Available online at www.sciencedirect.

com
Procedia
Engineering
Procedia
Procedia Engineering
Engineering 00 (2009)
2 (2010) 000–000
1849–1858
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Fatigue 2010

Experimental determination of the fatigue life of modified threaded

pipe couplings
Jeroen Van Wittenberghea, *, Jan De Pauwa, Patrick De Baetsa, Wim De Waelea, Magd
Abdel Wahaba and Guido De Roeckb
a
Ghent University, Laboratory Soete, Sint-Pietersnieuwstraat 41, B-9000 Gent, Belgium
b
K.U.Leuven, Department of Civil Engineering, Kasteelpark Arenberg 40, B-3001 Heverlee, Belgium

Received 26 February 2010; revised 10 March 2010; accepted 15 March 2010

Abstract

To assemble tubular structures such as pipelines, risers,… , threaded couplings can be used. In service these couplings are often
subjected to dynamic loads, and to maintain a secure connection, they are generally preloaded. The combination of preload and
external dynamic loads results in a multiaxial stress distribution over the connection, where the coupling’s threads act as stress
raisers, initiating fatigue cracks. In this study the fatigue life of the standard API Line Pipe coupling and two modified coupling
configurations is determined experimentally. This way the influence of the coupling’s global geometry on the connection’s
fatigue life is studied. Threaded pipe samples with a nominal size of 1” are tested on a four-point bending fatigue test setup to
obtain S-N curves. Both modified coupling configurations exhibit a fatigue life improvement. In the final part of this paper, these
experimental observations are explained by a combination of a damage evolution law and a finite element model.
c 2010 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Keywords: threaded coupling, threaded connection, fatigue, experiment, multiaxial, triaxiality, damage evolution law

Nomenclature

A damage parameter Sm mean stress [MPa]

D damage parameter [-] Sr stress range [MPa]
E Young’s modulus [MPa] STD standard deviation
F force [N] ȕ damage parameter [-]
N number of cycles [-] ıeq equivalent von Mises stress [MPa]
Nf number of cycles to failure ıH hydrostatic pressure [MPa]
R load ratio [-] ıy yield stress [MPa]

* Corresponding author. Tel.: +32-9-264 33 03; fax: +32-9-264 32 95.

E-mail address: Jeroen.VanWittenberghe@UGent.be.

1877-7058 c 2010 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2010.03.199
1850 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858

2 J. V. Wittenberghe et al./ Procedia Engineering 00 (2010) 000–000

RV triaxiality function [-] ıUTS ultimate tensile stress [MPa]

Sa stress amplitude [MPa] Ȟ Poisson’s ratio [-]
Sf fatigue stress [MPa]

1. Introduction

Threaded pipe couplings, as used in the oil and gas industry, are applied in particular to constitute casing strings,
drillpipe strings, production and transportation risers and pipelines. They have the advantage that they can be easily
coupled and uncoupled and the chance of introducing flaws during installation (as is the case for welding) is
negligible. Since exploration of new energy sources moves further to greater depths, deeper ocean water and into
harsher environments, this results in extreme requirements for the equipment used during exploration and
exploitation of the wells. As a consequence, several engineering challenges are introduced for the couplings.
The considered couplings comprise a male threaded element at the end of a pipe and a female threaded coupling.
The male part is commonly referred to as the pin, while the female part is called the box. Fatigue failure of threaded
pipe couplings is an important issue. In the case of drill strings, fatigue failure is caused by cyclic loads which
appear by bending of the drill string during operation. In other applications, dynamic loads can be caused by
environmental and operational conditions e.g. wind, waves, vortex induced vibrations, internal and external pressure
changes etc. Although much research has been done in this field, fatigue is still a very important cause of failure for
threaded couplings. E.g. it has been reported for drill strings, up to 73% of all defects are caused by fatigue cracks
[1]. Additionally, during the last years there is a trend towards using casing and tubing strings for drilling particular
wells [2-5]. This introduced enhanced fatigue requirements on threaded couplings that were conventionally used in
static applications. Therefore several couplings have been developed and patented, which propose modified
geometries, to improve the coupling’s fatigue life. However, few has been published about the underlying
mechanisms and the resulting fatigue life improvement.
During operation, an applied threaded pipe coupling should meet two basic requirements. First, the structural
integrity of the pipe string has to be assured. This means that the connection should not loosen and that no failure
should appear under the influence of external loads and vibrations. Second, a leak-tight joint should be provided. To
meet both requirements, the couplings commonly have a conical shape that enables them to be preloaded. This
preload is attained by applying a certain torque to the coupling. The so-called make-up torque has a specific value
for every coupling type and size. The combination of the make-up torque, external loads and the threaded geometry
of the coupling results in a multiaxial stress distribution over the connection. Local stress concentrations at the
thread roots can initiate fatigue cracks, in time causing the connection to fail. The highest stress concentration under
axial load can be expected at the last engaged thread (LET) of the pin [5]. This is caused by an unevenly distributed
load over the threads of the coupling [6], with the thread carrying the highest load being the LET of the pin.
In this study the authors tried to improve the fatigue life of a standard API threaded coupling, by improving the
coupling’s load distribution. This was done by reducing the stiffness of the box and hence, allowing the load to be
spread more evenly. In the following sections an overview of the experimental setup, samples and test results is
given. Tests were carried out on the standard coupling and two modified coupling configurations. Both
configurations exhibited an improved fatigue life. Afterwards the results are evaluated using a combination of a
damage evolution law and a finite element model.

2. Experimental work

2.1. Setup description

To determine the fatigue life of 1” API Line Pipe couplings (outer pipe diameter equal to 33.4 mm) and to assess
the influence of several modifications experimentally, four-point bending tests were carried out to develop S-N
curves. Tests were performed on an ESH 100 kN universal test machine and the used setup is shown in Fig. 1. The
 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858 1851

J. V. Wittenbergheet al. / Procedia Engineering 00 (2010) 000–000 3

test sample (schematically illustrated in Fig. 2) consists of a central box (1a) with two made-up threaded pipes (1b)
and is supported by the two outer load points (indicated by the blue arrows in the figure). A dynamic load is applied
by the hydraulic actuator (2) and transferred through the loadcell (3) to a transverse beam that carries the two inner
load points. To avoid a bending moment in the loadcell, it is connected to the transverse beam through a hinge (4).
Tests were carried out under load control with a load ratio R = Fmin / Fmax = 0.1 and at a testing frequency of
15 Hz. To enable the detection of through thickness cracks, the tube samples were pneumatically pressurized at a
pressure between 1 and 2 bar. When a crack grows through the wall thickness of a pipe, the pressure will drop below
a threshold and the test is automatically stopped. The number of elapsed load cycles up to this event, is considered
the fatigue life of the connection. To create two separate pressure chambers, two plugs (6) (see Fig. 2) are inserted
inside the box, between the two threaded pipe samples. Both free ends of the test specimen are closed using a special
end cap (7) with a pressure inlet. This way a crack in either pipe sample could be distinguished from each other. The
internal pressure in the two samples was continuously monitored using the pressure transducers (5) at both ends of
the test sample. After a pressure drop, the connection was disassembled and the exact crack location was found
using die penetrant. The failed pipe was then replaced by a dummy sample and the test continued until failure of the
second pipe specimen occurred. For all standard API Line Pipe specimens, these fatigue cracks initiated at the LET
of the pin.

1b 1a 1b

Fig. 1. Four-point bending fatigue test setup

110 mm
1b 6 1b 7
1a

277 mm

Fig. 2. Section view of test sample

1852 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858

4 J. V. Wittenberghe et al./ Procedia Engineering 00 (2010) 000–000

2.2. Coupling configurations

To assess the influence of the box stiffness on the fatigue life of a threaded coupling, three tests were carried out.
Firstly on standard 1” API Line Pipe connections according to the API spec. 5B [7], and afterwards on two Line
Pipe configurations with a modified box geometry. All couplings were manufactured from 1” API Line Pipe
couplings, consisting of grade B steel according to API spec. 5L [8]. The actual material properties were defined by
tensile testing of tubular specimens. Yield and ultimate tensile strength were found to be equal to ıy = 356 MPa and
ıUTS = 482 MPa respectively.
The two modified box configurations are shown in Fig. 3. In the first modification, the stiffness of the box is
reduced by globally reducing the wall thickness of the box by 1 mm, this is schematically shown in Fig. 3a. The
second modification consists of a local reduction of the recess length with 5 mm as illustrated in Fig. 3b. The recess
regions are the unthreaded parts at both sides of the box, which serve as a guide to assemble the coupling.

a) b)

Fig. 3. Box modifications: a) wall thickness reduction of 1 mm, b) recess length reduction of 5 mm.

2.3. Test results

The primary data obtained from the fatigue experiments are the applied stress amplitude Sa (which is the tensile
stress amplitude at the outer fiber of the lower side of the test specimen) and the corresponding number of cycles to
failure Nf. As described in the previous paragraph, failure is defined as the appearance of a through-thickness crack
that was detected by the pressure gauges. The results for the standard connection are shown in the S-N curve of Fig.
4a. In total 14 tests were carried out on the standard 1” API Line Pipe coupling, of which one specimen did not fail
after 1.5 million cycles. The mean S-N curve was found by regression analysis of the data points. The best least-
squares fit has a standard deviation of STD = 1.44 and can be expressed in exponential notation as:

Sa =238.4 · Nf -0.2233 (1)

The plotted design curve (for a load ratio R = 0.1), was calculated according to the standard BS7608 [9] as the
mean curve minus two standard deviations. For all curves a fatigue limit was assumed at 2 million cycles. Since no
specific fatigue curves exist for threaded pipe couplings, as a comparison the design curve given by BS 7608 for
threaded connections was added to the S-N graph of Fig. 4a. This design curve is strictly speaking only valid for
axially loaded bolts but can be expressed as a function of the tensile strength of the threaded component by equation
(2). In this equation, Sr is the applied stress range or Sr = 2Sa. Equation (2) is given for a load cycle with zero mean
stress (load ratio R = 0). Hence, the Goodman equation (3) was used to correct it to a load ratio R = 0.1. In equation
(3) Sf is the fatigue stress and Sm the mean stress.

3
§ Sr ·
¨ ¸ ˜N f 400 (2)
¨ı ¸
© UTS ¹
 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858 1853

J. V. Wittenbergheet al. / Procedia Engineering 00 (2010) 000–000 5

Sa § S ·
¨¨ 1  m ¸¸ (3)
Sf © ı UTS ¹

According to the BS 7608 design curve for R = 0.1, the fatigue limit of the stress amplitude at 2 million cycles of
a threaded part with a tensile strength of 482 MPa is 13.9 MPa, whereas the fatigue limit that was obtained from the
design curve of the standard 1” API Line Pipe connection in Fig. 4a is 23.9 MPa. Hence, application of the BS 7608
design curve would result in a too conservative design. It should be noted that, the design curve of BS 7608 does not
take into account the specific geometry of the threaded coupling; it is valid for threaded connections with a thread
angle of 60°. For this reason, it is not suitable to compare different box geometry modifications.
To determine the influence of box stiffness reductions on the fatigue life of the couplings, two box modifications
were tested. Firstly 7 couplings with a global wall thickness reduction of 1 mm were tested and afterwards 6 pipes
were assembled using couplings with a recess length reduction of 5 mm at both sides of the box. Two specimens
with reduced recess length did not fail after 1.5 million cycles. The test data together with the corresponding mean
curves are presented in Fig. 4b. The mean curve for the configurations with the wall thickness reduction and reduced
recess length can be expressed by equations (4) and (5) respectively. It can be concluded that both modified
configurations result in an improved fatigue life as compared to the standard API Line Pipe coupling. Currently
more fatigue tests are carried out on the modified configurations, the results will be presented during the conference.
In the following paragraphs, a damage evolution model is applied to affirm these experimental observations.

Sa =248.5 · Nf -0.2167 (4)

Sa =210.6 · Nf -0.2078 (5)

120 120

Experimental Data
Experimental WT -1 mm
100 Unfailed point 100 Experimental Recess - 5 mm
Mean Curve Recess - 5 mm unfailed
Design Curve Mean Curve WT -1 mm
80 Design Curve BS7608 80 Mean Curve Recess -5 mm
Mean Curve Standard
Sa [MPa]

Sa [MPa]

60 60

40 40

20 20

0 0
1.E+04 1.E+05 1.E+06 1.E+07 1.E+04 1.E+05 1.E+06 1.E+07
Nf Nf

a) b)

Fig. 4. S-N curve for load ratio R = 0.1 of a) the standard 1” API Line Pipe coupling, b) two modifications

3. Damage model

3.1. Damage evolution law

To assess the fatigue life of the threaded couplings, the damage evolution law as described in [10] was used. This
damage evolution law was derived using thermodynamic principles, assuming material plasticity at the crack
initiation location. Since plastic deformation is introduced at the thread roots from the moment the coupling was
1854 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858

6 J. V. Wittenberghe et al./ Procedia Engineering 00 (2010) 000–000

made up (see section 3.2), this assumption can be justified. Additionally this law takes into account the multiaxiality
of the stress state at the crack initiation place by introducing a triaxiality function.
The damage evolution law is described by the damage parameter D given in equation (6). D is 0 for the virgin
material and 1 for the completely damaged material. In this equation, A and ȕ are damage parameters and material
constants, m is the strain hardening exponent of the coupling material (m = 9.52 by fitting the experimentally
determined stress-strain curve of the grade B steel to the Ramberg-Osgood relation). ǻıeq is difference between the
maximum and minimum von Mises equivalent stress during a load cycle and RV is the triaxiality function as given in
equation (7) with E = 208 GPa the Young’s modulus and Ȟ = 0.3 the Poisson’s ratio of the steel. ıH is the hydrostatic
pressure.

1
D >
1  1  AE  m  1'V eq
E m E / 2
RV N @ E  m 1 (6)

2
2 ªı º
RV 1  Ȟ   31  2Ȟ « H » (7)
3 «¬ ı eq »¼

The number of cycles to failure Nf is determined from equation (6) by assuming a fully damaged situation (D = 1
and N = Nf). This results in the equation (8):

 E  m   E / 2
'V eq RV
Nf (8)
AE  m  1

In this study the material damage parameters A and ȕ are determined using the test data of the standard API Line
Pipe coupling. The stress components were calculated using the finite element model described in the following
section. The obtained damage parameters were then used to validate the experimentally proven fatigue life
improvement of the modified couplings.

3.2. Finite element model

As is common practice when modeling threaded couplings [11–13], a 2D axisymmetric finite element model is
used. This has the advantage of a drastically reduced calculation time as compared to full 3D models. Additionally,
the results of 2D axisymmetric models are in very good agreement with full 3D models as is known from [14] in the
analysis of bolts and from [15] for threaded pipe couplings.
In Fig. 5, a schematic representation of the standard API Line Pipe connection is shown. The red dashed square
indicates the 2D axisymmetric model. Only a half coupling is modeled since a plane of symmetry is assumed at the
centerline of the box (see the boundary conditions at the right in Fig. 6). The material model used, is based on the
results of the tensile tests on the API Grade B material. This resulted in a multi-linear elastic-plastic material model
with kinematic hardening. The coefficient of friction between the pin and box threads was determined to be 0.12
[16].
The model was generated using Python scripts in the software package ABAQUS® 6.8. This allows to easily
apply geometrical modifications to the model. The model’s mesh was optimized to give an accurate stress
distribution around the thread roots of the pin, since it is known that those locations are most susceptible for fatigue
crack initiation. Therefore a fine mesh was seeded with 22 elements along the pin thread roots. Detailed information
on the used modeling techniques and discussion of some results can be found in [17].
The equivalent von Mises stress distribution for make-up and for make-up with an additional axial tensile stress
of 150 MPa is given in Fig. 6. Note that even in the situation of make-up only, the stresses in the pin exceed the
material’s yield strength. This is mainly caused by high compressive hoop stresses introduced during make-up. The
stresses in the box remain much lower than in the pin, since the box is the more rigid part. When an additional axial
tensile stress is applied (see Fig. 6b), the highest stress concentration appears at the root of the LET of the pin
 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858 1855

J. V. Wittenbergheet al. / Procedia Engineering 00 (2010) 000–000 7

(indicated by the arrow). This corresponds with the observations from the experimental tests that fatigue cracks
initiate at this location. It is noted that the bending moment as applied during the experiments can not be introduced
in the 2D axisymmetric model. However, since the wall thickness to diameter ratio of the pipes is sufficiently small,
the situation where the axial tensile stress in the 2D axisymmetric model equals the bending stress at the outer fiber
of the test sample is representative for the maximum stresses occurring in a real test specimen during four-point
bending.
To evaluate the fatigue life of the threaded couplings, the triaxiality function Rv is calculated at a small distance
from the root of the LET of the pin. The exact location was determined as the distance at which the stress is
independent of mesh refinement, as is described in [10]. This is done to avoid a mesh dependent solution.

Fig. 5. Section view of the standard API Line Pipe mode; the red dashed square indicates the 2D axisymmetric model

150 MPa

Fig. 6. Equivalent von Mises stress distribution resulting from a) make-up; b) external axial tensile stress of 150 MPa

3.3. Results and discussion

Since the damage parameters A and ȕ from the damage evolution law are unknown material constants, the
experimental data from the standard connection is used to determine their value. This is done by evaluating the finite
element model at 8 different levels of stress amplitude between 45 and 93 MPa. For each of these stress levels, the
triaxiality function RV and the equivalent stress range ǻıeq are calculated and substituted in the damage evolution
law (8). The values of the damage parameters are then found using regression analysis to calculate the best least-
square fit between the calculated cycles to failure and the mean curve of the experimental data given by equation
(1). The mean curve of the experimental data for the standard connection together with the 8 points obtained with
equation (8) are plotted in Fig. 7. This methodology yields to the following values: A = 5.38 · 10-6 and ȕ = -8.50 and
the damage evolution law becomes:

1.02 4.25
Nf 9.20 ˜ 10 4 ǻı eq RV (9)
1856 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858

8 J. V. Wittenberghe et al./ Procedia Engineering 00 (2010) 000–000

With the damage evolution law (9), the fatigue life can be calculated for the two modified coupling
configurations. Again this was done by evaluating 8 stress amplitude levels for both configurations. The results of
these calculations are shown in Fig. 8. Additionally, numerical values for the standard coupling and both modified
configurations, evaluated at a stress amplitude of 67.5 MPa (ımin = 15 MPa, ımax = 150 MPa) are summarized in
Table 1. It can be seen that for both configurations the damage evolution law correctly predicts the increased fatigue
life of the couplings. However, for the configuration with reduced wall thickness, the fatigue life is underestimated
in the high stress region and overestimated in the low stress region. For the configuration with the reduced recess
length, a better correspondence can be observed. The discrepancy between the results may be explained by
following assumptions that were made. First, the fatigue of the couplings consists of two phases; crack initiation and
crack propagation. In this paper it is assumed that the crack initiation dominates the fatigue life of the connection
(this assumption is based on observations of the hydraulic cylinder displacement during the load-controlled test).
Second, in general the damage parameters can be influenced by the stress state, however this was not taken into
account and the damage parameters are considered to be constant.

120

100

80
Sa [MPa]

60

40

Mean Curve Experimental data Standard API

20
Damage evolution law

0
1.E+04 1.E+05 1.E+06
Nf

Fig. 7. Mean curve of the experimental data for the standard coupling together with the best fit of the damage evolution law.

120 120
Mean Curve Standard Mean Curve Standard

Mean Curve WT -1mm Mean Curve Recess Length -5mm

Damage Evolution Law

100 Damage Evolution Law 100
Regression Line Damage Evolution Law Regression Line Damage Evolution Law
Sa [MPa]

Sa [MPa]

80 80

60 60

40 40
1.E+04 1.E+05 1.E+06 1.E+04 1.E+05 1.E+06
Nf Nf

a) b)

Fig. 8. Damage evolution law (9) compared with the mean curves of the experiments on the configuration with a) a wall thickness reduction of
1 mm, b) recess length reduction of 5 mm.
 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858 1857

J. V. Wittenbergheet al. / Procedia Engineering 00 (2010) 000–000 9

Table 1. Values of triaxiality RV, equivalent stress range ǻıeq evaluated at a stress amplitude of 67.5 MPa and cycles to failure calculated by the
damage evolution law (8) and according to the mean curves from the experimental data.

Nf [-] Nf [-]
RV [-] ǻıeq [MPa]
by eq. (8) experimental
Standard API Line Pipe 1.94 24.2 6.7 · 104 8.4 · 104
Wall Thickness -1mm 2.01 9.0 20.5 · 104 14.5 · 104
Recess Length -5mm 1.97 21.2 8.2 · 104 11.1 · 104

4. Conclusions

In this study the fatigue life of threaded API Line Pipe couplings is determined experimentally. Additionally two
modified coupling configurations are presented that exhibit improved fatigue properties. The configuration with a
global wall thickness reduction of 1 mm showed the highest life improvement.
The experimental observations are evaluated using a damage evolution law together with the numerical results of
a finite element model. The necessary material damage parameters were obtained from the experimental results of
the standard API Line Pipe connection. These material parameters were further used in the damage evolution law to
asses the fatigue life of the two modified configurations. The experimental observations of fatigue life improvement
were confirmed by this analytical-numerical study.

Acknowledgements

The authors would like to acknowledge the financial support of the BOF fund (B/04939) of the Ghent University
and of the FWO Vlaanderen (3G022806).

References

[1] Vaisberg O, Vincké O, Perrin G, Sarda JP, Faÿ JB. Fatigue of Drillstring: State of the Art. Oil & Gas Science and Technology – Rev. IFP
2002;57(1):7–37.
[2] Santi NJ, Carcagno GE, Toscano R. Premium & Semi-premium Connections Design Optimization for Varied Drilling-with-Casing
Applications. Offshore Technology Conference 2005; OTC17221.
[3] Strickler RD, Wadsworth TM. Drilling with casing: Are you damaging your casing?. World Oil 2005;3:51–53.
[4] Carcagno GE, Bufalini A, Conde L, Toscano R. Low SCF Integral Premium Connections for Use in Highly Demanding Casing and
Tubing Drilling Operations. Offshore Technology Conference 2004;OTC16566.
[5] Griffin RC, Kamruzzaman S, Strickler RD. Casing Drilling Drill Collars Eliminate Downhole Failures. Offshore Technology Conference
2004;OTC16566.
[6] Newport A, Glinka G, Concentration of Cyclic Stresses in Preloaded Threaded Connections. J Eng Mech-ASCE 1991;117(6):1257–1273.
[7] API Specification 5B. Specification for Threading, Gauging and Thread Inspection of Casing, Tubing and Line Pipe Threads (U.S.
Customary Units). American Petroleum Institute 1996:fourteenth edition.
[8] API Specification 5L. Specification for Line Pipe. American Petroleum Institute 2000:forty-second edition.
[9] BS7608:1993. Code of practice for: Fatigue Design and Assessment of Steel Structures. British Standards Institution 1999.
[10] Abdel Wahab MM, Ashcroft IA, Crocombe AD, Shaw SJ. Predition of fatigue tresholds in adhesively bonded joints using damage
mechanics and fracture mechanics. J Adhesion Sci. Technol. 2001;15(7):763-781.
[11] Assanelli AP, Xu Q, Benedetto F, Johnson DH, Dvorkin EN. Numerical/experimental analysis of an API 8-round connection. J of
Energy Resources Technology-Transactions of the ASME 1997;119(2):81–88.
[12] Dvorkin EN, Toscano RG. Finite element models in the steel industry, Part II: Analyses of tubular products performance. Computers &
Structures, 2003;81(8–11):575–594.
[13] Guangjie Y, Zhenqiang Y, Qinghua W, Zhentong T. Numerical and experimental distribution of temperature and stress fields in API
round threaded connection. Eng Failure Analysis 2006;13(8):1275–1284.
1858 J.V. Wittenberghe et al. / Procedia Engineering 2 (2010) 1849–1858

10 J. V. Wittenberghe et al./ Procedia Engineering 00 (2010) 000–000

[14] Chen JJ, Shih YS. A study of the helical effect on the thread connection by three dimensional finite element analysis. Nuclear
Engineering and Design 1999;191(2):109–116.
[15] Zhong A. Thread Connection Response to Critical Pressures. Abaqus Users’ Conference. Paris, France 2007;690–706.
[16] Van Wittenberghe J, De Baets P, De Waele, W, Van Autrève S. Numerical and Experimental Study of the Fatigue of Threaded Pipe
Couplings, Contact and Surface Conference. In: De Hosson JTM, Brebbia CA, editors. Surface Effects and Contact Mechanics IX, Southampton:
WIT Press; 2009, p. 163–174.
[17] Van Wittenberghe J, De Baets P, De Waele, W. Modelling of preloaded threaded pipe connections. Proc. of the 8th Nat. Congress on
Theor. and App. Mechanics. Brussels, Belgium 2009; 149–156