Anda di halaman 1dari 9

Effect of Thread Pitch and Initial Tension on the Self-Loosening of Threaded Fasteners

Sayed A. Nassar Basil A. Housari


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

A mathematical model and an experimental procedure are presented to study the selfloosening phenomenon of threaded fasteners that are subjected to cyclic transverse loads. The study investigates the effect of thread pitch, initial bolt tension, and the amplitude of the external excitation on the loosening of a single-bolt joint. The rate of drop in the joint clamp load (fastener tension) per cycle, as well as the total number of cycles that would cause the complete loss of clamp load, are monitored. In the mathematical model, the differential equations of linear and angular motion of the bolt are formulated in terms of the system properties and the external cyclic transverse excitation. Numerical integration of the equation of angular motion provides the bolt rotation in the loosening direction, which causes the partial or full loss of the clamp load. An iterative MATLAB code is developed and used for the calculation of tension loss in the fastener tension due to the self-loosening. Analytical and experimental results are discussed. DOI: 10.1115/1.2349572 Keywords: Self-loosening, vibration loosening, threaded fasteners, bolted joints, thread pitch

Introduction
There have been several studies on the self-loosening of threaded fasteners; most of them are experimental. Hess 1 surveyed some studies on the vibration- and shock-induced loosening of threaded fasteners. Sakai 2, Haviland 3, Yamamoto and Kasei 4, Tanaka et al. 5, Vinogradov and Haung 6, Zadoks and Yu 7,8, and Junker 9 offered some insight into the effect of transverse load on the self-loosening of threaded fasteners. Sakai 2 presented a theoretical model of loosening based on slip at contact surfaces of the fastener. The results from his model suggested that the friction at the contact surfaces should drop to a very low value of 0.03, whereas the actual value of the friction coefcient for oiled steel is normally around 0.15. Haviland 3 stated that loosening happens as a result of the ratcheting action of the nut, but no mathematical model or experimental data were provided to prove his theory. Yamamoto and Kasei 4 proposed another mechanism, which suggested that loosening occurs due to the accumulation and release of potential energy due to the torsional bolt deformation. Tanaka et al. 5 conducted an axisymmetric nite element study for bolt loosening. Their model showed that the loosening process is the same as that presented by Yamamoto and Kasei. Vinogradov and Haung 6 presented a dynamic model to simulate the loosening under high-frequency loading 30 240 kHz, which is well beyond the normal operation range for most machine vibrations. Zadoks and Yu 7,8 presented a dynamic model to simulate the loosening process of screw in a tapped hole. Their friction model assumed that friction drops to zero in the circumferential direction when the friction is overcome in the transverse direction, this assumption was not a valid one. Junker 9 showed that the loosening of threaded fasteners was far more severe when the joint was subjected to transverse cyclic
Contributed by the Pressure Vessels and Piping Division of ASME for publication in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received June 27, 2005; nal manuscript received July 30, 2005. Review conducted by G. E. Otto Widera. Paper presented at the 2005 ASME Pressure Vessels and Piping Conference PVP2005, July 17, 2005July 21, 2005, Denver, Colorado, USA.

loads. In his work, he concluded that the self-loosening happens when slippage took place between engaged threads and/or under the bolt head/nut. This study proposes a mathematical model and an experimental procedure to study the effect of thread pitch, initial bolt tension, and the amplitude of the cyclic transverse load on the loosening of threaded fasteners. Both the numerical and experimental results are presented and analyzed.

Mathematical Model
Figure 1 shows a free body diagram of the threaded fastener, with various forces and moment/torque components. Linear and angular equations of motion are established for a mathematical model of threaded fasteners during the self-loosening. Eventual integration of the angular equation of motion, with small time steps, provides information on the bolt rotation that causes the loss of its tension self-loosening. The drop in the bolt tension is determined from the calculated angular displacement in the loosening direction and the tension-rotation relationship of the bolt. A MATLAB code is developed to calculate the clamp load loss in terms of the number of the transverse load cycles. Force and Moment Analysis. A model of a typical bolted joint system is shown in Fig. 2; the upper plate is clamped to a tapped block using a threaded fastener. The forces acting on the free body diagram Fig. 1 are detailed in Figs. 35. Figure 3 shows the underhead bearing surface or the threads contact area divided into a nite number n of small sectors with angle increments d. The differential force vectors acting on each sector of the contact area must be determined in order to calculate the differential bearing friction torque component dTb that must be overcome during the loosening of the bolt. The proposed self-loosening model takes into account some simplifying assumptions. First, it is assumed that when the bolt head slides over the top of the block, the rotation of the bolt head will be stopped if the shank of the bolt comes in contact with the edge of the hole. Second, it is assumed that the thread rotation is stopped when the bolt threads come in contact with the side of the tapped hole. The set of forces acting on the bolt head and on the threads are different before and during the side contact. Third, the Transactions of the ASME

590 / Vol. 128, NOVEMBER 2006

Copyright 2006 by ASME

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 1 Bolt-free body diagram

effect of damping is not considered in this study. Figure 1 shows the free body diagrams of the bolt segments with no side contact. On the bolt head segment, the forces and torques acting on the bolt bearing surface are the bolt tension F, the reaction force F from the upper plate, the friction force Fb between the bolt and the upper plate, the bolt shear force Fs created by bolt bending, the bending moment M b, and the torque Tl transmitted from the threads to the head which is caused by all forces acting on the threads. These forces and moments determine the motion of the bolt head during the loading cycle. It must be observed that only the torque component Tl will have the potential to rotate the bolt in the loosening direction. This means that the bolt has a potential to loosen if and only if the torque Tl is positive i.e., acting in the loosening direction as it overcomes the friction resistance under the bolt head. Figures 3 and 4 show the three different coordinate systems that are used in the modeling process. The rst is a global Cartesian , where X is in the system XYZ with respective unit vectors i, j, k positive direction of the external excitation, and Z is upward parallel to bolt axis. The second is a local coordinate system uvw, in which u is oriented in the radial direction, v is in the tangential direction, and w is parallel to the global Z direction. The third

Fig. 3 Forces on the bearing surface

coordinate system is u1v1w1, in which w1 is normal to the thread surface, and v1 is within the thread plane in the circumferential , v , w and direction. The relationship between the unit vectors u i, j, k is expressed in a matrix format as follows:

Fig. 2 Bolt motion under external transverse force

Fig. 4 Thread contact force components

Journal of Pressure Vessel Technology

NOVEMBER 2006, Vol. 128 / 591

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

v 1v 1 T w T = T11 u 1w u

Shear Force. Generally, a cyclic transverse excitation may be a force or a displacement. In this study, cyclic displacement is considered. The displacement of the upper plate in Fig. 1 is given by

t = sint i o = 2 f

8a 8b

t is the displacement vector as function of time, maxiwhere 0 mum displacement amplitude, is the angular frequency of the transverse loading, f is the frequency, and t is the time. Sliding of the head will take place when the bolt deection reaches a critical value cr, at which the force Fs overcomes the underhead friction. This critical displacement cr is approximated by

cr =

bF iL 3 3EI

where b is the underhead friction coefcient, Fi is the bolt tension, L is the grip length, E is the modulus of elasticity for the bolt material, and I is the moment of inertia of the cross section of the bolt. Depending on the friction coefcients under the bolt head and between the engaged threads, if motion between the threads happens before the sliding under the bolt head, then the critical displacement becomes:

crt =

bF iL 3 xtt 3EI

10

Fig. 5 Thread friction force components

T v w T = T u i jk

where xtt is the displacement of the thread in the x direction at time t. Equation 10 allows for three possible scenarios. In the rst scenario, the amplitude of the displacement cr, o cr. Then the bolt will be subjected to cyclic bending only. The corresponding shear force FsI is given by

where the superscript T indicates the transpose of a matrix, and T is a transformation matrix given by T = sin cos 0 0 0 1

tt t = 3EIt + x F sI 3 L

11

cos

sin 0

, as shown in i and u where is the angle between the unit vectors Fig. 3. The reverse transformation from the uvw system to the XYZ system is given by T = T1 u v w T i jk where the superscript 1 indicates the inverse of a matrix. , v , w are given by Similarly, the unit vectors u 1 T = T 1 u v 1w w T 1v u 4 the matrix elements in the rst and second rows in the transformation matrix T1 are obtained using the following geometric relationships: + cosv 1 = 0 u + sinw v 1 = tan2 + tan2 + 1 tanu + tanv +w 5 w where and are the thread prole angle and the thread lead angle, respectively, as shown in Figs. 4 and 5. The third row in the transformation matrix T1 is obtained by the cross product of the 1, v 1 unit vectors u 1=u 1 v 1 w Equation 4 may be expressed in an inverse way as follows: 592 / Vol. 128, NOVEMBER 2006 6 3

t. In the second tt always opposes the direction of where x scenario, the displacement amplitude is greater than the critical displacement cr and less than one half of the hole clearance . cr o / 2. This scenario may be divided into two phases. In the rst phase, the bolt will bend until the displacement cr is reached; in the second phase, when t cr, the bolt head will slide until the displacement reaches its maximum amplitude o. In the rst phase, the shear force is still given by Eq. 11; in the second phase, the shear force is given by t = 3EIcr t F sII 3 L t
12

t / t is the unit vector in the direction of the where the term . In the third scenario, the displacement amplitude shear force F s o is bigger than the critical value cr and is also larger than half of the hole clearance i.e., cr / 2 o. This scenario is divided into three phases. In the rst phase, the bolt will bend until the displacement reaches cr and the corresponding shear force is still given by Eq. 11. In the second phase, the bolt head will slide until the hole clearance is consumed, and the corresponding shear force is given by Eq. 12. In the third and nal phase, the shank of the bolt makes contact with the side of the hole in the upper plate, and more bending is exerted on the bolt shank. The corresponding shear force FsIII becomes t tt /2 t = 3EIt + x F sIII 3 L t
13

Transactions of the ASME

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Thread-Loosening Torque. The loosening torque transmitted to the bolt head segment from the engaged threads is determined from the free body diagram depicted in Fig. 1. The forces and moments acting on the threads include the bolt tension F, the bending moment M b, the shear force Fs, the contact force Fn normal to the thread surface, and the thread friction force Ft. The thread contact area is divided into a large number of small sectors as shown in Fig. 4. Although the bending moment exerted on the bolt shank increases the contact pressure on one side of the threads and reduces it on the opposite side, the pressure distribution on each side of the threads is assumed to be uniform. The change in the axial force on each sector can be given by Mb =

t = dR

u , v , w = T T F
1 1 1 1

+F F s + T1dF mt n

19 20

= dR dF n w1 t = dR + dR dF t1 u1 v1

21

Pr d r cos = 2
t t 0

/2

Prt2cosd

/2

, dR , dR are the components of dR in the u , v , where dR u1 w1 1 1 v1 is the thread friction force component w1, respectively, and dF t1 . that opposes the differential shear force dF s The maximum magnitude of the thread friction force is equal to times the coefcient of thread friction the normal force dF n t t dFt,maxt = tdF n
22

Mb =

Prt2

/2

cosd = Mb 2rt2

2 Prt2

/2

P=

Mb 2rt2

P=

= M b r 2 k = M b k = Pr dk dF mt t t n nrt 2rt2

14

where P is the contact pressure on the thread, rt is the mean thread is the change in bolt tension on each sector. The radius, and dF mt positive sign corresponds to the change in the axial force on the left half of the bolt when the moment is positive counter clockwise and to the right half of the bolt when the moment is negative clockwise, while the negative sign corresponds to the right half of the bolt when the moment is positive and to the left half of the bolt when the moment is negative. Similarly, the change in the reaction force on each sector under the bolt head is given by = 2 M b cosk dF mb nrb 15

where t is the thread coefcient of friction. When the magnitude + dR on one sector exceeds the maximum friction force of dR u1 v1 calculated in 22, the sector becomes free to move. The magnitude of the friction force on that sector is given by 22. On the + dR is less than dF other hand, if the magnitude dR u1 t,maxt, v1 will resist the in-plane then part of the thread friction force dF t1 forces, as shown in Eq. 21. The sector still can offer some friction resistance dFt2. The magnitude of the friction component is given by dFt2 = dFt,max dFt1 23

where rb is the mean bearing radius, which is determined more accurately by Nassar et al. 10; however, an approximate mean bearing radius is used in this study. , r , and r are given by Eqs. 1618, respectively, as M b t b follows:

= Lw s F M b
rt = rmaj + rmin 2

16 17

Figure 3 shows that the resultant force on sector A is greater than the resultant force on sector B. As the displacement increases, the shear force on each sector increases and sector A becomes free to move before sector B. In this case, the sector that has the minimum resultant force will act as an instantaneous center pivot point about which the bolt can rotate within the thread and the hole clearances. This suggests that the motion in the x direction is not pure sliding; it is more like rolling on one side of the bolt. For each sector, the vector rti points from the instantaneous center of rotation to the center of the sector under consideration. The direction of motion for each sector is in the loosening ti for the same sector. The instantadirection perpendicular to r neous center is found by locating the most constrained sector in ti can be obtained by adding two the thread area. The vector r vectors: the rst one is the vector from the instantaneous center to the center of the bolt, and the second is the vector from the center of the bolt to the center of the ith sector. The resultant differential force dF threadt on one sector of the thread is expressed in the uvw coordinate system as follows:
1 dF threadt = T1 dRt + dFt1t + dFt2t

rmaj + + ro rb = 2

24

18

where rmaj is the major bolt radius, rmin is the minor bolt radius, is the hole clearance, and ro is the outer radius of the contact area under the bolt head. The bolt tension F and the shear force Fs are assumed to be evenly distributed on all sectors. The differential components of these two forces and the force generated by the bending are expressed in the u1v1w1 coordinate system using Eqs. 26. The t on each sector has three components. The equivalent force dR rst component is normal to the thread surface and is equal and . The other two compoopposite to the thread reaction force dF n nents are in the thread sector plane and are equal and opposite to on the sector under conthe differential thread friction force dF t sideration. These components are given by Eqs. 1921 as follows: Journal of Pressure Vessel Technology

since the w component is parallel to the bolt axis, which is parallel to the axis of rotation, then only the u and v components are used of the resultant to calculate the differential thread torque dT t force on each sector. This torque is given by

t = dT rti dF t thread u + dFthread v

25

The resultant force F threads on the entire thread area is calculated by transferring the resultant force vectors on all sectors to the global XYZ-coordinate system, and adding them together as follows: F threadst =

T
1

dF threadt

26

The resultant torque on the threads is given by NOVEMBER 2006, Vol. 128 / 593

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

t = T t

t dT t

thread

t dT t

27

t = T b

t dT b

underhead

t dT
b 1

36

Underhead Friction Force and Torque Components. Differential underhead friction forces on each sector are used for determining the underhead friction torque component. In-plane forces acting on the bolt head include the shear force Fs, the bearing friction force Fb, and the torque Tt from the threads. It should be noted that the shear force acting on the bolt head is equal and opposite to the shear force acting on threads. The differential shear force on each sector under the bolt head is obtained by dividing the total shear force calculated by Eqs. 1113 by the total number of sectors n as shown by Eq. 27

The total force F bolt acting on the bolt is given by =F F bolt threads + Fb +T T=T t b
37 38

Finally, the total torque T acting on the bolt is determined by Equations 838 provide the force and torque components that are necessary for writing the linear and angular equations of motion of the bolt under the transverse cyclic displacement excitation t. Dynamic Analysis of Self-Loosening. The linear and rotational equations of motion of the bolt are given, respectively, by t Ft = mx 39 40

= Fs dF s n
The magnitude of the maximum friction force is given by dFb,maxt = b

28

t F t + dF mb n

29

t Tt = I
p

exceeds When the magnitude of the differential shear force dF s the maximum value calculated by Eq. 29, the sector becomes free to slide. The magnitude of the friction force on that sector is given by t = Ft + dF t dF b b mb n


dF s dF
s

where m is the effective mass of the bolt, I p is the effective polar is the linear acceleration of the moment of inertia of the bolt, x is the angular acceleration. The effective mass m bolt head, and of the bolt is equal to 33/140 of the shank mass m1 plus the full mass of the bolt head m2 11, as expressed in m= 33 m1 + m2 140 41

30

/ dF denes the unit vector in the friction where the term dF s s t. force direction dF t dF On the other hand, if dF s b,maxt, then part of the friction force dF will resist the shear force; this part is given by
b1 b

The polar moment of inertia I p about the bolt axis is given by 1 1 2 I p = m 1r 2 1 + m 2r 2 2 2 42

= dF dF b1 s

31

where r1, r2 are the shank and bolt head radii, respectively. The polar moment of inertia about the instantaneous center of rotation during the loosening process is given by I pc =

The sector still provides some frictional resistance dFb2, which is given by dFb2 = dFb,max dFs 32

1 1 2 2 m 1r 2 1 + m2r2 + mr1 2 2

43

The vector from the instantaneous center of rotation to the center bi. The vector r bi is obtained by adding two of the ith sector is r vectors; namely, the vector from the instantaneous center to the center of the bolt and the vector from the center of the bolt to the center of the ith sector. The resultant differential force on one sector under the bolt head is given in the uvw-coordinate system as follows

In this model, the friction and material damping effects are neglected due to the anticipated slow rate of motion. Additionally, the force and torque represented by Eqs. 39 and 40 are assumed to be constant for a differential interval of time dt, which permits the use of the following kinematic equations: t + dt = x dt + x t x tdt + xt t + dtdt2 + x xt + dt = 0.5x t + dt = dt + t t + dtdt2 + tdt = t + dt t = 0.5 44 45 46 47

t = TdF t + dF t + dF t dF b s b1 b2

33

The differential underhead torque dTb is determined by the following cross product:

t = r t bi dF dT b b

34

As the bolt rolls, it moves in the transverse direction and the maximum rolling angle before having side contact at the threads is affected by the thread clearance. This angle can be given by max = t rt 48

The resultant force on the bolt head is equal to the integration of all differential resultant forces on all sectors under the bolt head, or it can be approximated by the summation of all differential resultants as follows:

= F b

dF b

underhead

dF
1

35

where t is the thread clearance. The loosening angle of the bolt head reduces the bolt tension by F. The relationship between F and is given by 12 as follows: F =

The total underhead torque is equal to the integration of all differential torques on the bearing surface, and it can be approximated by the summation of all differential torques as follows: 594 / Vol. 128, NOVEMBER 2006

k bk c p = 2 k b + k c

49

where kb is the bolt stiffness, kc is the joint stiffness, p is the thread pitch, and is the slope of the bolt tension-rotation curve. Transactions of the ASME

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 6 Effect of thread pitch on the self-loosening

The slope may be determined either experimentally from the bolt tension-rotation curve shown in Fig. 6, or theoretically from Eq. 49. The bolt and joint stiffnesses are given by Ref. 13 as follows: kb = A bE b L A cE c L 50 51

kc =

where Ab is the bolt cross-sectional area, Eb is the modulus of elasticity of the bolt, Ac is the effective joint cross-sectional area, Ec is the modulus of elasticity of the joint, and L is the grip length of the bolt. The incremental reduction in the bolt tension is used to update the force and torque values and start the next iteration of calculations until the desired number of cycles is reached.
MATLAB Simulation Code. To obtain the reduction in bolt tension, a MATLAB code is written to numerically calculate the incremental angle of rotation and convert it to an incremental loss of bolt tension, due to self-loosening caused by the cyclic excitation. Figure 7 shows a ow chart of the MATLAB algorithm used to determine the tension drop in the bolt. The input information to the code include, bolt dimensions, external load information, friction coefcients between contact surfaces, bolt stiffness, joint stiffness, hole clearance, initial bolt tension, time increment, and the desired number of cycles, or until the fastener becomes completely loose. The MATLAB steps include iterative calculations of the bolt forces and torques using the equations derived in the previous sections and the calculation of linear and angular bolt accelerations, velocities, and displacements. The updated values of the bolt tension are calculated from the displacement information.

Fig. 7 Tension-rotation 1 / 2 in.-20 bolts

relationship

for

1 / 2 in.-13

and

data-acquisition system that records the bolt tension and the small head rotation loosening during the test. The eccentric mechanism can be set to provide various levels of the transverse displacement amplitude to the upper plate during the self-loosening test. Two displacement amplitudes are used in this study; namely, 0.03 in. and 0.05 in. The bolts and the tapped hole are ultrasonically cleaned and dried, then lubricated with SAE 5W-30 engine oil. The underhead friction coefcient b and the thread friction coefcient t are measured separately using FastLab Control Module fastening system 14. The experimental values of b and

Experimental Verication
An experimental procedure and test setup are developed in order to validate the results obtained from the new mathematical model that simulates the self-loosening of threaded fasteners. Figure 8 shows a schematic of the self-loosening testing machine that is constructed in a similar way to the Junker machine 9. The machine consists of a motor that drives a set of pulleys with an eccentric mechanism that reciprocates the upper plate of a bolted joint in order to cause self-loosening. A load cell is used to monitor the clamp load/bolt tension in real time. A socket cap is mounted on the bolt head and connected to a rotation sensor using a small belt. The load cell and the rotation sensor are wired to a Journal of Pressure Vessel Technology

Fig. 8 Flow chart for the

MATLAB

algorithm

NOVEMBER 2006, Vol. 128 / 595

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 9 Self-loosening test setup

t are input into the MATLAB code for the mathematical simulation of the self-loosening of threaded fasteners.

Fig. 11 Effect of thread pitch on the self-loosening

Results and Analysis


A typical self-loosening curve that shows the bolt tension versus the number of transverse load cycles is depicted in Fig. 9. The loosening curve can be approximated by a straight line, whose slope represents the rate of tension loss per cycle. The effect of the various factors on the self-loosening of threaded fasteners is discussed using the analytical and experimental data. Effect of the Thread Pitch. Two values of thread pitch are considered in this study; the values correspond to ne and course threads: 1 / 2 in.-20 and 1 / 2 in.-13 bolts, respectively. The pitch of the ne thread is 1 / 20 in. and is 1 / 13 in. for the coarse thread. The effect of thread pitch p may be pointed out in the model in three different places. First, changing the thread pitch means changing the threads lead angle , which is given by pitch p will affect the slope of the tension-rotation curve in accordance with Eq. 47. Neglecting the effect of the ne versus coarse thread pitch on the bolt stiffness, the slope of the tensionrotation is directly proportional to the pitch of the bolt. Hence, the slope for the 1 / 2 in.-13 bolt is almost 1.5 times the slope for the 1 / 2 in.-20 bolt. This means that ne bolts need to rotate 1.5 times the coarse ones to loose the same amount of tension. The experimental data, shown in Fig. 6, is in close agreement on the slope of the tension-rotation curves for the ne and coarse threads in both the tightening and loosening directions. Third, the slope of the loosening curve is also affected by the clearance between the threads. On the one hand, coarse thread bolts normally have larger clearance between the threads; consequently, they have steeper loosening lines because the larger clearance allows motion for a longer period of time before the threads of the bolt make contact with the side of the tapped hole. On the other hand, ne thread bolts normally have smaller thread clearances, that shortens the time duration of the sliding motion between the threads; this slows down the loosening of ne thread bolts. Figure 11 shows that the experimental and the analytical results are in good correlation for both coarse and ne threads. Effect of the Level of Initial Bolt Tension and Excitation Amplitude. The correlation between the amplitude of the excitation and the threshold level of the initial bolt tension, below which the process of self-loosening begins, is inherent in the proposed mathematical model. The friction forces between the engaged threads and under the bolt head are directly proportional to the bolt tension and to the friction coefcients. The expressions for these forces are given by Eqs. 22 and 29, respectively. Additionally the magnitude of the loosening force caused by the threads lead angle is also proportional to both the bolt tension and the lead angle, as shown by Eqs. 5, 19, and 21. The net effect of the forces given by Eqs. 21, 22, and 29 may be expressed in terms of two torque components, one in the loosening direction and an opposing component from the frictional resistance. The difference between the friction resistance and the loosening torque determines whether or not the bolt will start to rotate in the loosening direction. The friction resistance, under the bolt head and between the threads, is divided into two components; the rst component opposes the motion in the transverse direction, while the second component opposes the angular motion. Prior to applying the transverse displacement force, the rst component is zero. Hence, the entire friction resistance would be only opposing Transactions of the ASME

= arctan


p d

52

The angle is used in Eq. 5 to nd the transformation matrix, which is used in calculating the forces on the threads. This change in has a signicant effect on the magnitude of the loosening force; this means that the pitch value affects the threshold point at which loosening begins, as shown in Fig. 10. Second, the thread

Fig. 10 Typical self-loosening curve

596 / Vol. 128, NOVEMBER 2006

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

number of cycles to complete loosening from 60 to 120 cycles. Additionally, it is observed that the 0.05 in. amplitude of the transverse excitation is not sufcient to initiate any self-loosening when the initial bolt tension is increased to 9000 lb 40 kN, while it causes loosening when the initial bolt tension is reduced to 6000 lb 26.7 kN. Similarly, an excitation amplitude of 0.03 in. is not sufcient to cause loosening when the initial tension is 4000 lb 17.8 kN; it causes loosening at a lower level of initial bolt tension of 2000 lb 8.9 kN, however. In Fig. 13, the lower and upper horizontal lines represent bolt tensions that are higher than the threshold points for the 0.03 and 0.05 in. amplitudes, respectively. They show that loosening has not occurred because the initial bolt tension has exceeded the threshold value for each of the two indicated excitation amplitudes. The results from the mathematical model correlate well with the experimental data.

Conclusion
A new mathematical model is introduced and experimentally veried in order to study the effect of thread pitch and bolt tension on the self-loosening phenomenon in threaded fasteners under cyclic loads and/or displacements. A cyclic transverse excitation may or may not initiate the self-loosening of a tightened threaded fastener, depending on the amplitude of the excitation in relation to the level of bolt tension. For a known amplitude of the cyclic excitation, the bolt tension will have to be below a threshold value in order for the self-loosening to start. Once the loosening is started, the rate of clamp load loss in a bolted joint due to selfloosening is proportional to the thread pitch. The use of ne threads reduces the loosening rate per cycle, whereas coarse threads will loosen at a higher rate. The threshold level of bolt tension, at which the loosening begins under known excitation amplitude, is higher for ne threads as compared to coarse threads. For both coarse and ne threads, it has been found that once the bolt loosening has started, the number of cycles until complete loosening is reached is directly proportional to the initial bolt tension. Finally, the study introduces the correlation between the excitation amplitude and the required threshold value of bolt tension above which self-loosening may be prevented. The correlation concludes that bolts that are subjected to a higher level of excitation amplitude will have to be tightened to higher preloads, in order to avoid self-loosening.

Fig. 12 Effect of initial bolt tension on the self-loosening

the motion in the angular loosening direction. When the external transverse displacement force is applied, the rst component increases, while the second component decreases. At the same time the loosening torque is increased by the increase in the normal force on the threads. As the amplitude of the excitation is increased, the difference between the frictional resistance and the loosening torque becomes smaller and smaller, until the loosening torque overcomes the friction; then the bolt starts to rotate in the loosening direction. The loosening threshold point is a function of the friction coefcients between the threads and under the bolt head, initial bolt tension, thread lead angle, and the amplitude of the transverse excitation. Increasing the bolt tension over the threshold point will require larger excitation amplitude to initiate loosening. Once loosening has started, the number of cycles required to cause complete loosening is directly proportional to the initial bolt tension. Figures 12 and 13 show the experimental and analytical results for the effect of initial bolt tension and the amplitude of excitation, respectively, on the self-loosening of 1 / 2 in.-20SAE grade 5 plain bolt. Results show that the doubling of the initial bolt tension from 2000 lb 8.9 kN to 4000 lb 17.8 kN, doubles the

Nomenclature
E I F L R Fb Fb,max Fex Ff Fhead Mb M b,max Fi Fl Fn Fs Ft Kb Kc Tb Tl Tp bolt modulus of elasticity moment of inertia of the bolt cross section bolt tension grip length resultant force opposing the thread friction force bearing friction force maximum bearing friction force external excitation force friction force between the joint plates resultant force opposing the bearing friction force bolt bending moment maximum bolt bending moment initial bolt tension loosening force on the bolt head normal contact force on the threads bolt shear force resultant thread friction force bolt stiffness joint stiffness bearing friction torque component loosening torque pitch torque component NOVEMBER 2006, Vol. 128 / 597

Fig. 13 Effect of the excitation amplitude on the self-loosening

Journal of Pressure Vessel Technology

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Tt thread friction torque component dFb differential bearing friction force dFhead differential force opposing the bearing friction force dFn differential thread contact force dFl differential loosening force dFs differential bolt shear dFt differential thread friction force dFmb change in the underhead force due to bending dFmt change in the thread force due to bending dR subtotal differential force on one thread sector dTb differential bearing friction torque dTt differential thread friction torque dt incremental time f frequency of the external excitation m mass of the bolt m1 mass of the bolt shank m2 mass of the bolt head rb mean bearing contact radius p thread pitch P contact pressure rmaj major thread radius rmin minor thread radius ro outer bearing radius rt mean thread radius t time x linear displacement xt threads linear displacement linear velocity x linear acceleration x 2 thread prole angle lead helix angle hole clearance t thread clearance external displacement 0 external displacement amplitude sector angle in Fig. 3 slope of the bolt tension-rotation relationship

b t

static bearing friction coefcient static thread friction coefcient angular position angular velocity angular acceleration angular frequency of the excitation

References
1 Hess, D., 1998, Vibration and Shock Induced Loosening, Handbook of Bolts and Bolted Joints, J. H. Bickford, S. Nassar, eds., Marcel Dekker, New York, pp. 757824. 2 Sakai, T., 1978, Investigations of Bolt Loosening Mechanisms, 1st Report, On Bolts of Transversely Loaded Joints, Bull. JSME, 21, pp. 13851390. 3 Haviland, G. S., 1983, Designing With Threaded Fasteners, Mech. Eng. Am. Soc. Mech. Eng., 105, pp. 1731. 4 Yamamoto, A., and Kasei, S., 1984, A Solution for the Self-Loosening Mechanism of Threaded Fasteners Under Transverse Vibration, Bull. JSME, 18, pp. 261266. 5 Tanaka, M., Hongo, K., and Asaba, E., 1982, Finite Element Analysis of the Threaded Connections Subjected to External Loads, Bull. JSME, 25, pp. 291298. 6 Vinogradov, O., and Haung, X., 1989, On A High Frequency Mechanism of Self-Loosening of Fasteners, Proceedings of 12th ASME Conference on Mechanical Vibration and Noise, Montreal, ASME, New York, pp. 131137. 7 Zadoks, R. I., and Yu, X., 1993, A Preliminary Study of the Self-Loosening in Bolted Connections, ASME Nonlinear Vibrations, ASME Vol. DE-Vol. 54, 7988. 8 Zadoks, R. I., and Yu, X., 1997, An Investigation of the Self-Loosening Behavior of Bolts Under Transverse Vibrations, J. Sound Vib., 208, pp. 189 209. 9 Junker, G. H., 1969, New Criteria for Self-Loosening of Fasteners Under Vibration, SAE Trans., 78, pp. 314335. 10 Nassar, S. A., El-Khiamy, H., Barber, G. C., Zou, Q., and Sun, T. S., 2005, An Experimental Study of Bearing and Thread Friction in Fasteners, ASME J. Tribol., 127, pp. 263272. 11 Steidel, R. F., 1989, An Introduction to Mechanical Vibrations, 3rd ed., Wiley, New York. 12 Bickford, J. H., 1997, An Introduction to the Design and Analysis of Bolted Joints, 3rd ed., Marcel Dekker, New York. 13 Juvinall, R. C., and Marshek, K. M., 2000, Fundamentals of Machine Component Design, 3rd ed., Wiley, New York. 14 RS Technologies, 2004, Threaded Fastener Testing System User Manual, RS Technologies, Farmington Hills, MI.

598 / Vol. 128, NOVEMBER 2006

Transactions of the ASME

Downloaded 12 Dec 2011 to 195.83.11.66. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Anda mungkin juga menyukai