Engineering Structures 68 (2014) 14–23

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Rotational behaviour of column base plate connections: Experimental

analysis and modelling
M. Latour ⇑, V. Piluso, G. Rizzano
University of Salerno, Civil Engineering Department, Salerno, Italy

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

Article history: A research program aimed at the development of a mechanical model to predict the rotational behaviour
Received 19 March 2013 of base plate connections under cyclic loads has been recently undertaken at Salerno University. The eval-
Accepted 27 February 2014 uation of the accuracy of the component approach for predicting the rotational behaviour of base plate
Available online 24 March 2014
connections up to failure under monotonic loads has been carried out as first step of the planned research
activity. The accuracy in predicting the rotational stiffness, the flexural resistance and the overall
Keywords: moment–rotation curve of base plate connections by the component method given in Eurocode 3 is eval-
Experimental analysis
uated with reference to the experimental tests recently carried out at Salerno University. In particular,
Eurocode 3
Base plate connections
the specimens have been preliminarily tested in elastic range under different values of the column axial
Steel joints load and, successively, loaded up to collapse.
Modelling Such experimental tests are presented and discussed in this paper. In order to widen the investigated
Component method range of the geometrical and mechanical parameters affecting the behaviour of base plate connections,
Stiffness the comparison between the predictions coming from component method and experimental tests has
Resistance been extended not only to the authors’ own tests, but also to a significant number of tests collected from
the technical literature.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction
In last years, following the introduction in Eurocode 3 [5] of de-
tailed procedures for predicting the joint rotational behaviour, the
attention of researchers towards advanced modelling of steel
framed structures has gradually increased. In particular, with refer-
ence to seismic design issues, more and more interest has been de-
voted to the study of the cyclic rotational behaviour of steel
members and connections with the aim to develop reliable models
to be successfully used in the non-linear analysis of steel framed
structures subjected to earthquakes [2,18,15,19–21]. In addition,
significant research activity has been also carried out aiming to
the improvement of structural models for non-linear dynamic
analyses by reducing the epistemic uncertainties due to the predic-
tion of the cyclic behaviour of dissipative zones [19,20,8].
It is well known that accurate seismic non-linear analyses re-
quire an appropriate modelling of the hysteretic cyclic behaviour
of steel members and connections. These models must be able to
grasp with satisfactory accuracy the main phenomena characteris-
⇑ Corresponding author. Tel.: +39 089964342.
E-mail addresses: mlatour@unisa.it (M. Latour), v.piluso@unisa.it (V. Piluso),
g.rizzano@unisa.it (G. Rizzano).
ing the cyclic response under seismic actions, such as strength and
stiffness degradation and pinching phenomena.
Regarding connections, the modelling of the rotational behav-
iour under monotonic loading conditions has been extensively
studied both with reference to beam-to-column connections and
with reference to column–base connections leading to the devel-
opment of the so-called component method codified in Eurocode
3. The main advantage of the component method is its ability to al-
low the prediction of the rotational behaviour of different joint
typologies by means of a unified approach, starting from an appro-
priate identification and modelling of all the sources of strength
and deformability, i.e. all the components. Even though the method
has been codified dealing mainly with the prediction of the stiff-
ness and strength of the components starting from the values of
their geometrical and mechanical properties, the possibility of pre-
dicting also the component ductility has been recently pointed out
[25,12]. Therefore, it can be stated that the modelling of the whole
moment–rotation curve up to collapse can also be obtained, pro-
vided that rules for predicting the ultimate deformation of the joint
components are available, at least with reference to the weakest
joint component which governs the plastic rotation capacity of
connections.
Regarding the prediction of the cyclic behaviour, additional
studies and experimental tests are still needed in order to develop

http://dx.doi.org/10.1016/j.engstruct.2014.02.037
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
 M. Latour et al. / Engineering Structures 68 (2014) 14–23
accurate and reliable models. In particular, with reference to beam-
to-column connections, some works have been recently developed
aiming to the extension of the component method to the predic-
tion of the cyclic behaviour [23,19,20]. These studies have shown
that a reliable modelling of the cyclic behaviour of connections,
accounting for strength and stiffness degradation and pinching ef-
fects, is possible, provided that an accurate modelling of the exten-
sional (force versus displacement) behaviour of each joint
component is available.
It is evident that an accurate modelling of steel frames sub-
jected to seismic actions requires not only a reliable modelling of
the cyclic behaviour of beam-to-column connections, but also an
accurate modelling of the cyclic behaviour of column–base connec-
tions. Unfortunately, with reference to the cyclic behaviour of col-
umn base plate connections, predicting models based on the
component approach are nowadays not available in the technical
literature. Therefore, within this framework, the main purpose of
the research activity planned at Salerno University is the investiga-
tion and modelling of the cyclic behaviour of the base plate con-
nections. In particular, the whole research program on this topic
requires the development of the following steps:
 Step 1: Theoretical and experimental study of the monotonic
ultimate behaviour of base plate connections aiming to the
evaluation of the degree of accuracy of Eurocode 3 approach
for predicting the rotational stiffness and the flexural strength
of column–base plate connections.
 Step 2: Development of a method for predicting the plastic
rotation capacity of base plate connections starting from the
knowledge of the ultimate deformation of components [22].
 Step 3: Definition of new criteria for designing full-strength base
plate connections with adequate overstrength to account for
material strain hardening and uncertainties deriving from ran-
dom material variability of the connected column.
 Step 4: Modelling of the cyclic behaviour of base plate connec-
tions by the component approach and definition of practical
rules for designing dissipative column base plate connections.
In this paper, the results gained during the first step of the
planned research activity are presented and discussed. In particu-
lar, the accuracy of the component method suggested by Eurocode
3 for predicting the rotational stiffness and flexural strength of col-
umn–base connections is analysed starting from the experimental
test results of Salerno University and successively with reference to
those coming from a wide collection of tests available in the liter-
ature. Even though in the present work reference is made to the
European practice, it is important to underline that the new exper-
imental results herein presented can be also useful for further
investigations dealing with the accuracy of alternative analytical
models proposed in technical literature which are however out of
the scope of this work.
2. Prediction of the moment–rotation curve
The prediction of the rotational behaviour of base plate joint is
even more complex than that of beam-to-column connections due
to several factors, such as the bond between the anchor bars and
the concrete base material, the behaviour of the bedding grout
and the complex evaluation of the stiffness and resistance of the
concrete below the compressed flange, due to the contact phenom-
ena and the resulting stress distribution between the base plate
and the concrete. Moreover, the rotational response is controlled
by interactions between these components. Previous research has
sought to characterise these interactions and their effects on key
aspects of connection rotational response. Experimental studies
15
[6,29,3,1] have led to the development of methods for calculating
the flexural strength of these connections [7,13,9].
In addition, also the loading process plays an important role in
the behaviour of base plate joints. In the technical literature, two
different types of loading process are usually adopted: the first
one is characterised by a proportional loading process where the
vertical load and the bending moment are proportionally in-
creased, so that the eccentricity is constant during the loading pro-
cess; conversely, the second one is characterised by a load process
where the vertical load is applied in a first loading phase and, sub-
sequently, the application of an increasing bending moment is car-
ried out up to failure. In this case, the eccentricity is continuously
increasing giving rise to an additional non-linearity source in the
connection behaviour.
The application of the component method requires the develop-
ment of three steps:
 Identification of the deformability and strength of components.
 Mechanical characterisation of each component.
 Assembly of the mechanical model of the whole connection.
Several studies have been also carried out to develop proce-
dures to represent interactions between the various components
and for evaluating the rotational stiffness of column–base connec-
tions [30,31,16,27,6,17].
With reference to the procedure proposed by Eurocode 3, the
sources of deformability and resistance for a usual base plate joint
of ‘‘exposed’’ typology are: the concrete in compression, the base
plate in bending, the anchor bars in tension and the column flange
and web in compression. This latter component is involved only in
the resistance evaluation and does not provide a contribution to
the connection deformability.
In case of column–base connections subjected to centred axial
load, Eurocode 3 suggests to consider three compressed T-stubs lo-
cated according to the column flanges and web (Fig. 1a). Con-
versely, in case of column–base connections subjected to axial
load and bending moment, a simplified approach is proposed by
neglecting the contribution due to the web T-stub. Starting from
the translational and rotational equilibrium conditions, the con-
nection flexural resistance Mj,Rd can be expressed as a function of
the load eccentricity e as follows:
 
F cl;Rd z F cr;Rd z
low eccentricity e < zc;r M j;Rd ¼ min ;
zc;r =e  1 zc;l =e þ 1
ð1Þ
 
F tl;Rd z F cr;Rd z
high eccentricity e > zc;r M j;Rd ¼ min ;
1  zc;r =e zt;l =e þ 1
ð2Þ
where Fcl,Rd is the resistance in compression of the left T-stub, Fcr,Rd
is the resistance in compression of the right T-stub, Ftl,Rd is the resis-
tance in tension of the left T-stub, zcl, zcr, ztl are the distances from
the column axis of the components depicted in Fig. 2 and z is the
lever arm equal to zcl + zcr, in case of small eccentricity, and equal
to ztl + zcr, in case of high eccentricity.
The resistance of the base plate in tension Ftl,Rd can be modelled
by means of an equivalent T-stub according to the well-known
Eurocode 3 formulations. In particular, in case of base plates allow-
ing the development of prying forces, it corresponds to the mini-
mum value of the resistances corresponding to type 1, 2 and 3
collapse mechanisms; conversely, when the base plate is not stiff
enough, it is equal to the minimum resistance given by type 1
and 3 collapse mechanisms. Therefore, the resistance of the base
plate in tension can be evaluated by means of the following
relationships:
16 M. Latour et al. / Engineering Structures 68 (2014) 14–23

zt,l
(a) zc,l zc,r (b)

Left Right c
T-stub T-stub
Web
T-stub
leff

Concrete c
Footing

c c

beff

Fig. 1. Eurocode 3 model.

e < zc
e > zc
e
e

N
N

z z
zc,l zc,r
zt,l zc,r

Kc,l Kc,r Kt,l Kc,r

Fc,l Fc,r Ft,l Fc,r

Fig. 2. Mechanical model for base plate connections according to Eurocode3.

  
ð8n  2ew ÞMpl;1;Rd M c;Rd
with prying forces F t;Rd ¼ min ; F c;Rd ¼ min fjd beff leff ; ð5Þ
2mn  ew ðm þ nÞ ðhc  t cf Þ
P X 
2M pl;2;Rd þ n F t;Rd
; F t;Rd ð3Þ where fjd is the resistance in compression of foundation joint
mþn
material under concentrated loads accounting for the quality of
 X  the bedding grout and of the confinement effect due to unloaded
2M pl;1;Rd
without prying forces F t;Rd ¼ min ; F t;Rd ð4Þ concrete [4], beff and leff are the effective width and length of the
m
compressed T-stub flange (Fig. 1b), Mc,Rd is the column plastic mo-
where Mpl,1,Rd is the plastic bending moment of the base plate cor- ment accounting for the interaction with shear and axial stresses, hc
responding to type 1 collapse mechanism, Mpl,2,Rd is the plastic is the column section height and tcf is the column flange thickness.
bending moment corresponding to type 2 collapse mechanism, Ft,- Even though the present work is devoted to the component ap-
Rd, is the design resistance in tension of the anchor bars, m is the dis- proach as codified in Eurocode 3 [4], it is useful to note that a differ-
tance between the plastic hinge and the bolt axis, n is the distance ent approach is followed in US practice [11], where it is suggested
between the prying force and the anchor bar axis and ew is the the use of a rectangular stress block to represent the bearing stres-
washer diameter [4]. ses under the base plate. If the load eccentricity is low, the applied
The resistance of the components in compression Fcl,Rd and Fcr,Rd loading is resisted solely through the development of compressive
is given by the minimum value between the crushing resistance of bearing stresses under the base plate. Conversely, if the load eccen-
concrete and the crushing resistance of column flange and web, as tricity is high, the base plate will tend to uplift and loading must be
follows: resisted through a combination of bearing stresses on the
 M. Latour et al. / Engineering Structures 68 (2014) 14–23 17

Fig. 3. Test layout.

Regarding the rotational stiffness, according to Eurocode 3, it is

evaluated by means of the following relations as a function of the
eccentricity (Fig. 2):

Ez2 e zc;r kc;r  zc;l kc;l

low eccentricity e > zc;r Sj ¼   where ek ¼
l 1
þ k1c;r e  ek kc;r þ kc;l
kc;l

ð6Þ

Ez2 e zc;r kc;r  zt;l kt;l

high eccentricity e > zc;r Sj ¼   where ek ¼
l 1
kt;l
þ k1c;r e  ek kc;r þ kt;l

ð7Þ

where kc,l and kc,r are the stiffnesses of the components left and
right in compression, kt,l is the stiffness of the component in tension
resulting from the flexural stiffness of the base plate and the axial
stiffness of the anchor bars, l is a coefficient accounting for the stiff-
ness reduction in plastic range assuming a value greater than 1.0
when the bending moment is greater than 2/3 times the connection
moment resistance. In particular, the stiffness of the components in
tension and compression are given by:
pffiffiffiffiffiffiffiffiffiffiffiffiffi
Ec beff leff
kc ¼ ð8Þ
1:275E
!1
Fig. 4. Testing equipment. m3 Lb
with prying forces kt ¼ þ ð9Þ
0:85leff ;bp t 3bp 1:6Ab
compression side of the connection and tensile forces in the anchor
bars. Under the stress block assumption, vertical equilibrium equa- !1
tion and rotation equilibrium equation are simply exploited to com- m3 Lb
without prying forces kt ¼ þ ð10Þ
pute the bearing length and the tensile forces in the anchor bars. 0:425leff ;bp t3bp 2Ab
These actions are used to compute the bending moment per unit
of width in the base plate, so that the first yield moment of the where Ec is the concrete modulus of elasticity, leff,bp is the effective
connection is evaluated as the one leading to the attainment of base of the T-stub in tension, Ab is the anchor bar section, Lb is the
the capacity of the anchor rods or to the base plate yielding. effective length of the anchor bar equal to the sum of the thickness
18 M. Latour et al. / Engineering Structures 68 (2014) 14–23

LVDT 3

LVDT 6

LVDT 5
LVDT 1
LVDT 4

LVDT 2

Fig. 5. Location of LVDT transducers.

Table 1
Geometrical properties of specimens.

Specimen Bars tbp (mm) B (mm) L (mm) B0 (mm) L0 (mm) e1 (mm) e2 (mm) sf (mm) sw (mm)
HE 240 B-15 6 15 340 500 90 380 80 60 10 8
HE 240 B-25 6 25 340 540 80 440 90 50 10 8
HE 160 A-15 4 15 280 335 160 245 60 45 8 6

t piatto tpiatto
Barre M20
M20 Bars M20 Bars
e1

e1

Barre M20
B0
B0
B

B
B0

sw sw
e1
e1

sf HE 160 A
sf HE 240 B
e2 L0 e2
e2 L0 e2
L
L

Fig. 6. Definition of geometrical properties of specimens.

of half nut, washer, base plate, layer of mortar and eight times the
diameter of the anchor bar itself.
It is useful to note that an alternative approach has been re-
cently proposed [17] under the stress block assumption for bearing
stresses under the compression side of the base plate. However,
such approach herein is not investigated, because it is aimed at
the prediction of the secant rotational stiffness corresponding to
connection yielding. Conversely, the component approach as codi-
fied in Eurocode 3 is essentially devoted to the prediction of the
initial rotational stiffness.
3. Experimental tests
3.1. Test set-up
New experimental tests have been carried out at the Materials
and Structures Laboratory of Salerno University by means of a spe-
cifically developed test set-up. The vertical column has been con-
nected to a concrete base by means of an exposed a base plate
connection. The concrete base is fixed to the strong floor of the lab-
oratory by means of a reacting system constituted by two steel
beams connected to the strong floor with four high resistance
DYWIDAG bars properly preloaded in order to avoid the concrete
base rotation (Figs. 3 and 4).
The vertical axial load has been applied to the column by means
of a MOOG hydraulic actuator (maximum load equal to 2000 kN)
located on the top of the column and contrasted by a self-balanced
system constituted by a rigid steel beam connected to the concrete
base by means of two high resistance DYWIDAG bars (Fig. 4). The
horizontal load has been applied by means of an hydraulic actuator
MTS model 243 (maximum load ±250 kN, stroke ±125 mm). The
tests have been conducted under displacement control for mono-
tonic loading conditions.
Each specimen has been initially subjected to several tests in
elastic range aiming to the evaluation of the rotational stiffness
for different values of the applied axial load. In particular, such
tests have been performed applying seven values of the axial load
equal to 1%, 2%, 3%, 4%, 10%, 15% and 20% of the column squash
load. The values of the rotational stiffness have been evaluated as
the secant stiffness corresponding to three values of the top dis-
placement of the column equal to d = 1.5, 3 and 4.5 mm, which
have been selected sufficiently small to assure that the specimen
 M. Latour et al. / Engineering Structures 68 (2014) 14–23 19

HE 240 B - tbp= 15 mm – N=585 kN HE 160 A - tbp= 15 mm – N=34 kN

HE 240 B - tbp= 25 mm – N=585 kN HE 160 A - tbp= 15 mm – N=233 kN

Fig. 7. Experimental results and comparisons with Eurocode 3 model.

Table 2 rotational stiffness. Successively, a monotonic test up to failure

Comparison between experimental results and model predictions in terms of has been carried out to investigate the whole moment–rotation
dissipated energy. curve.
Specimen Epl,EC3/Epl,exp Eu,EC3/Eu,exp During the tests, the displacement and the force of both actua-
HE240B-15 1.11 0.75 tors have been acquires and, in order to monitor the displacements
HE240B-25 1.34 0.80 of each connection component, 6 LVDT transducers have been ar-
HE160A-15-33 1.31 0.87 ranged according to the scheme depicted in Fig. 5. In particular,
HE160A-15-233 1.10 0.60 the arrangement of the transducers was selected aiming to the
Average 1.21 0.76 evaluation of the deformability of the equivalent T-stub in tension,
the deformability of the concrete in compression and the elonga-
is in elastic range. Obviously, three different values of the bending tion of the anchor bars.
moment corresponding to such displacement levels are obtained Up-to-now 3 monotonic tests on the base plate connections have
leading to three different eccentricity values for each value of the been carried out. The specimens are composed by HE240B and
applied axial load. As seven values of the axial load have been HE160A column profiles connected to a 1400  600  600 mm con-
applied, a total of 21 eccentricity values have been investigated crete base by means of base plates with thickness equal to 15 and
by experimentally evaluating the corresponding values of the 25 mm. In addition, 3 cyclic tests have been planned and are

Table 3
Comparison experimental versus Eurocode 3 prediction for resistance.

Specimen MEC3 (kN m) Mexp (kN m) M EC3 Specimen MEC3 (kN m) Mexp (kN m) M EC3 Specimen MEC3 (kN m) Mexp (kN m) M EC3
M exp M exp M exp

W7-4.20-110 60.4 82.3 0.73 PB 1F 15.0 17.0 0.88 PB 13F 47.3 50.0 0.95
W8-4.20-0 44.9 55.0 0.82 PB 2F 15.0 18.0 0.83 PB 14CF 72.0 95.0 0.76
W9-4.20-0 44.9 55.0 0.82 PB 3F 15.0 18.0 0.83 PB 15CF 71.8 105.0 0.68
S220-010 41.4 58.3 0.71 PB 5CF 27.9 45.0 0.62 HE 240 B-15 116.7 116.0 1.01
S220-040 66.0 56.6 1.17 PB 8F 46.4 39.0 1.19 HE 160 A-15-34 53.1 54.2 0.98
S220-150 74.4 112.0 0.66 PB 9F 46.4 38.0 1.22 HE 160 A-15-233 68.1 83.0 0.82
S220-190 59.8 68.0 0.88 PB 10CF 69.2 71.2 0.97 HE 240 B-25 148.9 129.0 1.15
S140-010 23.5 40.0 0.59 PB 11CF 69.2 71.2 0.97
S140-100 31.7 32.0 0.99 PB. 12F 47.3 50.0 0.95
Mean value 0.89
Standard deviation 0.24

W [31]; V [28]; PB [24]; HE (authors’ tests).

20 M. Latour et al. / Engineering Structures 68 (2014) 14–23

Table 4
Comparison experimental versus Eurocode 3 prediction for stiffness.

Specimen SEC3 (kN m) Sexp (kN m) Sj;EC3 Specimen SEC3 (kN m) Sexp (kN m) Sj;EC3 Specimen SEC3 (kN m) Sexp (kN m) Sj;EC3
Sj;exp Sj;exp Sj;exp

W7-4.20-110 27,849 24,093 1.16 PB 1F 1868 1733 1.08 PB 13F 18,707 10,523 1.78
W8-4.20-0 15,255 11,500 1.33 PB 2F 1868 1733 1.08 PB 14CF 30,346 26,092 1.16
W9-4.20-0 15,255 11,500 1.33 PB 3F 1868 1733 1.08 PB 15CF 30,260 26,092 1.16
S220-010 3862 5061 0.76 PB 5CF 4924 8626 0.57 HE 240 B-15-585 68,416 59,193 1.16
S220-040 18,664 19,991 0.93 PB 8F 18,656 12,484 1.49 HE 160 A-15-34 11,336 7494 1.51
S220-150 85,300 70,404 1.21 PB 9F 18,656 12,120 1.54 HE 160 A-15-233 20,723 10,437 1.99
S220-190 83,693 53,000 1.58 PB 10CF 41,663 26,092 1.60 HE 240 B-25-585 80,508 43,851 1.84
S140-010 1897 1131 1.68 PB 11CF 41,663 26,092 1.60
S140-100 25,540 18,132 1.41 PB 12F 18,707 10,523 1.78
Mean value 1.35
Standard deviation 0.35

W [31]; V [28]; PB [24]; HE (authors’ tests).

Fig. 8. Comparison between experimental results and Eurocode 3 prediction.

ongoing. All the steel elements, columns and plates, are made of have been fixed to the concrete base by means of their insertion into
S275 steel grade while the concrete base is made of C20/25 class. a M24 hole, having a length equal to 400 mm, successively casted
The connection between the base plate and the concrete base is with a two-component resin FISV360S provided by Fischer. The bars
made by means of M20 threaded bars of 8.8 class. The anchor bars have been successively tightened with a torque wrench applying a
 M. Latour et al. / Engineering Structures 68 (2014) 14–23 21

M following). Successfully, it has been unloaded and then reloaded

up to failure with an axial load equal to 233 kN (referred as
M j,exp HE160A-15-233 in the following). For this specimen, the degree
of partial strength, evaluated according to Eurocode 3, is equal to
80% in the first loading phase and equal to 100% in the final phase.
experimental curve Specimens HE 240B-15, HE 240B-25 and HE 160A-15-34 have
been designed so that the weakest joint component is the base
Sj plate, while specimen HE 160A-15-233 has been designed by
S j /3 balancing the resistance of the component on the tensile side
and that on the compressed side. Regarding the failure mecha-
φ nisms, the experimental tests evidenced the good accuracy pro-
vided by Eurocode 3 model. In fact, in all the tests, according to
Fig. 9. Evaluation of the experimental flexural resistance. the prediction, collapse has occurred due to the fracture of the base
plate in the heat affected zone. Only in case of specimen HE-160A-
HE 240 B-15-Stiffness
15-233, premature local buckling of the column compressed flange
90000
anticipated the fracture of the base plate. In Figs. 6 and 7 the M–u
curves provided by the experimental tests are compared with
80000
those predicted by means of the application of Eurocode 3 model.
70000
A first quantitative comparison is also given in Table 2 where the
60000 ratios between the areas under the experimental curves and those
Sj,in [kNm]

Experimental Points
50000 under the corresponding predicted curves are evaluated. For each
EC3 Model
40000 experimental test, two values of the above ratio are shown in
30000
Alternative Model Table 2: the first one is evaluated considering the whole M–u
20000
curve up to the collapse and the second one considering only the
portion of the M–u curve up to the achievement of the plastic mo-
10000
ment. It can be observed that up to the achievement of the plastic
0
0 100 200 300 400 500 600 700 800 900 1000
moment, the error is equal to about 20% in average and it is mainly
due to the overestimation of the initial stiffness. The comparison
e [mm]
up to failure points out that the model provides a safe side predic-
tion, consistent with the aim of code provisions which simplifies
HE 240 B-25-Stiffness
140000 the complex non-linear response by adopting a perfectly plastic
branch, so that the capture of the hardening phase is out of the
120000 scope of the codified approach.
100000
4. Accuracy of Eurocode 3 approach
Sj,in [kNm]

80000
Experimental Points

60000
40000
20000
0 code 3 approach are clearly specified, a set of 25 experimental tests
0 200 400 600 800
e [mm]
Fig. 10. Accuracy of the predicting models.
tightening torque corresponding to a normal stress in the anchor bar
equal to 80% of the yield stress.
The contact between base plate and concrete was guaranteed
by means of epoxy expansive grout, Mapefill-R type, obtaining a
grout thickness equal to about 25 mm. The geometrical properties
of specimens are summarised in Table 1.
3.2. Experimental results and comparison with Eurocode 3 predictions
The specimens have been designed on the basis of the formula-
tions proposed by Eurocode 3. In particular, the base plate connec-
tions have been designed as partial-strength connections by
imposing a predefined ratio between the base plate connection
resistance and the column flexural strength, varying in the range
50–100%. In particular, specimens HE 240B-15 and HE 240B-25
have been designed with a partial-strength degree equal, respec-
tively, to 50% and 60%. Specimen HE 160A-15 has been tested
applying, in the first loading phase, an axial load equal to 34 kN
up to the rotation of 75 mrad (referred as HE160 A-15-34 in the
EC3 Model In order to widen the investigation dealing with the accuracy of
Alternative Model Eurocode 3 model, a more extensive comparison with experimen-
tal results collected from the technical literature has been
performed. In particular, considering only those experimental tests
for which all the parameters needed for the application of Euro-
1000 1200
have been identified [30,16,23,14]. All the tests deal with base
plate connections with a number of bolts in the tension zone rang-
ing from 1 to 3. The considered set includes column–base connec-
tions with anchor bars both outside and inside the column flanges.
The column profiles are double T profiles, whose height ranges be-
tween 100 and 220 mm, and tubular profiles whose size is equal to
150 mm. The anchors bars are made both with ‘‘washer plate’’ and
‘‘end hooks’’ and are characterised by diameter ranging between
M20 and M24 with resistance classes equal to 4.6, 8.8 and 10.9.
The range of plate thickness varies between 10 and 30 mm.
For each experimental test, the flexural resistance and the
initial stiffness have been evaluated both for proportional and for
non-proportional loading processes. In cases of non-proportional
loading, the evaluation of the initial stiffness has been performed
considering the eccentricity corresponding to 2/3 of the ultimate
moment. In Tables 3 and 4 and in Fig. 8, the experimental values
of stiffness and moment resistance are compared with those pre-
dicted by means of Eurocode 3. Regarding the evaluation of the
experimental value of the flexural resistance, according to
Eurocode 3 approach, a conventional value has been considered
as described in Fig. 9. The comparison with the experimental tests
shows that Eurocode 3 model provides sufficient accuracy in
predicting flexural resistance, giving an average value of the ratio
predicted/experimental value equal to 0.89, with a standard
22 M. Latour et al. / Engineering Structures 68 (2014) 14–23

Table 5
Comparison between experimental values of the rotational stiffness and the results obtained with the refinement of Eurocode 3 approach.

Specimen SEC3 (kN m) Sexp (kN m) Sj;EC3 Specimen SEC3 (kN m) Sexp (kN m) Sj;EC3 Specimen SEC3 (kN m) Sexp (kN m) Sj;EC3
Sj;exp Sj;exp Sj;exp

W7-4.20-110 20,533 24,093 0.85 PB 1F 1326 1733 0.77 PB 13F 18,441 10,523 1.75
W8-4.20-0 14,010 11,500 1.22 PB 2F 1326 1733 0.77 PB. 14CF 28,529 26,092 1.09
W9-4.20-0 14,010 11,500 1.22 PB 3F 1326 1733 0.77 PB 15CF 28,458 26,092 1.09
S220-010 2317 5061 0.46 PB 5CF 3379 8626 0.39 HE 240 B-15-585 55,505 59,193 0.94
S220-040 9660 19,991 0.48 PB 8F 18,461 12,484 1.48 HE 160 A-15-34 10,131 7494 1.35
S220-150 63,017 70,404 0.90 PB 9F 18,461 12,120 1.52 HE 160 A-15233 12,945 10,437 1.24
S220-190 60,369 53,000 1.14 PB 10CF 37,634 26,092 1.44 HE 240 B-25-585 41,452 43,851 0.95
S140-010 1543 1131 1.36 PB 11CF 37,634 26,092 1.44
S140-100 17,861 18,132 0.99 PB 12F 18,441 10,523 1.75
Mean value 1.09
Standard deviation 0.38

W [31]; V [28]; PB [24]; HE (authors’ tests).

deviation equal to 0.24. Conversely, the stiffness prediction is less
accurate providing a moderate overestimation. Such drawback has
been also pointed out by the results of the tests carried out at
different values of eccentricity.
The comparison between the results of experimental tests and
Eurocode 3 predictions for different values of the eccentricity eval-
uated (i.e. according to Eqs. (6) and (7)) is shown in Fig. 10. It can
be noted that all the tests show an overestimation of the stiffness
which is more significant in case of small eccentricity. The accuracy
provided by Eurocode 3 in predicting the rotational stiffness of
base plate connections can be generally considered satisfactory
provided that the attention is focused mainly on the ultimate
behaviour. Conversely, when structural deformability is mainly of
concern, a more accurate prediction can be useful. To this scope,
some simplifications introduced by Eurocode 3 for an easier appli-
cation can be removed. In particular, the definition of an effective
width to be specifically applied in stiffness calculation and, there-
fore, different from that used by Eurocode 3 for resistance evalua-
tion could allow a further improvement of the stiffness prediction.
Regarding the component ‘‘concrete in compression’’, according
to the approach suggested by Steenhuis et al. [27,26], the extension
of the concrete zone subjected to compression in elastic range can
be evaluated considering an equivalence between the behaviour, in
terms of maximum deformation, of a flexible plate supported by
springs with that of a rigid plate. From this analysis, the spreading
of the action transmitted by the compressed column flange, which
defines the effective width and length of the concrete in compres-
sion (Fig. 1), provides a value of the parameter c in elastic range
equal to 1.25 tbp.
In addition, with reference to the base plate in tension, accord-
ing to an approach already suggested for beam-to-column joints
[10], two different issues need to be considered: the first one re-
gards the definition of the effective width in elastic range which
can be defined assuming a 45° spreading of the concentrated bolt
action starting from the bolt head edge, the second one regards
the influence of bolt preloading, due to tightening, which can be
accounted for by means of the following relationships:
beff ;bp t3  1:28
w m3 bp t bp
kt ¼ 0:5 with w ¼ 0:57 pffiffiffi
g þ ð1  gÞw db  a
where g is the bolt preloading level expressed as a fraction of that
leading to a bolt tensile stress equal to 80% of the yield stress, db
is the bolt diameter and a = m/db.
The results of the comparison between the values of the stiff-
ness predicted by means of the proposed refinement and those
provided by the experimental tests are given in Table 5. It can be
observed that the introduction of a more appropriate evaluation
of the effective width both in compression and in tension allows
improving the predictions leading to an average value of the ratio
predicted/experimental value of the rotational stiffness of base
plate connections equal to 1.09, while the standard deviation is
almost unchanged.
5. Conclusions
The results of the first part of a research program aimed at the
prediction of the cyclic behaviour of base plate connections have
been presented. In particular, the experimental tests carried out
at the University of Salerno have been shown. In addition, the de-
gree of accuracy of Eurocode 3 component method for predicting
the rotational stiffness and the flexural resistance of column–base
connections has been investigated by comparing the numerical
results with those obtained from the presented test results and
from a collection of test results taken from technical literature. A
refinement of the approach for the prediction of the connection
rotational stiffness has been also suggested. Such refinement is
based on the definition of an effective width to be specifically ap-
plied for stiffness calculation, i.e. in elastic range. Conversely, for
sake of simplicity, the codified approach is based on the use of a
simple formulation of the effective width to be used both in stiff-
ness evaluation and strength evaluation. The proposed refinement
is made with reference to the components both in compression
and in tension, according to models already proposed in the
technical literature.
The comparison between experimental results and Eurocode 3
approach has shown that the codified version of the component
approach leads to a good accuracy in the prediction of the connec-
tion plastic flexural resistance defined as the one corresponding to
a secant rotational stiffness equal to 1/3 time the initial one. Con-
versely, some overestimation is obtained in the prediction of the
joint rotational stiffness.
Acknowledgments
This work has been partially supported with research Grant
DPC-RELUIS 2010–2013.
ð11Þ References
[1] Astaneh A, Bergsma G. Cyclic behavior and seismic design of steel base plates.
In: Proceedings, structures congress. ASCE; 1993. p. 409–14.
[2] Bernuzzi C, Zandonini R, Zanon P. Experimental analysis and modelling of
semi-rigid steel joints under cyclic reversal loading. J Constr Steel Res
1996;2:95–123.
[3] Burda J, Itani A. Studies of seismic behavior of steel base
plates. Reno: University of Nevada; 1999.
[4] CEN. Eurocode 3: design of steel structures – Part 1–8: design of joints; 2005.
[5] CEN. Eurocode 8: design of structures for earthquake resistance – Part 1:
general rules, seismic actions and rules for buildings; 2005.
[6] DeWolf J, Sarisley E. Column base plates with axial loads and moments. J Struct
Div 1980;106(11):2167–84.
 M. Latour et al. / Engineering Structures 68 (2014) 14–23
[7] Drake R, Elkin S. Beam–column base plate design: LRFD method. Eng J AISC
1999;36(1):29–38.
[8] Ermopoulos J, Stamatopoulos G. Analytical modeling of column base plates
under cyclic loading. J Constr Steel Res 1996;40(3):225–38.
[9] Ermopoulos J, Stamatopoulos G. Mathematical modeling of column base plate
connections. J Constr Steel Res 1996;36(2):79–100.
[10] Faella C, Piluso V, Rizzano G. Experimental analysis of bolted connections:
snug versus preloaded bolts. J Struct Eng 1998;124(7):765–74.
[11] Fisher J, Kloiber L. Design guide 1: base plate and anchor rod design. 2nd
ed. Chicago: American Institute of Steel, Construction; 2006.
[12] Girao Coelho A. Characterization of the ductility of bolted extended end plate
beam-to-column steel connections. PhD thesis. Universidade de Coimbra; 2004.
[13] Gomez I, Kanvinde A, Deirlein G. Exposed column base connections subjected
to axial compression and flexure. Chicago: American Institute of Steel
Construction; 2010.
[14] Grauvilardell J, Lee D, Hajjar J, Dexter R. Synthesis of design, testing and
analysis research on steel column base plate connection in high-seismic zones,
Minneapolis. Minnesota: s.n.; 2005.
[15] Iannone F, Latour M, Piluso V, Rizzano G. Experimental analysis of bolted steel
beam-to-column connections: component identification. J Earthq Eng
2011;15(2):214–44.
[16] Jaspart J, Vandegans D. Application of the component method to column bases.
J Constr Steel Res 1998;48:89–106.
[17] Kanvinde A, Grilli D, Zareian F. Rotational stiffness of exposed column base
connections – experiments and analytical models. J Struct Eng
2012;137(5):549–60.
[18] Kim K, Engelhardt M. Monotonic and cyclic loading models for panel zones in
steel moment frames. J Constr Steel Res 2002;58:605–35.
[19] Latour M, Piluso V, Rizzano G. Cyclic modeling of bolted beam-to-column
connections: component approach. J Earthq Eng 2011;15(4):537–63.
23
[20] Latour M, Piluso V, Rizzano G. Experimental analysis of innovative dissipative
bolted double split tee beam-to-column connections. Steel Constr
2011;4(2):53–64.
[21] Latour M, Rizzano G. Experimental behavior and mechanical modeling of
dissipative T-stub connections. J Struct Eng 2012;138(2):170–82.
[22] Latour M, Rizzano G. A Theoretical Model for Predicting the Rotational
Capacity of Steel Base Joints. J Constr Steel Res 2013;9:88–99.
[23] Nogueiro P, Simoes da Silva L, Bento R, Simoes R. Calibration of model
parameters for the cyclic response of end-plate beam-to-column steel–
concrete composite joints. J Steel Compos Struct 2009;9(1):35–58.
[24] Picard A, Beaulieu D. Behaviour of a simple column base connection. Can J Civ
Eng 1985;12(1):126–36.
[25] Piluso V, Faella C, Rizzano G. Ultimate behavior of bolted T-stubs. Part I:
theoretical model. J Struct Eng ASC 2001;127(6):686–93.
[26] Steenhuis C, Bijlaard F. Tests on column bases in
compression. Innsbruck: Institute for Steel, Timber and Mixed Building
Technology; 1999.
[27] Steenhuis C, Wald F, Sokol Z, Stark J. Concrete in compression and base plate in
bending. HERON 2008;53(1/2):51–68.
[28] Thambiratnam D, Paramasivan P. Base plates under axial loads and moments. J
Struct Eng 1986;112(5):1166–81.
[29] Vandegans D. Column Bases: Experimentation and Application of Analytical
Models. Research Center of the Belgian Metalworking Industry, MT 196, p. 80,
Brussels 1997.
[30] Wald F. Patky Sloupù – column bases. Praha: CVUT; 1995.
[31] Wald F, Sokol Z, Steenhuis M. Proposal of the stiffness design model of the
column bases. In: Proceedings of the third international workshop on
connections in steel structures, Trento, Italy; 1995.