Toshimichi Fukuoka

Professor
e-mail: [email protected]
Masataka Nomura
Associate Professor
e-mail: [email protected]
Faculty of Maritime Sciences,
Kobe University,
5-1-1 Fukaeminami,
Higashinada, Kobe 658-0022, Japan
Proposition of Helical Thread
Modeling With Accurate
Geometry and Finite Element
Analysis
Distinctive mechanical behavior of bolted joints is caused by the helical shape of thread
geometry. Recently, a number of papers have been published to elucidate the strength or
loosening phenomena of bolted joints using three-dimensional nite element analysis. In
most cases, mesh generations of the bolted joints are implemented with the help of
commercial software. The mesh patterns so obtained are, therefore, not necessarily ad-
equate for analyzing the stress concentration and contact pressure distributions, which
are the primary concerns when designing bolted joints. In this paper, an effective mesh
generation scheme is proposed, which can provide helical thread models with accurate
geometry to analyze specic characteristics of stress concentrations and contact pressure
distributions caused by the helical thread geometry. Using the nite element (FE) models
with accurate thread geometry, it is shown how the thread root stress and contact pres-
sure vary along the helix and at the nut loaded surface in the circumferential direction
and why the second peak appears in the distribution of Mises stress at thread root. The
maximum stress occurs at the bolt thread root located half a pitch from nut loaded
surface, and the axial load along engaged threads shows a different distribution pattern
from those obtained by axisymmetric FE analysis and elastic theory. It is found that the
second peak of Mises stress around the top face of nut is due to the distinctive distribution
pattern of
z
. DOI: 10.1115/1.2826433
Keywords: xing element, bolted joint, FEM, helical thread modeling, stress
concentration, contact pressure distribution
1 Introduction
Threaded fasteners are the most widely used machine elements
because they can repeatedly be assembled and disassembled by an
easy operation. Mechanical behaviors of the threaded fasteners,
such as the strength and the stiffness of bolted joints, have been
analyzed by experiment, theoretical analysis based on elastic
theory, and numerical method. Finite element method FEM is
found to be the most powerful numerical method for solving the
problems of bolted joints. The development of FEM made it pos-
sible to evaluate the stress concentration at the thread root with
high accuracy 1,2. The stiffness of bolted joints, which has a
dominant effect on its fatigue strength, and stress concentrations
of the bolt thread and the bolt head llet have also been studied
systematically with help of FEM 3,4. In the conventional studies
on the stress analysis of bolted joints, axisymmetric FEM has
mainly been used. In the case of three-dimensional analysis,
threaded portions were modeled by using the threads with axisym-
metric geometry, i.e., the effects of lead angle and the helix of
thread prole were neglected.
Recently, some researchers have started to use helical thread
models that were constructed with the advanced modeling func-
tions provided by a couple of commercial software 58. Some
studies have tried to elucidate the loosing phenomena of bolted
joints using the helical thread models thus obtained 9. However,
the aforementioned procedures do not necessarily provide helical
thread models adequate for analyzing the stress concentration at
the thread root and contact pressure distributions at nut loaded
surface, because of the complexity of thread prole and the limi-
tation of softwares functions. Meanwhile, a thread cross section
perpendicular to the bolt axis is identical at any position. Accord-
ingly, the thread prole can be dened mathematically using rig-
orous expressions by taking the effects of root radius, where the
cross section is divided into three portions.
In this paper, an effective modeling scheme for three-
dimensional FE analysis, which can accurately construct helical
thread geometry, is proposed using the equations dening the
thread cross section perpendicular to the bolt axis. The present
procedure has such benecial performances as modeling each
thread with one-pitch height independently and using ne meshes
only around threaded portions. Therefore, it is possible to con-
struct nite element models of bolted joints with high accuracy
and computation efciency. Using the FE models thus obtained,
the mechanical behavior caused by the helical thread geometry
has been evaluated, such as the distributions of the thread root
stress along the helix and nonsymmetric contact pressure distribu-
tions at the nut loaded surface. It is found that the maximum bolt
stress occurs at the thread root located half a pitch from nut loaded
surface, and the axial load along engaged threads shows a differ-
ent distribution pattern from the previous studies by taking the
helical thread geometry into account.
2 Mathematical Expressions of Thread Cross Section
Prole
The specications of thread proles are given in ISO 68, 261,
262, and 724. The thread root has an appropriate amount of round-
ness to avoid an excessive stress concentration. In Japanese Indus-
trial Standard JIS, it is recommended that the thread root radius
Contributed by the Pressure Vessel and Piping Division of ASME for publication
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received September 1,
2006; nal manuscript received December 8, 2006; published online January 17,
2008. Review conducted by Sayed Nassar. Paper presented at the 2006 ASME Pres-
sure Vessels and Piping Conference PVP2006, Vancouver, British Columbia,
Canada, July 2327, 2006.
Journal of Pressure Vessel Technology FEBRUARY 2008, Vol. 130 / 011204-1 Copyright 2008 by ASME
Downloaded 14 Feb 2013 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
should be more than 0.125P P: thread pitch for external threads
made of high strength steel. Figure 1 shows the cross sectional
prole along the bolt axis including the thread root radius. Assum-
ing that the rounded portion of the thread root is a part of a single
circle with diameter , the surface of external thread can be di-
vided into three parts such as A-B root radius, B-C thread
ank, and C-D crest. The thread prole perpendicular to the
bolt axis can be obtained by expanding those three parts into the
plane, as shown in Fig. 2. Its shape is naturally identical at any
cross section along the bolt axis. In the next section, helical thread
models are to be constructed by utilizing the characteristics ex-
plained here. The thread prole shown in Fig. 1 is expressed by
means of the following equations.
r =

d
2

7
8
H + 2
2

P
2
4
2

2
0
1

+
d
2

7
8
H
1

2

d
2

2

1
=

3
P

2
=
7
8

3
12
P H =

3
2
P
where d and H represent nominal diameter and thread overlap.
The prole of internal thread can be expressed in the same
manner.
r =

d
1
2
0
1

+
d
2

7
8
H
1

2

d
2
+
H
8
2
n
+
n
2

P
2
4
2

2

1
=

4

2
= 1

3
n
P

n

3
24
P
There are upper limits for the root radii of external and internal
threads, and
n
, appeared in Eqs. 1 and 2, in connection with
the thread geometry of minor and nominal diameters.
3 Proposition of Helical Thread Modeling With Accu-
rate Geometry
3.1 Conventional Methods. When analyzing the mechanical
behavior of bolted joints with three-dimensional analysis, it has
been a common practice that the threaded portion of the FE mod-
els has axisymmetric geometry, where the effects of lead angle are
neglected because of its small value. That is, external and internal
threads are modeled by stacking an appropriate number of threads
with axisymmetric geometry. Recently, some researchers start to
use helical thread models because of a growing recognition of the
importance of helical effects, e.g., loosening phenomena of bolted
joints. Their modeling procedures are classied roughly into three
categories in the case of external threads 59.
Type 1. Two-dimensional thread cross section model with one-
pitch height is rotated helically around the bolt axis 57. This
procedure inevitably generates a small hole around the bolt axis.
Type 2. Helical thread model made in the similar manner to
Type 1 is attached around a solid cylinder 8,9. Mesh patterns are
not coincident at the interface between helical threads and the
cylinder.
Type 3. Surface models of bolt and nut are made by means of a
sophisticated performance provided by commercial software, and
then the inside of the helical-shaped solid models of bolt and nut
is divided into three-dimensional elements using its automatic
mesh generation function.
In the case of Type 1, the effect of the small hole seems insig-
nicant. However, unfavorable meshes are to be generated due to
the helical rotation, especially around the far end thread and the
area connecting the thread runout and the bolt cylinder. The same
problem still remains in the case of Type 2. It is not an easy
practice even for Type 3 that highly stressed area is intensively
divided using small elements while the overall mesh pattern being
well balanced.
3.2 Helical Thread Modeling by Stacking Cross Sections
With Accurate Geometry. Helical thread modeling procedure
proposed in this paper is based on the fact that the shape of the
cross section perpendicular to the bolt axis is identical at any
position. The proles of external and internal threads are ex-
pressed mathematically by means of Eqs. 1 and 2. In Fig. 3,
illustrated are the real shapes of the cross section of external
threads with coarse pitch of P. In the following, it is shown how
the helical thread models of external thread with accurate geom-
etry can be constructed, where each ridge with one-pitch height is
divided into n thin plates with the same conguration. The proce-
dure consists of six steps.
Fig. 1 Thread cross section along the bolt axis
Fig. 2 Prole of the cross section of external thread perpen-
dicular to the bolt axis
011204-2 / Vol. 130, FEBRUARY 2008 Transactions of the ASME
Downloaded 14 Feb 2013 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Step 1. Cross section of external thread with accurate geometry,
shown in Fig. 3, is properly modeled using two-dimensional ele-
ments. This is a basic mesh model.
Step 2. The basic model is placed at the reference position, z
=0.
Step 3. Rotating the basic model anticlockwise by an amount of
2/ n, it is placed at the position of z=P/ n.
Step 4. Connecting the corresponding nodes of the two basic
models placed at z=0 and P/ n, a three-dimensional model with
P/ n thickness is obtained.
Step 5. One-pitch helical thread model is completed by repeat-
ing Steps 3 and 4 n times.
Step 6. An appropriate number of one-pitch model obtained in
Step 5 is stacked, according to the number of threads of the ob-
jective bolted joint.
If the one-pitch model is constructed by simply stacking the
basic model according to the aforementioned procedure, the ele-
ments around the threaded portion might have a large aspect ratio,
which causes low accuracy of the numerical analysis. In addition,
the mesh pattern around the bolt axis becomes ner than is nec-
essary. From the numerical accuracy and computation efciency
points of view, therefore, the nite element meshes for thread area
and bolt core portion should be constructed separately. Figure 4
shows an example of the mesh patterns of the cross sections per-
pendicular to and along the bolt axis. The circular area inside the
four arrows, shown in Fig. 4a, is divided by rather coarse
meshes, and the outside area is modeled by fairly ne meshes.
Therefore, the bolt core portion is simply modeled as a cylinder
and only the thread area is modeled following Steps 16. Figure
5a shows a one-pitch helical thread model thus obtained. The
mesh patterns of the two separate models are completely coinci-
dent at the interface. It follows that the helical model constructed
here is expected to attain both high accuracy and computation
efciency. Thread runout is modeled by gradually varying the
depth of the groove along the helix so as to be smoothly con-
nected with bolt cylinder. Following the above procedure, it is
possible to construct an entire bolted joint model only by eight-
node brick elements.
Internal threads can be modeled in the same manner. In this
case, the outside area of threaded portion is modeled as a hollow
cylinder. The outer surface of the nut is modeled as a cylindrical
shape for simplicity, although it is possible to construct a hexago-
nal nut. Figure 5b shows the cross section of the nut model with
helical geometry. Figure 6 is an example of the entire bolted joint
model, which is tightened by a single bolt with coarse thread of
M16. The total numbers of nodes and elements are 78,520 and
86,504, respectively. Numerical analysis with FE models con-
structed here can be implemented by standard FE analysis FEA
software packages.
4 Stress Analysis of Bolted Joints Using Helical
Thread Model
4.1 Numerical Models and Boundary Conditions. The
mesh generation scheme proposed here can be executed without
any help of commercial software. However, it is favorable to use
some sophisticated functions of commercial software for an effec-
Fig. 3 Accurate cross section prole of metric coarse thread
Fig. 4 Mesh patterns of cross sections of bolt model
Fig. 5 One-pitch model of external thread and cross section of
nut model
Journal of Pressure Vessel Technology FEBRUARY 2008, Vol. 130 / 011204-3
Downloaded 14 Feb 2013 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
tive modeling. In this study, Hyper Works is used for supporting
the mesh generation and the numerical calculations are conducted
as elastic problems by ABAQUS. Referring to the FE model shown
in Fig. 6, the axial bolt force is applied as a uniform displacement
at the lower end of the bolt cylinder. At the lower surface of the
fastened plate, axial displacements are completely restrained and
the circumferential ones are restrained at four nodes located
90 deg apart. The analytical objects are bolted joints tightened by
a single bolt of M16 or M12 with coarse thread. Bolt, nut, and
plate are supposed to be made of carbon steel whose Youngs
modulus and Poissons ratio are 200 GPa and 0.3, respectively. In
the case of M16, bolt hole diameters are changed as 17 mm,
17.5 mm, and 18.5 mm, which correspond to the rst, second, and
third classes specied in JIS, respectively. Coefcients of friction
are varied from 0.05 to 0.20 with an increment of 0.05 and
assumed to be identical at pressure ank of screw thread and nut
loaded surface. For a parametric study, a standard analytical con-
dition is dened as follows. Standard condition: M16, coefcient
of friction=0.15, bolt hole diameter=17.5 mm second class
4.2 Stress Distributions Along Thread Root. Many previous
studies have reported that the maximum stress occurs at the bolt
thread root located within one pitch of the nut loaded surface.
Most of the research on the stress concentration at the thread root
have been conducted using axisymmetric FE models. Even if in-
troducing such helical thread models explained in Sec. 3.1, it
seems difcult to evaluate the stress concentrations around the
thread root with practical accuracy. In this section, stress concen-
trations around the thread root are analyzed using the FE models
obtained in Sec. 3.2. Numerical calculations were performed us-
ing a single computer equipped with Pentium 4 of 3.4 GHz with
2 Gbyte RAM. CPU time changes from 3 h to 5 h as coefcient
of friction increases. It is shown in Fig. 7 how the maximum
stress, which occurred at the thread root, varies along the helix.
Mises stress at the thread root
eq
is normalized with respect to
the mean tensile stress
b
dened at the bolt cylinder. The ab-
scissa represents the distance from the nut loaded surface. The
maximum Mises stress
eqmax
occurs at half a pitch from the nut
loaded surface, as in the case of the previous studies 10, where
larger coefcient of friction produces higher peak stress. Then, the
stresses at the thread root gradually decrease toward the top face
of the nut, and they show a second peak. It is considered that this
phenomenon is caused by the low stiffness of the last engaged
thread for its bending deformation. In Fig. 8, shown are the effects
of friction coefcient and bolt hole diameter on the maximum
normalized stress
eqmax
/
b
. It increases slightly and almost lin-
early as coefcient of friction increases, and it almost decreases
linearly with increasing bolt hole diameter. As for the effects of
nominal diameter, larger bolt produces larger stress concentrations
as well as the previous studies 1.
4.3 Asymmetric Contact Pressure Distributions at Nut
Loaded Surface. The contact pressure at the nut loaded surface
decreases outward in the radial direction. It is predicted that the
contact pressure also varies in the circumferential direction,
though probably a small amount, because of the circumferential
variation of the stiffness of engaged threads adjacent to the nut
loaded surface. The latter phenomenon can be analyzed only when
introducing a helical thread model. Figure 9a shows the circum-
ferential contact pressure distributions at the nut loaded surface
for varying radial positions. The reference point of =0 is placed
at the plate top surface on which a fully formed nut thread with
one-pitch height exists. The magnitude of the contact pressure
varies in the circumferential direction, which is rather remarkable
along the bolt hole and at the outer end of the nut loaded surface.
In the radial direction, though not shown here, the contact pres-
Fig. 6 Fienit element model of entire bolted joint
Fig. 7 Mises stress distributions at the bolt thread root along
the helix
Fig. 8 Normalized maximum Mises stress occurred at the bolt
thread root
011204-4 / Vol. 130, FEBRUARY 2008 Transactions of the ASME
Downloaded 14 Feb 2013 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
sure decreases smoothly at any circumferential position. Shown in
Fig. 9b are the contact pressure distributions in the circumferen-
tial direction for third class bolt hole. The effect of the helical
shape of thread geometry appears similarly in the case of second
class. Such circumferential variation of the contact pressure dis-
tributions might cause various problems in the bolted joints.
4.4 Evaluation of Load Distributions Along Engaged
Threads. In this study, nite element meshes around thread ridges
have equal thickness in the axial direction, as shown in Fig. 4b.
Accordingly, the load distribution along engaged threads can be
evaluated by summing up the axial loads exerted on each thin
element with equal thickness. In a common bolted joint, bolt cy-
lindrical portion is subjected to axial bolt force F
b
at any position
between nut loaded surface and bolt head. Meanwhile, the axial
load F along the engaged threads gradually decreases from F
b
to
zero toward the top surface of the nut. It is well known that such
load distribution pattern causes various problems inherent to bolt-
nut connections. Yamamoto derives an equation for F along the
engaged threads based on the elastic theory, which shows that the
axial load F decreases following a hyperbolic function, sinhx,
where x denotes the distance from the nut loaded surface 11.
In Fig. 10, the load distribution along the engaged threads ob-
tained by the helical thread model is compared to that by Yama-
motos equation. Numerical result by axisymmetric FE analysis is
also shown, where the mesh pattern is the same as the cross sec-
tion of the helical thread model along the bolt axis. The axisym-
metric analysis gives a similar load distribution to that of concave-
shaped Yamamotos equation, except around the top face of the
nut. On the other hand, the numerical result by the helical thread
model shows slightly convex distributions both around the nut
loaded surface and the top face of the nut. In the cases of Yama-
motos equation and axisymmetric FE analysis, it is assumed that
every set of male and female threads is equally engaged. In the
actual engaged threads, however, the contact areas of engaged
threads rapidly decrease around the nut loaded surface and the top
face of the nut. It is considered that such effects could be repre-
sented by the helical thread models introduced here.
5 Discussions
A nut is classied into several kinds according to its shapes
around bearing surface and top face. The nut used here has a at
bearing surface that is completely in contact with the plate sur-
face. The threads at the top face of the nut are commonly cham-
fered, i.e., truncated at some angle, toward the bolt hole. The
effect of the chamfering is studied by FE analysis. Figure 11a
illustrates the nut cross section with and without chamfering. All
the numerical results presented so far are associated with the
chamfered nut models. Figure 11b represents the effect of the
chamfering on the stress concentrations at the thread root. It is
observed that for both chamfered and nonchamfered nuts,
z
shows characteristic stress distribution patterns, which steeply
vary between positive and negative values. Accordingly, it seems
that the second peak appearing in the Mises stress distribution is
caused by the distinctive distribution pattern of
z
. In the case of
nonchamfered nut, the second peak of Mises stress shows an un-
natural decrease compared to the case of chamfered nut. This
phenomenon can be mitigated by chamfering the top face of the
nut.
6 Conclusions
An effective three-dimensional thread modeling scheme, which
can accurately take account of its helical geometry, is proposed
using the equations dening the real conguration of the thread
cross section perpendicular to the bolt axis.
It is shown how the thread root stress varies along the helix and
that the maximum stress occurs at half a pitch from the nut loaded
Fig. 9 Circumferential contact pressure distributions at the
nut loaded surface
Fig. 10 Axial load distributions along engaged threads
Journal of Pressure Vessel Technology FEBRUARY 2008, Vol. 130 / 011204-5
Downloaded 14 Feb 2013 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
surface. The stresses at the thread root gradually decrease toward
the top face of the nut and they show a second peak because of the
low stiffness of the last engaged threads.
It is shown how the contact pressure at the nut loaded surface
varies in the circumferential direction due to the effect of the
helical thread geometry. The axial load distribution along engaged
threads analyzed by helical thread models shows a different dis-
tribution pattern from those obtained by axisymmetric FE analysis
and elastic theory.
The second peak appearing in the distributions of Mises stress
at the thread root is caused by the distinctive distribution pattern
of
z
.
Acknowledgment
The authors would like to acknowledge Mr. Yuuya Morimoto
DAIHATSU Motor Co. for his contribution to the numerical
calculations conducted in this research.
Nomenclature
D
i
bolt hole diameter
d nominal diameter
d
1
minor diameter
F axial load along engaged threads
F
b
axial bolt force
H thread overlap
P thread pitch
r radial coordinate
z axial coordinate
coefcient of friction
circumferential coordinate
,
n
root radii of external and internal threads

b
mean tensile stress dened at bolt cylinder

eq
Mises stress at thread root

eqmax
maximum Mises stress at thread root

z
axial stress
References
1 Fukuoka, T., 1997, Evaluation of the Method for Lowering Stress Concentra-
tion at the Thread Root of Bolted Joints With Modications of Nut Shape,
ASME J. Pressure Vessel Technol. 1191, pp. 19.
2 Fukuoka, T., and Takaki, T., 2003, Elastic Plastic Finite Element Analysis of
Bolted Joint During Tightening Process, ASME J. Mech. Des. 1254, pp.
823830.
3 Lehnhoff, T. F., Ko, K. I., and Mckay, M. L., 1994, Member Stiffness and
Contact Pressure Distribution of Bolted Joints, ASME J. Mech. Des. 1162,
pp. 550557.
4 Lehnhoff, T. F., and Bunyard, B. A., 2000, Bolt Thread and Head Fillet Stress
Concentration Factors, ASME J. Pressure Vessel Technol. 1222, pp. 180
185.
5 Chen, J., and Shih, Y., 1999, A Study of the Helical Effect on the Thread
Connection by Three Dimensional Finite Element Analysis, Nucl. Eng. Des.
191, pp. 109116.
6 Bahai, H., and Esat, I. I., 1994, A Hybrid Model for Analysis of Complex
Stress Distribution in Threaded Connectors, Comput. Struct. 521, pp. 79
93.
7 Rhee, H. C., 1990, Three-Dimensional Finite Element Analysis of Threaded
Joint, Proceedings of the Nintu International Confedence on Offshore Me-
chanics Arctic and Engineering, Vol. 3, Pt. A, pp. 293297.
8 Zadoks, R. I., and Kokatam, D. P. R., 1999, Three-Dimensional Finite Ele-
ment Model of a Threaded Connection, Comput. Model. Simul. Eng. 44,
pp. 274281.
9 Zhang, M., and Jiang, Y., 2004, Finite Element Modeling of Self-Loosening
of Bolted Joints, PVP Am. Soc. Mech. Eng., 478, pp. 1927.
10 Fukuoka, T., Yamasaki, N., Kitagawa, H., and Hamada, M., 1986, Stresses in
Bolt and NutEffects of Contact Conditions at the First Ridge-, Bull. JSME
29256, pp. 32753279.
11 Yamamoto, A., 1970, Theory and Practice of the Tightening Process of Screw
Threads, Yokendo, Tokyo, pp. 12 in Japanese.
Fig. 11 Effect of the chamfering of the nut top thread
011204-6 / Vol. 130, FEBRUARY 2008 Transactions of the ASME
Downloaded 14 Feb 2013 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm