Vibration Loosening PDF
Vibration Loosening PDF
Vibration Loosening PDF
BOLTED JOINTS
The present paper identifies critical parameters in preventing loosening and analyses two
possible mechanisms of vibration loosening. The mathematical models developed shed light on
the effect of various bolted–joint related parameters on vibration loosening and joint integrity.
It also develops simple rules-of-thumb for the prevention of vibration loosening. Finally, the
paper discusses the available locking devices and their correct usage and limitations.
2.0 NOMENCLATURE
D nominal bolt diameter
d0 amplitude of rectilinear vibration
d0(t) cyclic rectilinear displacement applied on the conveyor
F friction force on the particle
F0 friction force with respect to conveyor at relative rest of the particle
Fb bolt tension
Fi pre-tension of the bolt
f non-dimensional friction parameter
G non-dimensional inertial parameter 1
g gravitational acceleration
J polar moment of inertia
K nut factor
K1 thread pitch contribution of nut factor
K2 thread friction contribution of nut factor
K3 bearing friction contribution of nut factor
m mass of the particle on the conveyor
m1 mass of the transported component (bolt or nut)
P effective downward force
p thread pitch
R reaction force on the particle
rt effective radius of the thread
rb effective radius of the head/nut bearing surface
T tightening torque
Ti initial tightening torque
TI inertial torque
Tp prevailing torque
x displacement in x-direction
Page 2
Greek Symbols:
a thread flank angle
b inclined angle of the conveyor
d transformed time variable 2
G non-dimensional inertial parameter 2
g transformed time variable 1
l pre-load factor
mt thread friction coefficient
mb bearing friction coefficient
mI static friction coefficient of the conveyor surface
m dry friction coefficient of the conveyor surface
q angle of the applied rectilinear vibration to the conveyor surface
t non-dimensional time
t1 non-dimensional time when the particle at momentary rest
t2 non-dimensional time when the particle at momentary rest
w angular velocity of rectilinear vibration
x non-dimensional displacement
y angular rotation of the wheel
Axis Systems:
OXY inertial co-ordinate system placed on the conveyor
oxy relative co-ordinate system placed on the particle
Derivatives:
.
first derivative with respect to time - velocity
..
second derivative with respect to time – acceleration
Subscript:
mean time averaged
3.0 INTRODUCTION
Most bolted joints, especially the ones associated with machinery, are subject to significant
vibration levels during their life span. Rotating or reciprocating machines, such as gas/steam
turbines, electric motors and IC engines, are subject to vibration of relatively high frequency.
Gyratory crushers, jack hammers and so forth are subject to medium frequency vibrations.
Forging/stamping machines are subject to relatively low frequency, high amplitude vibrations.
Certain dynamic structures (eg. bridges and buildings subject to wind and cyclonic loads) also
undergo dynamic load fluctuations. Cyclic temperature variations may also cause dynamic
(very low frequency) load fluctuations in bolted joints. Although the frequencies of these
fluctuations are spread over a wide spectrum, the general effects of dynamic loading on bolted
joints are similar. The main effects are:
a) loosening of the nut/bolt and
b) fatigue failure.
In most situations, both of the above effects may occur in one joint.
It is common experience that vibration loosening of joints may occur when a bolted joint
undergoes dynamic load fluctuations1. There are a large number of thread-locking devices
available in the market in order to alleviate joint loosening problems. However, use of such
Page 3
devices needs to be done with care. It is very important to understand the root causes and
mechanisms behind the phenomenon of vibration loosening in order to decide the optimum
method of alleviating the problem. This paper discusses the mechanisms preventing loosening,
mechanisms of loosening and guidelines on using various thread-locking devices commonly
available in the market.
T = Fb .D.K ¾
¾®(2a )
Where K = K 1 + K 2 + K 3
p rm rm
K1 = ; K2 = t t ; K3 = b b
2pD D cos a D
where the geometric notation is as shown in Figure 1, Fb is the
resulting bolt tension, p the pitch of the thread, mt and mb are
the friction coefficient for thread and the under head bearing
surfaces respectively.
é-p rm rm ù
T p = Fi .D.ê + t t + b bú ¾
¾®(3)
ë 2pD D cos a D û
T p = Fi .D.(- K 1 + K 2 + K 3 ) ¾
¾®(4)
This Tp will be the initial torque required to loosen the joint. The above prevailing torque (Tp)
will be smaller than the initial tightening torque (Ti) by amount:
é pù
Ti - T p = Fi .ê ú ¾¾®(5)
ëp û
This indicates that the prevailing torque for a coarse threaded fastener will be smaller than that
for a fine threaded fastener if everything else remains the same. In fact, the prevailing torque
(Tp) will decrease linearly with the pitch (p) of the fastener. As is well known in the industry,
fine threaded fasteners are more suitable for high vibration applications1,2.
As shown in the above equation (3) the prevailing torque Tp is increasing with increasing pre-
tension load (Fi) and the increasing friction coefficients and radii mt , mb and rt, rb. As rt and rb
are proportional to the nominal diameter (D) of the fastener, both the tightening and loosening
torque will increase for a given pre-tension load when the fastener size is increased. Therefore,
Page 4
as the deficit between tightening and loosening torque is independent of the fastener diameter, it
is better to use a larger diameter fastener for an application subject to significant vibration.
Similarly, increasing thread angle (a) will increase both prevailing and installation torque if all
the other parameters are kept constant. Therefore a “buttress” thread (a=0°) will loosen easily
compared to a metric thread(a=30°).
The prevailing torque (Tp) increases linearly with the pre-tension load at a rate lower than the
increase of the tightening torque. Once an estimation can be made of the possible magnitude of
the loosening torque anticipated during the life span of the bolted joint, a minimum pre-tension
load can be determined to prevent vibration loosening.
The above analysis does not take into account any prevailing torque component due to thread
damage or plastic/elastic deformation of thread. The analysis is only a first order (linear)
analysis. When the fastener is tightened closer to the yield load, local plastic deformation
occurring at the threads will incorporate further prevailing torque.
Now that we have some understanding of the prevailing torque or the torque preventing
loosening, it is opportune to investigate the mechanisms of loosening.
If the particle is not in contact with the conveyor it is acted upon only by gravitational force
components -mgsinb and -mgcosb along OX and OY directions respectively.
If the static dry friction coefficient of the conveyor surface is m, and the reaction force is R, then
the friction force F:
F = - mR sgn x& ¾
¾®(8)
The direction of this friction force will always be opposite to the direction the particle tends to
move.
A simple harmonic displacement do(t) with an amplitude d0 and angular velocity w can be given
as:
d o (t ) = d 0 cos w t ¾
¾®(9)
By applying Newton’s Second Law to the particle, the following equations of motion can be
established.
For the particle not to depart the surface R>0 and y(t)=0. Hence the following inequality must
be satisfied:
R = mg cos b - md 0w 2 cos w t sin q > 0 ¾
¾®(14)
This will require;
g cos b
>1 ¾
¾®(15)
d ow 2 sin q
Similarly, when the particle is at state of relative rest with respect to the conveyor:
x& (t ) = &x&(t ) = 0
then,
md ow 2 cos w t cosq - mg sin b + F0 = 0 ¾
¾®(16)
Fo = mg sin b - md ow 2 cos w t cosq
Fo < m 1 R
mg sin b - md ow 2 cos w t cosq < m 1 (mg cos b - md 0w 2 cos wt sin q ) ¾
¾®(17)
where m1 is the static friction coefficient. If this condition is violated the particle starts to slip.
Page 6
The equation of motion for the state of relative slipping can be derived by substituting (8) and
(14) in (10):
&x& = d ow 2 cos w t cosq - g sin b - m (-d ow 2 cos w t sin q + g cos b ) sgn x& ¾
¾®(18)
In order to solve the above differential equation, introduce the following non dimensional
variables:
x
x = , ¾
¾®(19)
d o cosq
t = wt ¾
¾®(20)
Now;
g sin b æ g cos b ö
x&& = cost - + m tan q çç cost - ÷ sgn x& ¾
¾®(21)
2
d ow cos q è d ow sin q ÷ø
2
by substituting:
g sin b
G= ¾
¾®(22)
d ow 2 cosq
f = m tan q ¾
¾®(23)
g cos b
G= ¾
¾®(24)
d ow 2 sin q
in eqn. (21);
Depending on the values of above parameters G, f and G, various regular regimes of motion can
be established. For the current analysis we only consider the regime with two momentary stops
per period (2p) and no relative rest periods. The particle will slide in +ox direction for one part
of the cycle and in the opposite direction during another. Lets consider that the particle moves
forward in the time interval to to t1 and then backward to complete the cycle.
Now,
Similarly, the following equations can be derived for the backward motion:
x&& = (1 - f ) cost - (G - fG ) ¾
¾®(29)
x& = sin t - sin t 1 - f sin t + f sin t 1 - (G - fG)(t - t 1 ) ¾
¾®(30)
0 = sin t o - sin t 1 - f sin t o + f sin t 1 - (G - fG)(2p + t o - t 1 ) ¾
¾®(31)
Equations (28) and (31) can be solved for the two unknowns to and t1 . By substituting the
simple transformations:
t1 -t o
=g ¾
¾®(32)
2
t1 +t o
=d ¾
¾®(33)
2
we get
p
sin g cos d = Gg - (G - fG) ¾
¾®(34)
2
p
f sin g cos d = fGg + (G - fG ) ¾
¾®(35)
2
p (1 + f ) fG - G
g = . ¾
¾®(36)
2f G-G
æ p ö
ç fGg + (G - fG) ÷
d = cos -1 ç 2 ÷ ¾
¾®(37)
ç f sin g ÷
ç ÷
è ø
Now the mean velocity of the particle being conveyed in the forward direction is:
1 æç 1 & ö
t 2p +t o
x&mean = &dt ÷ ¾
2p çè tòo òt ÷ ¾®(38)
x dt + x
1 ø
Page 8
By substituting equations (27) and (30) with appropriate to and t1 derived from equations (36)
and (37) we get the mean speed of conveying:
2f éæ pö ù
x&mean = - cos g sin d - êç g - 2 ÷ cos g - sin g ú sin d ¾
¾®(39)
p ëè ø û
If the mean conveying speed is positive the particle will move upwards and if negative the
particle will move downwards.
A complete analysis of the vibratory conveyor mechanism is beyond the scope of this paper and
can be found in Bykhovsky 19723.
By applying the above analogy to the threaded fastener; the thread essentially acts as the ramp
with associated inclined plane and friction properties. The nut or the bolt – depending on which
top surface of the thread the load is bearing on – represents the materials being transported. For
example, if the nut/bolt assembly is in a vertical axis with the bolt head facing up, the load
bearing will occur on the top surface of the bolt thread, hence making the nut the material being
transported. Similarly, if the bolt head is facing downwards then the bolt will become the
material being transported. In either case negative mean conveying velocity will loosen the
joint.
The pre-tension will provide an additional reaction force on the head/nut joint interface and on
the thread (significantly larger than that caused by the mass of the nut), hence increasing the
friction forces. In this case, the resulting inertial forces due to vibration should overcome the
reaction force due to pre-tension before loosening can occur. It can be shown through an order
of magnitude analysis that the gravity force becomes insignificant once a reasonable clamping
force (pre-tension) is applied. If the top flank angle of the thread is a and the clamping force is
Fi and the mass of the transported component is m1, then the effective downward force (P) due
to clamping force and mass is:
( F + m1 g )
P= i ¾¾®(40)
cos a
for simplification, by substituting:
Fi = m1 l ¾
¾®(41)
now,
m (l + g )
P= 1 ¾
¾®(42)
cos a
Once a pre-load is applied to a bolted joint the bearing friction, under nut or under head friction,
will also help prevent it from loosening. The pre-load Fi will exert a frictional force of mbFi on
the bearing surface where mb is the under head/nut bearing friction coefficient. This can be
approximated by two forces mbFi tanb in the direction of the pre-load and mbFi/cosb in the
direction of friction force on the thread where b = tan-1(p/pD). Now the total effective
downward force is:
Page 9
m1 (l + m b l tan bsgnx& + g )
P= ¾
¾®(43)
cos a
For most cases b and mb are very small compared to the contribution from the pre-load l and g.
Therefore, for simplicity, we can omit the friction term from the above equation, hence making
it same as equation (42).
Now the equation of motion (10), (11) can be re-written for this case as:
m1 (l + g ) m F
m1 X&& = - sin b + F - b i sgnx& ¾
¾®(44)
cos a cos b
m (l + g )
m1Y&& = - 1 cos b + R ¾
¾®(45)
cos a
m1 (l + g )
R= cos b - m1 d 0w 2 cos w t sin q > 0 ¾
¾®(46)
cos a
This will require;
(l + g ) cos b
>1 ¾
¾®(47)
d ow 2 sin q cos a
Now the equation of motion that will govern the state of relative slip can be re-written as:
(l + g ) æ (l + g ) mbl ö
&x& = d ow 2 cos w t cosq - sin b - m t çç - d ow 2 cos w t sin q + cos b + ÷÷ sgn x& ¾
¾®(48)
cos a è cos a m t cos b ø
By using the same substitutions as in the previous case the corresponding equation of motion is:
by substituting:
(l + g ) sin b
G= ¾
¾®(50)
d ow 2 cosq cos a
f = m t tan q ¾
¾®(51)
1 æ (l + g ) cos b mbl ö
G= ç + ÷ ¾®(52 );
¾
d ow 2 sin q çè cos a m t cos b ÷ø
get;
x&& = cost - G + f (cost - G) sgn x& ¾
¾®(53)
Equation (53) is identical to the equation (25) for the conveyor case but with different values for
G, f and G.
Page 10
Hence, the solution to this equation will also be of the same form as that for equation (25) with
modified constants.
By using the above analysis, the trends of different variables of vibration loosening can be
established. As the above analysis only considers the specific case with upward and downward
slipping with momentary stops, the current trend analysis does not cover the full extent of the
solution domain.
Another mechanism of vibration loosening is the inertial torque. A wheel nut will be subject to
rotational inertia due to rotation of the wheel. The equivalent inertial torque (TI) on a nut with
polar moment of inertia J and a rotational acceleration y&& can be given as:
TI = Jy&& ¾
¾®(54)
For a right hand thread, an anti-clockwise acceleration of the wheel (stud/shaft) will cause the
nut to tighten while clockwise acceleration of the shaft will cause loosening of the wheel nut if
the above inertial torque TI exceeds the prevailing torque Tp described in equation (3). A
sudden anti-clockwise deceleration of the shaft could loosen a right hand wheel nut if adequate
pre-load is not provided. Similarly, if the shaft is rotating clockwise, screwing of any thing to
this shaft should be done with a left handed thread.
Rotational acceleration y
q&& may have a harmonic form if the wheel is subject to harmonic
motion. In this case, intermittent loosening may occur when the dynamic torque
momentarily exceeds the prevailing torque.
Once the polar moment of inertia of the nut and the possible acceleration levels are established
for a particular joint, the minimum prevailing torque requirement can be calculated using
equation (54). Then, using equation (3), the necessary minimum pre-load Fi to prevent this
form of loosening can be calculated. Please note that if the joint is subject to impact loads the
resulting accelerations could be very large.
Page 11
In the present analysis, rectilinear vibration has been considered. This assumes the in-phase
vibration in any two mutually perpendicular directions. For simplicity, two directions of
excitation (q), namely, lateral (q =-b) and longitudinal (q = 90-b) to the fastener axis have been
considered. These two cases provide uniform effects along the full extent of the thread.
Rectilinear vibration applied at any other direction will have different effects on diametrically
opposite sides of the thread. These cases are not covered under current analysis.
For an M12 coarse threaded fastener with under head/nut and thread friction coefficient of 0.15
subjected to a vibration amplitude of 1mm lateral to the fastener axis, the effect of frequency of
vibration at different pre-tension levels on vibration loosening is shown in Figure 3. As shown
in the figure, the required frequency of vibration for loosening is increasing with increasing pre-
tension. Increase in negative mean velocity indicates faster loosening. For a given pre-load,
with increasing frequency, the loosening velocity will reach a negative maximum value. When
there is no pre-tension vibration loosening will start at a very low frequency.
Figure 3: 2Effect
Figure of Frequency (M12, p=1.75,mt=ma=1mm,
Effect of Frequency (M12-Coarse, ut,b=0.15, b=0.15,Lateral)
d0=1,Lateral)
0
0 1000 2000 3000 4000 5000 6000
-0.0001
Pre-Load
-0.0002 0kN
0.1kN
0.5kN
1kN
-0.0003 2kN
4kN
8kN
16kN
-0.0004
32kN
-0.0005
-0.0006
Frequency (Hz)
When interpreting the results we should keep in mind that once the loosening starts at a given
pre-load the pre-load will progressively reduce. Once the pre-load is reduced the rate of
Page 12
loosening will increase until the failure occurs. Therefore, it is prudent in the design to
eliminate the possibility of any loosening.
Figure 4 shows the effect of amplitude of vibration on mean velocity (loosening is –ve) while
maintaining the lateral vibration frequency at 750Hz with all other parameters similar to the
previous case.
-0.002
Pre Load
-0.004 0
0.1kN
0.5kN
1kN
-0.006 2kN
4kN
8kN
16kN
-0.008 32kN
-0.01
-0.012
Amplitude (m)
As can be expected, this also shows that the increasing pre-load will increase the minimum
amplitude required to start loosening. Once loosening starts the pre-load will be lost and speed
of loosening further increases. A screw with no pre-load will start to loosen at a very small
amplitude at the frequency of 750Hz.
Figures 5 and 6 show the corresponding cases for longitudinal vibration. In comparison, it is
evident that the effectiveness of longitudinal vibration in loosening is significantly less than that
of lateral vibration. If the vibration is occurring in the longitudinal direction a relatively small
pre-load will assure that no vibration loosening will take place.
Page 13
Effect
Figure 5: Effect of of Frequency
Frequency (M12, p=1.75, mtut,b=0.15,a=1mm,
(M12-coarse, Longitudinal)
=mb=0.15, d0=1.0, Longitudinal)
0.0001
0
0 1000 2000 3000 4000 5000 6000
-0.0001
Pre-Load
-0.0002
0.001kN
-0.0003 0.1kN
0.5kN
-0.0004 1kN
2kN
-0.0005 4kN
8kN
-0.0006 16kN
32kN
-0.0007
-0.0008
-0.0009
-0.001
Frequency(Hz)
0.002
0
0 0.005 0.01 0.015 0.02 0.025
-0.002 Pre-Load
-0.004
.001kN
0.1kN
-0.006 0.5kN
1kN
-0.008 2kN
4kN
-0.01 8kN
16kN
32kN
-0.012
-0.014
-0.016
-0.018
Amplitude (m)
Page 14
Effectofofthread
Figure 7: Effect Threadfriction
Friction(M12,
(M12-coarse,
p=1.75,f=750Hz, ub=0.15,
mb=0.15, d0=1.0,a=1.0mm, Lateral)
freq=750Hz, Lateral)
0.0015
0.001
Pre-Load
0.0005
0.0kN
0.1kN
0.5kN
0
1kN
0 0.1 0.2 0.3 0.4 0.5 0.6
1.5kN
2kN
-0.0005
-0.001
-0.0015
Thread Friction Coe
Figure
Figure8:7 Effect of thread friction (M12, p=1.75, mb=0.15,
Effect of Thread Friction (M12-coarse, f=750Hz,
d0=1.0,
ub=0.15,
freq=750Hz, Longitudinal)
a=1.0mm,Longitudinal)
0.02
0.015
0.01
Pre-Load
0.005
0.001kN
0 0.1kN
0 0.1 0.2 0.3 0.4 0.5 0.6 0.5kN
1kN
-0.005 1.5kN
2kN
-0.01
-0.015
-0.02
-0.025
Thread Friction Coe
Page 15
Figure 8 shows no potential for loosening even at 0.1kN pre-load; however, it shows some
strange behaviour at no pre-load. This could be due to an instability in the solution at zero pre-
load and may be ignored. Further refinement in simplifying assumptions may be required to
avoid this instability.
0.0001
0
0 0.1 0.2 0.3 0.4 0.5 0.6
-0.0001
Pre-Load
0kN
-0.0002 0.1kN
0.5kN
1kN
-0.0003 1.5kN
1.8kN
-0.0004
-0.0005
-0.0006
Nut Bearing Fric Coef
0.01
0.008
0.006
Pre-Load
0.004
0.001kN
0.1kN
0.5kN
0.002
1kN
1.5kN
0 1.8kN
0 0.1 0.2 0.3 0.4 0.5 0.6
-0.002
-0.004
-0.006
Bearing Fric Coef
Page 16
As can be expected, and as seen in Figure 9, at zero pre-load the bearing friction coefficient has
no effect on loosening. This is due to no under head/nut contact at zero pre-load. With
increasing pre-load the minimum friction coefficient requirement goes down showing that a
pre-load of 2kN will not loosen the screw at any friction coefficient for the given vibration
condition.
The case for longitudinal vibration is shown in Figure 10. This shows a similar behaviour to the
effect of thread friction. At non-zero pre-load, loosening will not occur, and at zero pre-load
the results shown may not present a valid case as zero pre-load nullifies the effect of bearing
friction.
0.00005
0
0 500 1000 1500 2000 2500 3000
-0.00005
-0.0001
Coarse
fine
-0.00015
-0.0002
-0.00025
-0.0003
Frequency
Figureof11
Figure 12: Effect thread pitch
Effect (M12,
of Thread p=1.75,
Pitch mt=mb=0.15,
(M12,F=0.1kN, d0a=1.0mm,Longitudinal)
ut,b=0.15, =1.0, Fi=1.0kN, Longitudinal)
0.0008
0.0006
0.0004
0.0002
Coarse
0
fine
0 500 1000 1500 2000 2500 3000
-0.0002
-0.0004
-0.0006
-0.0008
Frequency
Page 17
As seen in Figure 11, loosening will start at the same frequency but the coarse thread will
loosen significantly faster than the fine thread under lateral vibration.
Figure 12 shows that a fine thread will cause the screw to tighten while a coarse thread will
cause it to loosen under longitudinal vibrations. This opens up a new possibility that vibration
tightening, through design, could be applied to machinery subject to severe vibrations. Further
research into this window of conditions may be of great importance for future developments.
0
0 10 20 30 40 50 60
-0.0001
Pre-Load
-0.0002
0.0kN
0.1kN
-0.0003 0.5kN
1kN
1.1kN
-0.0004
-0.0005
-0.0006
Flank Angle
0.0015
0.001
Pre-Load
0.0005
0.001kN
0.01kN
0 0.02kN
0 10 20 30 40 50 60 1kN
1.1kN
-0.0005
-0.001
-0.0015
Flank Angle
Page 18
From Figure 13 it is evident that the flank angle does not affect loosening under lateral
vibration. As discussed before, pre-load will have a significant effect on vibration loosening.
As seen in Figure 14, at very low pre-loads lower flank angles are susceptible to vibration
loosening. Figures 15 and 16 below show the effect of flank angle under different levels of
vibration (amplitude 1mm, increasing frequency) with thread and bearing friction coefficients
of 0.15 and a pre-load of 1kN for lateral and longitudinal vibration conditions respectively.
Figure 15:Figure
Effect14
of thread flank angle (M12, p=1.75, mt=mbub,t=0.15,
Effect of Flank Angle(M12-Coarse, Fi=1kN,
=0.15, da=1.0mm,
0=1.0, Fi=1.0kN,
Lateral)
Lateral)
0.00005
0
500 550 600 650 700 750 800 850 900 950 1000
-0.00005
Pre-Load
0
5
-0.0001
10
15
20
-0.00015
25
30
-0.0002
-0.00025
-0.0003
Frequency (Hz)
Figure
Figure 16: Effect 15 Effect
of thread flankofangle
Flank Angle(M12, Fi=0.1kN, ub,t=0.15, a=1.0mm, Longitudinal)
(M12, p=1.75, mt=mb=0.15, d0=1.0, Fi=1.0kN, Longitudinal)
0.0001
0
0 500 1000 1500 2000 2500 3000
-0.0001
Pre-Load
-0.0002
0
5
-0.0003 10
15
-0.0004 20
25
30
-0.0005
-0.0006
-0.0007
-0.0008
Frequency Hz
Page 19
As evident in Figures 15 and 16, lower flank angles again will start loosening at lower vibration
levels under both lateral and longitudinal vibration.
This is one of the reasons why “buttress” threads are not recommended for applications subject
to vibration excitation.
As discussed, all of the above considered parameters will have an impact on vibration
loosening. In general, as shown, adequate pre-load will solve almost all vibration loosening
concerns. Designers should consider these effects when designing machines subject to
significant dynamic loads and vibration conditions.
7.1 Pre-Load
Depending on the application, there are various methods of avoiding vibration loosening. As
discussed in the previous sections, the most reliable and economical method of preventing
vibration loosening is applying adequate pre-load. This will not only stop the chance of
vibration loosening but also provide a sound fatigue-tolerant joint. As shown in Figure 3 a
16kN pre-load on an M12 bolt requires a vibration amplitude of 1mm applied at approximately
2800Hz to start vibration loosening. Similarly as shown in Figure 4 for a frequency of vibration
of 750Hz an amplitude of 14.5mm is required to loosen a M12 bolt tightened to a load of 16kN.
For a Class 4.6 M12 bolt, 16kN is approximately 80% of the yield load. Same figures (3 & 4)
also show (by interpolation) that a pre-load of 13kN (65% yield load) would still require severe
vibration conditions to start loosening. This confirms our experience that a pre-load higher than
approximately 65% of the yield load of the fastener is adequate to prevent loosening at
vibration levels commonly experienced in the field. This is more so for larger diameter
fasteners. Using Equations 1 and 3 the tightening and prevailing torque correspond to 65%
yield load for this fastener is approximately 40Nm and 32Nm respectively. For this case, no
additional devices to prevent loosening are required.
There are situations that the joint may not function if a large pre-load is applied. If this joint is
subject to vibration an additional device to prevent loosening is required. There are a large
number of anti-loosening devices available4.
Lock-nuts (use of two nuts; one on top of the other) are commonly used as a device for
preventing loosening. It is our experience that most of the practitioners incorrectly use the lock
nuts. For example, it is common to use two identical nuts for this purpose. This method will
hardly ever work as the both nuts will rotate together in a vibration environment loosening the
joint. In this method, the interface between the two nuts will experience the pre-load and hence
be safe from loosening one nut with respect to the other; however, as the clamp interface will
not see a significant pre-load the joint could still loosen by turning both nuts together.
Page 20
Ideally two different nuts – one full nut and one half nut (Lock/Jam nut) – must be used in order
to achieve a sound locking mechanism to prevent vibration loosening. The Lock/Jam nut must
have a loose fitting thread and be made of a lower grade material. Here again we have seen that
some practitioners use the full nut first with the half nut on top, tighten the first nut to achieve
the desired tension (typically small) and then tighten the half nut to a greater tightness. This
procedure is not correct and will not create a sufficient locking mechanism. In this case again,
as the half nut cannot provide enough tensile force to create a
jam, both nuts will rotate together if the tension in the joint is
lost.
The correct method (Figure 17) is to use the half nut first and
snug tighten it (typically small tension). At this point the bolt
thread will be pulling the lock nut down and be bearing on the
upper flank of the nut thread. The full nut is installed next.
This nut will always bear on its upper thread flank while Figure 17: Correct use of lock nut
applying tension on the bolt. As this nut is further tightened,
the downward force applied on the lock nut by the full nut will push it down, imparting more
clamp load on the joint. This should be continued until the lower flank of the half nut bears on
the bolt thread. It is important that the half nut is held while the full nut is tightened.
In this case, the two adjacent nuts are bearing on opposite thread flanks and the threads are
deformed in opposite directions. Therefore, it will take a much larger torque to rotate them both
together, hence providing a non-loosening nut even if the tension in the joint is lost. There are,
however, a number of issues to keep in mind regarding this tightening method. Firstly, if the
half nut is over tightened at the start there is a great possibility that a larger than desired pre-
tension may end up in the bolted joint. Secondly, the standard torque tension relationships do
not apply for this tightening process. In general, if you have to use a lock nut, that implies that
the tension in the joint is not critical. If it is a critical joint a direct tension sensing bolt such as
SMARTBOLTä may be used.
As discussed, there are a large number of devices available for preventing vibration loosening.
Typically, these devices are necessary only if the joint is not capable of carrying enough pre-
load to prevent vibration loosening. If a minimum pre-load of » 65% Proof Load of the fastener
can be assured it is highly unlikely that anti-loosening devices will be needed.
8.0 CONCLUSION
Based on the above analysis and discussion the following generalized rules of thumb may be
established:
Bibliography:
1. Bickford, J.H., An introduction to the design and behaviour of bolted joints, Marcel Dekker
Inc., New York 1990.
2. Fernando, S. “An engineering insight into the fundamental behaviour of tensile bolted
joints”, Journal of the Australian Institute of Steel Construction, Steel Construction, Vol 35,
No 1, March 2001.
3. Bykhovsky, I., Fundamentals of Vibration Engineering, Mir Publishers, Moscow, Jan 1972.
4. Barret, R.T., “Locking Methods for Fasteners”, American Fastener Journal, Nov/Dec 1998.