Finite-Element-Based Methods for the Fatigue Design of Bolts and Bolted Joints

Author(s): Kai Buhr, Wolfgang Haydn, Manfred Bacher-Hoechst, Ulrich Wuttke and
Christina Berger
Source: SAE International Journal of Materials and Manufacturing , Vol. 2, No. 1 (2009), pp.
75-84
Published by: SAE International
Stable URL: https://www.jstor.org/stable/10.2307/26282736

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/10.2307/26282736?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

SAE International is collaborating with JSTOR to digitize, preserve and extend access to SAE
International Journal of Materials and Manufacturing

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
 2009-01-0041

Finite-Element-Based Methods for the Fatigue Design of Bolts and Bolted Joints

Kai Buhr, Wolfgang Haydn and Manfred Bacher-Hoechst

Robert BOSCH GmbH, Schwieberdingen, Germany

Ulrich Wuttke and Christina Berger

Institut für Werkstoffkunde, Technische Universität Darmstadt, Germany

Copyright © 2009 SAE International

ABSTRACT
FA working load
Due to the increase of computer performance, modern FSA additional bolt load
screw design is more and more based on finite element FM assembly preload FSA
analyses. Depending on the problem, different S elastic bolt resilience
possibilities of modelling bolted joints are reasonable. P elastic resilience of the
Traditionally, a nominal stress approach is used for the clamped parts
fatigue design of bolted joints. But, in modern lightweight
FAu=0
constructions it is more and more common to connect
threaded components directly without using a classic force F FAo=FA
bolted connection. In this case a local approach has to

FM
be used. In several investigations at the Fachgebiet und
S
Institut für Werkstoffkunde (Chair and Institute for
Materials Technology) in Darmstadt and the Robert
Bosch GmbH in Schwieberdingen different concepts for P
a local approach have been tested for the fatigue design
of threaded connections. As a result, an overview can be deformation f
given, which design concept and which model can be
properly used for calculations.

INTRODUCTION

Bolted joints have always been one of the most

δP δP
important and challenging fastening elements in the δS Φk =
design and construction of components. Especially the δP + δS
fatigue assessment is of vital importance, because
notched parts are far more sensitive to cyclic stress than
parts without notches. In order to use screws
economically regardless of their cyclic sensitivity, an
important goal of the design process is the minimization Figure 1: Graphic illustration of the mechanical load
of cyclic stresses in the screw. bearing behavior of an ordinary bolted joint described by
means of a simplified spring model /1/
As shown in Figure 1, the load distribution depends on
the joint partners’ elastic behaviour and is described by

SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1 75

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
the resilience factor k. Until opening of the joint starts, be evaluated. So, the limitations of the simplified
the working load is divided in an unloading of the analytical model of the guideline, especially referring to
clamped parts and an additional load FSA of the bolt. the nominal stress approach, can be avoided.
Because the bolt resilience is usually greater than the
resilience of the clamped parts, the major proportion of Since the grade of accuracy is strongly depending on
the working load is carried by the clamped parts. After the number, shape and local arrangement of the
the opening of the joint the additional bolt load is elements in the model, one can imagine that FEM
increasing. Therefore bolted joints usually are highly pre- models in general can become quickly non-applicable
loaded during assembly. for daily use. Especially for “normal” bolted constructions
the effort to model a bolted joint with the threads, the
The simple spring model of the VDI guideline 2230 /1/ material behaviour and the frictional contacts in all
may help to design bolted joints in many cases. Typically interfaces is, regarding to the efficiency, unacceptably
the nominal stress in the bolt is calculated to use the high (like in /2, 3, 4/ for example). Therefore two topics
nominal stress approach as it is described in the will be discussed within this article:
guideline. Similar to every model its use is limited by the
contained analytical and empirical approaches to the − The use of local concepts for the fatigue
following: assessment of threaded connections in general
and
− Valid for prismatic solids with limited lateral
dimensions and the assumptions, that the cross − A simplified local approach to use the advantages
sections remain flat under load and a linear stress of the local approach even for “normal” bolted
distribution exists. joints.

− The resilience factor Φ includes the elastic LOCAL FATIGUE ASSESSMENT OF BOLTS
resilience S of the screw and P of the bolted AND BOLTED JOINTS
parts. Both are calculated by the help of simplified
approaches, which can be incorrect or inaccurate A fatigue assessment in principle is based on the
and seldom be used for real cross-sections. comparison of failure relevant stresses to the fatigue
− Predicting the fatigue limit of screws is very strength. The failure relevant stresses are either
conservative since it is based on uncertain calculated or measured. Depending on the used
operands (for example special values in /1/ like approach for the fatigue assessment the fatigue strength
the distance of the line of action of the axial load can be measured on the whole structure, single parts
from the axis of the symmetrical deformation body (nominal approach) or material specimens (local
a, the distance of the bolt axis from the axis of the approach). Two approaches basically exist for a local
symmetrical deformation body ssym or the fatigue assessment of bolts:
resilience factorΦ ).
− the local elastic stress approach
− Inhomogeneous stress distributions, which
normally occur in components with bolted joints, − the local strain approach.
can only be considered to a very limited extent. For the local elastic stress approach, stresses are
− Especially the evaluation of multi-bolted joints computed at the failure relevant local spot of crack
contains so many assumptions that it inevitably initiation with the adoption of a linear-elastic material
leads to a conservative design. behaviour. The calculated elastic stresses are evaluated
with the fatigue strength determinate from force/stress
− The nominal approach in the guideline implies that controlled cyclic tests. Contrary to that, a fatigue
the fracture occurs in the first load-bearing thread assessment according to the concept of local strain is
turn of the bolt. Other notched regions or a based on the (overall) elastic-plastic strain amplitude in
variable load distribution along the bolts a notch root for example. The evaluated overall strain
longitudinal axis can not be evaluated. amplitude comprises both: the linear-elastic and the
The limitation of the analytical calculation model for elastic-plastic ratio /5/. The computation of the local
bolted joints in the guideline can partially be avoided by stresses and strains is done in consideration of the
means of the finite element method, because stiffness material yield behaviour specified by a cyclic stress
and load transmission within complex shaped and strain curve. The use of this concept assumes the same
loaded components can be calculated precisely. crack initiation behaviour for the material in the notch
Therefore stresses within these components can be root and a simple material specimen under the same
determined easily, if the corresponding FEM model is strain load.
only detailed enough.
Unfortunately the fatigue strength of metallic materials is
The result of a detailed FEM model, a model including strongly influenced by variables, such as mean stress /6/
the thread of the bolt and the contact in all load carrying or supporting effect. Both influences have to be included
interfaces, are local stresses and strains which have to in a fatigue assessment if values of this material

76 SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
property have to be compared or transferred from dimension M10 and a strength grade of 10.9. The tested
smooth or notched specimen to any other notched parts of flange and nut consisted either of heat treated
specimen or component with differing mean load. steel (42CrMoS4) or the aluminium wrought alloy (AW-
6082) and were assembled in a torque controlled
The tolerable stress- and strain amplitudes are tightening process. The torque is controlled by
influenced by the amount and nature of mean stress. measuring the elongation of the bolt. Overall, 6 series of
Generally, tensional mean stresses tend to decrease the bolted joints with clamped parts differing in their
tolerable amplitudes whereas compression mean tightening torques (MA) and materials have been tested
stresses increase them. Even if the VDI guideline 2230 for their corresponding fatigue limit. Table 1 gives an
/1/ disregards the influence of mean stress on the cyclic overview of the test results with values for the minimal
tolerable nominal stress amplitude SaD of bolts rolled load FU, the cyclic bearable testing force amplitude Fa50%
before heat treatment, investigations in /7/ as well as the range of scatter 1/TS and the standard deviation s. All
own tests prove such a dependency to a certain degree. series consisted of 25 test parts and were tested for
Approaches to consider a mean stress influence in the cyclic failure with test frequencies between 120 Hz and
concept of local strain are the so-called damage 150 Hz. The tests for fatigue limit were done by the
parameters. A well-known approach is the damage staircase method with a fatigue limit defining number of
parameter PSWT of Smith, Watson and Topper /8/ which 5x106 cycles. In the same way, series of single bolts and
factors the local strain amplitude εa and the maximum round material specimens have been tested. The
stress σo into the assessment: corresponding characteristics and results are listed in
Table 1.
PSWT = σ O ⋅ ε a⋅ ⋅ E = (σ a + σ m ) ⋅ ε a⋅ ⋅ E . (1)
Table 2 describes the characteristics of tested
specimen. The listed stress concentration factors Kt
Metallic notched specimens tolerate locally higher cyclic
were either calculated with a detailed FEM-model (index
stresses and strains than smooth specimens. This is
FEM) or analytically following /10/ (index FKM). Table 3
described by the supporting factor nσ which is the ratio of shows the positions of failure in the cyclic tested bolted
the stress concentration factor Kt and the fatigue notch joints and bolts. The majority of failures appeared in the
factor Kf and. The amount of support strongly depends first bearing thread and the transition between shaft and
on the local stress gradient, the strength and ductility of thread (shaft/thread). Some of the tested screws in the
the used material and the material class. A famous bolted joints failed in the transition between head and
example is the gradient concept of Siebel/Stiehler /9/. shaft (head/shaft). Numerical examinations in /12/ have
The influence of a local stress concentration is estimated shown, that this resulted from the process of tightening
with the so-called referred stress gradient χ*, which partially lead to slightly eccentrically tightened
bolted joints. A direct dependency of the location of
1 §dσ · (2) failure and the number of cycles to failure could not be
χ* = ¨ ¸
σ max © dx ¹ max found.

calculated from the local stress slope with reference to

the highest stress level max in the notch. This concept is
included in the FKM guideline analytical strength
assessment /10/. The supporting factor is defined as

Rm
− ( G + )
b G ⋅MPa
n = 1 + 4 χ * ⋅ mm ⋅ 10 , (3)

with the special material parameters aG und bG. A

modification of the supporting factor for a stress gradient
in the range 10 mm-1 und 100 mm-1 is described in /11/.

USE OF LOCAL CONCEPTS

The application of the local concept for a fatigue

assessment of bolted joints is presented. By means of a
comparison of local strength values calculated with finite Figure 2: Left to right: test device, tested structure
element analyses from nominal fatigue limits measured (bolted joints) and test part (single bolts)
on bolted structures (model bolted joints), single parts
(bolts) and material specimens.
With the help of finite element analysis the measured
As pictured in Figure 2 the tested bolted structure fatigue limits Fa50% were converted to local fatigue limits.
consists of a rotation-symmetric flange and nut, screwed Figure 3 shows exemplary the finite element model of
together by a bolt rolled before heat treatment with a the bolted joint. According to examinations in /12/, the

SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1 77

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
meshes of the axis-symmetric models (bolted joints, Figure 4 shows the calculated damage parameters PSWT
screws and specimen) were locally refined in the areas (as described in the previous section) in relation to the
relevant for computing local stresses and strains. Such corresponding supporting numbers nı at local hot spots
areas were the first bearing thread, the transition (notch between head and shaft, shaft and thread and
between shaft and thread or the transition between head within the first bearing thread) in the screws of cyclicly
and shaft. The contact behaviour between the bearing tested bolted joints and bolts. The spread of all
threads, in the interface between the clamped parts and parameters is about 25%. Leaving out the calculated
under the head was modelled by a finite sliding contact parameters in the notch between head and shaft, the
with coulomb’s friction. The elastic-plastic material spread reduces to 21%. Figure 5 overviews the means
behaviour was calculated using a simple model with a and spread of calculated damage parameters PSWT with
homogeneous, isotropic yield behaviour, calibrated with corresponding supporting factor nı at local hot spots in
a static stress-strain-curve and a Young’s modulus the tested specimens (notch) and screws (notch
E = 204 000 MPa. Further input parameters were the between shaft and thread and first bearing thread) of the
assembly preload FV, the minimal load FU, the cyclic bolted joints and bolts. The overall spread here is 22%.
bearable testing force amplitude Fa50%.

Table 1: Testing variants of bolts (B*) and bolted joints (BJ*) with corresponding results /12/

testing material matching

MA [Nm] FU [kN] Fa50% [kN] Sa50% [MPa] 1/TS s [kN]
variants flange nut
BJ1 AW-6082 AW-6082 38 1,00 12,54 - 1,27 1,17
BJ2 AW-6082 AW-6082 80 1,00 18,24 - 1,04 0,29
BJ3 42CrMoS4 AW-6082 38 1,00 13,09 - 1,05 0,25
BJ4 AW-6082 42CrMoS4 38 1,00 13,33 - 1,19 0,91
BJ5 42CrMoS4 42CrMoS4 38 1,00 14,59 - 1,19 0,97
BJ6 42CrMoS4 42CrMoS4 72 1,00 22,25 - 3,27 9,23
B1 AW-6082 AW-6082 - 26,20 5,28 91 1,05 0,24
B2 AW-6082 AW-6082 - 37,00 5,28 73 1,05 0,24
B3 42CrMoS4 42CrMoS4 - 26,2 5,04 87 1,31 0,53
B4 42CrMoS4 42CrMoS4 - 45,00 4,08 70 3,41 1,74

Table 2: Testing variants of cylindrical specimens with corresponding results /12/

testing variants according to /12/ R-A-06 R-V-06 R-V-02 R-V-01
sharpness of notch mild mild sharp sharp
notch geometry shouldered shaft V-formed V-formed V-formed
notch radius [mm] 0,6 0,6 0,2 0,1
stress concentration factor Kt,FEM 2,5 2,7 4,0 5,7
stress concentration factor Kt,FKM 2,2 2,6 3,8 5,1
supporting factor nı [-] 1,19 1,20 1,23 1,30
nominal lower stress SU [MPa] 622 837 779 643
nominal stress amplitude Sa50% [MPa] 144 168 139 75
range of scatter 1/TS 1,09 1,21 1,15 1,36
standard deviation s [MPa] 1,7 3,8 2,4 3

Table 3: Positions of failure of cyclical tested bolted joints (BJ*) and single bolts (B*) /12/.
testing variants according Bolted Joints Bolts
to /12/ BJ1 BJ2 BJ3 BJ4 BJ5 BJ6 B1 B2 B3 B4
head/shaft --- 11 1 3 --- 2 --- --- --- ---
shaft/thread 9 5 8 3 4 4 14 12 4 3
first bearing thread 4 1 3 4 10 5 --- 3 9 10

78 SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
Figure 3: Finite element model of the bolted joint

1,2
head/shaft shaft/thread first bearing thread
(hatched columns indicate position with most failures in the test)

1,0
(PSWT/nσ)/(PSWT,ref/nσ,ref)

0,8

0,6

0,4

0,2

0,0
BJ1 BJ2 BJ3 BJ4 BJ5 BJ6 B1 B2 B3 B4

Figure 4: Calculated damage parameters PSWT with corresponding supporting factor n at local hot spots in the screws of
cyclic tested bolted joints (BJ1-6) and bolts (B1-4)

SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1 79

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
 R-V/A
B1-4
BJ1-6

0,0 0,2 0,4 0,6 0,8 1,0 1,2

(PSWT/nσ)/(PSWT,ref/nσ,ref)

Figure 5: Means, maxima and minima for calculated damage parameters PSWT with corresponding supporting numbers
n at local hot spots in the screws of cyclic tested bolted joints (BJ1-6), bolts (B1-4) and specimens (R-V/A)

SIMPLIFIED USE OF LOCAL CONCEPTS

The dimensioning of “normal” bolted joints using the

local concept as it is described in the previous section is
very complex and expensive. To avoid this, a different
approach is necessary to combine the advantages of the
easy to calculate nominal stresses and the more
accurate local fatigue assessment. So, two problems
have to be solved:

1st to find a simplified FEM model of the bolted joint

to calculate nominal stresses correctly and
without the limitations of the VDI guideline 2230
/1/ for example and

2nd to give a fatigue assessment procedure using

the nominal stresses and the advantages of the
local concept.

The proceeding will be exposed on the example of a

multi-bolted joint: the connection between the cap and
the body of a high pressure fuel pump sample, see
Figure 6. The cap and the body are made of aluminium
2
alloy AW6082. The bolt has the dimension M6 and a
strength grade of 10.5.
3
In /13, 14/ is shown that, for the calculation of nominal
stresses using the FEM, a bolt can be represented by
beam elements. A complex three dimensional modelling
of the bolt is not needed to get a sufficient accuracy FB 4
respected to a nominal stress calculation. 1

As it is described in the introduction and illustrated in

Figure 1 the elastic axial and bending resilience is very
important for the calculation of the additional bolt load. In
general, values for the resilience can be estimated with Figure 6: Calculation example - top picture: high
the VDI guideline /1/. In the guideline calculation pressure fuel pump sample (Robert Bosch GmbH);
procedure different cross-sections with different stiffness bottom picture: pump sample with the assembly for the
were not taken into account. The total bolt resilience is cyclic tests (1: body, 2: cap, 3: fitting, 4: cylinder; bolts:
calculated as the sum of all resiliencies of the different M6 10.9, FB: total working load applied on the bottom of
bolt sections. So an “averaged” resilience is used. the cap)

80 SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
In most of all screw connections the bolt will get axial
and bending load. The bending stiffness of the clamped
parts is in general very much higher then the stiffness of
the bolt. For that reason the bending load of the bolt is
mainly defined by the deformation of the clamped parts.
Because a constraint condition is applied and not a
force, the resulting bending moment depends on the
distribution of the resilience along the bolt axis.
Therefore an “average” resilience for the whole bolt will
lead to incorrect results and the beam has to be
modelled with different cross sections.

In this case two different FEM models for the bolt are
used for the calculation:

− A detailed 3D model with contact conditions in all

contact surfaces (the thread and the contact
surface of the bolt head) and Detailed bolt FEM-model

− A beam model with a fix connection to the area of

the bolt thread and the contact surface of the bolt
head.
Idealised bolt
The bolt is idealized with one beam for the “cylindrical”
part (cross sectional area Aers and second moment of cx,K
area Iers) and two springs for the resilience of the bolt
cij,K
head cK and the resilience of the engaged thread cGM (as
described in /14/). The values are estimated from the
resiliencies given in /1/.
Iers
The different FEM models are shown in Figure 7. In Aers
both cases the calculation is performed under
consideration of linear elastic material behaviour and
contact conditions between the cap and the fitting.
cx,GM
cij,GM
Figure 8 shows the results of the bolt calculation with Bolt modeld as a beam
the beam model in comparison with a detailed FEM
model of the bolt to show the good correlation between Figure 7: Different FEM models: detailed bolt model
these two modelling methods. So, the nominal load can and bolt idealised with beam elements
be calculated with a much reduced FEM model. The
nonlinearity of the curves in Figure 8 is mainly a result
σ a + σ m = σ o ≈ RF (5)
of the nonlinear contact behaviour between the cap and
the fitting. Now, the next step is to define a simplified
local approach based on nominal stresses. ε a ⋅ E = σ a ≈ K t ⋅ S Sa (6)

Generally bolts will be highly prestressed. Because of PSWT = ( K tN ⋅ S SNa + K tM ⋅ S SMa ) ⋅ RF , (7)
the high notch factor the stresses will exceed the elastic
limit in the thread. If an elastic- ideal plastic material
(index N: relating to axial force; index M: relating to
behaviour is assumed the upper stress of a cyclic load is
bending moment). The elastic limit RF can be defined as:
defined as high as the elastic limit. In case of a bolt load
6
which leads to cycles in a range of N = 2 10 (endurance
Re + Rm (8)
limit as described in /1/) the local cyclic stress range can RF =
be idealised as linear. So the local cyclic stress 2
amplitude σa can be calculated with the elastic notch
factor Kt and the nominal stress amplitude SSa. Including (in /16/ stress defined for he calculation of the plastic
these two assumptions the damage parameter PSWT can collapse) with the elastic limit Re and the tensile strength
be simplified as follows /14, 15/: Rm.

Apart from the elastic limit only the nominal stresses and
PSWT = (σ a + σ m )ε a E (4)
the notch factors are needed for the calculation of PSWT.
Both values were evaluated separately for bending an

SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1 81

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
 6
FSA [kN]
S1 Balken
S1; beam model
5 S1 3D
S1; detailed model

0
0 10 20 30 40
FB=3 x FA [kN] Figure 9: FEM model for the calculation of the notch
factor
MSA [Nm]
3.5 (the index TG indicates, that the values are calculated
Msa_Pret_neu
S1; beam model for the first load bearing thread). In the present case the
3.0 value of the endurance limit as a result of cyclic tests is
Msadetailed
S1; 3D model very low (nominal value for the dimension M6 in
comparison with the endurance limit following the VDI
2.5
guideline). For that reason the resulting value of PSWT,D of
PSWT,D/PSWT,ref = 0,85 is at the lower scatter band in
2.0 Figure 5.

1.5 The fatigue limit can be calculated with the condition

1.0 PSWT = nσ ⋅ PSWT ,D . ( 10 )

0.5 The maximum value for PSWT results from the notch
effect of the different notched regions and the
distribution of the nominal stress. In the present case the
0.0
bending moment has an approximately linear distribution
0 10 20 30 40 along the bolt axis, with a maximum underneath the bolt-
FB=3 x FA [kN] head and a minimum near zero in the overlapping
threads. Contrary to this the maximum notch factor
Figure 8: Comparison of the additional bolt load (bolt exists in the first load-bearing thread turn of the bolt. The
S1 on the symmetric plane) with two different FEM conditions in this example lead to a maximum damage
models: detailed bolt model idealised bolt model (FB: parameter in the thread chamfer beneath the bolt head.
total working Load (see Fig. 6), FSA additional bolt load;
MSA additional bolt moment). Beside the material resistance and the resilience of the
different parts of the construction, the bolt preload is a
axial loading. The nominal stresses can be calculated very important parameter, which controls the opening of
with the beam model, the notch factor with a different the joint. In the case under consideration the preload will
detailed FEM-Model (see Figure 9). be applied torque-controlled. For that reason the
preload-values scatter over a much wider range. Beside
Normally the bolt fatigue strength is known in nominal the spread as a result of the tightening technique a
stresses and for a pure axial loading. With the simplified further reduction due to the embedding of the joint has to
equation for PSWT the endurable PSWT –factor is defined be taken into account. Because the joint has a very low
as: clamping length the amount of embedding is very
critical. In this example the minimum and maximum
1 preload FV is estimated by the VDI-Guideline 2230 /1/.
PSWT , D = K tN ,TG ⋅ S Sa , D ⋅ RF . (9)
The result in comparison with experimental values is
nσ ,TG
shown in the load-cycle diagram in Figure 10. Two main
conclusions can be drawn:

82 SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
 1st For cycles in the range of N = 2·106 (endurance the calculation as well. A fracture in the in the
limit according to /1/) the estimated value for the first load-bearing thread turn, the location of the
applicable service load is in a good agreement maximum notch factor, would occur at higher
with the experimental results. Especially the loadings.
calculated value for the minimum preload and a
probability of failure Pf of 1 % matches fine with In the range of low cycle fatigue (LCF) the simplified
the value calculated for the experiments. local approach had to be used with caution. The
assumption, that the local stress strain trail is nearly
2nd The location of the fracture can be precisely linear, is not correct any more. However, in this example
anticipated. In all the tests the bolt fractured the simplified approach seems to be usable even in the
underneath the bolt-head. This is the result of range to N = 105 cycles.

50
maximum working force FB[kN]

Comparison: experiment and calculation

Pf = 99 % high pressure fuel pump / M6 10.9
40

constant lower working force: FBu = 1,0 kN

30 tightening torque: MA = 8,5 Nm
Pf = 1 %

20

FV,max=13,0 kN; Pf = 99 %
2
experimental results 2
experimentally estimated F-N-curve 2
FEM based fatigue approach FV,min=5,0 kN; Pf = 1 % 3
10 3

4 5 6 7
10 10 10 10
number of cycles N

Figure 10: Comparison of the estimated life cycles based on the simplified local approach and experimental results (FV:
preload used in the calculation)

CONCLUSION and a higher confidence should be calculated as

described in /12/.
Based on the presented results a local fatigue
assessment of bolts, rolled before heat treatment, For the calculation of “normal” bolt connections the
seems to be possible. With such a concept a fatigue presented simplified local approach has an adequate
design can be done for bolted joints, which cannot be accuracy, especially if the high scatter of the nominal
designed following /1/. For example: threaded parts like stresses due to the scatter of the preload has been
thin-walled screws with bores next to the thread, taken into account. Even the prediction of the location of
eccentrically loaded bolted joints or just “normal” screws the fracture seems to be possible with this approach.
where several notched regions have to be evaluated. With the simplified local approach the nominal approach
The core of this local approach is a fatigue endurable for “normal” bolts can be easily enhanced to get a much
damage parameter PSWT of Smith, Watson and Topper, more accurate and safe design concept.
which can be calculated with a failure probability of 50%
from fatigue tests by means of finite element analyses. REFERENCES
The influence of the local notch sharpness on the
damage parameter is considered by using the specific 1 VDI-guideline 2230 part 1, Systematic calculation
local supporting factor nσ. For a reliable design process, of high duty bolted joints, Joints with one
local damage parameters with a lower failure probability cylindrical bolt, VDI, edition february 2003

SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1 83

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms
2 Tanaka, M., Miyazawa, H.; Asaba, E., Hongo, K., 11 Niessner, M., Ein Beitrag zur
Application of the Finite Element Method to Bolt- Festigkeitsbewertung scharf gekerbter Bauteile
nut Joints – Fundamental Studies on Analysis of unter statischer undzyklischer Belastung,
Bolt-nut Joints using the Finite Element Method. Veröffentlichung des Instituts für Stahlbau und
Bulletin of the JSME, Vol. 24, No.192, June 1981. Werkstoffmechanik, der TU Darmstadt, Heft 71,
2003
3 Tanaka, M., Yamada, A., Application of the
Boundary Element Method to Threaded 12 Buhr, K., Entwicklung von FEM-basierten
Connection. 2nd Report, In case of Threaded Konzepten für die schwingfeste Auslegung von
Connection with Free Flat Boundaries. Bulletin of Schraubenverbindungen, Dissertation, TU
the JSME, Vol. 29, No. 251, May 1986. Darmstadt, 2007

4 Fares, Y., Chaussumier, M., Daidie, A., Guillot, J., 13 Wuttke, U., Berger, C., Buhr, K., Bacher-Höchst,
Determining the Life Cycle of Bolts using a Local M., Haydn, W., Kreschel, H., Einsatzmöglichkeiten
Approach and the Dang Van Criterion, In: Fatigue der FEM zur Bewertung zyklisch beanspruchter
Fract Engn Mater Struct 29, 2006 Schraubenverbindungen, 8. Informations- und
Diskussionsveranstaltung Schraubenverbin-
5 Sonsino, C.M., Zur Bewertung des Schwing- dungen, Darmstadt, 2006
festigkeitsverhaltens von Bauteilen mit Hilfe
örtlicher Beanspruchungen, Konstruktion 45 14 Wuttke, U., Praxisgerechte Bewertung zyklisch
(1993) 1, S.25-33 beanspruchter Schraubenverbindungen mit Hilfe
der Finite-Elemente-Methode, Dissertation, Fach-
6 Radaj, D., Ermüdungsfestigkeit, Springer Verlag, gebiet und Institut für Werkstoffkunde, TU
Berlin Heidelberg New York London Tokyo, 1995 Darmstadt, 2007

7 Dünkel, V., Schwingfestigkeit von Schrauben- 15 Alt, A., Dauerfestigkeitsprüfung und Dauer-
verbindungen – Optimierte Versuchsdurchführung festigkeit von Schraube-Mutter-Verbindungen
und deren Anwendung bei der Untersuchung von unter kombinierter Zug- und Biegebelastung,
Randschicht- und Oberflächenzuständen, Dissertation, TU Berlin, 2005
Dissertation, Fachgebiet und Institut für
Werkstoffkunde, TU Darmstadt, 1999 16 Issler, L., Ruoß, H., Häfele, P., Festigkeitslehre –
Grundlagen, Springer Verlag, 1995
8 Smith, K. N., Watson, P., Topper, T.H., A stress-
strain function for the fatigue of metals, Journal of CONTACT
materials, JMLSA Vol. 5, No. 4, 1970, S. 767-778
Prof. Dr. Christina Berger
9 Siebel, E., Stieler, M., Ungleichförmige Span- Institut für Werkstoffkunde
nungsverteilung bei schwingender Bean- Technische Universität Darmstadt
spruchung, VDI-Z 5, S. 121-126, 1955 Grafenstraße 2
D-64283 Darmstadt
10 FKM guideline, Analytical strength assessment, Germany
Forschungskuratorium Maschinenbau, VDMA-
Verlag, Frankfurt, 2003 www.mpa-ifw.tu-darmstadt.de

84 SAE Int. J. Mater. Manuf. | Volume 2 | Issue 1

This content downloaded from

31.4.242.51 on Fri, 01 Oct 2021 15:22:25 UTC
All use subject to https://about.jstor.org/terms