Engineering Failure Analysis 13 (2006) 1410–1421

www.elsevier.com/locate/engfailanal

Effect of torque tightening on the fatigue strength of

bolted joints
a,* b
José Marı́a Mı́nguez , Jeffrey Vogwell
a
Departamento de Fı́sica Aplicada II, Facultad de Ciencia y Tecnologı́a, Universidad del Paı́s Vasco, Aptdo. 644, 48080 Bilbao, Spain
b
Department of Engineering and Applied Science, University of Bath, Bath, UK

Received 26 August 2005; accepted 8 October 2005

Available online 8 February 2006

Abstract

The effect of tightening torques on the life of plates bolted using single and double lap joints was investigated. The effect of
plate thickness using an aircraft grade aluminium alloy with double lap joints was also studied. Constant amplitude fatigue
tests, under load control, were carried out, with a near zero stress ratio, on plain specimens (for bench mark purposes) and on
both single and double lap joint specimens, for which several torque levels were applied on the bolted joints. The objective of
the fatigue tests was to demonstrate failure trends for each joint type, material thickness and torque loading, rather than the
generation of comprehensive S–N curves. Possible factors that affected the fatigue life of the bolted joints are discussed and
conclusions are drawn with respect to the beneficial effects of tightening torques on the bolted single and double lap joints.
Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: Bolted joints; Fatigue life; Stress concentration

1. Introduction

A major concern engineers have regarding the long service life of both commercial and military aircraft is
the detrimental ageing resulting from metal fatigue or corrosion. In fact, design philosophies have changed
over recent years in the way that fatigue life, durability and damage tolerance have replaced static strength
and safe life as the key elements of design. This, together with improved materials, sophisticated design anal-
ysis and thorough testing programmes, has significantly reduced the number of structural failures in service.
Aircraft structures are still largely constructed from high strength aluminium alloys, because the low den-
sity provides an optimum strength-to-weight ratio material. Connecting structural members is still widely
achieved by means of bolted joints. In this respect a number of authors [1–9] have detailed the performance
of bolted joints under static loading conditions.
Despite this, the behaviour of bolted joints under cyclic loading is still not well documented and fully
understood. The fatigue life performance of a bolted joint is dependent on several factors, such as size of

*
Corresponding author. Tel.: +34 94 601 3371/2000; fax: +34 94 601 5963/464 8500.
E-mail address: [email protected] (J.M. Mı́nguez).

1350-6307/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2005.10.012
 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421 1411

the bolt, the number and position of bolts used, the level of preload or torque tightening applied to the bolt,
the material plate thickness and surface roughness. How all these factors fully influence the fatigue life of a
joint is complex and has still not been thoroughly investigated.
Such analysis requires an accurate knowledge of fatigue and fracture relationships to achieve adequate
structural integrity. This has been the objective of the present work and of much of the abundant literature
on the subject [10–20].
This work, therefore, attempts to show how bolted joints in different states behave under cyclic loading. In
particular, the main objective of this project was to investigate the effect of bolt torque tightening and the
influence of material thickness, thus achieving different levels of joint preloading.
Following this goal, the trends of fatigue life curves were analysed at various levels of torque tightening on
both single and double lap joints, with different plate thicknesses, loading bolts in shear.

2. Load transmission in bolted joints

When the bolt is not torque tightened there is no load to cause compression of the members as the plates
press together. The joint members can freely move over each other and load is transferred from bolt to the
plates as shown in Fig. 1. If the stress levels within the joint material are sufficiently high to cause plastic defor-
mation in the material, elongation of the holes may occur due to the stress transmitted through the bolt into
the members and into the plate members.
In this case the main failure modes for a joint loaded in shear would be as shown in Fig. 2. These are: (a)
shear of the fastener, (b) tensile failure of the members or of the splice plates, (c) shear of the members, (d) tear
out of the members, and (e) bearing either of the fastener or of the members. In the case of the tensile failure
(b) of a double lap joint it is usual for the member to fail rather than the splice plate, as the member will be
subject to twice the load that the splice plate is subject to, if they are of equal thickness.
Alternatively, if the plate members are compressed by the action of a torque applied on the bolt, friction
occurs at the contact surfaces of the plates and thus member slip can be prevented. Then the load will be trans-
mitted through the joint directly from the member to the plate members by friction, instead of transferred
through the bolt to the plate members hole.
In this way, the stress concentrations caused by the bearing of the bolt on the edges of the holes are signif-
icantly alleviated, which is beneficial with respect to any failure mode shown in Fig. 2.
In practice it is not possible to accurately measure the bolt tension in a joint in service. To ensure that a
desired preload has been achieved with a bolt, it is more practical to use a torque wrench to apply the load
to the bolt through the nut.
The amount of torque T required to achieve a set amount of preload F in a bolt depends upon thread pitch,
bolt diameter and the friction coefficient between the nut and bolt. According to Shigley [2], to a first-order
level of accuracy, it can be shown that
T ¼ KFd; ð1Þ
where K is the thread coefficient and d the bolt diameter.
The thread coefficient may vary between 0.12 and 0.20, depending on many variables, such as thread mate-
rial, surface finish and coefficient of friction. From tests typically it is found that
5T
F ¼ . ð2Þ
d

Fig. 1. A bearing joint, with the bolt acting as a shear pin.

1412 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421

Fig. 2. Failure modes for joints loaded in shear.

This is suitable when using normal nuts and bolts, but, when nuts with antiloosening devices are employed,
the extra torque required to fasten the nut, usually called prevailing torque (TP), must also be considered and
included within Eq. (1), with which

T ¼ KFd þ T P . ð3Þ
The main effect of preloading a bolted joint with a torque and the subsequent compression of the members
is to produce sufficient friction between the plate and the members of the joint, thereby preventing their slip
and thus the bolt bearing on the edges of the holes. This is sometime called a friction weld joint.
Nevertheless, even when the total compression produced by the tension of the bolt on the members of the
joint might be known from Eq. (2), its distribution and so the distribution of the friction forces over the con-
tact surfaces of the members with the splice plates depends on many variables and is difficult to describe.
Therefore, it is hard to predict accurately the behaviour of friction joints up to the point that failure can occur
through either the gross section or the net section of the joint, for the highest stress concentration does not
necessarily appear in the net section, when friction is taking place.
However, Fig. 3 can illustrate, at least qualitatively, how a friction joint will behave under static loading in
shear. As the load is uniformly increased the joint as a whole will elastically deform, following section (1) of
the curve, until the critical load of the joint is reached, at which the joint starts slipping. At region (2) the joint
slips and the members and splice plate move until the bolts start bearing on the boltholes. Then elastic defor-
mation continues to occur in region (3), until yielding of the joint members or of the bolts occurs, with plastic
deformation, in region (4). Final failure of the joint will then happen at point (5).
This work studies how the torque applied to the bolt influences the behaviour of the joint and its fatigue
life, while transmitting a shear load.
 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421 1413

Fig. 3. Behaviour of a joint under increasing loading.

3. Single lap joints

This type of joint is more complex than the double lap joint, since the load transfer path is not symmetric
and a moment will be set up through the bolts and transferred into the plate (see Fig. 4). This means that there
is bending as well as shearing of the joint.
Effectively, as the joint is loaded, the bolts will pivot around the internal faces of the plate members. This
will cause the members and the splice plate to deform at the two bolt areas, as shown in Fig. 4.

Fig. 4. Deformation of a single lap joint under shear loading.

1414 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421

The bending moment that is applied to the members and the splice plate upon loading is approximately
given by
Pt
M¼ ; ð4Þ
2
where P is the shearing force and t the total thickness of the member together with the splice plate.
This means that the described distorting effect is bigger when the thickness of the joint members is
increased. For this reason, the single lap joint is only used with thin members and, for the same reason,
our tests with single lap joints were carried out only with members 2 mm thick.

4. Test specimens

The fatigue tests were carried out on specimens of the shape and size shown in Fig. 5, which were designed
to conform as closely as practicable to ASTM E466 – 82 and also to fit the INSTRON model 1332 test
machine with a 100 kN load capacity, which was to be used for the tests.
Tests were carried out on ‘‘dog bone’’ specimens with a centrally drilled hole, because these have a well
defined stress concentration factor. These specimens will be referred to as plain specimens and were used
for bench mark comparison purposes. Afterwards the specimens were split in two parts, and these linked with
splice plates, some by means of a double lap joint, as shown in Fig. 6, and some by means of a single lap joint,
as shown in Fig. 7.
The specimens were made of aircraft specification aluminium alloy designated BS L165, which was supplied
both as 2 mm sheet and as 5 mm sheet. In the two cases the aluminium alloy was in the T6 temper condition.
This allowed to have specimens with different thickness, with which the combined influence of the thickness
and of the applied torque on the fatigue resistance of the joint could be investigated.
The joints were constructed using class 12.9 M5 set screws, mild steel washers and aluminium ny-lock nuts.
The holes in the members and the splice plates were drilled and reamed to U 4.90 mm to suit the sank of the set
screws. Then ny-lock nuts were used to prevent relaxation of preload during fatigue testing.
During joint assembly the nuts were run down to 1 mm above the washer using a spanner, before the final
run-down torque was applied. Finally, the required torque loading was achieved using a torque wrench, RS
Stock No. 547-379. This was carried out for each specimen to ensure similar torque loading on the nuts was
achieved.

5. Experimental procedure

The experimental work consisted of initially testing some plain specimens with a hole (see Fig. 5). Then the
plate was divided in two parts and the two parts were assembled with a double lap joint (see Fig. 6) or with a
single lap joint (see Fig. 7).

Fig. 5. Plain specimen with a drilled hole (U 5 mm).

 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421 1415

Fig. 6. Double lap joint specimen.

Fig. 7. Single lap joint specimen.

All of them suffered a cyclic stress regime until failure occurred. Successive tests were carried out with an
increasing maximum stress, and the fatigue life was computed in each case by the cycles to failure. Therefore,
although the objective was not to accurately draw S–N diagrams, the trend of the fatigue diagrams could be
studied with different loads and under different preloading conditions. So the influence of the tightening torque
on the fatigue resistance of the joints could be investigated.
1416 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421

All fatigue tests were conducted using an INSTRON model 1332 test machine with a 100 kN load capacity.
The cyclic stress regime that was used was purely tensile with a sinusoidally alternating stress at a frequency of
20 Hz. This load regime is referred to as constant amplitude fatigue regime. The minimum stress was main-
tained as a small positive load, to prevent any compressive loading effects on the test specimens and on the
joints. This meant that the stress ratio rmin/rmax was maintained as near zero as possible and so the mean
stress equalled the stress amplitude:
rmax þ rmin rmax
rmean ¼ ¼ ¼ ra . ð5Þ
2 2
The 54 tests are numbered and summarised in Tables 1–4, in which the thickness of the plate is indicated in
each case, as well as the maximum and minimum loads of the cycling regime. The maximum stress and the
stress ratio, also shown in the tables, are calculated over the net section area of the specimens, considering
that in all cases the plate width is 25 mm and has a 5 mm hole, as it happens both in the tested joints, where
the hole is needed for the joining bolts, and in the plain plate, in which the only purpose of the hole is to make
the results consistent and comparable.
Tables 2–4 also show the torque applied to the joint in each test. The level of torque to be applied was
assessed by carrying out a series of run-down tests on a bolt and nut combination. Due to the fact that the
nuts were a ny-lock type to prevent loss of preload during testing, they required a level of prevailing torque
to overcome the frictional resistance of the nylon insert in the nut.
As a result of the run-down test, the lower torque level for fatigue testing of the joints was set at 1 Nm, to
ensure that the prevailing torque of the nylon insert was overcome and only a small amount of joint compres-
sion was achieved. This torque level resulted in a preload of 1 kN.
The upper torque level for the ny-lock nuts was set at 6 Nm to ensure that only elastic deformation of the
thread occurred and that no stripping of the thread from the inside nut took place. Then intermediate torque
levels were set at 2.3 and 3.5 Nm.
Finally, the use of plain stainless steel nuts enabled a torque of 8 Nm to be applied, as thread stripping
occurred at a higher torque level. In this case a lock nut was applied following the plain nut, to ensure no loss
of preload might happen during testing.

Table 1
Fatigue test results for plain specimens.
Test No. Thickness (mm) Pmax (kN) Pmin (kN) rmax (MPa) Stress ratio Cycles to failure
1 2 4.8 0.3 120.0 0.06 155,340
2 2 6.1 0.4 152.5 0.07 87,140
3 2 7.3 0.4 182.5 0.05 35,170
4 2 8.7 0.2 217.5 0.02 28,540
5 5 12.6 0.4 126.0 0.03 137,750
6 5 15.3 0.2 153.0 0.01 98,950
7 5 18.4 0.4 184.0 0.02 56,260
8 5 21.2 0.3 212.0 0.01 30,600

Table 2
Fatigue test results for 2 mm single lap joints
Test No. Torque (Nm) Pmax (kN) Pmin (kN) rmax (MPa) Stress ratio Cycles to failure
9 1.0 3.6 0.2 90.0 0.06 68,020
10 1.0 4.9 0.4 122.5 0.08 35,380
11 1.0 6.4 0.3 160.0 0.05 14,900
12 1.0 7.6 0.4 190.0 0.05 6,600
13 6.0 3.4 0.1 85.0 0.03 128,700
14 6.0 4.5 0.2 112.5 0.04 32,880
15 6.0 6.3 0.4 157.5 0.06 15,110
16 6.0 7.3 0.2 182.5 0.03 7,850
17 6.0 8.4 0.2 210 0.02 4,890
 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421 1417

Table 3
Fatigue test results for 2 mm double lap joints
Test No. Torque (Nm) Pmax (kN) Pmin (kN) rmax (MPa) Stress ratio Cycles to failure
18 1.0 5.1 0.4 127.5 0.08 33,320
19 1.0 6.4 0.2 160.0 0.03 14,760
20 1.0 7.5 0.3 187.5 0.04 9,910
21 1.0 8.7 0.4 217.5 0.05 5,770
22 2.3 4.8 0.0 120.0 0.00 502,860
23 2.3 6.0 0.1 150.0 0.02 223,410
24 2.3 7.4 0.5 185.0 0.07 101,620
25 2.3 8.6 0.3 215.0 0.03 72,150
26 3.5 4.9 0.2 122.5 0.04 619,980
27 3.5 6.3 0.3 157.5 0.05 320,360
28 3.5 7.4 0.5 185.0 0.07 194,700
29 3.5 8.8 0.6 220.0 0.07 125,390
30 6.0 5.2 0.5 130.0 0.10 966,000
31 6.0 6.5 0.5 162.5 0.08 376,190
32 6.0 7.6 0.1 190.0 0.01 153,410
33 6.0 8.6 0.4 215.0 0.05 117,530
34 8.0 4.9 0.5 122.5 0.10 949,360
35 8.0 6.3 0.4 157.5 0.06 364,740
36 8.0 7.4 0.2 185.0 0.03 241,610
37 8.0 8.7 0.4 217.5 0.05 168,470

Table 4
Fatigue test results for 5 mm double lap joints
Test No. Torque (Nm) Pmax (kN) Pmin (kN) rmax (MPa) Stress ratio Cycles to failure
38 1.0 10.5 0.5 105.0 0.05 40,680
39 1.0 12.2 0.1 122.0 0.01 114,050
40 1.0 13.7 0.5 137.0 0.04 17,030
41 1.0 15.6 0.0 156.0 0.00 10,790
42 1.0 18.5 0.1 185.0 0.01 7,120
43 1.0 21.4 0.3 214.0 0.01 5,310
44 1.0 23.4 0.1 234.0 0.00 3,420
45 3.5 12.5 0.4 125.0 0.03 974,860
46 3.5 14.8 0.3 148.0 0.02 616,140
47 3.5 16.6 0.4 166.0 0.02 145,050
48 3.5 18.2 0.2 182.0 0.01 25,460
49 3.5 21.0 0.2 210.0 0.01 31,390
50 6.0 12.9 0.3 129.0 0.02 1,098,010
51 6.0 15.4 0.2 154.0 0.01 777,390
52 6.0 18.5 0.6 185.0 0.03 445,600
53 6.0 21.6 0.3 216.0 0.01 96,340
54 6.0 24.3 0.3 243.0 0.01 79,940

6. Fatigue test results

Eight constant amplitude fatigue tests were carried out on plain specimens. Their fatigue lives are shown in
Table 1, as the cycles to failure.
Then nine fatigue tests were performed on the 2 mm single lap joint specimens, with torque loadings of 1
and 6 Nm on the bolts. The results are shown in Table 2.
Afterwards, twenty 2 mm double lap joints specimens were fatigue tested, with torque loadings of 1, 2.3, 3.5
and 6 Nm on the bolts. The corresponding fatigue lives appear in Table 3.
Finally, 17 fatigue tests were conducted on the 5 mm double lap joint specimens, with torque loadings of 1,
3.5 and 6 Nm. The results are shown in Table 4.
1418 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421

A limited number of tests were carried out for each specimen type, but each test series showed a close series
of trends and a minimum amount of scatter was observed. All the results from the 54 tests were regular, with
the only two exceptions of test 39, in which the fatigue life of the double lap joint was unusually long, and of
test 48, in which a premature failure took place, both compared to the other test results in their comparative
sets. Anyway, extra tests were carried out at intermediate maximum stress levels, to confirm the correct trend
of the test series, and so the two irregular tests shall not be considered.

7. Discussion

Although the objective of this work has not been the generation of full S–N curves, but to indentify trends
for each joint type, the influence of material thickness and torque tightening on joint fatigue life, the results
collected in Tables 1–4 are brought to S–N diagrams in Figs. 8–10.
Fig. 8 represents the fatigue life of a 2 mm single lap joint with different preloading torques, in comparison
with the behaviour of the plain plate.
Fig. 9 shows how a 2 mm double lap joint varies its behaviour when the preloading torque is increased. The
plain plate is taken as a reference.

Fig. 8. S–N curves for 2 mm single lap joints and plain specimens.

Fig. 9. S–N curves for 2 mm double lap joints and plain specimens.
 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421 1419

Fig. 10. S–N curves for 5 mm double lap joints and plain specimens.

Fig. 10 shows how the behaviour of a 5 mm double lap joint is improved when the applied torque is
increased. In this figure tests 39 and 48 are represented, but left apart, for being irregular from a statistical
point of view, as stated before.
With respect to the single lap joint, Fig. 8 reveals that the distorted joint, when transferring a shear load,
has a performance much poorer than the plain plate. This could be due to the distortion itself or to the slipping
of the members over the splice plate until the bolt bears on the hole edges, with which the stress concentration
factor becomes much greater and the fatigue life much shortened.
Nevertheless, the fact that increasing the tightening torque applied to the bolts does not have any effect on
the fatigue life for this type of joint reveals that the main cause of the poor performance of the single joint is
the distortion it suffers when loaded in shear, irrespective of the preloading torque, which would avoid the
mentioned slipping of the members. That distortion causes strong moments (see Eq. (4)) and so high bending
stresses, which accelerate the failure of the joint.
This is more certain when a high tightening torque is applied. Effectively, the failure mode for all the single
lap joints with a 6 Nm torque was through the splice plate at the edge of the washer area. A fatigue crack was
formed by the edge of the washer damaging the surface of the splice plate due to the bending moment exerted
on the washer by the bolt. This shows that by increasing the torque on the bolt the fatigue life is not improved,
but a standardised failure mode can be forced.
In the case of double lap joints things are different, since no geometry distortion is caused by shear loading,
for the shape of the joint is well kept unchanged for symmetry reasons.
Comparison of the 2 mm double lap joint S–N curves, shown in Fig. 9, demonstrates an improvement in
fatigue life when increasing the tightening torque on the bolts. At first, the joint with a low tightening torque of
1 Nm has a shorter fatigue life than the plain specimen, whereas increasing the preloading torque involves a
marked improvement in the fatigue life, which takes a large step from 1 to 2.3 Nm and up to 3.5 Nm. Nev-
ertheless, beyond that tightening torque there is not further improvement, but the fatigue life remains much
the same.
Curves of Fig. 10 demonstrate a similar trend when dealing with 5 mm double lap joints, with the only dif-
ference that, in this case, increasing the tightening torque beyond 3.5 Nm still brings about an improvement in
the performance of the joint.
The improvement in fatigue life can be attributed to the way the joint transmits the applied shear load.
When the low torque does not avoid slipping between the members and the splice plates, the load is transmit-
ted through the interaction of the bolt bearing against the bolthole, whereas, as the torque is increased, a lar-
ger part of the load is transmitted by friction between the members and the splice plates, before slip occurs in
the joint, and a smaller part requires to be transmitted by bearing the bolt on the bolthole.
1420 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421

Fig. 11. Stress concentration factors for test specimen sections.

This transfer of the load transmission from the bolt bearing against the hole to the friction between the joint
members and the plates represents a variation of the stress concentration factors, that explains the beneficial
effect of the tightening torque on the double lap joints. Effectively, according to Shigley [2] and as shown in
Fig. 11, when the load is transmitted through the plain specimen with a drilled hole, the stress concentration
factor is 2.5. Then, if the tightening torque preloading the double lap joint does not avoid slipping between the
members and the plates, the bolt goes against the bolthole and the full load is transmitted by the bolt bearing
on the hole, the stress concentration factor being increased to 5.5. This is the reason for the fatigue life being
shorter when the torque is 1 Nm in Figs. 9 and 10. Finally, if an efficient torque is applied on the joint, the
compression of the members helps a wider distribution of the load over the members and the stress concen-
tration factor may be lowered up to 1.6, in such a way that a greater load is transmitted by friction, before slip
occurs in the joint, and a bearing-type joint transfers a lower proportion of the load. This is reflected in Figs. 9
and 10, in which increasing the tightening torque is shown to prolong the fatigue life of the joints, making it
more than 10 times longer.
The improvement in fatigue life can also be attributed to the residual compressive stress surrounding the
bolthole caused by the compression of the member material by the bolt pretension. As the torque is increased
the pretension is also increased, so the member material is placed under greater compression. This residual
compressive stress, besides assisting in negating the stress intensity factor caused by the loading of the bolthole
by the bolt sank, will be beneficial to prevent fatigue cracks from growing from the bolthole.
Finally, the extension of the residual compression area in the material surrounding the bolthole, caused by
the torque applied to the bolt, may be greater the greater the rigidity of the splice plates and of the joint mem-
bers. This is the reason for the thicker joint (5 mm) benefitting more from a higher tightening torque than the
thinner joint (2 mm).

8. Concluding remarks

The experimental work carried out in this project demonstrates how both pretensioning the bolts and the
material thickness influence the fatigue life of double lap joints.
The tightening torque applied to the bolts results in a compression of the joint plate members, which causes
friction between them, and so prevents their relative slipping. Thus the bearing of the bolt against the hole
edges is avoided, or at least relaxed, and a high proportion of the shear load may be transmitted through
the joint by friction. In this way, the load is distributed over a larger area around the hole, the stress concen-
tration factor is diminished and so the fatigue performance of the joint is much improved.
 J.M. Mı́nguez, J. Vogwell / Engineering Failure Analysis 13 (2006) 1410–1421 1421

In fact, in all torque tightened joints a greater fatigue life resulted, typically greater that 10 times longer
than with the none torque tightened joints, which is very much in favour of applying tightening torques to
the joints.
Nevertheless, once the tightening torque is strong enough to prevent the slipping of the joint members and
the bearing of the bolt against the hole edges, its only benefit is to spread the load transmission over a wider
area. Then, as the rigidity of the joint members helps extend this area, a thicker joint may benefit from a high
tightening torque further than a joint between thinner members.

References

[1] Gould HH, Mikic BB. Areas of contact and pressure distribution in bolted joints. Trans ASME 1972;94(3).
[2] Shigley J. Mechanical engineering design. New York: McGraw Hill; 1977.
[3] Kulak GL, Fisher JW, Struik JHA. Guide to design criteria for bolted and riveted joints. New York: Wiley; 1987.
[4] Blake A. Practical stress analysis in engineering design. Marcel Dekker Inc; 1990.
[5] Carvill J. Mechanical engineer’s data handbook. Bath: Institute of Mechanical Engineers, Bath Press; 1993.
[6] Chen Wen-Hwa, Lee Shyh-Shiaw, Yeh Jyi-Tyan. Three-dimensional contact stress analysis of a composite laminate with bolted joint.
Compos Struct 1995;30:287–97.
[7] Cooper C, Turvey GJ. Effects of joint geometry and bolt torque on the structural performance of single bolt tension joints in
pultruded GRP sheet material. Compos Struct 1995;32:217–26.
[8] Sundarraj N, Dattaguru B, Ramamurthy TS. Analysis of a double shear lap joint with interference pin fit. Comput Struct
1995;55:357–63.
[9] Bickford JH. An introduction to the design and behaviour of bolted joints. Marcel Dekker Inc; 1997.
[10] Potter JM. Fatigue in mechanically fastened composite and metallic joints, American Society for Testing and Materials, Special
Technical Publication 927; 1986.
[11] Heller M, Jones R, Williams JF. Analysis of cold expansion for cracked and un-cracked fastener holes. Eng Fract Mech
1991;39:195–212.
[12] Horn WJ, Schmitt RR. Influence of clamp-up force on the fatigue strength of bolted composite joints. AIAA J 1992;32:665–7.
[13] Ossleton R. Optimisation of bolted aircraft materials with regard to fatigue, BEng thesis, University of Bath, Bath; 1997.
[14] Hibberd N. An Investigation into the optimisation of bolted aircraft materials with regard to fatigue, BEng thesis, University of Bath,
Bath; 1997.
[15] Lazzarin P, Milani V, Quaresimin M. Scatter bands summarizing the fatigue strength of aluminium bolted joints. Int J Fatigue
1997;19:401–7.
[16] Lazzarin P, Milani V, Quaresimin M. A re-analysis of fatigue data of aluminium alloy bolted joints. Int J Fatigue 1997;19:579–88.
[17] Ball DL, Lowry DR. Experimental investigation on the effects of cold expansion of fastener holes. Fatigue Fract Eng Mater Struct
1998;21:17–34.
[18] Szolwinski MP, Farris TN. Linking riveting process parameters to the fatigue performance of riveted aircraft structures. J Aircraft
2000;37:130–7.
[19] Gaerke J, Zhang X, Wang Z. Life enhancement of fatigue-aged fastener holes using the cold expansion process. In: Proceedings of the
Institute of Mechanical Engineers, vol. 14; 2000; p. 281–92.
[20] Vogwell J, Navid-Chakherlou T, Mı́nguez, JM. The effect of cold expansion on fatigue resistance of fastener holes. In: Proceedings of
the 10th international congress of fracture, Honolulu; 2001.