Professional Documents
Culture Documents
1 s2.0 S0168874X07000157 Main
1 s2.0 S0168874X07000157 Main
www.elsevier.com/locate/finel
Received 27 June 2006; received in revised form 18 December 2006; accepted 30 December 2006
Available online 7 March 2007
Abstract
Managing the axial stiffness of various components in a bolted joint is a major industrial concern for modelling different tightening processes
and for accurate fatigue dimensioning. This paper presents a new approach for calculating the axial stiffness of the several elements of a
bolt (the head and the engaged part), the nut and the fastened plates. Finite element modelling based on deformation energy improves the
existing models and take in to account the type of materials and coefficients of friction of various elements in contact. From these corrections,
approaches for axial stiffness calculation based on empirical formulas are proposed for easier application and for future FE modelling bolted
joints using beam elements. Finally, the theoretical study is validated by an original experimental approach.
䉷 2007 Elsevier B.V. All rights reserved.
Keywords: Bolted joint; Angle-controlled tightening; Axial stiffness; Bolt; Nut; Tapping; Contact; Modelling; Finite elements
Nomenclature
Ad3 thread root area (mm2 ) Kbolt.tapping tapped hole and thread stiffness (N/mm)
Ap equivalent cross-section of fastened parts Lp height of bolted parts (mm)
(mm2 ) L∗p dimensionless height of bolted parts
A∗p dimensionless equivalent cross-section of fas- L0 length of bolt cylindrical part (mm)
tened parts L1 length of bolt threaded part (mm)
As bolt stress area (mm2 ) p pitch of thread (mm)
A0 bolt nominal cross-section (mm2 ) We nut deformation energy induced in the nut
D nut nominal diameter (mm) (N mm)
d bolt nominal diameter (mm) We part deformation energy induced in the bolted
Da diameter under bolt head (mm) parts (N mm)
Dext nut external diameter (mm) bolt correcting factor of the engaged part of the
Dp fastened plates diameter (mm) bolt
Dp∗ fastened plates dimensionless diameter bolt.nut stiffness correcting factor of the nut and the
Dt bolt hole diameter (mm) engaged part of the bolt
Dt∗ bolt hole dimensionless diameter bolt.tapping stiffness correcting factor of a tapped hole
Ebolt bolt modulus of elasticity (N/mm2 ) and the engaged part of the bolt
Ehead bolt head modulus of elasticity (N/mm2 ) head stiffness correcting factor of the bolt head
Enut nut modulus of elasticity (N/mm2 ) threaded stiffness correcting factor of the bolt
Epart fastened plates modulus of elasticity (N/mm2 ) threaded part
Etapping tapped part modulus of elasticity (N/mm2 ) tapping stiffness correcting factor of a tapped hole
Ftot applied axial load (N) 0 displacement due to clamping (on preload-
H nut standardised height (mm) ing F0 ) (mm)
h height of bolt head (mm) coefficient of friction
Kbolt.nut stiffness of the nut and engaged part of the bolt eq equivalent coefficient of friction
(N/mm) 1 coefficient of friction between the bolt and
Khead bolt head stiffness (N/mm) the nut
Kp stiffness of fastened parts (N/mm) 2 coefficient of friction between the nut and
Kp∗ dimensionless stiffness of fastened parts the fastened plates
Ktapping stiffness of a tapped hole (N/mm) tightening angle (degrees)
2. Problem set up displacement along the axis, and consequently the length vari-
ation (see Fig. 2).
2.1. General approach Using the elastic deformation energy for each element, the
equivalent stiffness can be deduced. Applying the principle of
In the aim of developing a new general calculation method, energy conservation to the considered system (see Fig. 2a), and
the overall assembly is partitioned into functional parts: the considering a free unfrictional contact in the rigid plane (x, z):
threaded portion of the bolt, the bolt head, the threads engaged
2 F = WB + Wp + Wf ,
1
(1)
in the nut, the fastened plates and the threads engaged in the
tapped subassembly (see Fig. 1). The local approach consists of where WB is the elastic deformation energy of the bolt, Wp
calculating the local rigidity of each of these functional parts. the elastic deformation energy of the part, Wf the dissipated
It is then extended to a more global approach to calculate the frictional energy at the interface head-part, and 21 F the work
global stiffness of the bolted assembly. This classical approach of the external forces.
is similar to the one used in VDI 2230 recommendation [11]. Consequently, the two cases which can be investigated are
presented below.
2.2. Calculation of the equivalent stiffness
2.2.1. Unfrictional contact under head
2 F = WB + Wp .
Each subassembly is assimilated to a spring with an equiv- 1
(2)
alent stiffness (see Fig. 3). From finite element analysis, one
can conclude (see Fig. 3a) that under loading, the displacement Since the finite elements model show a linear behaviour, the
of the upper part of the screw () is proportional to the ap- stiffness of the springs can be calculated using the following
plied load F [12]. Calculating the stiffness KB and Kp of each relations:
element apart is justified since due to local deformations at F2 F2
KB = and Kp = . (3)
the interface head-part, it is difficult to accurately estimate the 2WB 2Wp
F. Alkatan et al. / Finite Elements in Analysis and Design 43 (2007) 589 – 598 591
fastened plates
0.00
−0.02
−0.04
−0.06
−0.08
−0.10
−0.12
−0.14
−0.16
Y
−0.18
Z X
−0.20
Wf = 21 F − (WB + Wp ). (4)
F F F
Bolt Nut z
θ r Contact
Webolt in contact
Wenut regions
Assembled part
3°
at the bottom of the fixation. This approach is advantageous types of materials, independently from the type of pitch. The
since several dimensional cases can be rapidly investigated. evolution of the ratio As /Aeq with p/d is almost linear. The
coefficient thread refines the equivalent length of the threaded
3. Results and analysis part for fine and large pitch screw. Thus, a valid formulation
for all types of screw is given:
3.1. Stiffness of the bolt’s elements
As p
thread = = 0.956 + 0.534 . (13)
Correcting factors are used to assimilate the threaded part, Aeq d
the nut and tapping to the equivalent length of a cylindrical bar
with cross-sections As and A0 for the bolt head. Fig. 6 shows 3.3. Stiffness of the bolt’s head
the current method [2,16] and the new modified model which
takes into consideration the influence of the threads, the head, Many studies have been conducted on the stiffness of the
and the part in bolt–nut contact. bolts head [7]. In this framework, German rules VDI 2230
The correcting factors are given by [4,11] recommend a corrective factor equal to 0.4 whereas Mas-
sol [7] introduced a ratio head related to the geometry. This
Ebolt · As study assumes perfect adherence and does not take into con-
bolt.nut = , (12a)
d · Kbolt.nut sideration the sliding of the head on the fastened plates and
Etapping · As consequently markedly modifies the zone of load introduction.
tapping = , (12b) Results presented in this paper show a weak influence of the
d · Ktapping
nominal diameter of the bolt and the diameter of the subassem-
Ebolt · As blies. Note that the correcting factor is about 4% greater for a
thread = , (12c) coefficient of friction close to 0. For a standardised head height,
L · Kthread
the influential parameters are the diameter of the tapped hole
Ebolt · A0
head = . (12d) and the subassemblies materials (see Fig. 8).
d · Khead For a coefficient of friction = 0.2 and a standardised head
height (h = 0.65d), the ratio head is given by
3.2. Stiffness of threaded portion of the bolt dt − 1.05d Epart
head = 0.47 + − 0.04 −1 . (14)
d Ebolt
Many studies were conducted concerning the stiffness of the
threaded portion of the bolt (see Fig. 6), which is classically During the tightening process, radial deformation is set free
assimilated to an equivalent section Aeq equal to the resistant as if the “radial” coefficient of friction is equal to zero [12].
section [2,11]. The purpose of this study is to refine this value Thus, the ratio head given by the empirical formula (14) should
by a weighting factor tread which corrects the threaded length be increased about 3% whereas for a head height equal to the
L1 without modifying the defined equivalent section. nominal diameter (optimum bolt head stiffness [12]), this ratio
Using the deformation energy method of several values for must be increased around 6%.
the undimensionless parameter p/d for large and fine pitch
screws (see Fig. 7), brings up results very close to those ob- 3.4. Stiffness of the nut and the engaged part of the bolt
tained in Section 1: the maximum relative variations, for sig-
nificant values of p/d are around 4.5%. It is important to accurately determine the axial stiffness of
Consequently, the traditional use of an equivalent section the nut since it dispatches the entire load through the bolt to
is justified. Practically, identical values are obtained for three the assembly Sawa and Maruyama [17] proposed a value of
a b c
αbolt-nut.d
0.8d or
αbolt-tapping.d
As
L1 As
L1+0.85d αthreaded.L1
Lp
L0 d A0
A0 L0+αhead.d
L0+ 0.4d
0.8d
d d
the bolt.nut factor equal to 0.7 based on theoretical 3D studies The evolution of bolt.nut with the elasticity coefficients of
and experimental results. materials is practically linear (see Fig. 9b). Concerning the
From a finite element results, Fukuoka [18] suggests a value coefficient of friction, the bolt.nut decreases for = 0.23 and
of 1.05 for a coefficient of friction equal to 0 and 0.65 for a then remains constant.
coefficient of friction equal to 0.4. However, he recommends For a bolt and a nut made of the same material, the value
a mean value equal to 0.85. Recently, Guillot [2] proposed a recommended by Guillot is found for = 0 and bolt.nut = 1.1.
value of 1.1d instead of 0.4d (see Fig. 6) to take into account This value can be used in the case of torque tightening, since
the stiffness of the threads. Genelot [19] validated this proposal orthoradial sliding yields to a free distension of the nut and
for low coefficients of friction and large hole diameters. thus to a theoretical coefficient of friction equal to 0 [12].
Note that the German VDI recommendation [4] distinguishes The mean value gbolt.nut = 0.85 recommended by VDI [4]
between the rigidity Kbolt of the portion of the bolt inside the Fukuoka [18] is also found. Finally, these results lead to a
nut and the rigidity Knut of the nut itself: more general empirical formulation that takes into account the
coefficient of friction and the ratio of elasticity coefficients:
Ebolt · Ad3 For 0 0.23:
Kbolt = for bolt = 0.5, (15)
0.5d
bolt.nut = A − B., (17)
Enut · A0
Knut = for nut = 0.4. (16)
0.4d where
For two identical materials, bolt.nut = 0.85. This value is A = 1.1 +
Ebolt
− 1 0.54
equal to the mean value suggested by Fukuoka [18]. Enut
This paper shows a relatively low influence of the nominal
Ebolt
diameter especially for a low coefficient of friction. Thus, a B = 1.65 + − 1 1.2
mean value is considered for the range of diameters between 6 Enut
and 36 mm. For 0.23:
Ebolt
bolt.nut = 0.71 + − 1 0.25. (18)
1.06
Enut
As/Aéq = 0.534*p/d + 0.956 One should know that friction coefficients bolt–nut 1 and
1.04 nut-fastened plates 2 do not have the same influence since
results show that the latter contributes around 60% in the cal-
As/Aéq
Fig. 7. Common function of variation of As /Aeq coefficient according to Numerous studies, based on experimental and digital tests,
p/d report ratio. have been conducted to investigate local behaviour at the thread
t 0.7
formula
0.7 αhead = (Dt-d)/d + 0.43
0.6
αhead
αhead
0.6 0.5
Fig. 8. head ratio in relation to the relative radial clearance (Dt − d)/d and the moduli of elasticity ratio Epart /Ebolt .
F. Alkatan et al. / Finite Elements in Analysis and Design 43 (2007) 589 – 598 595
αbolt-nut
αbolt-nut
1.4
0.8
αbolt-nut = -1.65μ + 1.1 1
global
αbolt-nut = 0.71 Linear (global)
0.6 0.6
0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 2.5 3
μ Ebolt/Enut
Fig. 9. Correcting factor bolt.nut in relation to the coefficient of friction : (a) and the moduli of elasticity ratio Ebolt /Enut ; (b) for a standardised nut.
0.95 Dext=48 mm
α
Dext=72 mm
Dext=120 mm
0.85 0.85
Fig. 10. Correcting factor in relation to the height and external diameter of a non-standardised nut.
root [8,14] to define the fatigue resistance and evaluate the max- The curves presented in Fig. 11 match with the results of the
imum stress taking into consideration the probability of local previous study. One can conclude that the coefficient of friction
plastification. However, the main objective remains the equiv- and the materials are less influential. This can be justified since
alent local stiffness of the portion of the bolt engaged in the the considered part has a large diameter, and thus the radial
tapped hole (see Fig. 5). VDI recommendation [4] and Thomala deformations near the threading are negligible independently
[20] are the only ones to give results for fastened plates. In this from the coefficient of friction and the material.
framework, a preliminary study of nuts with different heights Hence, the following empirical formula is presented:
and diameters showed that the correcting factor remains some- For 0 0.33:
how constant with large height and diameter of tapped parts
(see Fig. 10). Consequently, for stiffness calculation, one can bolt.tapping = A − B., (20)
consider: H = 3Hstandardised and Dext = 5Da . where
One should note that the portion of a bolt in a tapping is a
Ebolt
major constructive datum. However, a recent investigation [12] A = 0.78 + − 1 0.21,
showed that the stiffness of a tapped part barely varies for more Etapping
than seven engaged threads as shown by the results below, set
Ebolt
out from a study in the elastic range. Practically, some threads B = 0.27 + − 1 0.15.
Etapping
may plastify and consequently lead to a more uniform load
distribution: a study covering this phenomenon is actually under For 0.33:
development.
Ebolt
Since in the elastic domain, the coefficient bolt.tapping barely bolt.tapping = 0.7 + − 1 0.16. (21)
Etapping
varies with the nominal diameter, specifically for low coeffi-
cients of friction, it would be accurate to consider the mean Note that the objective of this study is to determine the
diameter. stiffness of the engaged portion of the bolt and the tapped
596 F. Alkatan et al. / Finite Elements in Analysis and Design 43 (2007) 589 – 598
1.2 1.4
1.1
αbolt-tapping = 1.016 1.2
αbolt-tapping = -0.57μ2 + 1.2
1 αbolt-tapping = 0.22Ebolt /Etapping + 0.56
αbolt-tapping
αbolt-tapping
Steel_Aluminium
0.9 1
Steel_Steel
0.8
αbolt-tapping = 0.69 0.8 αbolt-tapping = 0.17Ebolt /Etapping + 0.52
0.7 αbolt-tapping = -0.27μ2 + 0.78 2=0.1 2=0 2=0.3
2=0.2 2=0.4 2=0.5
0.6 0.6
0 0.1 0.2 0.3 0.4 0.5 1 1.5 2 2.5 3
μ2 Ebolt /Etapping
Fig. 11. Correcting factor tapping in relation to the coefficient of friction between threads 2 and the moduli of elasticity ratio Ebolt /Etapping .
Dp
Consequently, the equivalent undimensionless cross-section
is
F
Ap
A∗p = ,
Da2
LP1
where Da is the diameter of the bolt head.
LP LP The various geometrical magnitudes of the bolted joint are
LP2 Dt also defined in relation to Da by the following relations:
Dp Dt Lp
Dp∗ = ; Dt∗ = ; L∗p = . (24)
Da Da Da
Da F Ap
As previously mentioned, numerous studies concerning the
definition of assembled parts equivalent model with equivalent stiffness of fastened plates were conducted. The most
same stiffness
interesting approach is based on the notion of a deformation
Fig. 12. Equivalent calculation model of the fastened plates axial stiffness. cone with the same stiffness as the subassemblies [5,11]. This
description features an assembly of parts with different diam-
eters by a purely geometric approach. Consequently, the angle
part with large dimensions. However, concerning the area be- of the cone, considered as an additional geometric parameter is
tween the standardised nut (Hnut-N ; Dext nut-N ) and the part related to the various characteristics of the assembly [21–23] by
(H ; Dext ), a general model using the previously defined co- various empirical relations. For cylindrical parts with the same
efficients bolt.tapping and bolt.nut and based on the results of diameter, Rasmussen [24] presents an empirical formula (24)
Fig. 11 is used. Eq. (22) represents the generalised model for: which was modified by Massol [2,7] and used by Vadean [16].
bolt.nut < < bolt.tapping . This formula is consistent since it gives satisfactory results for
a coefficient of friction close to 0 (see Fig. 13).
1 0.8d/H
= bolt.tapping +
3 2 − 0.8d/H 1
A∗p = (1 − Dt∗2 ) + (Dp∗2 − 1)
2 1.15d/Dext 4 2
+ (bolt.nut − bolt.tapping ). (22) ⎡ ⎤
3 4 − 3.45d/Dext 0.35 L∗p + 1 + 2L∗2
⎢ p −1 ⎥
× tan−1 ⎣ ⎦, (25)
3.6. Stiffness of fastened plates 2(Dp∗2 − Dt∗2 )
Ap*
1
FEM_μ=0
2
0.8 Rasmussen (25)
Formula (26)
0.6
VDI_03 1
0.4 FEM_μ=0.2
0.2 0
1 2 3 4 5 0 2 4 6 8 10
Dp* Lp*
4. Experimental validation