Anda di halaman 1dari 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267612419

Clamp Load Loss in a Bolted Joint Model With Plastic Bolt Elongation and
Eccentric Service Load

Conference Paper  in  American Society of Mechanical Engineers, Pressure Vessels and Piping Division (Publication) PVP · January 2010
DOI: 10.1115/PVP2010-25813

CITATIONS READS
2 247

4 authors, including:

Sayed Nassar Zhijun Wu


Oakland University Oakland University
138 PUBLICATIONS   710 CITATIONS    20 PUBLICATIONS   71 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A Coupled Shear Stress-Diffusion Model for Adhesively Bonded Single Lap Joints View project

All content following this page was uploaded by Zhijun Wu on 08 December 2015.

The user has requested enhancement of the downloaded file.


Proceedings ofof
Proceedings the ASME
the 2010
ASME Pressure
2010 Vessels
Pressure and
Vessels & Piping
Piping Division
Division // K-PVP
K-PVP Conference
Conference
PVP2010
PVP2010
July 18-22, 2010, Bellevue, Washington, USA
July 18-22, 2010, Bellevue, Washington, USA

PVP2010-25813
PVP2010-25

CLAMP LOAD LOSS IN A BOLTED JOINT MODEL WITH PLASTIC BOLT


ELONGATION AND ECCENTRIC SERVICE LOAD

Xianjie Yang, Sayed Nassar, Zhijun Wu, Aidong Meng


Fastening and Joining Research Institute
Department of Mechanical Engineering
Oakland University
Rochester, MI, 48309, USA

ABSTRACT F2ecr Critical separating force for joint separation under


The nonlinear plastic deformation behavior of a clamped plastic deformation of bolted assembly under elastic
bolted joint model under a separating service load is tightening
investigated using analytical, finite element, and experimental Fecr Critical separating force for joint separation under
techniques. An elastic-plastic model is used for the bolt material
elastic tightening
while the joint material remains in the linear elastic range. Both
the analytical and FEA models investigate the variation in the F1′ecr Critical separating force for joint separation under only
tension of a preloaded bolt, and the corresponding change in the incremental elastic deformation of bolted assembly
joint clamp load, due to a separating service load that is placed under plastic tightening
away from the bolt center. Experimental verification is F2′ecr Critical separating force for joint separation under
provided for both the analytical and finite element results on the incremental plastic deformation of bolted assembly
bolt tension variation, clamp load variation and the clamp load under plastic tightening
loss caused by the incremental plastic bolt elongation under ′
Fecr Critical separating force for joint separation under
cyclic separating force. plastic tightening
Keywords: Bolted Joint Behavior, Yield Tightening, Fe max Peak separating force
Clamp Load Loss.
∆δ b Fastener elongation
NOMENCLATURE ∆δ b′ Fastener elongation at Fe = Fe max
Ab Effective cross-section area of bolt ∆δ b′′ Fastener elongation at Fe = 0 after the first cycle of
Fb Fastener tension separating force
Fb max Maximum fastener tension at peak separating force ∆δ c Joint compression
Fb′′ Residual fastener tension after the first cycle of ∆δ c′ Joint compression at Fe = Fe max
separating force ∆δ c′′ Joint compression at Fe = 0 after the first cycle of
Fc Clamping force separating force
Fc′′ Residual clamping force after the first cycle of Kb Bolt stiffness
separating force K b′ Equivalent bolt stiffness
Fi Initial fastener tension (preload) Kc Joint stiffness
Fe Separating force K w(thick ) Stiffness of thick washer
F1ecr Critical separating force for joint separation under K w(thin ) Stiffness of thin washer
elastic deformation of bolted assembly under elastic
K w(LC ) Stiffness of load cell washer
tightening

1 Copyright © 2010 by ASME


∆Fc′′ Clamp load loss caused by the incremental bolt plastic Recently, Nassar et al. [10-13] investigated the permanent
elongation under cyclic separating force Fe loss of clamp load after the removal of a separating force from a
joint in which the fastener preload was beyond its yield strength,
r Radial location of the bolt cross section
while the joint remained in the elastic range without accounting
L Fastener effective grip length
for the location far off the bolt axis. Variables studied included
E Modulus of elasticity for the fastener material
the joint to fastener stiffness ratio, the fastener preload level and
µt Thread friction coefficient the strain hardening of the fastener material. In the previous
σ Axial stress at the bolt cross-sectional area studies [10-13], the separating force was assumed not to lead to
σb Bolt nominal stress the separation of the joint members. However, the effect of
σs Bolt nominal initial yield stress tightening on the yield behavior of the bolt and clamp load loss
was not taken into account.
∆ε Total bolt strain
In this paper, an elastic-plastic model of bolt is proposed
∆ε p Bolt plastic strain for investigating the effect of the yield tightening on the joint
system behavior under service separating loads. Experimental,
1. INTRODUCTION analytical, and finite element methods are used for investigating
Bolted joints are often subjected to various forms of the nonlinear plastic behavior of the bolted joint under cyclic
service separating loads that may be static or dynamic in nature. separating tensile forces. The effect of the separating force level,
Both the type and the location of the separating load would thread friction coefficient, and the bolt preload level on the
normally affect the tension increase in the bolt and the variations of the clamp load, bolt tension and clamp load loss
corresponding reduction in the joint clamp load. When the are studied. Comparison between the experimental, analytical
bolted joint is tightened, the deformation behavior of the bolted and finite element results is provided.
joint system may be modeled as that of a one dimensional
spring. However, when the bolted joint is subsequently 2. DEVELOPMENT OF BOLT MATERIAL PLASTIC
subjected to an external separating force Fe , the joint MODEL AND EXPERIMENTAL VERIFICATION
deformation behavior is quite different and it will depend on the 2.1 Bolt Material Plastic Model
form, level, and location of the separating load. Even under a The following deformation assumptions of bolt are made:
pure separating tensile force, the plate bending and shear force (1) The normal strain ε due to bolt elongation is uniform over
distribution in a single-bolt joint will significantly affect the the entire bolt cross-section area; that is,
changes in the bolt tension and joint clamp load in a nonlinear ε = ε (r ) = ε 0 (1)
fashion [1-5]. where r is the radial location of the bolt cross section from
Kwiathowski et al. [6] and Lehnhoff et al. [7, 8] used finite the center in Fig.1.
element analysis and experimental methods to show that, in (2) The shear strain γ due to bolt twist is proportional to r ;
contrast with the conventional elastic theory, the joint stiffness
and the change in bolt tension are strongly dependent on the that is,
r∆θ b
magnitude of the external load. The external separating load γ = γ (r ) = = rφ (2)
causes an additional bending deformation that not only it L
reduces the joint stiffness, but also affects the separation of the where L is the bolt length, φ is the twist angle per unit bolt
joint. length, and ∆θ b is the twisting angle along the bolt length L .
The traditional linear elastic model [2] for the changes in For the plastic tightening of the bolt, the Von Mises yield
the bolt tension and the clamp load under separating loading criterion is
can be used just when the separating force is applied along the
bolt center to the bolt head or nut. In the literature [9], σ 2 + 3τ 2 = f 2 ( p ) (3)
experimental and finite element methods are used for where σ is axial stress and τ is the shear stress over cross-
investigating the nonlinear elastic behavior of the bolted joint section area, and p is the accumulated plastic strain. The
under cyclic separating tensile forces. The effect of the effective stress (Von Mises) is function of the radial location r .
separating force level, and location as well as the bolt preload Initial yielding occurs on bolt surface; subsequent yielding
level on the variations of the clamp load and bolt tension are due to increased loading will propagate inward towards the
studied. From the modeling, the separating force at the center of the bolt cross-section. If the radius at the
location out of the bolt axis will lead to the large decrease of the elastic/plastic boundary is rc as shown in Fig 1 and the initial
clamping force and the small increase of the bolt tension when
yield stress is σ s = f (0) , the stress condition at r = rc is
the separating force is relatively smaller than the bolt preload as
the bending effect of the joint members is too significant to be σ 2 (rc ) + 3τ 2 (rc ) = σ s2 (4)
ignored. The bolt-joint assembly cannot be simplified as a 1-D In the elastic region of cross-section ( 0 ≤ r ≤ rc ), Hooke’s
spring group just like the model shown in [2]. law may be expressed as follows

2 Copyright © 2010 by ASME


dσ = Edε (5) clamped single bolted joint as shown in Fig. 3. A high-precision
dτ = Gdγ (6) digital torque wrench is used for tightening the bolt to the
where E and G are, respectively, the Young’s modulus and desired preload level; a force washer (load cell) is used for
elastic shear modulus of the bolt material, dσ and dε are monitoring the bolt preload/clamp load. In the tests, two thick
steel washers matching the fixture and a thin steel washer are
the normal stress and strain increments, respectively, dτ
used as shown in Fig. 3. The thin washer has 13mm of internal
and dγ are the increments of shear stress and strain,
diameter, 25.4mm of external diameter and 1.94 mm of
respectively. The normal stress σ on the cross section of bolt thickness, and the other two are the thick one with 13mm of
is schematically shown in Fig.1. internal diameter, 25.4mm of external diameter and 9.54mm of
In the plastic region on the cross-section ( rc ≤ r ≤ a ), the thickness. Two kinds of bolts and nuts are used. One is made of
incremental plastic flow rule gives the incremental stress steel and another is made of aluminum alloy. The bolt and nut
components in the bolt as follows size is 1/2"-13. For steel bolt, its material grade is SAE Grade 5.
9G 2τ 2 (ri )dγ (ri ) + 3EGσ (ri )τ (ri )dε The tightening tests for the determination of the thread friction
dτ (ri ) = Gdγ (ri ) − (7) coefficient and the bolt tension-elongation under tightening
f ′( pi )σ eq
2
(ri ) + Eσ 2 (ri ) + 9Gτ 2 (ri ) were separately conducted using RS test machine. For the steel
E 2σ 2 (ri )dε + 3EGσ (ri )τ (ri )dγ (ri ) and aluminum alloy bolts, the detail on the mechanical
dσ (ri ) = Edε − (8) properties will be given in the next section.
f ′( pi )σ eq
2
(ri ) + Eσ 2 (ri ) + 9Gτ 2 (ri )
During bolt tightening, the collected experimental data
dσ (ri ) include the tightening torque T , bolt tension Fb , reaction
dε p (ri ) = dε − (9)
E (thread) torque Ttp , and the angle of turn θ . The angle of turn
τ (ri )
dγ p (ri ) = dγ (ri ) −
d
(10) θ includes both the bolt deformation angle and the relative
G angle of turn ∆θ bn between the engaged bolt and nut threads.
[dγ p (ri )]2 In the linear segment of the tension-turn ( Fb − θ ), the bolt
dp (ri ) = [dε p (ri )]2 + (11)
3 assembly can be assumed to exhibit elastic deformation only;
where dε p (ri ) is the normal plastic strain increment at the accordingly, the following equation may be utilized:
Ttp L
radius ri , dγ p (ri ) is the corresponding shear plastic strain ∆θ bn = θ − (15)
GJ p
increment, and dp (ri ) represents the equivalent plastic strain
increment. where G is elastic shear model of the bolt material, and J p
Using the numerical trapezoid integration, the bolt tension is the polar moment of inertia for the bolt cross-sectional area,
Fb and the twist torque Ttp are given by and L is the effective grip length of bolt.
The bolt elongation δ b is then given by
a
 n 

Fb = 2πrσ (r )dr = Eεπrc2 + πt rc Eε + aσ (a) + 2 riσ (ri )
 
∑ δb =
Fb ∆θ bn
=
F
p− b (16)
0 i=2 Kb 2π Kc
(12) where p is pitch length, K b is the bolt stiffness and K c is
a  n 
∑ the joint stiffness.
1

0

Ttp = 2πr 2τ (r )dr = Gφπrc4 + πt Gφrc3 + a 2τ (a ) + 2
2 
 i =2
ri2τ (ri )


During bolt tightening, the relationship between the bolt
tension and thread torque (reaction torque) is given by [14] as
(13) follows
t = (a − rc ) / n (14)
 p µr 
where t is the numerical integration step length, and n is the Ttp =  + t t  Fb (17)
 2π cos α 
number of step.
The nominal axial stress σ b in the bolt is simply equal to
2.2 Experimental Verification of Plastic Model the bolt tension Fb divided by the equivalent bolt cross-
The proposed plastic model in the paper should be sectional area Ab , and the bolt axial strain is ε . The
calibrated and be verified experimentally before it is used in the
experimental curve Fb − θ can be transferred as σ b − ε
bolt application. The experimental study by using MTS
machine and RS tightening machine will be conducted for the curve. On the other hand, the corresponding theoretical
experimental calibration and verification. predicted σ b − ε curves can be obtained by Eqs.(1-17) for the
The bolted joint assembly under eccentric separating force bolt under tightening and the ones under separating force. Figs.
Fe is schematically shown in Fig.2. A MTS Tensile testing 4-5 give the corresponding experimental and predicted curves.
machine is used to apply a cyclic separating force Fe to the In Figs. 4 and 5, “Exp” means experimental results, “ANA”

3 Copyright © 2010 by ASME


means the analytical results under pure tension or plastic tightening for thread friction coefficient µt = 0.49
tightening based on Eqs.(1) to (19), “SIM” means the simulated and µt = 0.286 . The predicted curves of the bolts under
results under plastic tightening and then separating loading
separating force are a little higher than those for pure tension.
based on Eqs.(1) to (19), the digital number means the thread
The higher the thread friction coefficient, the higher the
friction coefficient during tightening, and “SEP” means that
difference between the model and experimental results. When
under separating force.
the additional plastic deformation is introduced in the bolt, the
Figure 4 displays the experimental nominal axial stress vs.
difference diminishes to a near zero value.
nominal axial strain curves of aluminum bolt under various
When the thread friction coefficient is not high, the plastic
loading scenarios; namely, the pure tensile, bolt tightening and
the subsequent separating loading after bolt tightening with σ b − ε results with a separating force are nearly the same as
different friction coefficients. In the deformation range the axial those for pure tension. In this paper, the analytical model for the
bolt stress level σ b in pure tension is higher than that from bolt clamp load loss caused by the plastic deformation of the bolt is
tightening with non-zero thread friction coefficient; it is also based on an initial σ b − ε curve from tightening the bolt
shown that higher friction coefficient leads to lower plastic bolt followed by an additional σ b − ε curve segment under
nominal axial stress level at the same strain under bolt separating force Fe . The latter part is represented by the
tightening.
Fb
Figure 5 shows that for the experimental results in the function ∆ε bp = f1 ( ) = ε − ( Fb / Ab ) / E , which is dependent
plastic deformation range the nominal axial stress of the steel Ab
bolt with the thread frictional coefficient µt = 0.09 is higher on the thread friction coefficient µt and the bolt preload Fi .
than that in pure tension. As higher thread friction coefficient The σ b − ε correlation during the bolt torquing represented by
the function ∆ε bp = f1t (Fb / Ab ) = ε − ( Fb / Ab ) / E which is
leads to higher twist torque which is equal to the thread torque
in Eq. (17), the twist torque leads to lower nominal axial bolt
stress under plastic tightening than that in pure tension. This also dependent on the thread friction coefficient.
means that the twist torque has effect on the bolt axial stress Figure 6 displays the separating force Fe versus the
responses, which is dependent on the friction coefficient µt . displacement δ responses of aluminum bolt assembly under
Figures 4 and 5 show that the plastic strain of the bolt due different separating force levels of 10000N, 24000N and
to tightening will lead to the plastic strain hardening of the bolt 28000N for a bolt preload of 18000N. Results indicate that
material. During plastic tightening, higher thread friction when the separating force is 10000N, the structural system
coefficient will lead to larger plastic strain in the bolt than that stiffness is about 337350 N/mm. As the separating force is
produced by pure tension of bolt for the same desired bolt increased to Fe > 19000N, the complete separation of joint
tension. After initial plastic tightening, subsequent yield stress plates leads to a tangent stiffness value that is equal to the axial
improves significantly, depending on the amount of plastic bolt stiffness. When the bolt undergoes the significant plastic
strain in the bolt. The subsequent removal of the tightening deformation due to a large separating force of 28000N, the
torque causes the bolt twist torque to drop to an insignificant nonlinear plastic deformation of the bolt leads to the nonlinear
level for most practical applications. Subsequent application of Fe − δ curves as Fe is increased. As Fe is decreased in the
an external separating force Fe may lead to the increase of the range 15600 N ≤ Fe ≤ 28000 N (while the joint parts are still
bolt tension. However, only if the bolt axial stress is higher than separated) in the next 1/2 cycle, the slope dFe / dδ will be the
the yield stress hardened at the previous plastic tightening, the
new plastic deformation of the bolt occurs. This is very same as the elastic bolt stiffness for Fe ≥ 15600 N . As the
important for the determination of the clamp load loss under separating force is further reduced in the range
separating force. 0 ≤ Fe ≤ 15000 N , joint clamping is restored (though at a level
Based on uniaxial tensile testing of steel and aluminum that is lower than the initial preload), the slope dFe / dδ will
bolts used in this study, curve fitting the data gives the be the same as the initial structural system stiffness 337350
following correlation between the equivalent stress and N/mm.
equivalent plastic strain as follows
σ eq = 637.5 + 110[1 − e −1500 p ] + 8000 p for steel bolt (18) 3. ANALYSIS OF CLAMP LOAD LOSS
3.1 Scenario I: Separating Force Causing Plastic Bolt
σ eq = 180 + 60[1 − e −300 p ] + 1300 p for aluminum bolt (19)
Deformation after Elastic Tightening
where σ eq is the Von Mises stress. On the one hand, subsequent to an elastic tightening of the
bolted joint, if the separating force Fe is applied to joints
The analytical results from the plastic model developed
(Eqs. (1)-(19)) are shown in Figs.4-5. The figures show the without causing joint separation, the deformation compatibility
nominal bolt stress versus axial strain curves for the plastic condition is
∆δ b + ∆δ c = 0 (20)

4 Copyright © 2010 by ASME


where ∆δ b is the additional bolt elongation and ∆δ c is the As the bolt tension and the clamping load are not zero at Fe = 0 ,
decrease in joint compression (spring-back). the deformation compatibility condition given by Eq.(20)
Based on Eq.(20) and the axial force equilibrium condition becomes
Fb − Fc = Fe under the separating force, a critical value F F − Fb′′ Fi − Fb′′ 1 1
f1 ( b max ) L = i + = ∆Fc′′( + ) (29)
F1ecr of the separating force that would cause elastic joint Ab K b′ Kc K b′ K c
separation is given by Should the joint open (separates) under Fe max , the bolt
K + K b′ tension would be equal to the separating force ( Fb max = Fe max ).
F1ecr = c Fi (21)
Kc The clamp load reduction ∆Fc′′ , after the cyclic separating
where Fi is the initial bolt preload and K b′ is the equivalent force Fe has been removed, is given by
axial bolt stiffness which is given by Fb max
1 f1 ( )L
K b′ = (22) Ab
1 1 2 1 ∆Fc′′ = (30)
+ + + 1 1
K b K w( LC ) K w(thick ) K w(thin ) ( + )
K b′ K c
where K w(LC ) denotes the axial stiffness of load cell, and Should the additional plastic elongation in the bolt caused
K w(thick ) means the axial stiffness of the thick washer and by Fe max be very high and the joint remain open after the
K w(thin ) presents the one of the thin washer. separating force has been removed, the clamp load loss ∆Fc′′
On the other hand, if joint separation occurs under the would be equal to the bolt initial bolt even if ∆Fc′′ determined
plastic deformation of the bolt caused by the separating force, by Eq.(30) may be larger than Fi .
the critical separating force F2 ecr is given by
F2ecr F  1 1  3.2 Scenario II: Separating Force after Plastic Bolt
) L + 2ecr = 
f1 ( +  Fi (23) Tightening
Ab K b′ K
 b ′ K c  At the end of plastic bolt tightening while the joint remains
When the separating force reaches the peak value Fe max in the elastic range, the bolt and joint are respectively given by
without joint separation ( Fe max ≤ Fecr ) and with the joint F F
δ b = f1t ( i ) L + i (31)
remaining in the elastic deformation range while the bolt has Ab K b′
plastic deformation, the additional bolt elongation ∆δ b′ and Fi
δc = − (32)
the corresponding change in joint deformation ∆δ c′ are Kc
respectively given by As shown in Figs. 4 and 5, under the subsequent application of
Fb max F − Fi a separating force Fe , the new yield tension will be
∆δ b′ = f1 ( ) L + b max (24)
Ab K b′ increased from Fi to FNBy , which depends on the preload
( Fc min − Fi ) 2 (F − Fe max − Fi ) 2 Fi , the thread friction coefficient µt , the plastic deformation
∆δ c′ = − = − b max (25)
K c Fi K c Fi behavior of the bolt material and the bolt geometry as shown in
Based on the deformation compatibility condition, Eq.(20) the proposed plastic model of the bolt in section 2.1. From Figs.
becomes 4 and 5, the bolt will have just new elastic deformation when
the bolt force Fb under separating force Fe is smaller than
Fb max F − Fi ( Fb max − Fe max − Fi ) 2
f1 ( ) L + b max = (26) FNBy while the bolt will have new plastic deformation when
Ab K b′ K c Fi
For a cyclic load Fe , as the separating force decreases from the bolt force Fb under separating force Fe is higher
the peak Fe max to zero with the joint remaining clamped, the than FNBy .
new clamp load is Fc′′ and the bolt tension is Fb′′ (= Fc′′) , the Two possibilities (cases) are discussed for the joint
separation under the separating force Fe . When joint separation
following relationships are valid
Fb max − Fb′′ F F − Fi Fb′′ − Fb max occurs, Case 1 is for the just new elastic deformation of the bolt
∆δ b′′ = ∆δ b′ − = f1 ( b max ) L + b max + while Case 2 is for the new plastic deformation of the bolt.
Kb′ Ab Kb ′ K b′
For Case 1 ( F1′ecr < FBNy ), the joint separation will occur
(27)
F ′′− Fi before the bolt has new plastic deformation under separating
∆δ c′′ = b (28) force. The value F1′ecr of the separating force that would cause
Kc
joint separation would be determined by the following equation

5 Copyright © 2010 by ASME


F1′ecr  1 1  F Fb max F
) − f1t ( i )]L
=  +  Fi + f1t ( i ) L (33) [ f1 (
K b′ K
 b ′ K c  A b ∆Fc′′ =
Ab Ab
(41)
1 1
For Case 2, the bolt separation will occur after the value ( + )
F2′ecr of the separating force that would cause joint separation K b′ K c
would be determined by
4. ANALYTICAL RESULTS AND DISCUSSION
F′ F′  1 1  F
f1 ( 2ecr ) L + 2ecr =  +  Fi + f1t ( i ) L (34) The proposed model for clamp load loss is used to discuss
Ab K b′ K
 b ′ K c  Ab the effect of the preload level, the thread friction coefficient and
When the separating force reaches its peak the separating force level on the clamp load loss in this section.
′ ) without causing joint separation, the
Fe max ( Fe max ≤ Fecr For an aluminum bolt, Figure 7 shows that under the cyclic
incremental bolt tension elongation ∆δ b′ and the incremental separating force Fe = 14000 − 14000 cos(0.4t ) N, the preload
joint compression ∆δ c′ are respectively given by Fi = 18000N which is exceed the bolt yield, the thread friction
coefficient µt = 0.38 , after the first cycle, a clamp load loss of
 F F  F − Fi
∆δ b′ =  f1 ( b max ) − f1t ( i ) L + b max (35) 2700 N is observed; then the bolt tension variation remains
 Ab Ab  Kb ′
identical during subsequent cycles. This means that the axial
( Fi − Fc′ ) 2 ( F − Fb max + Fe max ) 2 plastic deformation of the bolt in the first cycle leads to the
∆δ c′ = − =− i (36) clamp load loss. The nominal yield stress of the bolt is
K c Fi K c Fi
increased from the initial yield stress 180.0MPa to the present
Deformation compatibility at Fe = Fe max yields yield stress 203MPa in the first cycle due to the plastic strain
 Fb max F  F − Fi (Fi − Fb max + Fe max )2 hardening. In all subsequent cycles, the bolt behaves elastically
 f1 ( ) − f1t ( i ) L + b max = and there is no further clamp loss beyond that of the first cycle.
 Ab Ab  K b′ K c Fi
Figure 7 shows a good correlation between the analytical model
(37) and the experimental results. Figure 8 displays the
The corresponding maximum bolt tension Fb max is corresponding experimental and analytical variation in bolt
determined from Eq. (37). tension (BT) and clamp load (CL) versus the cyclic separating
As the separating force is decreased from its peak level force as shown in Fig.7.
Fe max to zero, and as long as here is a non-zero residual Similarly, Figs. 9 and 10 give the corresponding
clamping in the joint at Fe = 0 , the further incremental bolt experimental and analytical results on the variation in bolt
tension and clamp load for a steel bolted assembly. Although
elongation ∆δ b′′ and the incremental joint deformation ∆δ c′′ the model and the experimental results seem to be in reasonable
would be given by agreement, the difference is more sensitive to the stress vs.
 F F  F − Fi Fe max − Fb′′ strain relationship in the range for the steel bolt.
∆δ b′′ =  f1 ( e max ) − f1t ( i ) L + e max − (38)
 Ab Ab  ′
Kb K b′
4.1 Effect of the Preload Level on Clamp Load Loss
Fb′′ − Fi Figure 11 shows the clamp load loss ∆Fc vs. the separating
∆δ c′′ = (39)
Kc force Fe for six different levels of bolt preload; namely,
The corresponding deformation compatibility condition leads to 30000 N, 50000 N, 55000 N, 60000 N, 65000N and 70000N,
F F F − Fb′′ Fi − Fb′′ 1 1 under a cyclic separating force of 37500 − 37500 cos(0.4t ) N
[ f1 ( b max ) − f1t ( i )]L = i + = ∆Fc′′( + )
Ab Ab K b′ Kc K b′ K c
for µt = 0.09 . The curves for Fi = 30000N, 50000N and
(40)
55000N are the nearly same. Obviously, the force Fe that
It must be emphasized again that, if the maximum
separating force Fe max leads to the complete opening causes the complete loss of clamp load would be different for
the different preload levels, because the bolt just has elastic
(separation) of the joint, the corresponding maximum bolt force
tightening for three preload levels even considering tightening
must be equal to the maximum separating force, therefore,
twist effect. On the other hand, if Fi = 60000,65000 and
Fb max = Fe max . Furthermore, should the additional plastic
70000N, the plastic tightening for the three preloads will lead to
elongation in the bolt due to Fe have caused the joint remain three different plastic strain values of the bolt, which are
loose after the removal of the separating force ( Fe = 0 ), the dependent on Fi and µt . The plastic strain will increase the
corresponding clamp load loss ∆Fc′′ would be given by yield stress of the bolt for the subsequent separating load. From
Fig. 11, the significant clamp load loss for higher preload
should need much higher Fe for the three high preloads.

6 Copyright © 2010 by ASME


Finally, the above results shows if the preload load level are 210GPa and 71GPa, respectively. The elastic Poisson ratios
can be reached by elastic tightening, the preload level will have for steel and aluminum alloy are 0.3 and 0.33, respectively.
no significant effect on the clamp load loss under separating Contact conditions are used at all interfaces between the
force. If the desired preload has been achieved by plastic nut/washer, bolt/washer, washer/load cell, joint/joint and
tightening, however, the increase of the preload level will joint/washer. The contact conditions are defined as a tangential
significantly reduce the clamp load loss caused by a known isotropic contact property with a friction coefficient of 0.2 at all
level of separating force. interfaces. The twist torque Ttp (during tightening) is applied
to the bolt through the shear stress on the top bolt head surface
4.2 Effect of the Thread Friction Coefficient on Clamp Load
and the displacement restraint condition on the nut. The
Loss
FEA model used 8-node linear brick elements with reduced
Figure 12 shows the clamp load loss ∆Fc vs. separating integration elements (C3D8R).
force Fe for six levels of thread friction coefficient µt for a Figure 14 displays FEA and experimental results on the
fixed preload of 50000N under a cyclic separating force Fe = variation of bolt tension and clamp load for the bolted assembly
37500 − 37500 cos(0.4t ) N. The ∆Fc − Fe responses for with an aluminum bolt, under a separating force Fe =
µt = 0.09 , 0.12 and 0.20 are nearly the same. The clamp load 14000 − 14000 cos(0.4t ) N and a bolt preload Fi = 18000N.
losses for µt ≥ 0.3 are significantly different from those for The clamp load loss based on FEA is 2600N as compared to
2700N from the experimental data as shown in Fig.8.
µt = 0.09 , 0.12 and 0.20. As the thread friction coefficient Similarly, Figure 15 gives FEA results for steel bolt under
increases, reaching the same preload by torquing would use the separating force Fe = of 30000 − 30000 cos(0.4t ) N with
higher thread torque. Therefore, if the preload is constant, when a bolt preload of 33500 N. The FEA results for the bolt tension
µt is high enough to make the plastic tightening, the increase and clamp load loss are in good agreement with the
of µt will improve the resistance to the clamp load loss after corresponding experimental ones.
the removal of the external load Fe . If, on the other hand, the Figure 16 shows the FEA and experimental results for a
preload is reached by elastic tightening, the thread friction 61500N preload and µt = 0.20 under the cyclic separating
coefficient will have no effect on the clamp load loss. force Fe = 35000 − 35000 cos(0.4t ) N. The analytical and
experimental results are in good agreement in predicting a
4.3 Effect of the Separating Force Level on Clamp Load 1500N clamp load loss, while the FEA model results predicting
Loss a 4500N clamp load loss has a little difference from the
Figures 11 and 12 show that the clamp load loss vs. experimental one because the over-simplified FEA geometry on
separating force, in which the clamp load loss is dependent on the corner of bolt underhead in the paper leads to a little higher
the yield stress in the bolt, which is dependent on the thread stress concentration than the real one, which leads to a little
friction coefficient and the preload level. When the cyclic peak higher bolt plastic deformation.
of the separating force is higher than the critical separating Figures 17-18 display the experimental and FEA
force Fe , that would begin to cause the clamp load loss upon the displacement responses at the two ends of the two bolt fixtures
under two cases of the cyclic separating forces
subsequent removal of the separating force, the higher the peak,
Fe = 14000 − 14000 cos(0.4t ) N and 12000 − 12000 cos(0.4t )
the larger the clamp load loss. If the clamp load loss is equal to
the full preload value, the joint becomes completely loose after N, respectively, for a bolt preload Fi = 18000N of aluminum
the full removal of the separating force. From Figures 11 and 12, bolt. Fig.17 shows that when the displacement is smaller than
the increases of the separating force higher than the critical one 0.05mm, the relationship between the separating force and the
will lead to significant clamp load loss upon the separating displacement is linear; when the separating force leads to
force removal. separation of the joints and the plastic deformation of the bolt,
the nonlinear hysteresis loop of the Fe − δ responses can be
5. FINITE ELEMENT SIMULATION observed under either experimental or FEA results. On the other
A static 3-D finite element model is constructed using hand, after the first cycle of the cyclic separating force, the
ABAQUS/Standard 6.7 commercial software. The FEA model Fe − δ curve can be considered as two main segments: one is
includes the joint, bolt, washers and the load cell as shown in
for the elastic stiffness of the bolted assembly without
Fig.13. Both the bolt and nut are modeled as one part in order
separation of the joints, and another is for the bolt stiffness with
to apply bolt load to the fastener section. In the FEA model, the
the separation of the joints. The bolted assembly has just
bolt is assumed to be elastic-plastic, and the other components
elastic deformation as the plastic strain of the bolt increases the
remain in the linear elastic range based on the experimental set-
bolt yield stress from the initial yield stress 180MPa to the
up. The elastic Young’s moduli for steel and aluminum alloy
maximum bolt nominal stress 228.16 MPa at Fe max = 28 000N

7 Copyright © 2010 by ASME


in the first cycle. From Figure 18, when the peak separating [5] Sawa, T., 2006, “New design formula for bolted joints
force is 24000N, the Fe − δ responses has very little hysteresis under tensile loads,” SAE 2006 World Congress, 2006-01-
loop, this can be approximately considered as the elastic 0989, April 3-7, 2006, Detroit, Michigan, USA.
deformation of the bolt. From Figs.17-18, the Fe − δ [6] Kwiathowski, J. K., Winnicki, L. A., and Krzyspiak, A.,
1986, “Stress analysis of bolted tensile end plate
responses from finite element simulation are in good agreement
connections,” Rozprawy Inzynierskie Engineering Trans.,
with the experimental ones. This means that the finite element
34(1-2), pp.113-117.
method can characterize the deformation behavior of the bolted
[7] Lehnhoff, T. F., Ko, K. I., and Mckay, M. L., 1994,
assembly caused by the plastic deformation of the aluminum
“Member stiffness and contact pressure distribution of
bolt.
bolted joints,” ASME J. Mech. Des., 116, pp.550-557.
[8] Lehnhoff, T. F., and Bunyard, B. A., 2001, “Effects of Bolt
6. CONCLUSIONS
Threads on the Stiffness of Bolted Joints,” ASME J.
This study provides an experimental, analytical and finite
Pressure Vessel Technol., 123, pp.161-165.
element investigation of the behavior of a bolted joint model
[9] Nassar, S. A., Xianjie Yang, Satya Vijay Teja Gandham,
due to a cyclic separating tensile load that is placed out off the
and Zhijun Wu, “Nonlinear Deformation Behavior of
bolt axis. Results are presented and discussed in terms of the
Clamped Bolted Joints under a Separating Service Load”,
increase in the bolt tension, the reduction in the clamp load, and
ASME Journal of Pressure Vessels Technology, 2010, in
the clamp load loss caused by the plastic elongation of the bolt
press.
under separating force. The comparison among the analytical,
[10] Nassar, S. A., and Shoberg, R. S., 1992, “Effect of Fastener
FEA results and the experimental results verified that the
Tightening Beyond Yield on the Behavior of Bolted Joints
proposed analytical model and the finite element method using
under Service Loads”, Proceeding of the 18th Annual
ABAQUS can predict the bolt tension, clamp load variation and
Conference of AMSE, Michigan State University, East
clamp load loss very well. Analysis of the results leads to the
Lansing, USA, pp.1-17
following conclusions:
[11] Nassar, Sayed A., and Martin Payam H., 2004, “Fastener
(1). After tightening, the experimental observation shows that
Tightening Beyond Yield”, Proceedings of the ASME
the further incremental plastic bolt elongation caused by the
Pressure Vessels and Piping Conference, July 25-29, 2004,
cyclic separating force in the first cycle will lead to the clamp
San Diego, California
load loss. The clamp load loss is proportional to the incremental
[12] Nassar, Sayed A., and Martin, Payam H., 2006, “Clamp
plastic bolt elongation during the first cycle. The increase of the
Load Loss due to Fastener Elongation Beyond Its Elastic
preload level can reduce the clamp load loss under the same
Limit”, ASME Journal of Pressure Vessels Technology, 28,
cyclic separating force level only if the tightening leads to the
pp.379-387
plastic strain of bolt for the preload.
[13] Nassar, Sayed, A., and Matin, Payam, H., 2006, “Nonlinear
(2). If the preload is just achieved only by the plastic tightening,
Strain Hardening Model for Predicting Clamp Load Loss in
the larger plastic deformation of the bolt caused by larger thread
Bolted Joints”, ASME Journal of Mechanical Design,
friction coefficient under tightening will increase the resistance
128, pp.1328-1336
of the clamp load loss under cyclic separating force at the same
[14] Motosh, N., 1975, “Load Distribution on Threads of
preload.
Titanium Tension Nuts and Steel Bolts,” ASME J. Eng.
(3). The proposed analytical model and the finite element
Ind., 97(1), pp.162-166
method can characterize the plastic deformation behavior of the
bolted assembly, and experimental verification shows that they
can predict the clamp load loss under cyclic separating force
quite accurately.

REFERENCES
[1] Sawa, T., and Maruyama, K., 1977, “Force ratio of bolted
joints: The case where clamped parts are hollow cylinders,”
Bulletin of the JSME, 20 (141), Paper No 141-15.
[2] Juvinall, R.C., and Marshek, K, M., 2000, Fundamentals of
Machine Component Design, 3rd ed., Wiley, New York.
[3] Grosse, R. I., and Mitchell, L. D., 1990, “Nonlinear axial
stiffness characteristics of axisymmetric bolted joints,”
ASME Journal of Mechanical Design, 112, pp.442-449.
[4] Zhang, O., 2004, “New Analytical Model of Bolted Joints,”
ASME Journal of Mechanical Design, 126, pp.721-728.

8 Copyright © 2010 by ASME


300

Plastic Zone 250


a

Nominal Bolt Tension Stress (MPa)


σσ 200

µt
Elastic Zone 150
Exp-Tightening-0.49
Exp-Pure tension
100 SIM-0.49SEP

r ANA-pure tension
ANA-0.49 Tightening
rc 50 SIM-0.286SEP
ANA-0.286Tightening

0
0.000 0.005 0.010 0.015 0.020 0.025 0.030
Bolt Axial Strain

Fig.4 Nominal bolt tension stress versus bolt axial strain for aluminum
Fig.1 Schematic of the elastic and plastic deformation zones within the alloy fasteners under tightening and separating force
bolt cross section
900

Nominal Bolt Tension Stress (MPa) 800

700
µt
600
Exp-Tightening-0.09
500
Exp-Pure Tension
400
Fitting-Pure tension
300 ANA-Tightening-0.09

200 SIM-0.09-SEP

100

0
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014
Bolt Axial Strain
Fig.5 Nominal bolt tension stress versus bolt axial strain for steel alloy
fasteners under tightening and separating force
30000

Fig.2 Bolted Joint Schematic


25000
Separating Force (N)

20000

15000
Fe max

10000 24000N
28000N

5000 10000N

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Displacem ent (m m )
Fig.3 Experimental Setup
Fig.6 Separating force versus the overall tensile displacement
responses of aluminum bolt assembly for a bolt preload Fi = 18000N

9 Copyright © 2010 by ASME


30000 70000

28000
60000 Exp-BT
26000
ANA-BT

Bolt Tension, Clamp Load (N)


24000 50000 Exp-CL
Bolt Tension (N)

22000 ANA-CL
40000
20000

18000 30000

16000
20000
14000
Exp
12000 ANA 10000

10000
0
10 15 20 25 30 35 40
0 10000 20000 30000 40000 50000 60000 70000
Time (s)
Separating Force (N)
Fig.7 Bolt tension variation due to cyclic separating force Fe for Fig. 10 Bolt tension and clamp load variation for experimental and
aluminum alloy assembly ( Fi = 18000N, analytical results of aluminum alloy assembly due to cyclic separating
Fe = 14000 − 14000 cos(0.4t ) N) force Fe = 30000 − 30000 cos(0.4t ) N
30000 70000
Preload Level
60000
25000
Fi=30kN
Bolt Tension, Clamp Load (N)

Clamp Load Loss (N)

50000
Fi=55kN
20000
40000 Fi=60kN

15000
Fi=70kN
30000
Exp-BT Fi=50kN
ANA-BT Fi=65kN
10000 20000
Exp-CL
ANA-CL 10000
5000

0
0 0 20000 40000 60000 80000
0 5000 10000 15000 20000 25000 30000
Separating Force (N)
Separating force (N)
Fig.11 Effect of the preload level on the clamp load loss due to cyclic
Fig. 8 Bolt tension and clamp load variation for experimental and
separating force for thread friction coefficient µt =0.09
analytical results of aluminum alloy assembly due to cyclic separating
force Fe = 14000 − 14000 cos(0.4t ) N
70000
60000
Thread friction coefficient
60000 50000
ut=0.4
Clamp Load Loss (N)

50000 40000 ut=0.09


Bolt Tension (N)

ut=0.20
40000
30000
ut=0.5
30000 ut=0.3
20000
ut=0.12
20000 Exp
Ana 10000
10000

0
0 0 20000 40000 60000 80000
0 2 4 6 8 10 12 14 16
Separating Force (N)
Tim e (s)
Fig.9 Bolt tension variation due to cyclic separating force Fe for Fig.12 Effect of the thread friction coefficient on the clamp load loss
due to cyclic separating force for a bolt preload Fi = 50000N
steel alloy assembly ( Fi = 33600N, Fe = 30000 − 30000 cos(0.4t ) N)

10 Copyright © 2010 by ASME


80000

70000

60000

Bolt Tension (N)


50000

40000 Exp-BT
Ana-BT
30000
FEA-BT
20000

10000
Fig.13 Finite element modeling
0
30000
Bolt Tension (BT)/Clamp Load (CL) (N)

0 20000 40000 60000 80000


FEA-BT
FEA-CL
separating force (N)
25000
EXP-BT Fig.16 Bolt tension variation due to a cyclic separating force
EXP-CL Fe = 35000 − 35000 cos(0.4t ) N, Fi = 61500 N for steel alloy bolt
20000 30000

15000
25000

10000
Separating Force (N)

20000

5000
15000 Exp
FEA
0
10000
0 5000 10000 15000 20000 25000 30000
Separating Force (N)
5000
Fig.14 Bolt tension and clamp load variation due to cyclic separating
force ( Fe = 14000 − 14000 cos(0.4t ) N, Fi = 18000N, µt = 0.38 for
0
aluminum alloy bolt) 0 0.05 0.1 0.15 0.2 0.25 0.3
Displacem ent (m m )
60000
Fig.17 Separating force vs. displacement due to a cyclic separating
force Fe = 14000 − 14000 cos(0.4t ) N, Fi = 18000N for aluminum
50000
Bolt Tension, Clamp Load (N)

bolt
30000
40000
Exp
25000
FEA
30000 Exp-BT
Separating Force (N)

FEA-BT 20000
Exp-CL
20000
FEA-CL 15000

10000
10000

0 5000
0 10000 20000 30000 40000 50000 60000
Separating Force (N) 0
Fig.15 Bolt tension and clamp load variation due to a cyclic separating 0 0.05 0.1 0.15 0.2
force Fe = 30000 − 30000 cos(0.4t ) N, Fi = 33500N for steel alloy Displacem ent (m m )
bolt Fig.18 Separating force vs. displacement due to a cyclic separating
force Fe = 12000 − 12000 cos(0.4t ) N, Fi = 18000N for aluminum
bolt

11 Copyright © 2010 by ASME

View publication stats

Anda mungkin juga menyukai