The Southampton test results for embedded connections in shear [2,3].
where fcu is the cube strength of concrete. For f cu = 40 and 50 N/mm2 , Eq. (4) gives
nominal shear strengths Fu = 168 kN and 183 kN respectively. BS5400 also requires that
the diameter of stud connectors should not exceed 2t when welded to a non-tensile flange
plate, and except for t = 15 mm the test plate thicknesses do not satisfy this requirement.
Nevertheless, the shear strength given by Eq. (4) will be compared with the test results in
the next section.
3. Experimentation
The shear test specimens consisted of a Bi-steel unit and a 200 mm cubic concrete core
(Fig. 3). The nominal plate thicknesses were 6, 8, 10, 12 and 15 mm. The dimensions were
chosen to allow testing in a standard test machine, which could apply both static and fatigue
loading. 24 specimens were tested statically.
The steel specimens were fixed in moulds, and filled with ready-mix concrete of Grade
40/50, which would nominally give a minimum cylinder strength f ck = 40 N/mm2 and
cube strength fcu = 50 N/mm2 . The actual concrete and steel properties are given in
Table 1.
The testing arrangement is shown in Fig. 4. The total load 2F from a spherical bearing
was transmitted to a 40 mm thick loading block (120 mm × 120 mm) and then applied
to two 25 mm thick packing plates to obtain an approximately uniform load over the top
surface of the concrete, resulting in a shear force F acting on each bar/plate interface. The
bottom plate edges bear upon a plate which is supported by the moving platen of the testing
machine.
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 629
Table 1
Details of embedded shear specimens
Specimen Plate thickness t (mm) Plate width B (mm) f yP (N/mm2 ) f uP (N/mm2 ) f cu (N/mm2 )
Notes: bar diameter d = 25 mm, yield stress f yB = 560 N/mm2 , ultimate strengh f uB = 586 N/mm2 .
630 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
(a) Elevation.
Horizontal external clamps (‘A’ in Fig. 4(a)) were used to restrain concrete splitting,
representing the restraint existing in a real structure, in which the tendency of the
connectors to cause splitting of the concrete is resisted by adjacent connectors and
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 631
eventually by closure plates. The 25 mm thick plates above and below the concrete
block were joined by external bars (‘B’ in Fig. 4(a)), so the concrete can grip the test
bar over its length. Also, external clamps (‘C’ in Fig. 4(a)) were used to prevent plate
bending/buckling. All clamps were initially “finger tight”.
The test rig shown in Fig. 4 cannot reproduce the asymmetric S-shape deformation of
the bar which can be expected in a Bi-steel beam, but the small bar curvature at mid-
length is realistic and symmetry does result in the weaker of the two connections causing
failure. The bar tension for symmetrical deformation should approximate that for anti-
symmetric deformation, but the asymmetry of a Bi-steel beam with tensile cracking of the
concrete on one side, compression on the other side is not reproduced. The conditions
are similar to those in a conventional push test, in which restraint against splitting is
provided by reinforcement. Reasonable comparisons with the behaviour of conventional
shear connectors can therefore be made.
Load was applied and controlled by an Amsler cabinet, and recorded by a pressure
transducer. The slip between steel and concrete was measured approximately by a
displacement transducer (LVDT) below the bottom surface of the moving loading platen
of the test machine, as shown in Fig. 4.
Two specimens for each plate thickness were fitted with a strain gauge bonded by resin
injected into 2 mm holes on the centreline at each end of the bar, as illustrated in Fig. 5.
The gauge length was 6 mm. The magnitudes of applied load, slip and bar strains were
recorded by the data acquisition equipment DATASCAN, and logged using the software
DALITE which runs on a PC platform.
4. Test results
The test results for the 24 static shear specimens are summarised in Table 2. Fig. 6
shows the variation of the ultimate load with the plate thickness. It is evident that for
d = 25 mm and t ≥ 10 mm the shear strength does not increase with thickness.
632 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
Table 2
Results of static shear tests on embedded connections
Specimen Fu (kN) τBu (MPa) δu (mm) K s (kN/mm) Failure mode (see Fig. 7)
6S/1 162.2 330.5 2.47 85 Plate tearing (Mode S1). Failed by the bar
6S/2 181.4 369.5 6.24 104 tearing a through-thickness plug from the
plate, starting from the tension side.
6S/3 190.0 387.1 3.29 100 Plate tearing (Mode S1), occurred on the
6S/4 185.0 376.9 2.83 108 tension side of the bar. The load was
removed when it had fallen to about 0.9Fu
after reaching its peak value Fu .
8S/1 189.9 386.9 8.51 86 Plate tearing (Mode S1), occurred on the
8S/2 192.3 391.8 8.24 91 tension side of the bar. The load was
removed when it had fallen to about 0.9Fu
after reaching its peak value Fu .
8S/3 172.8 352.0 8.16 90 Plate tearing (Mode S1). Failed by the bar
8S/4 197.6 402.6 8.60 100 tearing a through-thickness plug from the
plate, starting from the tension side.
10S/1 203.4 414.2 3.75 114 Plate tearing (Mode S1). Fracture occurred
within the plate thickness and started on the
tension side.
10S/2 213.2 434.2 4.92 118 Plate tearing (Mode S1), within its thick-
ness, starting from the tension side.
10S/3 209.0 425.8 3.85 116 Plate tearing (Mode S1), occurred on the
10S/4 223.8 455.8 3.68 114 tension side of the bar. The load was
removed when it had fallen to about 0.9Fu
after reaching its peak value Fu .
10S/5 247.9 505.0 13.60 105 Bar shear fracture (Mode S2), occurred
across the bar section at the top of the weld
flash.
10S/6 212.6 433.1 11.41 116 Plate tearing (Mode S1), occurred on the
10S/7 211.8 431.4 10.32 116 tension side of the bar, almost through the
10S/8 225.8 460.0 15.61 90 thickness of the plate. The load was removed
when it had fallen to about 0.9Fu after
reaching Fu .
12S/1 243.3 495.6 12.89 128 Bar shear fracture (Mode S2), occurred
across the bar section at the top of the weld
flash.
12S/2 208.4 424.4 2.70 179 Interface shear fracture (Mode S3). The
bar connector was sheared off at one
connection along the interface. The weld
flash stayed on the bar.
12S/3 226.8 461.9 6.86 127 No fracture could be observed after the load
12S/4 221.7 451.6 6.92 175 was removed when it had fallen to about
0.95Fu after reaching Fu . Shear deformation
was seen at both bar ends.
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 633
Table 2 (continued)
Specimen Fu (kN) τBu (MPa) δu (mm) K s (kN/mm) Failure mode (see Fig. 7)
15S/1 209.3 426.4 9.21 113 Plate tearing (Mode S1), occurred on the
15S/2 218.4 444.9 6.14 134 tension side of the bar. The load was
15S/3 212.5 433.0 9.46 119 removed when it had fallen to about 0.9Fu
after reaching its peak value Fu .
15S/4 220.4 448.9 10.67 125 Interface shear fracture (Mode S3). The
bar connector was sheared off at one
connection along the interface. The weld
flash stayed on the bar.
Notes: (1) The details of each specimen are listed in Table 1.
(2) Connection fracture occurred when the load was falling after passing its peak value.
Fig. 6. Ultimate shear load applied to one embedded bar–plate connection. For equations: E c = 3 × 104 N/mm2 ,
f ck = 34 N/mm2 , σuB = 586 N/mm2 , σ0P ≥ 355 N/mm2 .
(a) Plate tearing. (b) Bar shear fracture. (c) Interface fracture.
Fig. 7. Typical static failure modes for the embedded connections subjected to bar shear.
bending (Fig. 8). In the pure shear tests, the bar sheared off the plate along the interface,
and the ultimate shear force does not vary significantly with the plate thickness.
The pure shear strength varies from 506 to 541 N/mm2 , with an average value of
527 N/mm2 . The tensile strength of the bar material is 586 N/mm2 . In the embedded
tests interface failure (mode S3) occurred in two specimens, at 424 and 449 N/mm2 . In
these cases shear was accompanied by bending and tension.
The average tearing failure stress (mode S1) for t ≥ 10 m is 436 N/mm2 and the
average bar shearing stress (mode S2) is 500 N/mm2 .
The bar shear strengths predicted using Eqs. (1), (2) and (4) are shown in Fig. 6,
assuming E c = 30 GPa and fck = 34 N/mm2 . For t ≥ 10 mm, compared to the strength
given by the experimental Eq. (5), the strength given by Eq. (1) is 10% greater, whereas
the values from Eqs. (2) and (4), which account for the compressive strength of concrete,
are 13% and 20% lower respectively.
Strength variations can be caused by variations in the through-thickness plate strength,
the quality of the weld, the concrete strength, the compaction and the local disposition of
the aggregate.
In several instances the load was removed when the resistance dropped to 90–95% of
the maximum, in order to reveal the development of failure. Table 2 indicates when this
procedure was adopted; for these specimens the F/δ curves in Fig. 9 are curtailed.
The measured shear slip δ versus the applied shear force F is plotted in Fig. 9, and
the slip δu , at the ultimate shear load, Fu , is listed in Table 2. The static test specimens
were made in two groups representing separate deliveries of the steel units. The concrete
cube strength is 42 N/mm2 for the first group of specimens, and 40 N/mm2 for the second
group. However, the average strength of the second group 10 mm specimens is 6% greater
than for the first group. The steel properties did not vary significantly but the weld quality
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 635
might have been better for the second delivery. Perhaps that can explain the greater ductility
of the second group 10 mm plate specimens.
The linear parts of the experimental F–δ curves were used to derive the connector
shear stiffness K s , and the values obtained are plotted in Fig. 10. The shear stiffness of
the connectors K s can be estimated approximately from the following equation:
K s = 3.2 t + 75 kN/mm for d = 25 mm. (6)
The reduced initial stiffness is caused by the bedding down of the loading platen.
The axial strains in ten bar connectors were measured by means of bolt strain gauges
embedded at each end of the connector along its centreline (Fig. 5). The centre of the strain
gauge was 12 mm away from the inner surface of the plate. The measured strains at both
ends of the gauged connectors are plotted in Fig. 11 against the shear force F applied to
one connection.
The inequality in bar axial strains at each end of the bar up to about 2Fu /3 may be
attributable to differences in rotational stiffness of the bar/plate connections. After bar
636 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
Fig. 10. The shear stiffness of Bi-steel bar connectors. Eq. (6).
yielding, plane sections will not remain plane, and that could explain the erratic changes
in ε1 and ε2 . At the elastic limit of the slip curve bar axial centreline strains are generally
around 500 µε (σB ≈ 100 N/mm2 ). However, the strain increases rapidly thereafter.
Fig. 11. Measured bar axial strains. The strain gauge locations are shown in Fig. 5.
test specimen. Fig. 12 shows the FE mesh in which the steel plate, the bar connector and
the concrete are all modelled using three-dimensional solid elements (ABAQUS Standard
C3D8 and C3D6 elements). As indicated in Ref. [1] and Report BS3 [6], the weld flashes
are not effectively connected to the plate, and their only effect is to increase the concrete
bearing area, so the flashes are not included in the FE model. A finer mesh is used in the
region around the bar–plate weld connection. The mesh consists of 10 406 nodes and 8752
elements.
The contact between the plate and concrete and that between the bar and concrete
are modelled using the contact pair approach in ABAQUS. This approach is based on
a master–slave concept, and the contact problem is solved with the Lagrange multiplier
method. A friction surface interaction model was introduced at the contact surfaces, and
solutions were obtained for µ = 0 and µ = 0.5.
Boundary conditions were chosen to represent the specimens tested in the rig as shown
in Fig. 4. The end of the bar was fully connected to the plate. Symmetrical constraints were
applied to the two planes of symmetry, and the bottom edge of the plate was prevented from
moving vertically (axis 3, Fig. 12) and in the bar direction (axis 1). To model the action of
the external clamp in preventing plate buckling, the plate nodes on the internal surface at
the bottom edge of the concrete were prevented from moving away from the concrete. A
uniformly distributed load was applied to the top of the concrete.
The stress/strain curve for steel, based on material tests, is shown in Fig. 13.
638 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
Fig. 13. The stress–strain relationship for steel adopted in the FE models.
In a preliminary study, the smeared crack concrete model and the concrete damaged
plasticity model were applied, but the analysis terminated prematurely due to severe
discontinuities. To enable the analysis to be carried out up to failure, the concrete was
then modelled as an elasto-plastic material using the ABAQUS cast iron plasticity model
in which a different stress–strain relationship for tension and for compression can be
specified. In the parametric study the stress–strain relationship given in Fig. 14 is adopted.
The tensile strength of concrete is taken as 10% of the ultimate compressive strength
0.85 f cu and fcu = 42 N/mm2 . Poisson’s ratio of concrete is taken as ν = 0.15.
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 639
Fig. 14. The stress–strain relationship for concrete adopted in the FE models (—— FE model and - - - - - CEB
1973 [7]).
Table 3
Comparison of the maximum shear forces per connection Fu (kN)
(a) 8 mm plate.
(b) 15 mm plate.
Fig. 15. Deformed profiles of the shear test specimens when approaching/at failure. FE results. The friction
coefficient µ = 0.5. The displacement magnification factor = 10.
The shear slips obtained from the FE analyses with µ = 0 and 0.5 are compared with
the measured values in Fig. 16. There are various factors that may contribute to the greater
calculated shear stiffness. The concrete is modelled as an elasto-plastic material (Fig. 14),
and discrete concrete cracking and crushing are not considered, because of numerical
difficulties. Other factors that can affect the shear deformation of the embedded connection
include the concrete modulus, the coefficient of friction between steel and concrete, and
residual welding stresses. It was found that by introducing axial springs between the bar
and the plate, the finite element F–δ curve could be brought closer to the experimental
curve. This suggests that perhaps the through-thickness steel stiffness at the connection
is less than the in-plane stiffness. The initial low stiffness is caused by bedding down of
the loading plates. The slip stiffness is assumed to be that given by the linear part of the
load/slip graph. According to the FE analysis the ratio K s(µ=0.5)/K s(µ=0) is about 1.6 for
the three values of the plate thickness t.
Fig. 17(a) shows for t = 8 mm and µ = 0.5 the variation in the axial stress along the
centre axis of the bar connector with the magnitude of the applied shear force. When the
shear force F reaches 202 kN, the end of the bar has already yielded in shear. The contour
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 641
diagram of the axial stress in the bar connector for F = 164 kN, t = 8 mm is presented in
Fig. 17(b).
Fig. 18 demonstrates the effect of the friction coefficient µ on the axial stress
distribution in the bar connector. The bar is subject to predominant shear at its ends, and
the shear deformation causes the bar to stretch, resulting in a bar tension which increases
quickly within the first 20 mm from the weld connection. Without friction between steel
642 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
and concrete, the bar centre axial stress reaches a constant value at a section about 12 mm
from the inside surface of the plate.
In ten specimens, bolt strain gauges were fitted as shown in Fig. 5. The numerical
solutions give the axial stress at the gauge position and at the mid-length of the bar.
The former is compared with measurements in Fig. 19, which shows good agreement
between the calculated bar forces and the measured values for a shear force up to about
150 kN.
The measured values for specimens 15S/3 and 15S/4 differ greatly; the numerical results
are closer to the measured values of 15S/3. The FE results show that the bar axial force
at the mid-length of the connector is 20–40% of the applied shear force up to around
F = 150 kN.
The comparisons made in the previous section have demonstrated that the finite element
model established here provides a reasonable representation of the embedded shear test
specimens. In this section, finite element analyses, assuming that the plates and bars are
elastic, are used for parametric studies. These analyses are especially relevant to fatigue
failure mechanisms. The concrete stress/strain curve is assumed to be that defined in
Fig. 14. It is seen in Section 6.2 that a zero friction coefficient produces a better agreement
with the experimental results. Also, contact surfaces under cyclic loading are less prone to
friction. Therefore it is assumed that µ = 0.
The local plate bending caused by the shear force applied to the bar connector is
dependent on the plate thickness and the stiffness of the concrete, and the rotation at the
bar end affects the shear slip and shear stiffness. Fig. 20 shows numerical slip and shear
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 643
Fig. 19. Comparison of measured bar axial forces with the numerical values calculated from the FE solutions for
the axial stress at the gauge position and at the mid-length of the bar.
stiffness curves. The non-linearity of the slip curves is caused by the non-linear behaviour
of the concrete (that is, the difference between E c and E 1 —see Fig. 14). The slips between
F = 100 and 200 kN were used to calculate the shear stiffness K s , which is then compared
in Fig. 20(b) with the shear stiffness given by the empirical Eq. (6), which is based on test
results.
The numerical values of K s are about 28% greater than those from Eq. (6). Fig. 10
shows that the experimental value of K s varies by 16–40% for specimens having the same
644 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
nominal plate thickness, and the maximum value K s = 178 kN/mm for t = 12 mm is
greater than the corresponding numerical value K s = 146 kN/mm. It should be noted that
the finite element model assumes that E = 205 000 N/mm2 prevails throughout the region
of the connection.
The finite element models are also used to examine the bar tension caused by the shear
force applied to the bar. For t = 8 mm, the distribution of the bar tension along its length
is the same as the curve for µ = 0 shown in Fig. 18. In Fig. 21, the tensile force TB at
the mid-length of the bar connector is expressed as a ratio of the applied shear force F.
It can be seen in Fig. 21(a) that the TB /F ratio increases with the increasing shear force,
at a rate that reduces gradually. The values of TB /F at F = 100 kN range from 0.33 for
t = 15 mm to 0.43 for t = 8 mm. Fig. 21(b) shows that the TB /F ratio decreases linearly
with the increasing plate thickness t. The shear strength of the connection is governed by
shear and bending at the bar ends.
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 645
Fig. 23 shows the distribution of the longitudinal stress in the plate, caused by bar shear
forces. Fig. 23(a) presents the longitudinal plate stresses σP in the inner, middle and outside
plate surfaces on a longitudinal section through the centre of the bar. A large local tensile
stress exists around part of the boundary of the bar connector on the inner surface.
646 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
Fig. 22. The maximum principal stress in a bar–plate connection. The concrete stress versus strain curve is given
in Fig. 14. Elastic steel. Friction coefficient µ = 0.
Fig. 24. Variation in plate stresses with plate thickness. FE results. The concrete stress versus strain curve is given
in Fig. 14. Elastic steel. Friction coefficient µ = 0.
As shown in Fig. 24, the maximum principal stress σmax and the longitudinal plate stress
σP at the boundary of the connector on the inner plate surface decrease nearly linearly with
the increasing plate thickness t. However, the stress magnitude indicated will vary with the
mesh size.
The effect of the concrete modulus E c on the shear stiffness of bar connectors is studied
by using a bilinear stress versus strain relationship for concrete (Fig. 25(a)), with the
modular ratio m = E/E c varying from 6.4 to 15. As can be seen in Fig. 25(b), the modular
ratio affects the initial slope of the shear force versus slip curve, because the concrete is
elastic when the applied shear force is small. As the shear force increases, the concrete
at the ends of bar connectors reaches its compressive strength f ck , and consequently the
influence of the elastic modulus of concrete diminishes. For design of Bi-steel sandwich
construction the effect of E c on the shear stiffness can be neglected.
8. Conclusions
(1) Three failure modes were observed in the push tests: plate tearing either through or
within the plate thickness, shear failure through the cross-section of the bar and failure
across the bar/plate interface. The failure modes are illustrated in Fig. 7 and recorded
in Table 2.
(2) The test results show that the ultimate shear strength of the embedded connection in-
creases by about 25% when the plate thickness increases from 6 mm to 10 mm. Further
increases in plate thickness do not increase the shear strength (Fig. 6 and Eq. (5)).
(3) The highest shear strength was achieved when the shear failure occurred across the bar
section, and was about 85% of the tensile strength of the bar connectors and 95% of
the pure shear strength.
648 M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649
(4) For t ≥ 10 mm plates, the bar shear strength given by the experimental Eq. (5) is
about 10% lower than that given by the Bi-steel guide Eq. (1), but about 20% higher
than those given by Eqs. (2) and (4).
(5) The shear stiffness of the embedded connection is derived from the measured shear
slip versus shear force curves (Figs. 9 and 10 and Eq. (6)) and from finite element
analyses (Fig. 20).
(6) The strain measurements on the centre axis of the bar connectors indicate significant
tension caused by bar shear (Fig. 11). The FE solutions show that the bar axial tension
is small at the weld connection, but increases quickly within the first 20 mm from the
connection (Figs. 17 and 18). This indicates that the embedded bar shear strength is
not reduced by tension and partly explains why the shear strength is about 73% of the
tensile strength. The axial strains vary according to the relative stiffnesses of the bar,
the steel plates and the concrete.
(7) FE analyses for shear slip and axial bar force agree reasonably well with test measure-
ments, having regard to the variation between nominally identical specimens, and are
considered to be suitable for parametric studies (Figs. 16 and 19 and Table 3).
M. Xie et al. / Journal of Constructional Steel Research 61 (2004) 625–649 649
Acknowledgments
The work reported here was sponsored by the Engineering and Physical Sciences
Research Council (EPSRC) and by Corus. The authors would like to thank Mr Hugh
Bowerman (Corus) for his technical support.
References
[1] Xie M, Chapman JC. Static and fatigue tensile strength of friction-welded bar–plate connections embedded
in concrete. Journal of Constructional Steel Research [in press].
[2] Moy SSJ, Xiao RY, Lilistone D. Tests for British steel on the shear strength of the studs used in the Bi-steel
system. Department of Civil and Environmental Engineering, University of Southampton; 1998.
[3] Clubley SK, Moy SSJ, Xiao RY. Shear strength of steel–concrete–steel composite panels, Part I—testing and
numerical modelling. Journal of Constructional Steel Research 2003;59(6):781–94.
[4] Bowerman HG, Gough MS, King CM. Bi-steel design and construction guide. Scunthorpe (London): British
Steel Ltd; 1999.
[5] British standard institution. Steel, concrete and composite bridges, BS5400: Part 5. BSI, London: Code of
Practice for Design of Composite Bridges; 1979.
[6] Xie M, Foundoukos N, Chapman JC. Shear and bending tests on friction welded bar–plate connections.
CESLIC Report BS3, Department of Civil and Environmental Engineering, Imperial College London; 2002.
[7] Comite Europeen du Beton (CEB). Deformability of concrete structures—basic assumptions, Bulletin
D’Information No. 90, 1973.