Journal of Constructional Steel Research 145 (2018) 70–81

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Limit analysis for stiffened eight-bolted base plate connections in EHV

transmission towers
Hongjun Liu ⁎, Yazhou Liu, Zhengliang Li, Haibin Huang
College of Civil Engineering, Chongqing University, Chongqing 400045, China
Key Laboratory of Construction and New Technology of Mountainous Region Cities (Chinese Ministry of Education), Chongqing University, Chongqing 400045, China

a r t i c l e i n f o a b s t r a c t

Article history: Theoretical models based on yield-line theory are presented for predicting the ultimate uplift capacity of stiffened
Received 6 June 2017 eight-bolted base plate (EBBP) connections in EHV transmission tower. Based on the test observations, the four
Received in revised form 2 January 2018 models for rigid connections consider combined effects of the base plate, stiffeners and anchor bolts and the
Accepted 12 February 2018
model for flexible connections merely considers the base plate. The presented models are validated using exper-
Available online 29 March 2018
imental data from an associated test program and compared with the results calculated by methods in a corre-
Keywords:
sponding code, and reasonable accuracy has been demonstrated. A parametric analysis shows that the
EHV transmission tower presence of the stiffeners makes numerous contribution to the strength of rigid connections in particular cases.
Base plate © 2018 Elsevier Ltd. All rights reserved.
Stiffener
Yield-line theory
Uplift load

1. Introduction
In a large proportion of extra-high voltage (EHV) transmission
towers it is common practice to employ tower base connections to
transmit the load from tower legs to concrete footings (see Fig. 1a, b).
They are typically fabricated by welding a steel base plate across the
boot plate end and using four or eight anchor bolts placed around the
base plate. The design of boot plates, base plates, stiffeners, and anchor
bolts is generally determined by the capacity to resist the uplift force
due to wind load and broken cable conditioning. Simultaneously with
the increasing application of EHV transmission towers in the Chinese
electronic industry, eight-bolted base plate (EBBP) connections have
been utilized instead of the conventional four-bolted base plate
(FBBP) connections in order to withstand the increasing magnitude of
uplift load. Currently, there are no industry standards that provide spe-
cific requirements for the analysis of EBBP connections in latticed steel
transmission structures. The design method of FBBP connections is di-
rectly implemented to EBBP connections, which is unrealistic [1].
Thus, it is critical to conduct further investigations on the mechanical
properties of EBBP connections.
The large number of completed studies on steel column bases and
bolted flange-plate connections have laid a reasonable foundation for
the study of EBBP connections in EHV transmission towers. Two models
with three-dimensional yield-line patterns were developed by Kato and
⁎ Corresponding author at: College of Civil Engineering, Chongqing University,
Chongqing 400045, China.
E-mail address: LHJ20040308@126.com (H. Liu).
Mukai [2] in their study of flange-plate connections with bolts along
four sides of a hollow structural section. Packer et al. [3–5] conducted
a series of investigations on the behaviour of bolted rectangular hollow
section (RHS) flange-plate connections under axial tension. Subse-
quently investigations on the same connections subjected to bending
were conducted by Wheeler et al. [6,7], in which a straightforward
and accurate model for predicting the strength of these connections
was proposed using a modified stub-tee analogy coupled with yield-
line analysis. Wang et al. [8] proposed a practical design model for
four basic types of bolted flange-plate connections with and without
stiffeners under pure bending. In these investigations on bolted
flange-plate connections, prying forces were considered. However, pry-
ing forces in base plate connections are generally marginal and
neglected during calculations because of the limited stiffness of grout
and larger elongation of anchor bolts in comparison with the stiffness
of base plates. Hon and Melchers [9] conducted a series of tests on
column-bases under axial and bending load and three base plate
yield-line mechanisms contingent on base plate thickness were ob-
served. References [10,11] proposed analytical models for column-
bases under tension load, in which tee-stub analogy coupled with
yield-line analysis was used as well.
There are two representative types of EBBP connections in EHV
transmission towers [1]: a) rigid EBBP connections and b) flexible
EBBP connections, as illustrated in Fig. 2. In comparison with steel col-
umn base plate and tubular flange-plate connections, due to the pres-
ence of stiffeners and the three-dimensional configuration of these
connections, it is not convenient to establish theoretical models based
on the conventional stub-tee analysis.

https://doi.org/10.1016/j.jcsr.2018.02.006
0143-974X/© 2017 Elsevier Ltd. All rights reserved.
 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81 71

Fig. 1. Plot of a representative type of EBBP connections in EHV transmission tower.

This study is devoted to developing a theoretical analysis method
based on yield-line theory to predict the ultimate strength of EBBP con-
nections under uplift load. A total of 15 specimens were tested. Accord-
ing to the test results, high attentions have been paid to the effects of
stiffeners for rigid EBBP connections. Four yield-line models for rigid
EBBP connections considering the contributions of stiffeners and anchor
bolts and one for flexible EBBP connections are presented.
2. Experimental work
2.1. Test specimens and testing procedure
Two sets of tests were carried out. Overall, 15 specimens were tested
under quasi-static tensile loading in a 2000 t servohydraulic test ma-
chine (see Fig. 3). In order to investigate the static behaviour of EBBP
connections under uplift load, the horizontal shear forces induced by
the inclined tower leg were neglected. Thus, the boot plates were de-
signed to be symmetrical and perpendicular to the base plate and
jointed to a pair of vertical stub angles by bolts. The nominal geometric
dimensions and measured material properties of these test specimens
are presented in Tables 1 and 2, where t is the thickness of base plate,
S is the distance between the bolt hole centre and the centre line of ad-
jacent boot plate, D is the diameter of anchor bolt, tb and hb are the thick-
ness and depth of the boot plate, ts and hs are the thickness and depth of
the stiffener. Six rigid EBBP specimens were designed according to
Ref. [1], considering the variation of thickness and width of base plates,
diameter of anchor bolts, and thickness and depth of stiffeners (see
Table 1). For the design of the nine flexible EBBP specimens, only varia-
tions of geometric properties of base plates and anchor bolts were con-
sidered (see Table 2). The anchor bolts connecting the specimens to a
concrete footing were 8.8 grade high strength bolts [12] of diameters
ranging from 42 to 60 mm. The nominal throat thickness of the fillet

Fig. 2. Plot of detailing for EBBP connections.

72 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81

Fig. 3. Test set for EBBP connections.

welds connecting the boot plate to base plate and those connecting the
stiffener to base plate and to boot plate was 8 mm.
In order to measure the local strain values during the tests, strain
gauge groups designated No.1–6 and No. 7–12 were installed on each
rigid and flexible EBBP specimen, respectively (see Fig. 2):
1) Strain gauge groups designated No. 1–3 and No. 7–9 were installed
in the intermediate zones between each bolt hole and adjacent
boot plate for rigid and flexible EBBP connections, respectively. The
direction of measurement for strain gauge groups designated No. 1
and 7 was parallel to the adjacent boot plate, for No. 3 and 9
perpendicular to the adjacent boot plate and for No. 2 and 8 inclined
at 45°.
2) To study the stress distribution of the stiffeners of the rigid EBBP
specimens, strain gauge groups designated No. 4–6 were installed
at the middle depth of the stiffeners at distance S from the centre
line of the boot plate, with three directions of measurement:
throughout the depth of the stiffeners (No. 4), inclined at 45° (No.
5) and along the longitudinal direction of the stiffeners (No. 6).
3) Strain gauge groups designated No. 10–12 were installed at the loca-
tions adjacent to the base plate centre.
Linearly varying displacement transducers (LVDTs) were mounted
to measure the displacement of the base plate centre and edge in the
test procedure.
2.2. Test results and observations for rigid EBBP connections
Owing to the limitation of the loading capacity of the test instru-
ment, only G2 and G3 were loaded to failure; they underwent signifi-
cant deformation in both base plates and stiffeners. The load versus
centre-displacement curves for rigid EBBP connections are plotted in
Fig. 9, one point need to be noted is that parts of the displacement
data of G2 were lost and its ultimate strength was determined by visible
deformation as shown Fig. 4. The experimental ultimate strengths Tu of
the rigid EBBP specimens are listed in Table 3. From the test results, cer-
tain critical observations can be obtained as follow:
1) Looking at the test results of specimens G1 and G3, with pronounced
variation of stiffener dimensions, the ultimate strength of G3 is ap-
parently lower than G1 (see Table 3), while the thickness of base
plate of G3 has an increment of 4 mm. Therefore, it is indicated
that the stiffener dimensions exert a significant influence on the
strength of rigid EBBP connections under uplift load which will be
considered further in the yield-line analysis in Section 4.
2) According to the ultimate deformation of specimens G2 and G3, il-
lustrated in Figs. 4a and 5, negligible deformation was observed in
the boot plates. The base plates exhibited significant out-plane
Table 2
Geometry and material properties of flexible EBBP specimens.

Specimen Base plate Bolt Boot Stiffener Material properties

plate
Table 1 t S S1 L D tb hb ts × hs fy (MPa) fys (MPa)
Geometry and material properties of rigid EBBP specimens.
R1 28 80 100 95 36 12 500 8 × 140 352.1 364
Specimen Base Bolt Boot Stiffener Material properties R2 22 80 160 95 42 12 500 357.1 371.8
plate plate R3 20 80 160 95 42 14 500 358.4 371.8
R4 36 100 120 115 42 14 500 333.1 371.8
t S D tb hb ts hs fy (MPa) fys (MPa)
R5 32 100 120 115 36 14 500 336.9 371.8
G1 20 80 42 12 500 10 150 356.8 375.2 R6 28 100 120 115 42 14 500 352.1 371.8
G2 20 120 42 12 500 10 150 356.8 375.2 R7 40 100 120 115 42 14 500 358.4 371.8
G3 24 80 42 12 500 8 60 352 363.5 R8 28 120 140 135 42 12 500 352.1 371.8
G4 28 120 56 14 600 12 170 345.5 366.2 R9 32 120 140 135 42 14 500 336.9 371.8
G5 24 120 52 12 600 12 170 352 366.2
Note: The yield strengths of base plates (fy) and stiffeners (fys) were tested according to the
G6 32 120 60 14 600 12 150 330.9 366.2
procedure mentioned in Ref. [15].
 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81 73

(a) Base plate (b) Stiffeners

Fig. 4. Ultimate deformation of specimen G2.

flexure near the interfaces between base plates and boot plates and yield-line mechanism model, which is mainly based on experimental
the end of the interfaces between base plates and stiffeners. observations. Virtual work principles are used to obtain the analytical
3) For the stiffeners, the in-plane deformation mainly occurred around expression of the ultimate strength for each of the models. The external
the interfaces between boot plates and stiffeners, as illustrated in virtual work done by the uplift force is expressed as
Fig. 4b.
4) The measured strain of strain gauge groups designated No. 1–3 for G2 δW E ¼ Tδ=4 ð1Þ
and G3 is over 2500 εμ, as illustrated in Fig. 7a, which implies that the
where δ is the virtual displacement at the boot plate end, T is the uplift
zones between bolt holes and boot plates have already yielded.
load subject to the whole connection. The internal virtual work dissi-
5) In comparison with G1, the large amount of strain in the stiffeners of
pated by the base plate can be given as
G3 indicates that the parts of stiffeners near the boot plates have
undergone numerous plastic strain (see Fig. 7b). Moreover, the X
δW I;base ¼ mp li jθi j ð2Þ
measured strain along the longitudinal direction of stiffeners
(No. 6) is higher than that in the other two directions (No. 4 and 5).
where mp, li, |θi| are plastic moment per unit length, length of yield-line i
in the base plate, absolute value of virtual rotation angle respectively. By
2.3. Test results and observations for flexible EBBP connections solving T from the following equation

The load-centre displacement curves for flexible EBBP connections
are plotted in Fig. 10. The experimental ultimate strength Tu of flexible
connections is listed in Table 3, and only specimens R2, R3, R8, and R9
were tested to failure. The following observations were obtained:
1) Marginal deformation in both boot plates and edge stiffeners was
observed (see Fig. 6), whereas numerous vertical displacement
was observed in the base plate zones around bolt holes.
2) The measured strain of strain gauge group designated No. 9 is over
2500 εμ, while that of the rest groups is less than 1500 εμ (see
Fig. 8). This implies that the zones between bolt holes and boot
plates have approached the yield condition.
3. Yield-line analysis and work equations
δW E ¼ δW I ⇒Tδ=4 ¼ δW I;base þ δW I;s þ δW I;b ð3Þ
where δWI,s and δWI,b, denoting the internal virtual work dissipated by
stiffeners and anchor bolts respectively, are added to derive the whole
internal virtual work, an upper limit for the uplift resistance of EBBP
connections can be achieved.
4. Models for rigid EBBP connections with stiffener deformation
4.1. Assumptions
The two theoretical models proposed in this section are applicable to
such a connection that incorporates flexible stiffeners. The assumptions
adopted to the analysis of these models are as follows:

A quarter of the EBBP connection is modelled due to symmetry con- 1) The in-plane plastic deformation of the stiffeners occurs solely
sideration. The major component in yield-line analysis is the assumed around the interfaces between boot plates and stiffeners, which is

Table 3
Results of rigid connections.

Specimen Actual strength Ref. [1] Predicted ultimate load (kN) TR,1/Tu Tpred/Tu TR,2/Tu
Tu(kN) TR,1(kN)
Model 1 Model 2 Model 3 Model 4 Tpred
(TR,2)

G1 N3210 1701.2 3699.1 3360.6 2928.2 3063.5 2928.2 0.53 b0.91 b0.95
G2 3200 1705.2 2979.6 3091.6 3048.8 3184.1 2979.6 0.53 0.93 1.00
G3 2850 2401.0 2500.5 3345.8 4039.7 4352.1 2500.5 0.84 0.88 1.53
G4 N3650 3236.2 5167.7 5449.1 5559.8 5890.6 5167.7 0.89 b1.42 b1.61
G5 N3500 2433.0 4438.3 4393.8 4247.3 4441.9 4247.3 0.70 b1.21 b1.27
G6 N3720 4048.3 5472.2 6252.6 6847.5 7314.2 5472.2 1.09 b1.47 b1.97
Mean for G2 and G3 0.67 0.91 1.27
Mean for all b0.76 b1.14 b1.39

The bold data represents the predicted critical model.

74 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81
Fig. 5. Ultimate deformation of specimen G3.
Fig. 6. Ultimate deformation of specimen R8.
simplified to a triangular deformation zone HAH′ as illustrated in
Fig. 11.
2) The out-plane plastic deformation of the stiffeners is represented as
the rotation of the trapezoid ADGI about yield-lines AE and ED and of
the triangular AIH about yield-lines AH and AI.
3) The boot plates are of sufficient stiffness so that virtual displacement
of their end connected to the base plate is of a constant value δ.
4) As illustrated in Fig. 11, the virtual displacement of corner D of the
stiffener is zero, while that of corner A equals δ.
5) For simplification of the analysis, the influence of the thickness of
stiffeners, boot plates and fillet welds is ignored when calculating
(a) Measured strain on baseplate for G2 and G3
Fig. 7. Measured strain for rigid EBBP connections.
Fig. 8. Measured strain for flexible EBBP connection (R7).
the internal work dissipated by the base plate δWI,base (see Fig. 12).
This assumption is suitable for all the models presented in Sections
4 and 5.
4.2. Analysis of Model 1
This model is applicable to such a connection that incorporates flex-
ible anchor bolts and flexible stiffeners. The base plate, stiffeners and an-
chor bolts will simultaneously undergo plastic deformation when the
connection attains its ultimate strength.
4.2.1. Stiffeners
According to assumption 1 in Section 4.1, the stiffener undergoes
both in-plane deformation and out-plane torsion. For in-plane deforma-
tion, the stiffener is assumed to be a quasi-mechanism [13], in which a
yielding tension zone develops between lines AH and JK throughout
the depth. The tensile zone AJKH, which is initially a rectangle, un-
dergoes in-plane deformation and becomes a trapezoid (AJKH′). There-
fore, the internal virtual work dissipated by the in-plane deformation of
stiffeners (for one quarter of the rigid EBBP connection) is
Zhs 2
f y;s hs t s δ
δW in
I;s ¼ 2 f y;s t s lθdl ¼ ð4Þ
4S
0
where fy,s, hs, ts are the yield strength, depth and thickness of stiffeners, S
is the distance between the bolt hole centre and the centre line of the
adjacent boot plate, the in-plane rotation angle of the whole stiffener
θ = δ/(4S).
The out-plane deformation of the stiffener is represented herein as
the rotation of parts AIH and ADGI. Subsequently, by summing over
(b) Measured strain on stiffeners for G1 and G3
 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81 75

4000
work dissipated by the out-plane deformation of stiffeners can be re-
3500
written as

3000 δW out
I;s ¼ 0 ð6Þ

2500
4.2.2. Base plate
Load/kN

2000
As illustrated in Fig. 12, the rotation angles of the yield-lines of the
base plate are expressed by
1500
G1 pffiffiffi pffiffiffi
δ 2δ 2δ
G2
jθAB j ¼ jθA0 B0 j ¼ ; jθAO j ¼ jθA0 O0 j ¼ jθD0 O j ¼ jθDO0 j ¼ ; jθAA0 j ¼
1000 G3 2S 4S 4S
G4
G5
ð7a–cÞ
500 G6
and the plastic moment of each yield-line in the base plate is expressed
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 by
Displacement/mm

1
Fig. 9. Load-deformation curves of rigid EBBP specimens. mp ¼ f t2 ð8Þ
4 y

Substituting Eqs. (7a–c) and (8) and the lengths of yield-lines into
the yield lines AH, AI, AE and ED, the internal virtual work dissipated by Eq. (2), the internal virtual work dissipated by the base plate is obtained
the out-plane deformation of stiffeners is given as as
X
δW out
s
msp li j θsi j; ði ¼ AH; AI; AE; EDÞ pffiffiffi !
I;s ¼ 2 ð5Þ 5 2d
δW I;base ¼ − f y t2 δ ð9Þ
where the superscript s means the parameters are taken in the 4 8S
stiffeners. Based on the method to determine the reduced plastic-
moment capacity of an inclined yield-line under axial force [14], the
plastic moment for yield-lines AH and AI, which are in the yield tensile 4.2.3. Anchor bolts
zone, is equal to zero and the rotation angles of yield-lines AE and of As illustrated in Fig. 13, the anchor bolt undergoes both flexural
ED are zero based on the kinematic properties. Thus, the internal virtual and tensile deformation induced by the rigid displacement of the

3500 3500

3000 3000

2500 2500

2000 2000
Load/kN

Load/kN

1500 1500

1000 R1 1000 R6
R2 R7
R3
R8
500 R4 500 R9
R5

0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 1 2 3 4 5 6 7
Displacement/mm Displacement/mm

(a) (b)

Fig. 10. Load-deformation curves of flexible EBBP specimens: (a)R1–R5 and (b) R6–R9.

Fig. 11. Plastic mechanism of the stiffener.

76 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81

Fig. 12. Plastic mechanism of Model 1.

base plate. Consequently, the internal work dissipated by the anchor an approximate linear relation curve is proposed to represent the theo-
bolts is given as retical one as follows:
  !
Tbδ M δ Tb f y;b d
3
Tb
δW I;b ¼ 2 þ pbffiffiffi ð10Þ cos3 α ¼ 1− ⇒M b ¼ 1− ð13Þ
2 2 2S Ab f y;b 6 Ab f y;b

where Tb, Mb are the tension and bending moment of each anchor
bolt. As the prying forces are neglected, the tension of each anchor Substituting Eqs. (11) and (13) into Eq. (10), WI,b can be derived as
bolt can be expressed by 3
!
T f y;b d Td
W I;b ¼ þ pffiffiffi − pffiffiffi δ ð14Þ
T b ¼ T=8 ð11Þ 8 6 2S 12 2πS

Based on the assumed stress distribution of anchor bolt illustrated in Finally, by substituting Eqs. (4), (6), (9) and (14) into Eq. (3), the ul-
Fig. 13(a), Tb and Mb can be written as. timate strength T for Model 1 is derived as

3 " pffiffiffi #, pffiffiffi !

Tb ¼
α þ sin α cos α
Ab f y;b ; Mb ¼
f y;b d
cos3 α ð12Þ pffiffiffi d 2
2 f y;s hs t s 2 2 f y;b d
3
2d
π=2 6 T¼ 10− 2 f y t2 þ þ 1þ ð15Þ
S S 3 S 3π S

where α is the geometric parameter shown in Fig. 13(a), the sectional By defining a set of dimensionless parameters in terms of the geo-
area of anchor bolts Ab = πd2/4 (d = anchor bolt diameter), fy,b = bolt metrical variables as follows:
yield stress. Combining the equations above, the theoretical relation
curve between the dimensionless factors cos3α and Tb/(fy,bAb) can be hs ts d S
η¼ ;ξ ¼ ;γ ¼ ;λ ¼ ð16a–dÞ
calculated, and it is plotted in Fig. 13(b). For convenience of calculation, S t S t

Fig. 13. Formulation of the strength for anchor bolt: (a) stress distribution of bolt shank and (b) moment-tension relation curves.
 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81 77

Eq. (19a,b) can be rewritten in terms of η, ξ, γ, λ as Substituting Eq. (22a–c) into Eq. (5), the internal work dissipated by
" pffiffiffi #, pffiffiffi ! the out-plane deformation of stiffeners is obtained as
T  pffiffiffi  f y;s 2 2 2 f y;b 2 3 2
¼ 10− 2γ þ 2 λη ξ þ λ γ 1þ γ ð17Þ 1
f y t2 fy 3 fy 3π δW out f t s 2 δ;
I;s ¼ ð23Þ
2 y;s

Then the whole internal virtual work dissipated by stiffeners is given

4.3. Analysis of Model 2
by
This model is applicable to such a connection that incorporates rigid 2
anchor bolts and flexible stiffeners. Plastic deformation will occurs f y;s hs t s δ 1
δW I;s ¼ þ f y;s t s 2 δ: ð24Þ
solely in base plates and stiffeners when the connection attains its ulti- 4S 2
mate strength.
Substituting Eqs. (21) and (24) into Eq. (3), the non-dimensional ul-
4.3.1. Base plate timate strength for Model 2 is derived as
As illustrated in Fig. 14, according to assumptions 1 and 4 in
Section 4.1, the interface lines AD and A′D′ between base plates and stiff- T f y;s 2 f y;s 2
¼ 17:366−3:311γ þ 2 ξ þ λη ξ: ð25Þ
eners remain straight when the connection attains its ultimate strength; f y t2 fy fy
thus, the virtual displacement of point E equals δ/2. Similarly, the rota-
tion angle of each yield-line in the base plate can be derived as
pffiffiffi 5. Theoretical models for rigid connection without
2δ δ stiffener deformation
jθAA0 j ¼ ; jθAE j ¼ jθA0 E0 j ¼ pffiffiffi ð18a; bÞ
4S 2S
pffiffiffiffiffiffi Two theoretical models are proposed in this section, which is appli-
34δ δ cable to such a connection that incorporates rigid stiffeners. Thus, the
jθAO j ¼ jθA0O0 j ¼ jθAB j ¼ jθA0B0 j ¼ jθCO j ¼ jθC 0 O0 j ¼ ð19a; bÞ
4S S virtual displacement of interface lines AD and A′D′ is assumed to be con-
pffiffiffi pffiffiffi stantly δ.
δ 2δ 2δ
jθEO j ¼ jθE0O0 j ¼ ; jθD0O j ¼ jθDO0 j ¼ ; jθEF j ¼ ð20a–cÞ
2S 4S 2S 5.1. Analysis of Model 3

Substituting Eqs. (8), (18a,b)–(20a–c) and yield-line lengths li into

As illustrated in Fig. 15, this model is applicable when the stiffeners
Eq. (2), the internal work dissipated by the base plate for Model 2 is ob-
are thin but of large depth. The rotation angle of each yield-line in
tained as
the base plate connecting the corresponding parts can be calculated as
" pffiffiffi pffiffiffiffiffiffi! # follows:
 pffiffiffi pffiffiffiffiffiffi 3 2 þ 34 d
δW I;base ¼ 9 þ 3 2 þ 17 − þ mp δ ð21Þ pffiffiffi
2 4 S
δ 2δ
jθAE j ¼ jθA0 E0 j ¼ jθAB j ¼ jθA0 B0 j ¼ ; jθAO j ¼ jθA0 O0 j ¼ jθEO j ¼ jθE0 O0 j ¼
S S
ð26a; bÞ
4.3.2. Stiffeners
The internal work dissipated by the in-plane deformation of stiff- pffiffiffi
2δ 2δ
eners for Model 2 can be obtained from Eq. (4) through the same proce- jθCO j ¼ jθC 0 O0 j ¼ ; jθEF j ¼ ð27a; bÞ
S S
dure as for Model 1, except that the rotation angle of yield-line ED in
Fig. 11, which connects the stiffener to the base plate, varies from that Substituting Eqs. (26a,b) and (27a,b) and yield-line lengths li into
in Model 1 owing to the different rigid displacement of base plate Eq. (2), the internal virtual work dissipated by the base plate for
parts OD′FE and O′DFE. The rotation angles and plastic moment of Model 3 is obtained as follows:
yield-lines AE and ED in the stiffener for Model 2 are given by
  pffiffiffi d
 s   
θ  ¼ 0; θs  ¼ δ ; ms ¼ 1 f t s 2 ð22a–cÞ δW I;base ¼ 22− 2 þ 2 2 mp δ ð28Þ
AE ED
2S p 4 y;s S

Fig. 14. Plastic mechanism of Model 2.

78 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81

Fig. 15. Plastic mechanism of Model 3.

The rotation angle of the yield-line ED (ED′) connecting the 5.2. Analysis of Model 4
base plate part CD′FEO (C′DFEO′) and the corresponding stiffener is
expressed by For Model 4, yield-lines are distributed solely in the base plate as il-
lustrated in Fig. 16. This model is applicable when the stiffeners are thick
 s   s  δ but of small depth. The rotation angles of yield-lines for Model 4 are
θ  ¼ θ  ¼ ð29Þ identical to those for Model 3, except that yield-lines DE and D'E appear
ED ED0
S
in the base plate instead of the stiffeners. Substituting Eqs. (8), (26a,b),
(27a,b), (29) and yield-line lengths li into Eq. (2), the internal virtual
Substituting Eqs. (22a–cc), (29) and corresponding yield-line work dissipated by the base plate is obtained as follows:
lengths into Eq. (5), the internal virtual work dissipated by stiffeners 
is obtained as follows:
 pffiffiffi d 1

d

δW I;base ¼ 24− 2 þ 2 2 mp δ≐ f y t 2 δ 24−4:828 : ð32Þ
S 4 S

δW out 2
I;s ¼ f y;s t s δ ð30Þ Subsequently, by combining with Eq. (3), the non-dimensional ulti-
mate strength for Model 4 can be derived as follows:
According to Eqs. (3), (28) and (30), the non-dimensional ultimate T
strength for Model 3 is derived as ¼ 24−4:828γ: ð33Þ
f y t2

T f y;s 2
¼ 22−4:828γ þ 4 ξ ð31Þ 6. Theoretical models of flexible EBBPs
f y t2 fy
From the experimental observations, it is revealed that both the boot
plates and stiffeners of flexible EBBP connections did not undergo in-
plane deformation. Thus, the vertical displacement of the interfaces be-
tween boot plates and base plate and that between edge stiffeners and
base plates is assumed to be constantly δ. Besides, for simplification of
analysis, the influence of the thickness of edge stiffeners and boot plates
and fillet welds is neglected, and the centre lines of the edge stiffeners
are assumed to be localized at the edges of the base plate (see Fig. 17).
The rotation angle of each yield-line connecting the corresponding
base plate parts and yield-line lengths li is derived as follows:
pffiffiffi
2δ δ
jθAA0 j ¼ jθOO0 j ¼ l; jθAB j ¼ jθA0 B0 j ¼ ; jθAO j ¼ jθA0 O0 j
2S þ S1−a S
LAO δ
¼ ; ð34a–cÞ
Sð2S þ S1−aÞ

LBO δ LAO δ
jθBO j ¼ jθB0 O0 j ¼ ; jθAO j ¼ jθA0 O0 j ¼ ; jθBO j
LS Sð2S þ S1−aÞ
LBO δ
¼ jθB0 O0 j ¼ ; ð35a–cÞ
LS

δ δ
jθBD j ¼ jθB0 D0 j ¼ jθEO j ¼ jθE0 O0 j ¼ ; jθED j ¼ jθED0 j ¼ ; ð36a; bÞ
L cos β S
pffiffiffi pffiffiffi
Fig. 16. Plastic mechanism of Model 4. lAA0 ¼ 2a; lOO0 ¼ 2S1−d; ð37a; bÞ
 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81 79

Substituting Eqs. (34a–c)–(39a,b) into Eq. (2) and according to

Eq. (3), the non-dimensional ultimate strength for flexible EBBP con-
nections is obtained as follows:
pffiffiffi
T a þ S1−d= 2 LAO 2 −LAO d=2 S þ S1 þ L a
¼ 2½ þ þ − − tanβþ
f y t2 2S þ S1−a Sð2S þ S1−aÞ S S
2   ð41Þ
LBO −LBO d=2 S S1 þ L
þ þ −d=2 =ðL cosβ Þ
LS L cos2 β cosβ þ sinβ

where a and β are geometric parameters as illustrated in Fig. 17. Com-

bining Eqs. (40a,b) and (41), the ultimate load T is presented as a func-
tion of the parameters a and β. According to yield-line theory, these two
parameters are obtained by minimizing the feasible value of the yield
load in the plastic model considered. Therefore, the parameters a and
β can be obtained by the following minimum conditions:

∂T ∂T
¼ 0 and ¼0 ð42a; bÞ
∂a ∂β

A closed-form solution could not be obtained for Eq. (41); therefore,

the resource is made to numerical integration.

7. Results of analysis
Fig. 17. Plastic mechanism of flexible EBBP.
7.1. Comparison for rigid connections
lAO ¼ lA0 O0 ¼ dAO −d=2; lAB ¼ lA0 B0 ¼ S þ S1 þ L−a−S tan β; ð38a; bÞ
Table 3 presents the measured and predicted strengths for rigid
S1 þ L EBBP connections. In addition, the results calculated by models recom-
lBO ¼ lB0 O0 ¼ dBO −d=2; lEO ¼ lE0 O0 ¼ −d=2; ð39a; bÞ
sin α þ cosα mended in Ref. [1] are also included. As these design models have as-
sumed factored resistance value for properties, all resistance or safety
where dAO and dBO are the distances from bolt hole centres to points A factors have been removed for this evaluation.
and B, respectively, whence. The yield-line analysis demonstrates reasonable correlation be-
tween the experimental and predicted results. The mean of predicted-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi to-test ratios for the specimens tested to failure (G2 and G3) was 0.91
dAO ¼ ðS þ S1−aÞ2 þ S2 ; dBO ¼ ðL−S tan βÞ2 þ S2 ð40a; bÞ and for all lower than 1.14. The predicted critical plastic models (bolded

Fig. 18. Non-dimensional ultimate strength of rigid EBBP connections.

80 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81

Table 4
Results of flexible connections.

Specimen Actual strength Ref. [1] Predicted ultimate load

Tu/kN TR/kN a β Tpred/kN TR/Tu Tpred/Tu

R1 N3350 1282.9 119.9 0.62 3870.7 0.38 b1.16

R2 2850 817.2 163.9 0.59 2849.5 0.29 1.00
R3 2800 682.7 163.9 0.59 2361.5 0.24 0.84
R4 N3020 1986.1 145.9 0.61 6038.8 0.66 b2.00
R5 N3080 1565.9 145.2 0.61 4902.0 0.51 b1.59
R6 N3080 1270.1 145.9 0.61 3861.6 0.41 b1.25
R7 N3300 2636.1 145.9 0.61 8015.1 0.80 b2.43
R8 2880 1234.6 171.3 0.60 3906.1 0.43 1.36
R9 3440 1551.4 171.3 0.60 4884.4 0.45 1.42
Mean for R2, R3, R8, and R9 0.35 1.15
Mean for all b0.46 b1.45

Fig. 19. Non-dimensional ultimate strength of flexible EBBP connections.
in Table 3) match the actual failure models of both G2 and G3. Compar-
ing the measured and predicted results of G1 and G3, the significant ef-
fect of stiffener dimensions is satisfactorily predicted: the increment
rate of the measured strength of G1 (with a thinner base plate but sub-
stantially stronger stiffeners, as presented in Table 1) to G3 is higher
than 12.6%, while that of the predicted strength is equal to 17%. The
major reasons for the lower theoretically estimated values of specimens
G1, G2, and G3 are as follows:
1) The yield-line analysis provides an estimate of the yield strength Ty,
which is defined as the intersection of the initial connection stiffness
and strain hardening stiffness [7]. Thus, the strain hardening action
is not considered in this analysis.
2) The membrane action is neglected, particularly when calculating the
internal virtual work dissipated by the out-plane deformation of
stiffeners. Actually, the tension zones in the stiffeners underwent a
large amount of membrane action when the connection attains its
ultimate strength.
3) The thickness of stiffeners, boot plates and fillet welds does influence
the yield-line distribution of the base plate, which is neglected ac-
cording to assumption 5 in Section 4.1.
The results TR,1, TR,2 are calculated by the two models (see Appendix)
recommended in Ref. [1], respectively. The mean of TR,1 to Tu ratios for
the specimens tested to failure (G2 and G3) was 0.67 and for all lower
than 0.76; whereas the mean of TR,2 to Tu ratios for the specimens tested
to failure was 1.27 and for all lower than 1.39. In comparison with these
design models, it is implied that the theoretical method proposed herein
provides a more accurate prediction for rigid connections.
of this design model is attributed to the fact that the whole base plate
is merely simplified to several independent cantilevered plates.
7.3. Parametric analysis of models
Using the theoretical models 1–4, the non-dimensional ultimate
strengths of rigid EBBP connections with various values of parameters
η, ξ and γ are calculated and plotted in Fig. 18a–c. The ultimate strength
was determined as the lowest of results of the four presented models.
From the results depicted in Fig. 18a and b, the following observations
can be made. The parameters η and ξ, which are corresponding to the
depth and thickness of stiffeners respectively, are important parameters
for determining the ultimate strength of rigid EBBP connections. The
critical model, for calculating the final result, depends on the value of
γ when the stiffeners are flexible: the ultimate strength is determined
by Model 1 when γ = 0.4, while Model 2 is the critical model when
γ = 0.6. The presence of the stiffeners increases the strength of rigid
connections by an increment ratio ranging for 22.1% to 66.8%, and
when the stiffness of stiffeners approaches an appropriate value, the
strength will not increase. As illustrated in Fig. 15c, when η = 0.8 or
1.4, the curves consist of two segments, which were obtained from
Models 1 and 3, respectively; while η = 2.0, Model 3 is the sole critical
model.
According to Eq. (41), the non-dimensional ultimate strength of
flexible EBBP connections with various values of parameters S1/S and
L/S has been calculated and plotted in Fig. 19. The parameter S1/S is
the most critical parameter for determining the ultimate strength. For
each group of flexible EBBP connections with identical L/S value, it ap-
pears that the increasing ultimate strength is approximately linear.
8. Conclusions

7.2. Comparison for flexible connections
The measured and predicted strength of flexible EBBP connections is
presented in Table 4, including the results TR calculated by Eq. (A2). The
mean value of predicted-to-test ratios for the specimens tested to fail-
ure (R2, R3, R8, R9) was 1.15 and for all lower than 1.45. Moreover,
Eq. (41) provides a relatively precise prediction for the specimens
with thinner base plates (R2 and R3), whereas for those with thicker
base plate, it overestimates the ultimate strength. The main reason for
the theoretical overestimation of connections with thicker base plate
is that the edge stiffeners are assumed to be of sufficient stiffness,
which is not realistic due to the large stiffness of the base plate com-
pared with the edge stiffener. Since the variation of the dimensions of
edge stiffeners is not considered when designing flexible EBBP speci-
mens (see Table 2), the influence of edge stiffeners was not observed
in the test.
The mean value of TR to Tu ratios for the specimens tested to failure
was 0.35 and for all lower than 0.46. The substantial underestimation
Four theoretical models based on yield-line theory have been pre-
sented for predicting the strength of rigid EBBP connections under uplift
loading. With these models, effects of stiffeners and anchor bolts can be
considered, by incorporating the contribution of their plastic deforma-
tion. The reliability of these models have been demonstrated by exper-
imental results. Compared with the results of design models in the
corresponding code, the present models appear to be more accurate.
From the parametric analysis, it is implied that the presence of the stiff-
eners can efficiently increase the strength of the rigid connection and
when the stiffeners are of sufficient stiffness, this increasing effect
disappears.
The presented model for flexible EBBP connections has also been
verified by experimental results, it appears that this model gives a
more accurate prediction for flexible connections with thinner base
plate. More accuracy has been demonstrated in comparison with results
from the corresponding code as well. According to the parametric anal-
ysis, the strength of flexible connections can be improved by optimize
the location of anchor bolts. Since the behaviour of connections with
 H. Liu et al. / Journal of Constructional Steel Research 145 (2018) 70–81 81

flexible edge stiffeners (or thick base plate) is not included herein, fur-
ther investigations need to be conducted on that topic.

Acknowledgements

The research reported herein was sponsored by the Natural Science

Foundation of China (No. 51508054) and Chongqing Basic and Frontier
Research Project (cstc2015jcyjA00041). The writers would like to pay
their gratitude to all the sponsors for their financial aid.

Appendix A. Design models for EBBP in ref. [1]

A.1. Rigid EBBP connections

The fist design model for rigid EBBP connections assumes that the
tension Tb of each anchor bolt is withstood by the three shadow parts
1−3 identically (see Fig. A1), each of which is assumed to be an inde-
pendent cantilevered plate fixed to the stiffener or boot plate end. The
design thickness of the base plate is obtained from the maximum
value calculated from each of the three parts. In the case herein, by re-
moving the resistance and safety factors and using the plastic moment
mp = fyt2/4, we obtain
Fig. A2. Design model for flexible connections in [1].

1 6D
T Y ¼ mp D⇒T ¼ 8T b ¼ f t2 ðA1Þ
3 b Y y
References
where Y = the distance between the bolt hole centre and stiffener or [1] DT/T5154-2012. Technical Code for Design of Tower and Pole Structures of
boot plate, D = the width of cantilevered part for calculation. Overhead Transmission Line, China Planning Press, Beijing, 2012 (In Chinese).
The second model assumes that each rectangular zone surrounded [2] B. Kato, A. Mukai, Bolted Tension Flanges Joining Square Hollow Section Members,
CIDECT Rep. Univ. of Tokyo, Tokyo, 1982.
by stiffeners and boot plates with bolt hole is in such a boundary condi-
[3] S. Willibald, J.A. Packer, R.S. Puthli, Design recommendations for bolted rectangular
tion that with three edges clamped and one edge free, which is identical HSS flange-plate connections in axial tension, Eng. J. 40 (2003) 15–24.
to Model 4 in Section 4. Thus, the results by this design model are iden- [4] S. Willibald, J.A. Packer, R.S. Puthli, Experimental study of bolted HSS flange-plate
connections in axial tension, J. Struct. Eng. 128 (2002) 328–336.
tical to those by Model 4.
[5] A.P. Jeffrey, B. Luigi, C.B. Peter, Limit analysis of bolted RHS flange plate joints,
J. Struct. Eng. 115 (1989) 2226–2242.
A.2. Flexible EBBP connections [6] A.T. Wheeler, M.J. Clarke, G.J. Hancock, FE modeling of four-bolt, tubular moment
end-plate connections, J. Struct. Eng. 126 (2000) 816–822.
[7] A.T. Wheeler, M.J. Clarke, G.J. Hancock, T.M. Murray, Design model for bolted mo-
The design model for flexible EBBP connections assumes that the ment end plate connections joining rectangular hollow sections, J. Struct. Eng. 124
tension Tb is transmitted solely to the shadow zone shown in Fig. A2. (1998) 164–173.
Thus, using the similar procedures above, we obtain [8] Y.Q. Wang, L. Zong, Y.J. Shi, Bending behavior and design model of bolted flange-
plate connection, J. Constr. Steel Res. 84 (2013) 1–16.
[9] K.K. Hon, R.E. Melchers, Experimental behaviour of steel column bases, J. Constr.
2D
T b Y ¼ mp D⇒T ¼ 8T b ¼ f t2 ðA2Þ Steel Res. 9 (1988) 35–50.
Y y [10] F. Wald, Z. Sokol, J.P. Jaspart, Base plate in bending and anchor bolts in tension,
Heron 53 (2008) 21–50.
[11] F. Wald, V. Bouguin, Z. Sokol, J.P. Muzeau, Effective length of T-stub of RHS column
base plates, ROANOKE (2000) 393–402.
[12] High Strength Bolts With Large Hexagon Head for Steel Structures, China Standard
Press, Beijing, 2006.
[13] N.W. Murray, P.S. Khoo, Some basic plastic mechanisms in the local buckling of thin-
walled steel structures, Int. J. Mech. Sci. 23 (1981) 703–713.
[14] X.-L. Zhao, G.J. Hancock, A theoretical analysis of the plastic-moment capacity of an
inclined yield line under axial force, Thin-Walled Struct. 15 (1993) 185–207.
[15] Steel and Steel Products-location and Preparation of Test Pieces for Mechanical Testing,
China Standard Press, Beijing, 1998 (In Chinese).

Fig. A1. Design model for rigid connections in [1].