ASE 7-3_02-238 7-6-04 11:51 am Page 245

Finite Element Analysis and Tests on Bolted End-Plate

Connections in Steel Portal Frames

Gang Shi, Yongjiu Shi*, Yuanqing Wang, Shaofu Li and Hong Chen
Department of Civil Engineering, Tsinghua University, Beijing 100084, China.

(Received 14 May 2003; received in revised form 8 September 2003; accepted 18 September 2003)

Abstract: Bolted end-plate connections are being widely used in steel portal
frames. In this paper, some tests on beam-to-column and beam-to-beam
end-plate connections are reported and the corresponding finite element
analysis (FEA) is carried out to analyze their loading resistance and rotational
stiffness characteristics. The design method about extended end-plate
connections in current design code is discussed, and the necessary improvements
are proposed. Especially, some precautions about the classification criterion of
the joint stiffness in steel structures are recommended. From the results of
comparison, one new model of the bolt tension force distribution and the
corresponding design method are presented. It has been verified by the tests
that this method can calculate the actual stresses of the bolts and the end-plates
more accurately and mobilize their loading capacities more efficiently with
sufficient safety and reliability.

Key words: connection, portal frame, extended end-plate, rigidity of connection, finite element analysis.

1. INTRODUCTION 2003). The end-plate can be installed vertically,

1.1. End-Plate Connections horizontally or aslope (Figure 2).
Lightweight steel portal frames have been widely used
in industrial and civil engineering structures for years. It 1.2. Load-Carrying Properties of
has been indicated by a large number of research and End-Plate Connections
engineering practice that the bolted end-plate connection The load-carrying property of bolted end-plate
is the most economical connection form in lightweight connections is usually described by using its moment-
steel portal frames. It is normally used as moment- rotation characteristic. Typical moment-rotation
resistant connections. There are two types: flush end- characteristic of connections is shown in Figure 3. There
plate connections and extended end-plate connections are three main performance indicates, namely, ultimate
(Figure 1). moment resistance Mj,R, initial rotational stiffness Sj,i,
It is suggested to use the end-plate connection as a rotational ability ΦCd.
primary form of rigid joints in the Chinese code Connections in frames can be classified by its
Technical specification for steel structure of light- rotational stiffness as: rigid, semirigid or nominally
weight buildings with gabled frames (CECS102:2002, pinned. The initial rotational stiffness Sj,i is the main

*Corresponding author. Email: [email protected]; Fax: +86-10-6278-8623; Tel: +86-10-6279-2330.

Advances in Structural Engineering Vol. 7 No. 3 2004 245

ASE 7-3_02-238 7-6-04 11:51 am Page 246

Finite Element Analysis and Tests on Bolted End-Plate Connections in Steel Portal Frames

(a) extended (b) extended with stiffener (c) flush

Figure 1. End-plate connections

(a) vertically (b) horizontally (c) aslope (d) beam splice

Figure 2. Forms of end-plate connections in lightweight steel portal frames

classifying standard. The division of joint rotational connects under load. For the beam-column end-plate
stiffness is shown in Figure 4. The connection is rigid connection, joint rotation Φ usually includes two parts:
when the initial rotational stiffness is bigger than some rotation caused by the relative deformation between the
specific value, which is 25EIb/Lb for unbraced frames as end-plate and the column flange, rotation caused by the
defined by Eurocode 3, where EIb/Lb is the linear stiffness shearing deformation of the panel zone. For the beam
of the beam. The connection is nominally pinned when splice connection, joint rotation Φ only includes the
the initial rotational stiffness is smaller than some specific rotation caused by the relative deformation between the
value, which is defined as 0.5EIb/Lb in Eurocode 3 (1993). two end-plates.
Because the moment-rotation curves of connections are
usually nonlinear, the initial rotational stiffness Sj,i is 1.4. Problems in Current Chinese Design Code
often taken as the secant stiffness corresponding to a (1) The design of semirigid connections is not advocated
certain moment, which is 2/3 of the design plastic in current Chinese design code yet. In portal frames,
moment resistance of the beam in Eurocode 3 (1993). the end-plate connection between the beam and
the column is calculated and designed as a rigid
1.3 Joint Rotation of End-Plate Connections connection which requires the connection to have
Joint rotation Φ is defined as the change of the included enough rigidity. But the constructional detail of
angle between the axes of the members that the joint end-plate connections sometime can not satisfy the

M M
Sj,i 1. rigid
2. semirigid
3. pinned
Mj,Rd 1

2
Moment
Moment

Rotation ΦCd Φ Rotation Φ

Figure 3. Moment-rotation characteristic of connection Figure 4. Classification of connections by rigidity

246 Advances in Structural Engineering Vol. 4 No. 3 2004

ASE 7-3_02-238 7-6-04 11:51 am Page 247

Gang Shi, Yongjiu Shi, Yuanqing Wang, Shaofu Li and Hong Chen

requirement for rigid connections (Jing 1999). Table 1. Sectional dimension of beam and
Generally speaking, the rigid-joint assumption will column (unit: mm)
lead to an underestimate of frame drift and
Web Web Flange Flange
overestimate of frame strength (Lui and Chen depth thickness width thickness
1986). Particularly in lightweight steel portal
frames, deflection is a major factor considered in Beam 388 3 160 6
Column 400 8 160 10
structural design, even as the controlling factor.
(2) No method for evaluating is recommended in current
Chinese design code for actual rotational stiffness
of connections. There is no classification criterion 300

for rotational stiffness of connections either. How 250

stress (MPa)
to make the stiffness of connections complying 200
with the assumption of the calculation model, and 150
how to accurately estimate deflection of the 100
structure are questions unanswered. 50
(3) In drafting Chinese current design code for bolted 0
end-plate connections in lightweight steel portal 0 0.002 0.004 0.006 0.008 0.01
frames, experimental researches are insufficient strain
(Li 1998). But there is adaptability and limitation
Figure 5. Stress-strain curve of the steel
in the design method of each country. The relevant
research has also indicated that there is irrationality
in assumptions and calculations (Jing 1999; Shi
and Jing 1999). in Figure 5, which was obtained from material property
Also since flexibly-connected frame behavior is, to a tests. Flanges and webs of the members are all welded
large extent, affected by connection behaviour, additional to end-plates by full penetration groove welds.
and systematic research into connection behaviour is According to CECS102:2002, the bolted end-plate
deemed necessary (Lui and Chen 1986). So, in this connection must be designed to resist the maximum
paper, experimental research and finite element analysis internal force, and when the internal force is small, the
(FEA) on loading capacity and rotational stiffness of resistance should be larger than half of loading capacity
extended end-plate connections are presented and a new of the weaker member connected. In these two tests,
design method is proposed. the resistance of connections are designed to be equal to
that of the connected beam. The test specimens are
2. EXPERIMENTAL PROGRAM assembled with 14 mm thick end-plates and 8 M16 high
2.1. Design of Test Specimens strength friction grip bolts (Class 10.9) are installed on
Two tests were carried out to examine the strength and 4 rows with pretension force at 100 kN (Figure 6).
stiffness of typical extended end-plate connections. The
purpose of these two tests is to verify the feasibility and
35

54

reliability of applying finite element analysis to simulate

and analyze end-plate connections.
54
35

During the connection tests, specimens were

100

54

designed to be rigid following the current Chinese

design code. The main purpose of the tests is to verify
the feasibility and reliability of applying FEA to
695
695

simulate and analyze end-plate connections in steel

portal frames.
Test specimen 1 is a beam-to-column eave
35 100

54

connection with extended end-plate placed aslope. Test

54

specimen 2 is a ridge connection (beam-to-beam) with

extended end-plate placed vertically. Section dimension
35

54

40 80 40 40 80 40
of the beam and the column in both specimens are 160 160

shown in Table 1. The steel is grade Q235 with the true

(a) test 1 (b) test 2
yield strength fy = 270 MPa and the elastic modulus
E = 2.06 × 105 MPa, and its stress-strain curve is shown Figure 6. Configurations of connections (unit: mm)

Advances in Structural Engineering Vol. 7 No. 3 2004 247

ASE 7-3_02-238 7-6-04 11:51 am Page 248

Finite Element Analysis and Tests on Bolted End-Plate Connections in Steel Portal Frames

jack 3. FINITE ELEMENT ANALYSIS (FEA)

3 3.1. Finite Element Model
7 4
1
Large-scale general-purpose finite element software
ANSYS is used for the analysis and calculation of the

400
6
two test specimens. In the FEA models, all components
1
of the two specimens are meshed by 8-node solid
5 10 2
2 element SOLID45; the interfaces between the end-
200
8
plates are simulated by creating contact pairs with 3-D
1400

10 target surface elements TARGE170 and 3-D 8-node

200
surface-to-surface contact elements CONTA174.
High strength friction grip bolts are firstly meshed by
4 3 8-node solid element SOLID45. And then, the PSMESH
command is used to define the pretension sections in the
middle of the bolt shanks and generate the pretension
420 1000 100
elements PRETS179 through which the pretension force
of the bolts are applied by using the command SLOAD.
Figure 7. Test setup of test specimen 1
The FEA model of a high strength friction grip bolt
jack jack is shown in Figure 9. And the FEA models of test
specimen 1 and 2 are given in Figure 10 and Figure 11.
5 7
3 9
1
10 1 6 11 3.2. Material Properties
9 3
6 8 In the FEA models, the adopted stress-strain relationships
5
10
2
4 7
2
8 10
12
4 of materials are elastic-perfectly plastic. Yield strength of
100 1000 500 500 1000 100 high strength friction grip bolts (including bolt heads,
shanks and nuts) is 990 MPa, and that of the steel plates is
Figure 8. Test setup of test specimen 2

The configurations of tests 1 and 2 are given in

Figure 7 and Figure 8 respectively. In both figures, No.
2~10 designate displacement transducers to measure the
deformation of the specimens. Other numbers indicate
strain gauges to record the stress of specimens. Figure 9. FEA model of a high strength friction grip bolt

2.2. Test Process

Monotonic load was applied up to failure of the specimens
by hydraulic jacks. The out-of-plane deformation in the
specimens was restrained. The failure mode of test
specimen 1 is that the compressed flange near the column
base evaginated and the web of the column buckled. The
failure mode of test specimen 2 is that the upper flange
near the loading point buckled. The measured maximum
loads are listed in Table 2. The load-displacement curves
are shown in Figures 12 and 17.

Table 2. Comparison of loading capacity of FEA

and test values

Test value FEA value FEA value

(kN) (kN) test value
Specimen 1 143.27 162.82 1.14
Specimen 2 167.19 166.49 1.00
Figure 10. FEA model of test specimen 1

248 Advances in Structural Engineering Vol. 7 No. 3 2004

ASE 7-3_02-238 7-6-04 11:51 am Page 249

Gang Shi, Yongjiu Shi, Yuanqing Wang, Shaofu Li and Hong Chen

capacity from FEA is 162.82 kN which is close to the

measured value, and the corresponding deflection at the
loading point is 60.00 mm. The load-displacement curves
of the loading point from the test and the FEA are shown
in Figure 12. From comparison of these two curves, it can
be seen that the initial stiffness of specimen 1 from FEA
Figure 11. FEA model of test specimen 2
is larger than that of the test, the reasons may be:
(1) In FEA, it is assumed that the column base is
270 MPa, which were obtained from material properties initially ideally rigid. In fact, during the test
tests. The elastic modulus is taken as 2.06 × 105 MPa, the there was possibly existing rotation, and a very
Poisson’s ratio is 0.3. The pretension force applied on each minor rotation of the column base can lead to
high strength friction grip bolt is 100 kN, and the friction large vertical deformation at the loading point.
coefficient of contact surfaces between end-plates is 0.35. (2) Because of tolerance in fabrication and
assemblage, the end-plates can hardly be in
3.3. Analysis Program contact completely. So, there are minor gaps
During the analysis and solution of the two FEA between the end-plates which will reduce the
models, displacements were applied according to the rotational stiffness of the connection and increase
test measurements. The maximum displacement applied rotational deformation.
at the actual loading point in Test 1 is 60 mm which The moment-rotation curves of the end-plate
was applied through 15 load substeps. The maximum connection of specimen 1 from FEA and test are shown
displacement applied at the actual loading point in Test in Figure 13. According to the definition of the joint
2 is 16 mm which was applied through 7 load substeps. rotation above, the joint rotation of test specimen 1 can
The analysis type is “Large Displacement Static” which be measured by No. 5, 8, 10 displacement transducers.
is second-order elasto-plastic analysis. The material yield 2/3 of the design plastic moment resistance of the beam
criterion is von Mises yield criterion, and the flow rule is 70.40 kN-m, and the initial rotational stiffness is taken
is adopted after yielding. as the secant stiffness corresponding to this moment.
The linear stiffness of the beam EIb/Lb is 7586 kN-m
(Lb = 2.42 m). The initial rotational stiffness Sj,iF of the
4. COMPARISON BETWEEN THE connection from FEA is 15460 kN-m/rad which is 2 times
RESULTS OF FEA AND TESTS of EIb/Lb. The initial rotational stiffness Sj,iT of the
4.1. FEA and Test Results of Specimen 1 connection from test is 3811 kN-m/rad which is 0.5 times
The measured ultimate loading capacity of specimen 1 of EIb/Lb, so this connection can be classified as semirigid
is 143.27 kN, and the corresponding deflection measured according to Eurocode 3. Because of limitation of test
at the loading point is 57.04 mm; the ultimate bearing conditions, the net length of the beam in test 1 is only

180

150

120
Load (kN)

90
Test
FEA
60

30

0
0 10 20 30 40 50 60
Displacement (mm)

Figure 12. Load-displacement curves of specimen 1

Advances in Structural Engineering Vol. 7 No. 3 2004 249

ASE 7-3_02-238 7-6-04 11:51 am Page 250

Finite Element Analysis and Tests on Bolted End-Plate Connections in Steel Portal Frames

200

Moment (kN-m) 160

120
FEA
Test
80

40

0
0 0.02 0.04 0.06 0.08 0.1
Rotation Φ

Figure 13. M-Φ curves of specimen 1

From FEA results, curves of bolts tension force

against loads are shown in Figure 15, in which the first,
second, third and fourth bolt-row is sequential from the
out side of the beam tension flange to that of the beam
compression flange. The tension force distribution of
the four bolt-rows is shown as Figure 16, in which curves
1, 2 and 3 indicate the initial stage, the intermediate
stage (elastic stage) and the ultimate stage of loading,
the loads of which are 87.72 kN, 144.06 kN and
162.82 kN separately. In addition, curve A is the tension
force distribution of the four bolt-rows under the
Figure 14. Deformation of the end-plate of specimen 1 moment corresponding to the design value of the moment
resistance (resistance stage). Because the pretension force
of 100 kN is applied at each bolt, in Figure 16, the bolt
is in tension zone if its tension force is more than 100 kN,
1.00 m while the beam of the same section is often and vice versa.
used in lightweight steel portal frames with typical From Figure 16, on the elastic stage and resistance
span as long as 12 m. So the linear stiffness of the beam stage, the tension force of the second bolt-row (the first
EIb/Lb is 1530 kN-m, and Sj,iF is 10EIb/Lb and Sj,iT is bolt-row in the inner side of tension flange) is maximal,
2.5EIb/Lb, and hence this connection is still semirigid to and it is significantly larger than that of the first bolt-
the European code classification standard. row (the bolt-row at the out side of tension flange). Only
The ultimate deformation status of the end-plate from when the load is approximate to the ultimate bearing
FEA is shown as Figure 14. It can be seen that, under the capacity of specimen 1, the tension force of the first bolt-
ultimate load, the relative deformation at the middle of row (the bolt-row at the out side of tension flange) is
the end-plates is relatively large, and there are obvious developed fully and exceeds the second bolt-row (the
gaps with a maximum width of 0.7 mm. This is mainly first bolt-row in the inner side of tension flange). During
because no bolt is placed there and deformation is not the loading process, the rotation center of the connection
restrained. In addition, the part of the end-plates where is generally on the center line of compression flange,
the tension flanges of the column and beam are complying with the design assumptions of the European
connected tends to separate, and the gap width is 0.3 mm. code (Eurocode 3 1993).
From the distribution of stress on the end-plates, the
stress where the end-plates connect with the tension 4.2. FEA and Test Results of Specimen 2
flanges of the column and the beam is maximum and it The measured ultimate loading capacity of specimen 2
yielded. is 167.19 kN, and the mid-span deflection measured is

250 Advances in Structural Engineering Vol. 7 No. 3 2004

ASE 7-3_02-238 7-6-04 11:51 am Page 251

Gang Shi, Yongjiu Shi, Yuanqing Wang, Shaofu Li and Hong Chen

120

Bolt-row tension force (kN)

115 first bolt row
second bolt row
third bolt row
110 fourth bolt row

105

100

95
0 50 100 150 200
Load (kN)

Figure 15. Bolt-row tension force curves of specimen 1

400

300

200
Bolt-row positioin (mm)

100

0
95 100 105 110 115 120
−100

−200 1
2
−300 3
A
−400
Bolt-row tension force(kN)

Figure 16. Bolt-row tension force distribution of specimen 1

18.23 mm. The ultimate bearing capacity from FEA is 6120 kN-m (Lb = 3 m). The initial rotational stiffness
166.49 kN and the corresponding mid-span deflection Sj,iF of the connection from FEA is 318708 kN-m/rad
is 17.44 mm, which both agree with the measured value, which is 52.1 times of EIb/Lb. The initial rotational
and it has less variation compared with specimen 1. stiffness Sj,iT of the connection from test is 156171 kN-
The load-displacement curves of the mid-span point of m/rad which is 25.5 times of EIb/Lb. This connection is
specimen 2 from the test and FEA are shown in Figure17. rigid according to Eurocode 3. If the tested beam is used
The moment-rotation curves of the end-plate in lightweight steel portal frames with typical span
connection of specimen 2 from FEA and test are shown about 12m, the linear stiffness of the beam EIb/Lb is
in Figure 18. According to the definition of the joint 1530 kN-m, and Sj,iF is 208EIb/Lb and Sj,iT is 102EIb/Lb,
rotation above, the joint rotation of test specimen 2 can and hence this connection can be considered as rigid.
be measured by No. 5, 6 displacement transducers. 2/3 According to the FEA results, tiny gaps occur only in
of the design plastic moment resistance of the beam the very local area of the end-plates around the tension
is still 70.40 kN-m, and the initial rotational stiffness flange, and the end-plates contact closely in most part.
is taken as the secant stiffness corresponding to this Corresponding to this, the distribution of flexural stress
moment. The linear stiffness of the beam EIb/Lb is on the end-plates is that, the stress where is near to the

Advances in Structural Engineering Vol. 7 No. 3 2004 251

ASE 7-3_02-238 7-6-04 11:51 am Page 252

Finite Element Analysis and Tests on Bolted End-Plate Connections in Steel Portal Frames

180

150

120
Load (kN)

FEA
90
Test

60

30

0
0 4 8 12 16 20
Displacement (mm)

Figure 17. Load-displacement curves of specimen 2

180

160

140

120
Moment (kN-m)

Test
100 FEA
80

60

40

20

0
0 0.001 0.002 0.003 0.004 0.005 0.006
Rotation Φ

Figure 18. M-Φ curves of specimen 2

150

140 first bolt-row

Bolt-row tension force (kN)

second bolt-row
130 third bolt-row
fourth bolt-row

120

110

100

90
0 30 60 90 120 150 180
Load (kN)

Figure 19. Bolt-row tension force curves of specimen 2

252 Advances in Structural Engineering Vol. 7 No. 3 2004

ASE 7-3_02-238 7-6-04 11:51 am Page 253

Gang Shi, Yongjiu Shi, Yuanqing Wang, Shaofu Li and Hong Chen

400

300 1
2
Bolt-row position (mm) 200 3
A
100

0
90 100 110 120 130 140 150
−100

−200

−300

−400
Bolt-row tension force (kN)

Figure 20. Bolt-row tension force distribution of specimen 2

tension flanges of the beam and the lower two bolt-rows tension force is the first bolt-row on the inner side
is the largest and it yielded. of tension flange and the rotation center of the
From the FEA results, the bolt tension force against connection is generally on the center line of
applied load is shown as Figure 19, in which the first, compression flange. Some existing researches have
second, third and fourth bolt-row is sequential from the pointed out that, in order to ensure the end-plate
top to bottom. Similar to Figure 16, the tension force rotates as a plane and increase the tension force of
distribution of the four bolt-rows under different loading the bolt-row on the out side of tension flange so as
is shown as Figure 20, and the loads of curves 1, 2 and to increase moment resistance and improve the
3 are 55.98 kN, 147.25 kN and 166.49 kN separately. earthquake-resistant performance of bolted end-
From this, during the loading process, the tension plate connections, the solution is either using much
force on the third bolt-row (the first bolt-row in the inner thicker end-plate or applying rib stiffener on the
side of tension flange) is maximal, and it is significantly extended part of the end-plate (Tsai and Popov
larger than that of the fourth bolt-row (the bolt-row in 1990; FEMA-355D, 2000) .
the out side of tension flange), and the rotation center of (3) From the results of the FEA and tests, it can be
the connection is generally on the center line of predicted that, the extended end-plate connection
compression flange, which both agree with the design in most cases can satisfy the requirement of rigid
assumptions of Eurocode 3 (1993). connection if it is designed appropriately. But in
many cases it should be as a semirigid connection.
5. DISCUSSION Its detailed mechanical characteristics and design
(1) Table 2 gives the loading capacity of the two method are to be studied further.
specimens from FEA and the tests. From the results Especially, the differences between the test and
of the comparison and some other research findings, FEA values of connection rotational stiffness of the
it can be concluded that FEA can be applied to above two specimens indicate that it should be taken
calculate the ultimate bearing capacity and the into account that error of fabrication and assemblage
rotational stiffness of the end-plate connection very will lead decrease of connection rotational stiffness
well (Sherbourne and Bahaari 1994, 1997; Bahaari when the connection classification standards are
and Sherbourne 1997). regulated.
(2) The current Chinese design code assumes that the (4) In CECS102:2002, the formulae for calculating
end-plate is stiff enough and rotates as a plane thickness of the end-plate are deduced according to
about its central axis and the outermost bolt-row on the yield limit analysis theory in which the ultimate
the tension side bears the largest tension force. tension force of the bolt is replaced with the design
Neither assumption coincides with the tests or value of the tension force. Even the tension force
FEA. In fact, the bolt-row which resists the largest distribution model of the bolts adopted is traditional

Advances in Structural Engineering Vol. 7 No. 3 2004 253

ASE 7-3_02-238 7-6-04 11:51 am Page 254

Finite Element Analysis and Tests on Bolted End-Plate Connections in Steel Portal Frames

triangular distribution which assumes that the end-

N1

h1
plate is a rigid plate. This is inconsistent with the
yield limit analysis theory, consequently causing
deviation of calculating results, which often gives N2
much thicker end-plate. For example, when the
Ni
extended part of the end-plate is a projecting plate
where the tension force of the bolt-row is largest,

y1
h
y2
this tension force requires extremely thick end-

yi
plate to resist. But from the FEA and tests results,
the tension force of the outermost bolt-row is much
less but the actual tension force of the first bolt-row
in the inner side of tension flange is larger than the
calculated result from triangular distribution.
Therefore, when calculating the thickness of the
end-plate, instead of adopting triangular distribution to
determine the bolt-row tension force, the model of the Figure 21. Tension force distribution of bolts proposed
bolt-row tension force distribution introduced in the
following text is recommended. So the appropriate
thickness of the end-plate can be obtained in the design line of compression flange, and the tension force of the
of bolted end-plate connections ensuring strength and first bolt-row in the inner side of tension flange is
stiffness of the connection. maximal, so the tension force of each bolt can be
calculated from Eqns (1) to (5). It is required that the
6. PROPOSED DESIGN METHOD calculated maximal tension force of the bolt must not
The design procedure of the bolted end-plate connection exceed the design value of the tension resistance of
is normally suggested as: firstly, determine the tension single bolt.
force distribution of the bolts according to the applied
 n

load, so as to calculate the tension force Ni on each bolt-
row; secondly, define appropriate rule and method to
N2 = My2 m y1 y2ξ1 ξ2 +

∑y
i=2
i
2


(1)

determine how the total shear force V distributes among

all the bolts based on the type of the high strength bolts, N1 = N2ξ1 ξ2 (2)
and finally check the shear and the tension resistance
of the bolts. It is necessary for the bolted end-plate Ni = N2 yi y2 (3)
connection design to adopt a reasonable mechanical
model and to determine the force distribution of the ξ1 = (h1 + h − y1 ) h1 (4)
bolts taking into account the characteristic of the
lightweight steel structure design work. ξ2 = y2 h (5)
According to the above FEA and tests as well as
the existing relative research findings, especially the This paper compares the proposed design method
research on the tension force distribution of the bolts with Chinese current design code on the basis of the two
and the end-plate deformation, the tension force tests as shown in Table 3. The test value is the ultimate
distribution model of the bolts is proposed for the load of the test.
extended end-plate connection without stiffener on the The design and calculation of the thickness of the
extended part of the end-plate as shown in Figure 21. end-plate can refer to Eqns (7.2.9-1) to (7.2.9-4) in
The rotation center of the connection is on the center CECS102:2002 (2003).

Table 3. Comparison of test value, code value and proposed value of the connection moment resistances

Test value Code value Proposed value Test value Test value
(kN-m) (kN-m) (kN-m) Code value Proposed value
Specimen 1 173.4 64.4 131.1 2.69 1.32
Specimen 2 167.2 64.5 116.0 2.59 1.44

254 Advances in Structural Engineering Vol. 7 No. 3 2004

ASE 7-3_02-238 7-6-04 11:51 am Page 255

Gang Shi, Yongjiu Shi, Yuanqing Wang, Shaofu Li and Hong Chen

7. CONCLUSIONS Shi, Y.J. and Jing, J. (1999). “Design moment resistance of end plate
(1) For the extended end-plate connection without connections”, The Second International Conference on Advance
stiffener on the extended part of the end-plate, the in Steel Structures, Hong Kong, China, December, Vol. 1,
first bolt-row on the inner side of tension flange pp. 277-282.
resists the largest tension force and the rotation center Sherbourne, A.N. and Bahaari, M.R. (1994). “3D simulation of
of the connection is generally on the center line of end-plate bolted connections”, Journal of Structural Engineering,
compression flange. Vol.120, No.11, pp. 3122-3135.
(2) In many cases, the extended end-plate connection Sherbourne, A.N. and Bahaari, M.R. (1997). “Finite element
should be regarded as a semirigid connection. Its prediction of end plate bolted connection behabior I: parametric
detailed mechanical characteristics and design study”, Journal of Structural Engineering, Vol. 123, No. 2,
method are to be studied further. pp. 157-164.
(3) When the connection rigidity classification Bahaari, M.R. and Sherbourne, A.N. (1997). “Finite element
standards are regulated, the influence of fabrication prediction of end plate bolted connection behavior II: analytic
and assemblage error should be taken into account. formulation”, Journal of Structural Engineering, Vol. 123, No. 2,
(4) It can be seen from Table 3, the design method of pp. 165-175.
the connection moment resistance in current Chinese Tsai, K. C. and Popov, E.P. (1990). “Cyclic behavior of end-plate
design code is too conservative, while the proposed moment connections”, Journal of Structural Engineering,
method above in this paper can calculate the actual ASCE, Vol. 116, No. 11, pp. 2917-2930.
stress of the bolts and the end-plate more accurately Federal Emergency Management Agency. FEMA-355D (2000).
and mobilize their loading capacities efficiently. State of the Art Report on Connection Performance.
(5) It has been validated that FEA can simulate and
analyze end-plate connections with sufficient NOTATION
feasibility and reliability. In addition, the The following symbols are used in this paper:
experimental research and FEA on cyclic behavior Mj,R = ultimate moment resistance of the connection,
of end-plate connections are now being conducted, kN-m;
and the corresponding research findings will be Sj,i = initial rotational stiffness of the connection,
published soon. kN-m;
ΦCd = rotational ability of the connection, rad;
ACKNOWLEDGEMENTS E = elastic modulus, MPa;
This work was supported by the Outstanding Teachers Ib = inertia moment of the beam section, m4;
Supporting Program of Chinese Ministry of Education. Lb = length of the beam, m;
fy = yield strength, MPa;
REFERENCES Sj,iF = initial rotational stiffness of the connection
CECS102:2002 (2003). Technical Specification for Steel Structure from FEA;
of Light-Weight Buildings with Gabled Frames. (in Chinese) Sj,iT = initial rotational stiffness of the connection
Eurocode 3 (1993). Design of Steel Structures: Part 1.1 General from test;
Rules and Rules for Buildings. yi = distance from the center line of ith bolt-row to
Jing, J. (1999). Behavior and Design Method of Bolted End-Plate the center line of compression flange;
Connection, Master thesis, Department of Civil Engineering, M = design value of the moment on the connection;
Tsinghua University, Beijing, China, (in Chinese). n = number of the bolt-rows;
Lui, E.M. and Chen, W.F. (1986). “Analysis and behaviour of m = number of the bolt columns;
flexibly-jointed frames”, Engineering Structures, Vol.8, No.4, h = distance between the center lines of tension
pp. 107-118. flange and compression flange of the beam;
Li, S. (1998). “Bolted end-plate connection of steel structure”, h1 = distance from the center line of tension flange to
Building Structures, No.8, pp. 24-26. (in Chinese). the edge of the extended part of the end-plate.

Advances in Structural Engineering Vol. 7 No. 3 2004 255

ASE 7-3_02-238 7-6-04 11:51 am Page 256

Finite Element Analysis and Tests on Bolted End-Plate Connections in Steel Portal Frames

Gang Shi is a PhD graduate student at the Department of Civil Engineering, Tsinghua
University. His topic of dissertation is loading capacity and seismic behavior of steel
structural semirigid end-plate connections.

Yongjiu Shi is Professor at the Department of Civil Engineering, Tsinghua University. In

1989, he got his PhD from Heriot-Watt University, Edinburgh, UK. In the following three
years, he worked in The Steel Construction Institute as a postdoctoral fellow. His main
research field is the mechanical behavior and design method of steel structures.

Yuanqing Wang is Associate Professor at the Department of Civil Engineering, Tsinghua

University. His main research field is the mechanism of brittle fracture under temperature
of steel and design method of steel structures.

Shaofu Li is Professor at the Department of Civil Engineering, Tsinghua University. He has

been working in the steel structures area for more than 40 years.

Hong Chen is post doctorate at the Department of Civil Engineering, Tsinghua University.
He got his PhD from the China University of Mining and Technology. His main research field
is the seismic behavior of steel structures

256 Advances in Structural Engineering Vol. 7 No. 3 2004