Journal of Constructional Steel Research 85 (2013) 1–11

Contents lists available at SciVerse ScienceDirect

Journal of Constructional Steel Research

Behaviour of bolted flange joints in tubular structures under monotonic,

repeated and fatigue loadings I: Experimental tests
Hoang Van-Long ⁎, Jaspart Jean-Pierre, Demonceau Jean-François
ArGEnCo Department, University of Liège, Belgium

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

Article history: This paper presents a test program on bolted flange plate connections used in circular tubular structures. In the
Received 15 May 2012 tested joints, the connected tubular elements are made of high strength steel (TS590) whilst normal steel grade
Accepted 26 February 2013 (S355) is used for the flanges. The tests were performed under monotonic loadings to obtain the mechanical
Available online 27 March 2013
properties of the connections and under cyclic loadings to characterise the behaviour of these connections
for low cycle and high cycle fatigue. The test results are first presented herein, including a critical analysis of
Keywords:
Flange bolted joints
the latter; then, design methods available in literature are applied for the tested configurations with the objective
Tubular elements to investigate the accuracy of these methods through comparisons with the experimental results.
High-cycle fatigue © 2013 Elsevier Ltd. All rights reserved.
Low-cycle fatigue
Hot-spot stress
Experimental tests

1. Introduction extreme loading conditions” and funded by RFCS (Research Fund for
Coal and Steel) was launched for three years; one of the objectives of
In practice, circular tubes are used in various types of structures, this project was to investigate the behaviour of the previously mentioned
such as buildings, bridges, wind mill towers, radio/television mats, off- connection configuration under monotonic and cyclic loadings. In partic-
shore structures, etc. The external loads acting on these structures are ular, a test program has been performed in Liège on bolted flange connec-
obviously from different types (for instance, exploitation loads, wind tions between HSS tubes. The tested specimens were made of circular
loads, wave loads, etc) and from different natures (monotonic, cyclic tubes with HSS and flanges with “normal” mill steels (S355). During
or repeated); according to the considered loads, the latter may induce the tests, the connections have been subjected to axial tension forces
low cycle or high cycle fatigue phenomena in structural details. applied monotonically, to obtain the mechanical properties of the joints,
Also, bolted joints using circular flanges are frequently met within and cyclically, to characterise the behaviour of the tested connections
the mentioned structures, in particular for column splices. Such a type under low-cycle and high-cycle fatigue.
of connections has been the subject of many researches during the In the present paper, the performed test campaign is first described
past three decades, e.g. [1–4,8,10,13,16]. Through these researches, de- with a critical analysis of the obtained results (Section 2). Then, in a
sign models have been proposed and/or improved using experimental, second step, a summary of current methods for the design of the tested
analytical and/or numerical approaches, leading to different guidelines connection configuration is reported in Section 3, with comparisons
for the design of such connections. However, significant differences with the previously presented experimental results.
between the results obtained through these methods are observed. In a companion paper entitled “Behaviour of bolted flange plate
Moreover, most of the performed studies have been dedicated to the connections in tubular structures under monotonic, cyclic and fatigue
behaviour of such connection under monotonic loads whilst the behav- loadings. II: Numerical investigations” and published in the same jour-
iour of the latter under cyclic or repeated loads have been rarely consid- nal, numerical studies performed on the investigated connections are
ered, in particular for connections with High Strength Steel (HSS) tubes. also presented. The two companion papers constitute a global overview
Besides that, Eurocode-3 part 1.8 [5] (dealing with the design of of the investigations conducted in Liège, in the framework of the previ-
joints) doesn't give yet design guidelines for bolted joints using circular ously mentioned European project, with the objective to derive design
flanges leading to difficulties for the practitioners which have to deal recommendations for bolted flange connections.
with such connections.
It is the reason why, in 2008, a European project entitled “HITUBES — 2. Performed test program
Design and integrity assessment of high strength tubular structures for
The present section details the test program performed at the M&S
⁎ Corresponding author. Tel.: +32 4 3669614. laboratory (laboratory for Material mechanic and Structures, University
E-mail address: [email protected] (H. Van-Long). of Liège, Belgium). The tested connection configurations will be first

0143-974X/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcsr.2013.02.011
2 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11

Fig. 2. Instrumented bolts.

from the tubes and the flanges were performed following the recom-
mendations given in ISO 6892 [11]. The mean values of the main charac-
Fig. 1. Specimen description. teristics of materials are presented in Table 2, more information can be
found in [8].

described in detail; then, the obtained test results will be presented 2.1.3. Measurement of the initial geometrical deformations of the flanges
and discussed. As illustrated in Fig. 3, a significant deformation of the flanges has
been observed; this deformation is due to the heat-effect of the welding
2.1. Preparation of the tests process. The deformations of each flange at 12 points on the flange edges
and uniformly distributed around the tube were measured using an
2.1.1. Description of the tested specimens and of the testing strategy electronic calliper and one straight ruler (Fig. 3). The measured values
Two bolted joint configurations were tested with the objective (two flanges per specimen) are given in Table 3; the deformations for
to highlight different collapse modes. The properties of the tubular specimens M1 and M2 were not measured.
elements (356 × 12 mm circular tube), the diameter of the circular
flanges (556 mm) and the number of bolts (12 bolts) are the same; 2.1.4. Instrumentations of the specimens
only the thickness of the flanges and the size of bolts are varied Four types of strain gauges were used to record the deformations on
between the two configurations (Fig. 1): the tubes, on the flanges and in the bolts, as reported in Table 4. The
locations of the strain gauges on the specimens are presented on
– in Configuration 1, the flanges thickness is 15 mm and M27 8.8 bolts
Figs. 4–7. The strain gauges were placed so as to measure deformations
are used;
in critical zones, where the development of cracks/failure could have
– in Configuration 2, the flange thickness is 20 mm and M20 8.8 bolts
been expected, as in the bolt shanks, on the tubes at the weld toe or
are used.
on the flanges at the weld toe. The strain gauges are only located on
Full penetration welds were realised between the flanges and the one side of the connections as the latter are symmetric. Amongst the
tubes for both configurations. As mentioned previously, the steel grade 12 bolts present in each tested joint, there are 4 bolts are instrumented
for the tubes is TS590 (i.e. nominal yield stress equal to 590 N/mm2) as illustrated in Figs. 4–7. These instrumented bolts have also been used
whilst the steel grade for the flanges is S355 (i.e. nominal yield stress to control the applied torque moment during the bolt tightening
equal to 355 N/mm2). procedure.
Each configuration was tested under monotonic, high cycle fatigue In order to obtain the global deformation of the joints, four displace-
and low cycle fatigue loadings. Therefore, in total, six specimens were ment transducers are set-upped as shown on Fig. 8.
fabricated and tested as presented in Table 1. Beside the main tests
reported in Table 1, some supplementary tests aiming at obtaining infor- 2.1.5. Tightening procedure
mation about the hot-spot stress distribution in critical zones were also A dynamometric wrench was used to tighten the bolts. The applied
performed in the elastic linear domain. torque force was controlled using the measurements got from the strain
gauges embedded in the instrumented bolts. According to Eurocode-3,
2.1.2. Tests on base materials part 1.8 (Article 2, Section 3.6.1) [5], the design preloading has to be
Tensile tests were carried out on coupons extracted from the tubes taken as equal to Fp,Cd = 0.7fubAs/γM7, in which fub, As and γM7 are the
and the flanges composing the tested specimens. Also, tensile tests on ultimate strength of the bolt (nominal value), the threaded section
bolts were performed; amongst the tested bolts, some were instrumented of the bolt shank and the safety coefficient (=1.1) respectively. With
using the BTM-6C embedded strain gauges (Fig. 2) in order to have more this force, the stresses in the threaded and unthreaded portions of
information on the stress-strain curve. The tests on the coupons extracted the bolts are 509 N/mm2 and 407 N/mm2 respectively. Assuming
a Young modulus equal to E = 2.1 × 105 N/mm2, a strain equal to

Table 1
List of performed tests.
Table 2
Test number Specimen name Configuration (Fig. 1) Loading type Results of coupon tests.
1 M1 1 Monotonic loading
Parameters Plate 15 mm Plate 20 mm Tube M27 bolts M20 bolts
2 M2 2
3 H1 1 High cycle fatigue loading Yield strength 387 384 822 857 850
4 H2 2 (N/mm2)
5 L1 1 Low cycle fatigue loading Utimate strength 544 529 881 930 930
6 L2 2 (N/mm2)
 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11 3

Fig. 3. Measurement of the initial deformation of the flanges.

Table 3
Initial deformations of the flanges.

Spec. Flange Measured value in mm at the points given in Fig. 3

1 2 3 4 5 6 7 8 9 10 11 12

H1 1 1.79 2.99 3.28 2.39 3.88 2.92 1.39 2.88 3.10 2.87 3.65 3.03
2 2.25 2.67 2.65 2.35 2.84 2.78 2.57 2.72 2.92 2.12 3.26 2.79
H2 1 1.98 2.11 2.07 1.57 1.72 1.59 1.22 1.39 1.57 1.30 1.63 1.59
2 1.86 2.46 2.30 1.52 2.35 2.29 2.20 1.96 1.82 1.82 1.56 1.71
L1 1 1.84 0.82 1.62 2.03 1.30 1.89 1.93 0.95 1.94 2.47 1.37 1.98
2 1.68 2.80 2.5 1.63 2.56 2.49 1.59 2.98 2.82 1.62 2.68 2.68
L2 1 4.37 3.65 3.38 2.48 2.92 3.25 2.56 2.65 2.38 1.97 3.47 4.44
2 1.74 2.55 2.95 1.67 2.45 2.60 2.30 2.83 2.57 1.66 2.05 2.21

1.9 × 10−3 can be computed for the unthreaded part, i.e. where the so the applied loading may be considered as a quasi-static loading too.
strain gauges are placed in the bolts. A “round-by-round” procedure is A frequency of 1.4 Hz was applied for the high cycle fatigue tests
used for the tightening, until the value in the strain gauges in the (Specimens H1 and H2). A constant load variation of 900 kN with a
instrumented bolts reach the previously computed value of strain maximal value of 1000 kN is maintained for specimen H1 (configu-
(1.9 × 10−3). During the tightening procedures, the indications from ration 1), whilst two constant load bands are used for Specimen H2
all the strain gauges were recorded. (configuration 2): (1) first band with a load variation of 600 kN
(maximal load of 700 kN), and second band with a load variation of
2.1.6. Test set-up 1100 kN (maximal load of 1200 kN), as the show in Fig. 10.
A SHENCK machine with a capacity of 2500 kN in both tensile and
compression is used for the tests. The tests set-up is illustrated in Fig. 9. 2.2. Test results and discussion
Two tensile plates welded to the tubes are used to connect the specimens
to the machine and to introduce the applied load. The distance between 2.2.1. Test results
the tensile plates and the flanges of the tested connections is equal to The measured load–displacement curves are presented in Figs. 11–14
1000 mm in order to avoid affecting the stress distribution in the tubes for the monotonic and low cycle fatigue tests (the displacement
near the tested connections. Accordingly, with this test set-up, the load reported in these figures is the average value of the measurement,
acting on the specimens may be considered as a perfect axial load, in sup- ((D1 − D4) + (D2 − D3))/2, — see Fig. 8). The test on Specimen M1
posing that the tolerances are small. had to be stopped because the load reached the maximal capacity of the
testing machine (i.e. 2500 kN), but after reaching a significant deforma-
2.1.7. Loading procedure tion of the tested specimen (see Fig. 11). 37 cycles and 3 cycles of loading
A quasi-static loading is adopted for the monotonic tests (Specimens were applied to Specimens L1 and L2 respectively (see Figs. 13 and 14).
M1 and M2). For the low cycle fatigue tests (Specimens L1 and L2), dis- For Specimen L2 (cyclic loading on configuration 2), the application of
placement control is used and a velocity of about 0.4 mm/s was applied; cycles with a constant displacement was initially planned. However,
after the first cycle, no residual deformation was observed (see Fig. 14);
accordingly, it was decided to increase the applied displacement for the
Table 4 second cycle but the same phenomenon was observed. At the third
Used strain gauges. cycle, the displacement reaches the maximal point reached during the
monotonic test, and the test was stopped. For Specimen L1 (configuration
Type of strain gauge Description Symbol Location
1 under repeated loading), a typical behaviour for bolted joints (that have
BTM-6C Axial strain gauge for BT In the bolts (Figs. 4–7)
been presented in some researches, e.g. [13,15]) was observed with the
bolts
FRA-6-11 0°/45°/90° 3-element FR On the flanges (Fig. 4) development of hysteresis (see Fig. 13).
rosette For the high cycle fatigue tests, 960,000 cycles of loading were
FLA-6-11 Single element FL On the tubes (Figs. 4 & 5) applied to Specimen H1 (configuration 1) whilst 2,404,000 cycles of
QFXV-1-11 5-element single-axis QF On flanges and tubes the first loading band and 277,600 cycles of the second loading
(Figs. 5 & 6)
band (Fig. 10) were applied to Specimen H2 (configuration 2).
4 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11

Fig. 4. Location of strain gauges on specimens M1 and M2 (monotonic tests).

The failure modes of all tested specimens are presented in Figs. 15–17; joints for the monotonic tests. Indeed, the ultimate load capacity of the
the corresponding observations are reported in Table 5. According to joints was equal to less than 30% of the axial resistance (under monotonic
the T-stub concept as defined in the Eurocodes [5], a failure mode 2 was loading) of the tube (=10,000 kN). However, the cracks develop in the
observed for configuration 1 (yielding of the flange plate + failure of tubes during the fatigue tests meaning that the tubes were not over-
the bolts) whilst a failure mode 3 occurred for configuration 2 (failure designed as it looks like.
of the bolts with no yielding in the flange plate). This observation is not
in agreement with the predicted failure modes obtained through the an- 2.2.2. Discussion
alytical procedure used for the specimen design. The analytical methods Some specific aspects about the behaviour of flange bolted joints
will be summarised in Section 3.2. observed have been observed during the tests:
By the limitation of space, the strain gauge records are not reported
herein. However, the tendency of the strain gauge records at critical ­ The initial stiffness of the joints is very high, and the linear branch
zones is illustrated in Fig. 18. Noting that with respect to fatigue investi- observed when unloading the specimens in the non-linear domain
gation, the stress variation (stress range) is more important than the is not parallel to the initial branch (Figs. 11 and 12).
absolute value of stress. ­ The stress in the bolts and the stress in the flange at the weld toe are
It is also interesting to highlight the following observation: the plates almost constant in the linear domain whilst the stress in the tube at
and the bolts of the joints were designed so as to have failure in the the weld toe considerably increases (Fig. 17). The concentration of

Fig. 5. Location of strain gauges on H1 and H2 (fatigue tests).

 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11 5

Fig. 6. Location of strain gauges on specimen L1 (cyclic test).

stresses in the tube near the weld toe is less important during the the stresses in the bolts and the stresses in the flanges at the weld toes
loading state in comparison with the tightening state. The way the are almost constant in the linear domain. Also, the bending moment in
stresses develop needs to be accurately understood and predicted the tube wall is almost constant but the axial force is progressively in-
as it influences the fatigue strength of the joints. creases due to the application of the external load; this leads to the stress
ranges in the tubes which are significantly higher than the ones in the
The above phenomenon may be explained as explained here below. bolts and on the flanges.
Due to the initial deformation of the flanges as previously described, In order to clarify the above explanations, the behaviour of the joints
the pre-stressing of the bolts during the tightening state induces pre- can be illustrated as shown in Fig. 19, in which two cases (with and
stresses not only in the bolts but also in other components: without initial deformation of flanges) are distinguished. The forces
– the bolts are principally pre-loaded in the axial direction; applied to the joints are (see Figs. 19 and 20): P, the external load, F1,
– the flanges are principally pre-loaded in bending; the prying force, F2, the sum of the force in the bolt and of the contact
– the tube walls are principally pre-loaded in bending, in the longi- force between the two flanges around the bolts, and F3, the sum of the
tudinal direction. contact forces between the two flanges inside the tubes. F3 is equal to
0 in the case of perfect flanges, i.e. with no initial deformation, whilst
It is obvious that the stress components already pre-stressed in the it appears in joints with deformed flanges. This force (F3) increases
tightening state will slowly increase in the beginning of the loading from zero to a maximal value at the end of the tightening state. In
state, i.e. when the tensile load is applied to the specimen. Therefore, the loading state, F3 progressively decreases, and may vanish at some
moments. As the increase of the external load is almost equilibrated
by the decrease of F3, F2 almost doesn't change at the beginning of the
loading state, leading to the non-variation of the moment in the flanges
at the weld toe. Concerning the effects of the welding deformation
of flanges on the joint behaviour, interesting discussions can be also
found in [1].

Fig. 8. Location of displacement transducers on the specimens (four displacement

Fig. 7. Location of strain gauges on specimen L2 (cyclic test). transducers are used (D1, D2, D3 and D4)).
6 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11

3.1. Existing methods to determine the resistance of joints

As the objective is to compare the method predictions to the exper-

imental results, all the safety factors are taken equal to 1.0 and they are
not appearing in the formulas reported here below. Moreover, the yield
strength and the plastic moment are replaced by the ultimate strength
and the ultimate moment respectively as the objective is to make com-
parisons on the predicted ultimate resistance of the joints.

3.1.1. Notices
Geometrical dimensions (Fig. 22):
A: cross-sectional area of a tube;
As: threaded area of a bolt;
d: diameter of a bolt;
D: outside diameter of a tube;
Dp: diameter of bolt pitch circle;
Df: diameter of a flange;
aw: throat thickness of a weld;
e1: distance from the bolt axis to the outside surface of tube;
e2: distance from the bolt axis to the flange edge;
t: tube thickness
pffiffiffi
m ¼ e1 −0; 8 2aw ;
Fig. 9. Test set-up. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e; ¼ D2f =4−m2 −Dp =2;
nb: number of bolts.
In summary, the influence of the initial deformation on the stress
Geometrical coefficients:
development in the different joint components is illustrated in Fig. 21. D
If the initial deformation of flanges is omitted, as it is traditionally done k1 ¼ ln D−tp
;
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
for design, the stress range in the flange at the weld toe is very important 2
k2 ¼ 2k11 2 þ k1 þ k1 þ 4 ;
in comparison with the stress range in the tube at the weld toe, and
maybe the flanges would be identified as the critical component under D =2þn
k3 ¼ ln D=2−t=2
p
.
the fatigue loading, which may not fit with the actual response of the
joint including the initial deformation of the flanges. (n is the distance from the prying force to the bolt axis, see [2])
Mechanical characteristics:
B: total force in all the bolts;
3. Comparison of the test results vs. predictions from existing B0: total preload in all the bolts;
design methods Bu: ultimate resistance of all the bolts;
T: applied tensile force;
In this section, existing methods for the design of bolted joints fu,f: ultimate stress of a flange;
are firstly summarised. The comparison between the predictions mu: ultimate moment of a flange (per unit length);
obtained through these methods and the experimental test results is
ν: Poisson's ratio.
then presented.
Fatigue parameters:
Under monotonic loading, the global behaviour of joints is described
by the load–displacement curve characterised by its main parameters, Dd: fatigue damage;
i.e. the initial stiffness, the plastic resistance and the ultimate resistance. ΔσC: detail category;
As the plastic resistance is strongly influenced by the “convention” used ΔσR: stress range (ΔσRi is the stress range in loading band i);
for its estimation when analysing test results, only the ultimate resis- NRi: endurance (in cycles) of detail with the loading band i;
tances are compared and discussed here after. nl: number of loading bands.

Fig. 10. Loading procedures for high cycle fatigue tests.

 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11 7

Fig. 11. Load–displacement curve for Specimen M1 (configuration 1, monotonic loading).

Fig. 13. Load–displacement curve for Specimen L1 (configuration 1, repeated loading).

3.1.2. Igarashi method [10]
By applying limit analysis for the flanges, Igarashi proposed ana-
lytical formulas to determinate the flange thickness and number of
bolts to be used as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi with μ, the slope of the curve depending of the geometrical dimensions
2T of the joints (detailed formulas for the computation of μ can be found
tf ≥ ;
f u;f πk2 in [3]).
2 3 Secondly, the bending moment in the flange at the weld toe and at
1 1
Bu ≥T 41− þ  5: the bolt pitch circle may be determined as follows:
k2 k ln D =D
2 f p 2   !3
−T 4 D2f −ðD−t Þ2 − D2f −D2p μ Df Df
M ri ¼ ð1−υÞ þ 2ð1 þ υÞ ln −μ ln 5
For a joint where the geometry and the material are defined, the 8π D2f D−t Dp
" ! #
capacity of the joint (T) can be obtain from above equations. T
2
Df −Dp
2
ðD−t Þ2 Df
The Igarashi development has been adopted by CIDECT as a guide- M rp ¼ ð1−υÞ μ− þ 2 ð1 þ υ Þ ðμ−1 Þ ln
8π D2f D2p Dp
line for designing bolted circular flange joints [16].
With a joint for which the geometry and the material are defined,
3.1.3. Cao and Belle method [3,4]
the ultimate capacity of the joint is limited by the ultimate resistance
Cao and Bell use a model in which the material is assumed to be linear
of the bolts (Bu) and the ultimate bending moment in the flanges (Mu):
elastic; the prying forces and bolt forces are assumed to be uniformly
distributed at the flange edge and at the bolt pitch circle respectively.
Analytical formulations for bolted joints were developed on the basis B≤Bu ;
of these assumptions and are reported here below. Tables and charts M r ≤M u :
were also established for practical purpose.
Firstly, the bolt force (B) is assumed to vary linearly with the applied 3.1.4. “Eurocode” concept [12]
tension force T following two straight lines: Referring to Eurocodes, no specific guidelines for the design of
bolted flange joints are given. However, using the concept of T-stub
B ¼ 1:1B0 ðfor B≤1:1B0 Þ as given in Eurocode-3, part 1.8 [5], CIDECT Research Project 5BP
B ¼ μT þ 0:1B0 ðfor B > 1:1B0 Þ

Fig. 12. Load–displacement curve for Specimen M2 (configuration 2, monotonic loading). Fig. 14. Load–displacement curve for Specimen L2 (configuration 2, repeated loading).
8 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11

Specimen M1 Specimen M2

Fig. 15. Specimens after monotonic testing.

Specimen H1 Specimen H2

Fig. 16. Specimens after fatigue testing.

[12] proposes an approach for flange bolted joints fully in line with are proposed to determine the resistances of the joints (defined as the
the Eurocode philosophy. Within the proposed approach, three minimum of the resistances of the three possible collapse modes):
failure modes are distinguished, and formulas for each collapse mode
Mode 1 resistance — yielding of the flange (thin flanges):

 
D þ Dp
T 1 ¼ 2πmp 1 þ ;
2m

Mode 2 resistance — yielding of the flange + failure of the bolts

(intermediate flanges):

2πmu þ nBu
T2 ¼ ;
mþn

Mode 3 resistance — failure of the bolts (thick flanges):

T 3 ¼ Bu :

Crack in the flange of Specimen A bolt of Specimen

L1 L2 3.1.5. Couchaux method [2]
Recently, Couchaux [2] proposed an alternative procedure for
Fig. 17. Specimens after cyclic testing. the design of flange bolted joints, combining the Igarashi model and
 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11 9

Table 5
Description of tested specimens at the end of the tests.

Specimens Load type Referent figures Component states

Tubes Flanges Bolts

M1 Monotone Fig. 15 No particular sign Plastic deformation at weld toe Plastic deformation occurs
M2 Monotone Fig. 15 No particular sign Very small deformation All bolts failed
H1 Fatigue Fig. 16 Crack at the weld toe No particular sign No particular sign
H2 Fatigue Fig. 16 Crack at the weld toe No particular sign A bolt is completely cracked, others are partially cracked
L1 Repeated Fig. 17 No particular sign Crack at the weld toe Plastic deformation is observed
L2 Repeated Fig. 17 No particular sign Very small deformation Plastic deformation is observed

Fig. 18. Illustrative view of the stress evolution in joints vs. the applied external load.

the T-Stub concept as suggested in the Eurocodes. Five (5) failure 5 are related to the connected tubes and the welds not decisive for
modes are identified and characterised to compute the joint resis- the joint configurations considered herein.
tance. Only Modes 1, 2 and 3 are presented here after; Modes 4 and
Mode 1 resistance (thin flanges):
 
2
T 1 ¼ 2πmu 1 þ ;
k1  
2e;
but T 1 ≤2πmu nb min 2; 1 þ :
πm

Mode 2 resistance (intermediate flanges):

   
1 k
T 2 ¼ 2πmu 1 þ þ Bu 1− 1 ;
k3 k3

Mode 3 resistance (thick flanges):

T 3 ¼ Bu :

3.2. Fatigue strength according to Eurocode-3 [6,14]

As Eurocode-3, part 1.12 [7] doesn't give specific rules for the fatigue
resistance of high strength steel, the rules given for normal carbon steel
in Eurocode-3, Part 1.9 [6] are used.
According to the latter, the fatigue strength of joints may be calculated
through the following steps:

­ Identify the critical zones where cracks may develop;

­ Assign to each critical zones a detail category as reported in
Eurocode-3, part 1.9 (Tables 8.1–8.10 and Table B1 in [6]);
­ Calculate for each detail category the reference stress ranges
Fig. 19. Forces acting on the joint. (ΔσC), corresponding to a fatigue strength at two million cycles;
10 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11

Fig. 20. Development of the forces acting on joint.

­ Determine the stress ranges (ΔσR) associated to the loading (the the investigated joints can be obtained. The so-obtained results are
nominal stress range concept is used for the details of Tables summarised in Tables 6 and 7.
8.1–8.10 given in [6] whilst the hot-sport stress ranges is adopted To compute the fatigue resistance of the tube, the constructional de-
for the details of Table B.1 given in [6]); tail number 11 in Table 8.5 of Ref. [6] is used (tube socket joint with 80%
­ Compute the endurance (in cycles) of each detail with a loading full penetration butt welds). Even if a ring flange (with a central hole) is
band for which the stress ranges are constant: considered for this detail, it is identified as the closest detail for the
investigated connection. The stress range in the tube is calculated as
 
Δσ C 3 
6 6
 the nominal stress by dividing the axial force by the tube area. When
NRi ¼ 2  10 for N Ri ≤5  10 ;
Δσ the hot-spot stress concept is adopted, the detail category of 100 is
 Ri 5   used (Table B1 of Ref. [6]) whilst the stress range is deduced from the
6 Δσ C 6
NRi ¼ 2  10 for NRi > 5  10 measured deformation given by the strain gauges (FL1, FL2 and QF1,
Δσ Ri
QF2 on Figs. 4, 5 and 6).
­ Calculate, finally, the joint damage through the following formula:
3.3.1. Some remarks can be drawn
­ The results from the current methods and the experimental tests are
X
nl
n Ei quite different, in both ultimate capacity and failure mode (Table 6).
Dd ¼
i
NRi It can be observed that the ultimate capacity of the joints is always
underestimated through the analytical models, what is on the safe
with nEi is the number of cycles associated to the stress range ΔσRi side. However, as the failure mode is not well predicted, it can lead
for loading band i (in case of constant loading, i is equal to 1). to troubles in the design process; indeed, when the designer is
looking for a ductile mode of failure (for instance, Modes 1 and 2
3.3. Comparison to the experimental results in the Eurocode concept), the fact that the actual failure mode is
not well predicted can lead to a lack of ductility.
With the geometric and material data of the considered joints ­ There are significant differences between Eurocode and the test
(Section 2) and using the previously mentioned formulas, the ultimate results concerning the fatigue strength of tube at the weld toe
resistances under monotonic loading and the fatigue strength of (Table 7). It is important to notice that the fatigue strengths given

Fig. 21. Comparison of the evolution of stresses in different components.

 H. Van-Long et al. / Journal of Constructional Steel Research 85 (2013) 1–11 11

Table 7
Comparison of the method (fatigue resistance - high cycle loading).

Method Damage index

Configuration 1 Configuration 2

Eurocode (nominal stress) [6] 0.45 0.38

Eurocode (hot-spot stress) [6] 1.32 1.67
Experimental 1.00 1.00

However, the more important is the difference concerning the failure

modes between the existing methods and the test results, which can
lead to troubles in the design process as a ductile failure mode may be
predicted through these methods whilst it is a brittle one which is actu-
ally occurring.
With respect to the fatigue resistance check for the tube at the weld
toe by using Eurocode [6], some inconsistencies within the recommen-
dations in Eurocode were identified; in particular, it was highlighted
that the effects of the geometrical dimensions on the stress distribution
are not well taken into account. To improve the prediction given by the
Eurocode, one solution would be to use the hot-spot stress (computed
through FEM analyses or analytical approaches) in the concerned detail
instead of the nominal stress as actually recommended in the Eurocode;
this solution is already suggested in other researches [2,9].
The above remarks will be also clarified in the companion paper
Fig. 22. Geometrical parameters of the considered joint. where numerical investigations conducted on the joints studied herein
are developed.
by Eurocode are not conservative in comparison with the test
results. Even the geometries (ring/full flanges) and the material Acknowledgements
(normal/high-strength steel) are not corresponding between the
used detail and that of the tested specimens, it cannot explain the This work was carried out with a financial grant from the Research
observed differences. In fact, some inconsistency within the rules Fund for Coal and Steel of the European Community, within the
given in the Eurocode for this detail can be identified; indeed, the HITUBES project: “Design and Integrity Assessment of High Strength
stress range in the tube at the weld toe is very sensitive to the Tubular Structures for Extreme Loading Conditions”, Grant No
geometries of the different components (tubes/flanges), what is RFSR-CT-2008-00035.
not taken into account within the procedure of Eurocode-3, part
1.9 [6]. To improve the prediction given by the Eurocode, one References
solution would be to use the hot-spot stress (computed through [1] Bouwan LP. Bolted connections dynamically loaded in tension. J Struct Div ASCE
FEM analyses or analytical approaches) in the concerned detail 1982;108.
instead of the nominal stress as actually recommended in the [2] Couchaux M. Comportement des assemblages par brides circulaires boulonnés.
PhD thesis. France: INSA of Rennes; 2010.
Eurocode. The last opinion has been also recommended in some [3] Cao JJ, Bell AJ. Determination of bolt forces in a circular flange joint under tension
researches, e.g [2,9]. force. Int J Press Vessel Pip 1996;68:63–71.
[4] Cao JJ, Packer JA. Design of tension circular flange joints in tubular structures. Eng
J First Q 1997:17–25.
4. Conclusions [5] Eurocode 3: Design of steel structures — Part 1-8: design of joints. EN-2005.
[6] Eurocode 3: Design of steel structures — Part 1-9: fatigue. EN-2005.
[7] Eurocode 3: Design of steel structures — Part 1-12: additional rules for the exten-
Results of a test program on bolted joints with circular flanges were sion of EN 1993 up to steel grades S700. CEN-2007.
presented. The behaviour of these joints under monotonic, high cycle [8] HITUBE project “Design and integrity assessment of high strength tubular struc-
tures for extreme loading conditions”, Grant N0 RFSR-CT-2008-00035. Deliverable
and low cycle loadings was recorded. The measured mechanical proper- D.4: Test data on bolted tubular connections, 9-2011.
ties such as initial stiffness, ultimate capacity, failure mode, stress distri- [9] Hobbacher A. Recommendations for fatigue design of welded joints and compo-
bution and fatigue strength in the joint behaviour were given. The nents. IIW document IIW — 1823-07, ex XIII — 2151 r4 — 07/XV — 1254r4-07; 2008.
[10] Igarashi S, Wakiyama N, Inoue K, Matsumoto T, Murase Y. Limit design of high
application of the current methods for the investigated joints was also strength bolted tube flange joint, Part 1 and 2. , Journal of Structural and Construction
presented and discussed. Engineering Transactions of AIJJapan: Department of Architecture reports, Osaka
It was observed that these existing guidelines give different values University; August 1985.
[11] ISO 6892-1. Matériaux métalliques- Essai de traction, Partie 1: Méthode d'essai à
for the ultimate capacity of joints, but all methods are conservatives.
température ambiante; 2009.
[12] Jaspart JP, Pietrapertosa C, Weynand K, Busse E, Klinkhammer R. Development of a
Table 6 full consistent design approach for bolted and welded joints in building frames and
Comparison of the methods (monotonic loading). trusses between steel members made of hollow and/or open sections: application
of the component method. Volume 1: practical guidelines. CIDECT Report 5BP —
Method Configuration 1 Configuration 2 4/05; 2005.
[13] Nemati N, Le Houedec D, Zandonini R. Numerical modelling of cyclic behaviour of the
Ultimate Critical Ultimate Critical
basic components of steel end plate connections. Adv Eng Softw 2000;31:837–49.
capacity (kN) components capacity (kN) components [14] Nussbaumer A, Borges L, Davaine L. Fatigue design of steel and composite struc-
Igarashi [10] 1466 Flanges 1801 Bolts tures. ECCS — European Convention for Constructional Steelwork; 2011.
Bell and Cao 1493 Flanges 1596 Bolts [15] Piluso V, Rizzano G. Experimental analysis and modelling of bolted T-stub under
[3,4] cyclic loads. J Constr Steel Res 2008;64:655–69.
[16] Wardenier J, Kurobane Y, Packer JA, Dutta D, Yeomans N. Design guide for circular
“Eurocode” [12] 1944 Flanges 2249 Flanges + bolts
hollow section (CHS) joints under predominantly static loading. CIDECT; 2008.
Couchaux [2] 1561 Flanges 2161 Flanges + bolts
Experimental >2500 Flanges + bolts 2320 Bolts