See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/269100489

End-plate steel joint with four bolts in the row

Conference Paper · January 2006

CITATIONS READS

5 730

2 authors:

Zdzisław Pisarek Aleksander Kozlowski

Rzeszów University of Technology Rzeszów University of Technology
9 PUBLICATIONS 19 CITATIONS 67 PUBLICATIONS 70 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

stability View project

joints View project

All content following this page was uploaded by Aleksander Kozlowski on 15 May 2015.

The user has requested enhancement of the downloaded file.

 Progress in Steel, Composite and Aluminium Structures – Giżejowski,
˛
Kozłowski, Śleczka & Ziółko (eds)
© 2006 Taylor & Francis Group, London, ISBN 0-415-40120-8

End-plate steel joint with four bolts in the row

Z. Pisarek & A. Kozłowski

Rzeszów University of Technology, Rzeszów, Poland

ABSTRACT: In steel framed structures, bolted extended end-plate moment connections are often applied.
These types of joints are widely used to connect beams to columns as well as beam and column elements
together. Design rules given in Eurocode 3 (EN 1993-1-8 2005) relate to bolted end-plate joints with only two
bolts in each bolt row. In case of heavy loaded steel frames end-plate joints with four bolts in the row are
used. Different methods are used for the calculation of beam-to-column joints with four bolts in each row. The
component method is recently the most popular method for the estimation of the moment capacity, initial stiffness,
and rotation capacity of the joints. This method has been recently largely developed, but still the resistance and
stiffness of some components are not evaluated. This mainly regards the T-stubs with four bolts. An analytical
model for T-stub with four bolts, which can be used for prediction of the joint resistance and initial stiffness has
been elaborated and compared with tests results.

Keywords: Steel joints, semi-rigid joints, bolted end-plate connection, component method, experimental test.

1 INTRODUCTION a) b)
In steel framed structures, bolted end-plate moment
connections are often used. These types of joints are
widely applied to connect beams to columns and as
beam and column splices. The most common end-
plate connections utilize only in their tension zone two
bolt rows with two bolts in each row. One bolt row is
positioned outside the tension flange of the beam, on
an extended portion of the end-plate. Second row is
positioned inside the beam flange. When the depth of
the beam is large, the moment capacity of the joint
with only two bolt rows in tension is not sufficient to
carry the external moment. Increase of the resistance Figure 1. Modelling of an extended end-plate joint by
of the joint can be achieved by applying additional separate T-stubs.
bolt rows; however additional bolt rows have a rather
small participation in the resistance of the whole joint,
because of their reduced associated lever arm. When end-plate and is also limited to the tensile resistance
greater resistance of joints is required, designers apply of the bolts including prying forces. For joints with
joints with four bolts in each row. Design rules given only two bolt rows, resistance of tension zone can be
in Eurocode 3 (EN 1993-1-8 2005) relate to bolted predicted by modelling behaviour of end-plate with
end-plate joints with only two bolts in each bolt row. the use of the equivalent T-stub model (Fig. 1).
The resistance of such joints is determined by the If inside the beam flanges there are two or more bolt
resistance of the tension, compression or shear zones rows with four bolts in each row (Fig. 2), the modelling
of the connection. The resistances of the shear and of the end-plate by typical T-stubs cannot be applied.
compression zones of joints with four bolts in each row The analytical models used to predict the resistance
can be predicted according to the European Standard of the multiple rows end-plate joints are generally
(EN 1993-1-8 2005) like for joints with two bolts in based on experimental tests. Most of the tests were
each row. The resistance of tensile zone depends on performed for beam splices. In some models, the dis-
the resistance of the beam flange, column flange and tribution of the internal forces amongst the bolts have

257

Figure 2. Multiple rows extended end-plate joints with four
bolts in each row.
been arbitrarily established and regardless of dimen-
sions of the joint. The same problems existing for
beam-to-column connections with unstiffened column
flanges.
2 REVIEWS OF THE ANALYTICAL MODELS
The analytical models of the end-plate multiple rows
joints with four bolts in the row are based on three main
methods (Kozłowski & Pisarek 2005): yield line the-
ory, bolt force determination and mechanical models
using component method.
2.1 Method of yield line theory
The yield line analysis is based on the virtual work
principle. The yield line pattern for an end-plate with
three-bolt rows and four bolts in each row is illustrated
in Figure 3.
Utilizing yield line patterns, the design resistance
MRd of the end plate joint is given by (Sumney &
Murray 2003):
258
tf pf0
pf1
pb
s
h0 g0 g g0
h1
h
h2
tw
bp
Figure 3. Yield line pattern.
where fy,p = the yield strength of the end-plate steel;
tp = the thickness of the end-plate, and Y = the geo-
metric yield line parameter.
The geometric yield line parameter for end-plate
with three-bolt rows and four bolts in each row is
given by:
where: bp = the width of the end-plate; g, g0 = the
horizontal spacing between bolts in one bolt row;
pb = the vertical spacing between bolt rows inside
beam flange; pf 0 , pf 1 = distances between beam flange
and adjacent bolt rows; s = the range of the yield line
pattern, and h0 , h1 , h2 = distances of bolt rows to the
center of compression.
2.2 Bolt force determination
In this method two cases of the end plate behaviour are
considered, namely thick and thin end-plates (Kennedy
& Vinnakota & Sherbourne 1981). In case of a thick
end-plate, the end-plate is stronger than the bolts and
its deformation is small. The resistance of the joint is
determined by tensile resistance of the bolts loaded
only by normal forces.
The design resistance of the joint with thick end-
plate can be predicted from:
where: Bt,Rd = the bolt tensile resistance: ni = the
number of bolts in bolt row i and hi = the distance
from the centerline of the compression flange of the
beam to the centerline of the bolt row i.
 In case of a thin end-plate, the behaviour of the joint
is characterized by end-plate yielding with large plate
deformations. The resistance of the joint is determined
by end-plate deformations and the tensile resistance
of the bolts additionally loaded by prying forces. For
joints with thin end-plates the moment capacity of the
joint is given by:

where: Bt,Rd = the bolt tensile resistance; Tb = the Kennedy et al. Kennedy et al.
preload force in the bolt; ni = the number of bolts in thick end-plate thick end-plate
bolt row i; hi = the distance from the centerline of the
compression flange of the beam to the centerline of
the bolt row i and Qmax,i = the maximum value of the
prying force in bolt row i.
The maximum value of the prying force can be
predicted from equation (Sumney & Murray 2003):

where: fy,p = the yield strength of the end-plate steel;

tp = the thickness of the end-plate; w′ = the net width
Polish Standard
of the end-plate per bolt; an = lever arm for the prying
force and Fi′ = the additional force from: Figure 4. Bolt resistance contribution for end-plate joint.

2.3 Component method

where: fy,p , tp , w′ = as above; bp = the width of the end-
plate; ni = the number of bolts in bolt row i; db = the
diameter of the bolt; fu, b = the ultimate tensile strength
of the bolt and pf , i = the distance between the bolt and
the stiffener.
The simplification of this method leads to the pro-
cedure which allows to determinate of the internal
forces distributions to the individual bolts. A method
presented in Polish Standard (PN–90/B-03200 1990),
based on experimental tests of splices of high beams
(Łaguna 1984), has adopted a simplified method of
force determination. The resistance of the joint can be
predicted from:
where: Bt,Rd = the bolt tensile resistance; ni = the
number of bolts in bolt row i; hi = the distance from
the centerline of the compression flange of the beam to
the centerline of the bolt row i and ωi = the coefficient
of load distribution.
The values of load distribution coefficient used
in Polish Standard (PN–90/B-03200 1990) and
(Kennedy &Vinnakota & Sherbourne 1981) are shown
in Figure 4.
Moment resistance of the joints is based on the low-
est resistance of the constitutive components such as
column web, column flange, beam flange, bolts and
end-plate.
The moment resistance of the joint can be calcu-
lated as:
where: Fr,Rd = the effective design tension resistance
of bolt row r allowing resistance of all components and
hr = the distance from the centerline of the compres-
sion to the centerline of the bolt row r.
The end-plate is divided in separated T-stubs.
The resistance and deformation of the equivalent T-
stubs can be predicted according to different methods
(Krumm 1991), (Ungermann & Schmidt 2005). The
difference between these methods is mainly in the
statical scheme of parts of the end-plate.
3 PRELIMINARY EXPERIMENTAL TESTS
To verify analytical methods, it was decided to conduct
tests for T-stubs with four bolts. The test specimens
consist of two T-elements screwed together by four
M12 bolts. The mode of failure for such a connection

259
 1/2 HEB240

17,3
90

160
4φ13

80

Figure 5. Geometry and dimensions of T-element.

depends on the tensile resistance of the bolts in relation Figure 6. View of a specimen on the testing machine.
to the resistance of the end-plate. To reduce variable
factors, the geometry of the T-elements was the same
for all test specimens. The dimensions of the test speci-
men are presented in Figure 5. The T-elements were
made of steel grade S235. As a variable factor the bolts
grade was adopted, namely grade 8.8 for specimen “A”
and grade 5.8 for specimen “B”.
The tests were carried out on a testing machine
ZD100.A view of the specimen during testing is shown
in Figure 6.
The main aim of the experimental test was to observe
the behaviour of the T-stub under increasing loading
to identify the mode of failure and verify the analyti-
cal model for this component. During the test the load
and the displacement of theT-elements were measured.
The first investigated “A” specimen, where only one Figure 7. Shear of the thread in the nut.
nut was applied on each bolt, failed by shear of the
thread in the nut (Fig. 7).
In the other specimen “A” tests, two nuts on each
bolt were applied. The mode of failure for this connec-
tion was bolt rupture with partial yielding of the T-stub
flange (Fig. 8).
The tensile capacity obtained for the specimen “A”
was 195,2 kN. After the rupture of the bolts adjacent to
the web of the T-element, the joint still demonstrated a
resistance associated to the remaining bolts. The load-
deflection characteristic is presented in Figure 9.
Although in test specimen “B” weaker bolts were
applied than in specimen “A”, the mode of failure of the
connection was the same, i.e. bolt rupture with partial
yielding of the T-stub flange (Fig. 10).
The tensile capacity for the specimen “B” was Figure 8. Bolt failure with partial yielding of the T-stub
110,4 kN. In contrast to the specimen “A”, after flange.

260
 F 200
[kN]
150
100
50
0 0
0 5 10 15 20
δ[mm]
Figure 9. Load displacement characteristic for speci-
men “A”.
Figure 10. The failure of the specimen “B”.
the rupture of the bolts adjacent to the web of the
T-element, the joint had not a clear resistance, because
remaining bolts were already plastifying. The load-
deflection characteristic is presented in Figure 11.
4 PROPOSED ANALYTICAL MODEL FOR
JOINT RESISTANCE
Based on the observed behaviour of the test specimens,
a simple beam model was developed to determine the
behaviour of the T-stub. The resistance of the T-stub
depends on the resistance of the bolts, thickness of the
end-plate and geometrical dimensions of the speci-
men. The possible models of failure are presented in
Figure 12.
Final resistance of the equivalent T-stub in tension is
the minimum value of the resistances obtained for all
failure modes.The bolts forces are determined by static
conditions, admitting elastic distribution of internal
forces to the bolts. The lowest components resistance
governs the ultimate resistance of the bolt rows and
the whole joint.
The effective resistance of the equivalent T-stub can
be received from:
261
F 120
[kN] 100
80
60
40
20
0 5 10 15 20
δ [mm]
Figure 11. Load displacement characteristic for speci-
men “B”.
where: FT .1.Rd = the resistance due to bolt failure;
FT .2.Rd = the resistance due to bolt failure with partial
yielding of the T-stub flange. FT .3.Rd = the resist-
ance due to bolt failure with yielding of the T-stub
flange; and FT .4.Rd = the resistance due to the complete
yielding of the T-stub flange.
4.1 Mode I – bolt failure
This mode of failure takes place if a very thick T-stub
flange is used. The bolts are loaded by the external
force without prying forces. It is a result of the large
plate stiffness. The resistance of the T-stub in this case
can be predicted as a sum of the tensile resistances of
the bolts:
where: Bt.Rd = the tensile resistance of the bolt,
considering tensile resistance of the bolt shank and
punching shear resistance of the bolt head.
4.2 Mode II – bolt failure with partial flange
yielding
If the thickness of the T-stub flange is reduced, the
bolt failure with partial yielding of the T-stub flange
occurs. The resistance of the T-stub can be obtained
from the equilibrium of the external and internal
works using virtual work method for model shown in
Figure 13.
For small angle, it is possible to write:
Virtual work principle can be written as:
where: WE = external loads work, WI = internal loads
work.
 Mode I Mode II Mode III Mode IV

n
Bt Bt,III Mpl,III Bt,III
p
Bt Bt,II Bt,II Mpl,II Bt,II
m
FT FT FT FT
m Mpl,I Mpl,I Mpl,I
Mpl,II
Bt Bt,II Bt,II Bt,II
p
Mpl,III
n Bt Bt,III Bt,III

Figure 12. T-stub model for end-plate with four bolts.

Q Assuming that all bolts in the joint are of the same

grade and diameter, the resistance of the bolts is the
ϕ same:
n
Bt,III
δIII
p then:
Bt,II
δII
m Finally, the resistance of the T-stub can be predicted as:
FT
δI
m Mpl,I

Bt,II
p where: Mpl,I ,Rd = the yielding moment of T-stub
flange;m = reduced distance between the bolts and
the web; p = the bolt spacing; n = minimum distance
Bt,III between the bolt and the edge of the T-stub flange
n Q limited to 1,25 · m; and Bt.Rd = as above.

4.3 Mode III – bolt failure with flange yielding

Figure 13. Simply supported beam model of the bolt failure
with partial yielding of the plate.
When the thickness of the T-stub flange is lower, plas-
tics hinges appear close to the edge bolts. Using the
model of failure as shown in Figure 14, the resistance
of the T-stub can be obtained as follows:

Substituting eqs. (13) and (14) to (12) we obtain:

Equilibrium of the external and internal works leads to:

262
 Q

n
Bt,III Q
Mpl,III
p
ϕ
Bt,II
?
δII
m
FT
δI Mpl,I
m

Bt,II
p
Mpl,III
Bt,III Q
n Q

Figure 14. The simply supported beam model of the bolt Figure 15. Application of T-stubs in extended end-plate.
failure with yielding of the plate.

4.5 Generalization of the model

Finally, the resistance of the T-stub can be calcu-
lated as:
where: Mpl,III ,Rd = the yielding moment for T-stub
flange close to the edge bolt;Mpl,I ,Rd ; m; p; and
Bt.Rd = as above.
4.4 Mode IV – complete yielding of the flange
For thin T-stub flanges, the complete yielding of the
T-stub flange occurs. The resistance of the T-stub in
this case can be predicted like for T-stubs with only
two bolts:
where: Mpl,II ,Rd = the yielding moment for T-stub
flange in place of the bolt adjacent to the web;Mpl,I ,Rd ;
and m = as above.
The yielding moments can be determined as:
For end-plate moment connections, the equivalent
T-stubs are adopted for each bolt row and effective
lengths are assumed for individual bolt row and/or
for group of the bolt rows. The effective length of the
T-stubs can be taken as for T-stubs with two bolts only
for end-plate with bolts adjacent to the web and/or
flange. For other parts of end-plate, such effective
length can not be applied, because the interaction
between perpendicular T-stubs are still not elaborated
(Fig. 15).
5 STIFFNESS OF THE JOINT
Amongst the methods presented in. §2 only the com-
ponent method can be applied to determine the rota-
tional stiffness of the joint. The mechanical model for
calculation of the initial stiffnes (Fig. 16) is based on
a flexibility of the components.
The stiffness of the joint can be predicted from (EN
1993-1-8 2005), as:

where: leff = the effective length (assumed as the width

of the T-stub flange for examined T-stubs); t = the where: E = the elastic modulus; z = the level arm;
thickness of the T-stub flange; fy = the yield resistance µ = the stiffness ratio and ki = the stiffness coefficient
of T-stub; γM 0 = partial safety factor. for basic joint component i.

263
 Q
k3 k4 k5 k10 keq
n ϕ Bt,III n ϕ kbIII
δIII δIII
h1 p p
h2 Bt,II kbII FT/2 FT/2
h3 δII δII
k1 m m
k2 FT FT FT
m δI Mpl,I m
Bt,II FT/2
p ∆δI ∆δI
1 - the column web in shear 4 - the column flange in bending
2 - the column web in compression 5 - the end-plate in bending Bt,III
3 - the column web in tension 10-bolts in tension n Q

Figure 16. Mechanical model for end-plate joint.

Figure 17. The model for calculation of the end-plate
deflection.
The stiffness coefficients for column web in all
zones can be taken as for joint with two bolts in the Assuming that:
row. The flexibilities of the column flange, end-plate
in bending and bolts in tension should be modelled
differently. In the component method procedure, as
introduced by Eurocode 3 (EN 1993-1-8 2005), the
end-plate and bolts are considered separately. Because
of different loading of the particular bolts in T-stub
with four bolts, a modified model considering both
bending of the end-plate and elongation of the bolts and taking into account that:
should be applied.
The stiffness coefficients are obtained for each of
the modes of failure as shown below.

5.1 Mode I we obtain:

The displacement of the T-stub depends only on the
elongation of the bolts. The stiffness coefficient of the
T-stub with four bolts can be taken as twice value of
T-stub with two bolts: Assuming that all bolts in the joint are of the same
grade and diameter, the stiffness of the bolts is the
same:

where: As = the tensile stress area of the bolt; and

p
Lb = the bolt elongation length, taken as a sum of the
thickness of plate and washer, plus half of the heights
of the bolt head or nut. From eq. (34) one can obtain:

5.2 Mode II
The elastic deformation of the T-stub in tension
depends on the bending of T-stub flange and bolts Substituting eq. (36) to (32) we obtain:
elongation (Fig. 17).
Deflection δI in line of the web can be determined
as a sum of the elongation of bolts adjacent to web δII
and flexure of internal part of T-stub 1δI :

The deflection of internal part of T-stub 1δI is obtained

as for simple supported beam from equation:
From static condition:

264
 Table 1. Collection of resistance and stiffness coefficients of T-stubs with four bolts.
Finally, the deflection of the T-stub can be predicted as:
The stiffness coefficient is obtained from:
If the influences of the prying forces are considered,
according to Eurocode 3 (EN 1993-1-8 2005), the
stiffness coefficient is given by:
p presented in Table 2.
where: As ; Lb ; leff ; n; p; m; and t = as above.
5.3 Mode III
The stiffness coefficient can be received as:
p
where: As ; Lb ; leff ; m; and t = as above.
265
5.4 Mode IV
The elastic deformation of the T-stub in tension
depends only on bending of the T-stub flange. The stiff-
ness coefficient can be obtained like for T-stub with
only two bolts, from:
where: leff ; m; and t = as above.
For end-plate moment connections the stiffness
coefficient adopted for each equivalent T-stub is
assumed as for an individual bolt row.
The equations for proposed resistance and stiffness
coefficients are collected in Table 1.
6 COMPARISON OF ANALYTICAL MODEL
WITH TEST RESULTS
Using measured values of the geometrical charac-
teristics of the tested samples and characteristic val-
ues of the material properties, the resistance of the
T-stubs was calculated according to proposed models.
The results of the resistance of the specimens received
from analytical model and from experimental test are
The resistance for the specimen “A” obtained from
analytical model is smaller ca. 11% than from test, and
for specimen “B” is greater (ca. 13%). This difference
can be ensued from dispersions of the properties of the
bolt steel.
7 CONCLUSIONS
The design methods existing up to now for determi-
nation of the resistance of the joint with four bolts in
 Table 2. Comparison between results of the tests and ana- REFERENCES
lytical model.
Kozłowski, A., Pisarek Z. 2005. Characteristics of bolted
Specimen “A” Specimen “B” end-plate joints with four bolts in the row. Proc. of 10th
[kN] [kN] Scient. Conf. Rzeszow-Lviv-Kosice, State of Art, Trends of
Development and Challenges in Civil Engineering, 11–13
Analytical model September, 2005, Kosice, Slovakia.
Mode I 269,8 168,6 EN 1993-1-8. Eurocode 3. 2005. Design of Steel Structures.
Mode II 185,6 126,4 Part 1.8: Design of joints. CEN.
Mode III 174,5 144,0 Sumner, E.A and Murray, T.M. 2003. Behavior and design of
Mode IV 234,2 234,2 multi-row extended end-plate moment connections. Pro-
Experimental test 195,2 110,4 ceedings of the International Conference on Advances in
structures. (ASCCA ’03). Sydney.
Łaguna, J. 1984. Resistance of the end-plate moment con-
nections preloaded by high grade bolts. PhD Thesis, ITB,
each row give results which differ very much. Design Warsaw, (in Polish).
rules given in Eurocode 3 (EN 1993-1-8 2005) relate to Kennedy, N.A., Vinnakota, S., and Sherbourne, A. 1981. The
bolted end-plate joints with only two bolts in each bolt split-tee analogy in bolted splices and beam-column con-
row. The analytical model for assessment of the resist- nections. Joints in Structural Steelwork. John Wiley and
ance and stiffness of T-stub with four bolts presented in Sons, New York, pp. 2.138–2.157.
the paper can be used in the component method. Com- PN-90/B-03200. 1990. Polish Standard. Steel Structures.
parison of the resistance of the joints from proposed Design Rules. PKN Warsaw, (in Polish).
Krumm, R. 1991. Calculation of Rigid Face Plate Connec-
method and test results shows that analytical models
tions According to the DSTV/DASt Guidelines. Stahlbau
rather well agree with experimental results. The com- Vol. 60/3. Berlin.
ponent method has been largely developed in the last Ungermann, D., Schmidt B. 2005. Moment Resistance of
years but still some components are not evaluated. This Bolted Beam to Column Connections with four Bolts in
mainly regards the interaction of perpendicular T-stubs each Row. Proceedings of IV European Conference on
and determination of the effective length of bolts not Steel and Composite Structures. Eurosteel 2005. Maas-
adjacent to the web. Therefore, a research program for tricht.
modelling of end-plate connections with four bolts in
each row is now being initiated.

266

View publication stats