Load Distribution Between Threads in Threaded Connections

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

A. S.

Grewal
Product Design Engineer,
Light Truck VC, Ford Motor Company,
CAE Technology,
Livonia, Ml

M. Sabbaghian
Professor,
Department of Mechanical Engineering,
Louisiana State University,
2508 CEBA,
Baton Rouge, LA 70803-6413

Load Distribution Between


Threads in Threaded
Connections
Threaded connections are commonly employed in axial load-bearing equipment and
pressure vessel components. There are a number of parameters that affect the load
distribution between the threads and the stress concentration at the thread roots.
These include the thread form, the thickness of walls supporting the threads, the pitch
of threads, number of threads engaged, and the boundary conditions. In this paper,
the influence of these parameters on the load distribution between threads is reported.
Load distribution analyses in threaded connections is performed by analytical and
by finite element methods. Square and buttress-type threads have been considered.
Three-dimensional nonlinear finite element analyses on threaded connections have
been performed using MSC/NASTRAN finite element code. The effect of clearance
between the nonmating faces of threads as well as the presence of a flexible media
between the mating faces of threads are investigated.

Introduction
The load distribution among the threads in a threaded connection is not uniform. Most of the applied load is carried by the
first three engaged threads. The load distribution depends upon
a number of parameters. These include the form of threads, the
thickness of the walls supporting the threads at the threaded
section, the pitch of the threads, the number of engaged threads,
and the boundary conditions.
The well-accepted analytical model to predict the load distribution between the threads was developed by Sopwith (1948)
during World War II. His efforts led to a second-order differential equation to predict the load distribution between the threads.
Gray (1954) performed the photoelastic studies to check the
validity of Sopwith approach. Sopwith (1948) and Hey wood
(1948) outlined an approach that can be used to estimate the
stresses at the root of the first loaded thread.
Kenry and Petterson (1985) found that Sopwith results
were in agreement with their experimental results, except at
the loaded face of the nut where the thread was not fully
formed. Using finite element analysis, they included (1986)
thread deflection factors to be used in conjunction with Sopwith equations, thus accounting for thread at the loaded face.
Ezzat and Jean (1994) developed a nonlinear analytical
model for determination of the additional load in the bolted
connections subjected to high eccentric loading. Although
Sopwith's method for obtaining load distribution in threaded
connections is well accepted for nuts and bolts, very little
has been published on its application when designing the
threaded end closure of high-pressure vessels. Warnke
(1967) used Sopwith's analysis in the design of an isostatic
compaction vessel that had failed. Burn and Chabaan (1988)
applied Sopwith's work to buttress threads. They compared
Sopwith's prediction with different finite element predictions
and concluded that Sopwith's method overestimated the load
level carried by the first three threads in the case of more
than 12 active threads, and underestimated that in case of
fewer active threads.

Finite element analyses on threaded connections of thick


wall pressure vessels have been performed by different investigators such as Chaaban and Jutras (1988, 1992), Pick and
Burn (1971), Perez and Sloan (1987), Burn et al. (1987),
Kenny and Patterson (1985), and Fasiczka (1988). Crum
(1988) proposed an analytical method based upon the relative
stiffness of the vessel body, the closure, and the threads. His
results for the case of 15 engaged threads predicts fairly
accurately the same load distribution as those obtained using
finite element method. However, in the case of less than 15
engaged threads, his model predicts, for the first threads, a
load well below that predicted by finite element methods as
indicated by aforementioned references. In this publication,
Crum's analytical model is modified.
Analytical Models
Pressure Vessel. A pressure vessel with an internal plug
is shown in Fig. 1. It can be modeled by spring elements as
shown in Fig. 2. Applying the constraint at a-a' to the assembly,
the matrix form of force compliance relationship for JV engaged
threads (Crum 1988) can be written in matrix form as

PLUG

VESSEL

Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the PVP

Division, April 11, 1995; revised manuscript received April 16, 1996. Associate
Technical Editor: W. K. Liu.

Fig. 1 Pressure vessel with internal plug

FEBRUARY 1997, Vol. 1 1 9 / 9 1

Journal of Pressure Vessel Technology

Copyright 1997 by ASME

Downloaded 29 Jan 2009 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

'K0~
C + C c
C + Cc

2(C + Cc) + T
2(Cb + Cc)

2Cb + Cc
3(C + Cc) + T

2Cb
3Cb + 2CC

Cb + Cc

2(Cb + Cc)

3(Cb + Cc)

4(C + Cc))

where Cb and Cc are compliance of any vessel body spring and


plug spring, respectively, and T is twice the compliance of any
thread spring.
Mechanical Screw Jack. For a mechanical screw jack,
threaded connection shown in Fig. 3 can be modeled by spring
elements shown in Fig. 2. However, in this case, the model is
constrained at b-b' instead of a-a'. The matrix form of force
compliance relation in a similar manner can be written as

NCb + Cc + T
(N-l)Cb
+ Cc
(N - 2)Cb + Cc

(N-DQ
(N - \)Cb + 2CC + T
(N - 2)Cb + 2CC

Cb + Cc

Cb + 2CC

Finite Element Models


Figure 4 shows the finite element model of pressure vessel
threaded connections having square threads. Twenty noded isoparametric solid elements (HEXA) were adopted for nonlinear
finite element analysis using MSC/NASTRAN finite element
code (v67.5). In the threaded end region, adaptive frictional
elements (GAP) were used in order to account for various levels
of friction between threads. Figure 4 is the segment cut out
from complete 360-deg models. Due to the disk space limitation
in the computer system, only up to four fully engaged threads in
the case of 360-deg model could be analyzed. In the segmented
models using cylindrical coordinates, hoop displacements at the
segment boundary were constrained. In each case, at the inner
edge of the solid end of the plug, radial displacements as well
as the rotations about the axial and radial axes were constrained.
By comparing the results of the full 360-deg model with the
segmented models subjected to aforementioned constraints, it
was observed that the ratios of load distribution among threads
were almost the same. Hence, the load distribution analysis was
performed on the segmented models. Also, in all these cases,
the first pair of engaged threads is the pair of mating threads
closest to the face of inner body on which the pressure acts.
The length of cylindrical pressure vessel before the first engaged
thread was taken as approximately four times the thickness of
the cylinder, and all the displacements at that section were
constrained. The finite element model of a mechanical screw
jack is similar to the model of a pressure vessel, with the only
difference being that, in this case, the length of the outer body
after the last engaged thread was taken as four to five times
of its thickness and all the displacements at that section were
constrained.

FN.

Ko

(1)

-Ko.

Results and Discussion


Pressure Vessel Threaded Connection. The predicted
thread load values from the nonlinear finite element analysis of
the threaded assembly with Y' = 2.1 and Y = 1.55 are summarized in Fig. 5, where
Y' = DpIDi
Y = Do/Dp

(N - 2)Cb - Cc
(N - 2)Ch + Cc
(N - 2)Cb + 3CC + T

where Cc is one-half of the compliance of any inner body spring


and the rest of variables are the same as those in Eq. (1). Given
a set of values of K0, one can solve the foregoing matrices to
obtain the ratio of forces FJFi. Based on such results and the
equilibrium equation, the thread load distribution for a given
threaded connection can then be obtained. The results of these
models are discussed later on.

92 / Vol. 119, FEBRUARY 1997

F2
F,

(N-3)C~2CC
(N-3)Cb

(N - 3)C + 2CC
Cb + 4CC

~Fi~
F2
F3

fs.

~Ko
Ko
Ko

(2)

.Ko.

As shown in Fig. 1, Dp is the pitch diameter of engaged threads,


Di is the inner diameter of plug, and Do is the outer diameter
of pressure vessel. All the results discussed in this paper are
for a friction coefficient of 0.15 between the engaged threads.
It is noted that the load carried by the first thread increases
slightly as the coefficient of friction increases. The load distribution on threads is also affected by the parameters Y and J", as
shown in Table 1. Comparing the results obtained from these
analyses, it is observed that for a constant value of Y', the load
carried by the first thread decreases as the value of Y increases.
However, beyond Y = 1.55, the variation in the load among

Fig. 2 Spring model of threaded connection

Transactions of the ASME

Downloaded 29 Jan 2009 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 5 Thread load distribution in pressure vessel ( V = 2.1, Y = 1.55)


Fig. 3

Mechanical screw jack assembly

the threads becomes less than one percent of the total load. In
a similar manner for a constant Y, the load on the first thread
decreases with increasing Y'. Again, as Y' approaches and
exceeds the value of 2.1, the variation in load becomes very
small. From this observation of the load distribution pattern
among threads with the change of thickness of the walls at
the threaded section, one can conclude that the optimal load
distribution is achieved with Y = 1.55 and Y' = 2.1. Any further
increase in either of the values of Y or Y' beyond those stated
has very little effect on the thread load distribution, while any
decrease in either of these values results in an increase in the
load carried by first engaged thread.
To verify the validity of the analytical model, it was applied
to several different combinations of Y and Y'. The results of
the analytical model are summarized in the last three rows of
Table 1. In the first case, the outer, pitch, and inner diameters of
the threaded connection were 31.0, 20.0, and 9.5, respectively.
Hence, Cb, Cc, and T in this case were 1.97785e~*IE,

I- '!

1.99567e" 3 /', and 4.7746e~ 2 /, respectively, where E is the


modulus of elasticity of the threaded connection material. In
the last two cases, the outer, pitch, and inner diameters of the
threaded connection were 22.5, 15.5, and 8.5, respectively.
Hence, for this connection Ch, Cc, and T were 5.08787e~ 3 /,
4.21491 e~*IE, and 6.16084e~ 2 /, respectively. By comparing
these results with finite element results, it becomes apparent
that the thread load distribution predicted by analytical model
compares very well with those obtained by finite element
method.
For the same values of Y and Y' comparing the results of
square threaded connection and buttress threaded connection, it
is found that there is no significant difference between the two
cases. Also, if there is clearance between the nonmating faces
of threads, then load on the first thread increases from one to
two percent. By comparing the results of Table 1 with the 2-D
axisymmetric analyses (Chaaban and Jutras, 1992), it appears
that 2-D axisymmetric analyses predict fairly accurate load distribution between threads.
Mechanical Screw Jack Threaded Connection. The predicted thread load distributions for a mechanical screw jack in
the case of Y' = 1.8, Y = 1.45 from nonlinear finite element
analyses are summarized in Fig. 6. As shown in the figure, the
load carried by the last engaged thread is significantly higher
compared to the load carried by the second or third engaged
thread, and in some cases it can be the highest. The load distribution on the thread in a mechanical screw jack also depends upon

Table 1 Mean fraction of load carried by first three threads*


Y

No of
threads

F:/FT

V T

F.E.

1.55

F.E.

1.55

2.1

0.345

0.224

0.173

2.1

0.289

0.181

F.E.

0.136

1.30

1.8

0.422

0.239

0.162

F.E.

1.45

1.8

0.404

0.228

0.156

F.E.

1.45

1.8

0.387

0.209

0.144

F.E.

1.65

1.8

0.400

0.225

0.163

F.E.

1.65

2.1

0.342

0.223

0.175

I
'

Method

F.E.

'

'

Fig. 4 A 45-deg segment of finite element model of pressure vessel

Journal of Pressure Vessel Technology

F 3 /F T

1.65

2.1

0.284

0.179

0.135

ANALYT

1.55

2.1

0.321

0.233

0.176

ANALYT

1.45

1.8

0.381

0.245

0.162

ANALYT

1.45

1.8

8
0.234
0.371
is the i otal axial lo ad applied o the threacled connection

0.147

* FT

FEBRUARY 1997, Vol. 1 1 9 / 9 3

Downloaded 29 Jan 2009 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 3 Fraction of load carried by first three threads with flexible media
between the mating faces
No of
threads

F,/FT

F ^ T

F3/FT

1.55

2.1

1.2x10"

0.234

0.242

0.209

1.55

2.1

8.0xl0 7

0.220

0.240

0.213

1.55

2.1

4.0x10'

0.194

0.233

0.219

1.55

2.1

4.0x10'

0.163

0.220

0.224

1.55

2.1

1.2x10"

0.198

0.197

0.164

1.55

2.1

8.0x10'

0.186

0.196

0.166

1.45

1.8

1.2x10*

0.372

0.233

0.161

1.45

1.8

4.0x10'

0.361

0.232

0.164

1.45

...

1.2x10"

0.278

0.254

0.211

* F T is the total axial load applied to tne pressure vessel

Effect of Flexible Media

4
5
Thread Number

As observed from Table 1, the load distributed on the first


thread in the pressure vessel threaded connection having Y 1.55 and Y' = 2.1, is around 34.5 percent. This high percentage
of load causes high stress concentration at the root of first
the stiffness of the wall of screw jack body, the stiffness of the thread. There are many documented failure of pressure vessels
threads, and the stiffness of the inner body. As shown in Table subjected to cyclic loading (Warnke, 1968). The fatigue crack
2, the load carried on the first thread decreases and increases in a threaded connection usually initiated at the root of first
on the last thread as the value of Y decreases. In the case of Y thread because of the high load carried by the first thread and
= 1.3, the load carried by the last thread is much higher than the stress concentration.
that carried by the first thread. One may observe from Table 2,
To explore the effect of a new parameter, the pitch of the
in the case of the mechanical screw jack, that the increase in
threads
was increased, maintaining the previous conventional
the value of Y' also makes the load distribution more uniform.
However, the effect of an increase in Y' diminishes and be- thickness of threads. Then a flexible media was introduced becomes insignificant beyond Y' = 2.1. Hence, any increase in tween the mating faces of threads. The thread load distribution
the value of Y' above 2.1, has practically no significant effect for such a case is shown in Table 3. The tabulated results are
media. The units of
on the thread load distribution in either the case of pressure in terms of stiffness (K) of the introduced
3
stiffness
used
in
this
section
are
lb
/in
,
i.e.,
the stiffness per
f
vessel or of the mechanical screw jack.
unit mating face surface area. The exact choice of flexible mateTo verify the validity of the analytical model, several differ- rial will depend upon the operating conditions of the threaded
ent combinations of Y and Y' were tried. The results are shown connection. With a given material, the dimensions, particularly
in Table 2. In the first case, the outer, pitch, and inner diameters the thickness of the material, could be selected to obtain the
of the threaded connection were 14.5, 10.0, and 5.5, respec- desired stiffness. As observed by comparing Table 1 and Table
tively. Hence, Cc, C, and T were 8.488e~3/, 1.2732*-2/ 3, in the case of Y = 1.55 and Y' = 2.1, the maximum load on
E, and 9.5493e~2/E, respectively, where E is the modulus of the thread decreases by 9 to 10 percent of the total load, by
elasticity of the threaded connection material. In the last two putting a flexible material of stiffness 1.2e + 8 lbf/in3 between
cases, only the outer diameter of the connection was changed the threads. It was observed from the 2-D axisymmetric finite
to 16.5 and the other dimensions were same as in the first case. element analyses of the threaded assembly having Y = 1.55 and
Hence, the value of Cb in this case was 7.859e ~ 3 /. From Table 1" = 2.1, that the maximum load carried by the first thread is
2, it is clear that predicted thread load distribution by analytical 20 percent in the case of 15 to 16 engaged threads. But with
model again compares very well with the finite element method. the introduction of flexible media, the maximum load on the
The results obtained by finite element analyses for the screw first thread is reduced to 19.8 percent for the case of 8 engaged
jack threads having clearance between the nonmating faces of threads. Hence, in this case, up to 30 to 40 percent of threaded
the threads, shows that the loads carried by the first engaged section material can be saved by putting suitableflexiblematethreads were higher by four to five percent of the values shown rial between the mating faces of threads.
in Table 2.
The percentage of reduction of maximum load on a thread
reduces with decrease in the value of Y' and Y. In the case of
Table 2 Mean fraction of load carried by first, second, and last engaged
assembly having Y' = 1.8 and Y = 1.45, the load decreases by
thread
about 3 percent, using the same flexible media (K = 1.2e + 8
lbf/in3) as before. The load distribution in such a case can be
Method
No of
Y
y
F
/
F
F
/
F
*VF T
Fs/F T
made further uniform by modifying the inner plug, based upon
Threads
the previous studies of the behavior of load distribution pattern.
5
1.80
F.E
1.45
0.253
0.171
0.259
...
One such load distribution is shown in the last row of Table 3.
F.E.
8
1.80
0.137
0.202
0.231
0.073
1.46
In this case, the plug was solid up to two engaged threads, and
F.E.
5
1.80
0.205
0.162
1.30
0.289
after that 1" was 1.8. Hence, by introducing the flexible media
between threads, the load and stress distribution in a threaded
F.E.
5
1.65
1.80
0.300
0.192
0.206
...
connection can be more uniform compared to the conventional
F.E.
5
2.75
0.229
0.166
1.45
0.268
...
threaded connections.
0.214
F.E.
8
2.75
0.128
1.45
0.216
0.079
Fig. 6 Thread load distribution in screw jack ( V = 1.8, Y = 1.4S)

ANALYT

1.45

1.80

0.281

0.162

0.057

ANALYT

1.45

1.80

0.304

0.193

0.203

ANALYT

1.80
0.328
1.65
* t'j is t h e totai axial lot d a'ppliec1 to the JEck

0.210

0.167

94 / Vol. 119, FEBRUARY 1997

0.166

...

Conclusions
1 In the case of a pressure vessel with conventional
threaded connections, there will be an optimum load distribution
Transactions of the ASME

Downloaded 29 Jan 2009 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

among threads if Yand Y' are at least 1.55 and 2.1, respectively.
Furthermore, there will be no significant change in the load
distribution for Y or Y' larger than the stated values.
2 For the same values of Y and Y\ there is no significant
difference in the thread load distribution between the square
threaded connections and the buttress threaded connections.
3 The presence of clearance between nonmating faces of
threads increases the load on the first thread by only about two
percent in the case of a pressure vessel and by aboutfivepercent
in a mechanical screw jack.
4 In a mechanical screw jack, in some cases, the distributed
load on the last engaged thread can be higher than it is on the
first engaged thread.
5 The maximum load carried by the engaged thread can be
reduced significantly by introducing a suitable flexible media
between the mating faces of engaged threads. The percentage
of reduction of maximum load on the engaged threads reduces
with decrease in the value of Y or Y' from their respective
optimum values.
6 In a threaded connection of a thin wall pressure vessel,
the distributed load on the first thread can be reduced significantly by making the plug or cap solid up to two to three
engaged threads and putting a suitableflexiblemedium between
the mating faces of threads.

References
Bakhiet, E. M., and Guillot, J. F 1994, "Analytical Model of Bolted Joints
Subjected to High Eccentric Loading," ASME PD-Vol. 6 4 - 8 1 , Engineering System Design and Analysis, Vol. 8, Part A, p. 117.
Burn, D. J Chaaban, A., Mraz, G and Neubard, P. D., 1987, "Load stress
and Fatigue Analysis of Threaded end Closure'' High Pressure Engineering and
Technique, ASME PVP-Vol. 125, pp. 63-71.

Journal of Pressure Vessel Technology

Chaaban, A., and Jutras, M., 1988, "Static and Fatigue Analysis of Buttress
Threads of Thick Walled Vessel Using the Finite Element Methods," High Pressure Technology: Material, Design, Stress Analysis and Applications, ASME
Pressure Vessel and Piping Conference, Pittsburgh, PA, June, p. 35.
Chaaban, A., and Jutras, M., 1992, "Static Analysis of Buttress Threads using
the Finite Element Method," ASME JOURNAL OF PRESSURE VESSEL TECHNOLOGY,
Vol. 114, p. 209.
"Compendium of Gas Autoclave Engineering Studies," 1965, Y-12 Engineering Division, Union Carbide Report, No. Y-1478.
Crum, A. S. D., 1988, "The Distribution of Load Over Threads of a Threaded
Closure, Including Temperature Effects and Local Imperfections," High Pressure
Technology, Material Design Stress Analysis and Applications, ASME PVP-Vol.
148, p. 43.
Fasiczka, R. G., 1988, "Quantitative Significance of the Design Parameters of
Threaded End Closure," High Pressure Technology: Material Design, Stress
Analysis and Application, ASME PVP-Vol. 149, p. 139.
Gray, R. M 1954, "Preliminary Investigation of Load Distribution in Threaded
Connections," Report No. WAL 730/562-47, Purdue Research Foundation, Contract No. DA-11-022-DRD-17.
Heywood, R. B., 1948, "Tensile Fillet Stresses in Loaded Projections," Proceedings of the Institution of Mechanical Engineers, Vol. 159, p. 384.
Kenry, B., and Patterson, E. A., "Load and Stress Distribution in Screw
Threads," Experimental Mechanics, Vol. 25, 1985, pp. 208-213.
MSC/NASTRAN Finite Element Program, The MacNeal Schwendler Corporation U.S.A., Version 67.5, 68.
Patterson, E. A., and Kenry, B., 1986, " A Modification to the theory for the
Load Distribution in Conventional Nuts and Bolts," Journal of Strain Analysis,
Vol. 21. No. 1, p. 17.
Perez, E. H Sloan, J. G., and Keller, K. J., 1987, "Application of Fatigue
Crack Growth to an Isostatic Press,'' High Pressure Engineering and Technique,
ASME PVP-Vol. 125, pp. 53-61.
Pick, R. J., Burn, D. J., 1971, "Finite Element Analysis of threaded end closure
of Thick walled Vessel,'' Proceedings of the Institute of Mechanical Engineers,
London, U.K., pp. 15-25.
Sopwith, D. G 1948, "The Distribution of load in Screw Threads," Proceedings of the Institution of Mechanical Engineers, Vol. 159, p. 373.
Warnke, E. G 1968, "Design and Fatigue Life of Operable High Pressure
Vessel," Conference on High Pressure Engineering, Proceedings of Institution
of Mechanical Engineers, Vol. 182, p. 47.

FEBRUARY 1997, Vol. 1 1 9 / 9 5

Downloaded 29 Jan 2009 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

You might also like