The Effect of Bending on the

Normalized Stress at Roots of

Threaded Connectors
R. L. Burguete
E. A. Patterson
Department of Mechanical and
Process Engineering,
University of Sheffield,
Sheffield, England

Three-dimensional photoelasticity was used to analyze the effect of bending on the

normalized stress at the roots of threaded connectors. Loading was effected by steel
cages and a combination of eccentric weights (to provide the bending load) and
concentric weights {to provide the axial load). The ratio of the bending stress to
the axial stress was determined and various levels of this stress ratio, Ra, were tested.
The connections were analyzed by taking thin slices in the plane of bending and
perpendicular to it. The position of the maximum fringe order at the roots was
determined using Mesnager's theorem and the maximum fringe order found by
Tardy compensation. The fringe orders were normalized using the nominal axial
stress and the total nominal stress (bending plus axial stress), which were calculated
from the loads applied. The results, when normalized using the nominal axial stress
and compared to those in connections without bending, exhibit a lower and broader
peak of normalized stress values plotted against the helix length. The normalized
stress values are also periodic in relation to the bending plane due to the variation
in stress around the longitudinal axis of the bolt. It was found that bending in
connectors will affect the normalized stress and that it is possible to determine this
effect in a similar way to the method used for axially loaded connections.

Introduction
The analysis of the normalized stress at the root of screw
threads has traditionally only considered bolts loaded axially.
With the advent of new methods of analysis of threaded connections, which show that most bolts are loaded eccentrically,
the need for a more realistic analysis has arisen.
The stress concentration at the thread root has a major
influence on the fatigue life of the connection (Patterson, 1990).
A detailed knowledge of the stress concentrations and the way
in which they are affected by the applied loads is crucial for
good design. It is known that eccentric loading increases the
load on a bolt, but the effect on the thread root stress is not
known. It is important to find a relation between the thread
loads and/or stresses, and the bending loads that affect the
connection.
The approach used to do this was divided into two stages:
the first stage will be described as simple bending, and the
second stage is called joint bending. This paper describes only
simple bending. The reason for this is in part to simplify the
analysis so that the effect of bending can be broken down into
components. In the second stage, bending is applied more
realistically, but analysis is more complex. This will be covered
in a separate paper.
The method used to apply bending is similar to the methods
used in Patterson and Kenny (1987), with an additional load
applied to create a bending moment. The main benefit of this
Contributed by the OMAE Division for publication in the JOURNAL OF OFFSHORE MECHANICS AND ARCTIC ENOINEERINO. Manuscript received by the OMAE

method is that a simple relation between bending load and the

normalized stress distribution in the thread can be sought.

Experimental Setup
The bolts that were tested were scaled Ml2 double ended
bolts according to BS 3643 (ISO, 1981). All the dimensions of
the Ml2 bolt were doubled, to give a bolt with nominal diameter of 24 mm and a pitch of 3.5 mm. The material used
was epoxy resin MY750 prepared according to Kenny (1965).
The dimensions of the bolts, nuts, and washers are shown in
Fig. 1.
The apparatus used to load the bolts is shown in Fig. 2. The
upper half, which is not shown, is symmetrical to the lower
half with the central tie bar used to hang the rig inside the
oven.
The bending of the bolt was affected by applying an eccentric
force to the cage. The applied loads were specified so that the
nominal bending stress at the thread roots (i.e., excluding
concentration effects) was a specific fraction of the nominal
axial stress. The applied loads are listed in Table 1.
The parameter Ra has been defined for these experiments
to allow simple categorization of the results. This parameter
is defined by Eqs. (3) and (4). The values aa and ab are the
axial and bending stresses, respectively, and are defined as
follows:

Division, March 14, 1994. Associate Technical Editor: S. Liu.

Journal of Offshore Mechanics and Arctic Engineering

Copyright 1994 by ASME

ua = 4(Fa + Fb)/ird2s

(1)

AUGUST 1994, Vol. 116/163

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

54.00

Table 1 Applied loads for the various values of R tested

Eccentric load, N
Axial load, N
Re
7.0
0.47
126
10.6
0.96
89
12.8
1.42
68
14.2
1.90
53

minor diameter of the bolt. The value of <jmax (0.57 MPa) was
obtained by preliminary experiments to ascertain the stress level
which provided an adequate number of fringes, suitable for
easy measurement without inducing failure.
Ob

= R

(3)

All dimensions in mm

Fig. 1 Dimensions of test specimens--(a) double-ended bolt, (b) washer,

(c) nut
Epoxy Resin Bolt

ab + aa = a m a x

j2

Epoxy
Resin
~ Washer

Epoxy
Resin
Nut
Load
Eccentricity
20mm

~CT
jmi

or

where
Fa =
Fb =
d =
ds =

Steel Cage

Eccentric Load

Bolt loading apparatus

T3

.w
"ra

o
2

0.0
-14.0

-7.0

0.0

7.0

14.0

21.0

Distance f r o m loaded face of nut, m m

Fig. 3 Normalized thread root stress distribution for bolt with R,

ff* = =

32Fj,e
vd3

= 0.5

(2)

where the load Fa is the weight of the cage plus the load
suspended from it.
The axial and bending stress are calculated using the actual
1 6 4 / V o l . 116, AUGUST 1994

32e

32e

Tie Bar -

Fig. 2

(4)

Using these criteria, formulas were developed for the axial

and eccentric loads as functions of Ra, amm and the bolt dimensions, as shown in Eqs. (5) and (6)

Rn+l

Rn+l

(5)
(6)

axial load
eccentricity of load providing bending moment
minor diameter of bolt
stress diameter of bolt (BS 3643Note: For the calculation of ds, the value d has been substituted for e?3
as used in the standard, as d is the real minor diameter,
not one based on the nominal major diameter of 24
mm)

Experimental Procedure
The process of stress freezing required the specimens to be
heated in an oven using a carefully controlled heating cycle,
as described in (Patterson and Kenny, 1987). The oven temperature was raised to 140 C at a rate of approximately 1.5 C
per h. The specimen was "soaked" at this temperature for 6
h, and then the temperature was reduced to approximately
50C at a rate of 3C per h.
A tensile calibration specimen was loaded with the test specimen, which had a known cross-sectional area and was loaded
in pure tension. This specimen is required for the calibration
of the fringe values that are measured in the test specimen.
The stress-frozen specimen was removed from the oven and
the nut was glued to the bolt to maintain its relative orientation.
Eight equispaced meridional slices were removed from the nut
bolt connector, with the first slice being aligned with the neutral
plane.
The analysis of the slices was performed in a polariscope
with white light, with the slice immersed in a tank containing
benzyl benzoate. The image of the slice was produced on a
ground glass screen at a magnification of four times. This was
then viewed through a traveling microscope to observe the
fringe pattern.
The highest fringe order at the root of the thread had to be
located. This was done using Mesnager's theorem (Hollister,
1967). Once the point was located the fringe order was measured using Tardy compensation. The measurements taken were
from three pitches below the loaded face of the nut to five or
six pitches above, for the principal planes, i.e, the neutral plane
and the bending plane. Readings for up to two pitches either
side of the loaded face were taken in the planes at 45 deg to
the principal planes.
The tensile specimen is necessary to calculate the nominal
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

4.5 a

4.0 Q C

(/)
CD
~ 3.0 -

mm

mm

3 2.5 -

CO

E 2.0 o
2
1.5 1.0 -

&

0.0 -J
-14.0

-7.0

0.0

7.0

14.0

21.0

Distance from loaded face of nut, m m

Fig. 4 Normalized thread root stress distribution for bolt with /?= 1.0

4.5 -

E3

-m

-7.0

m @

0.0

7.0

a
1 -

14.0

21.0

Distance from loaded face of nut, m m

Fig. 6 Normalized thread root stress distribution for boit with fl = 2.0

i1

T3

2.5 -

CO

m
m

E 2.0 k_

in

if"
13
B

"-

"

1.0

m
B

ft

0.5 -

-14.0

j.

0.0 -

3.0 -

1.5 - "

3.5 -

zo

-a

4.0 -

<U

-14.0

&

0.5 -

1
-7.0

0.0

7.0

p-

14.0

21.0

Distance from loaded face of nut, mm

Fig 5 Normalized thread root stress distribution for bolt with /?= 1.5

-14.0

-7.0

0.0

7.0

14.0

21.0

Distance from loaded face of nut, m m

Fig, 7 Normalized thread root stress distribution for axially loaded bolt

fringe order in the plain section of the bolt. This is so that the
values of measured fringe order in the specimen can be normalized. The tensile specimen was sliced at 45 deg to its face
and the maximum fringe order measured in the same polariscope using Tardy compensation.

for bolts loaded with only axial load (Fig. 7). Most importantly
though, the data for the eccentrically loaded bolts shows peaks
of maximum normalized stress which are broader than the
axially loaded bolts (compare with Fig. 7). The peak for an
axially loaded bolt is usually found within one pitch from the
loaded face of the nut, and decreases rapidly. In eccentrically
loaded bolts, this peak is broader and flatter. In general, the
Results
maximum is lower in magnitude, and to be found within the
The data were analyzed using the same method as described first two loaded pitches.
by Patterson and Kenny (1987). The data were plotted as shown
The periodicity observed in the stress concentration data by
in Figs. 3 to 6. The stress used for the normalization of the Fessler and his co-workers (1983, 1984) and Patterson and
stress measured at the thread root is the axial stress aa as Kenny (1987) are also present in this new data. The effect of
defined in the foregoing. An example of the results from a bending appears to be to accentuate this fluctuation. This
concentrically loaded bolt is shown in Fig. 7.
probably arises from two sources: firstly, the bending adds
Discussion
to the eccentricity of the load paths in the nut generated by
The most noticeable thing about the data as shown in Figs. the deflection of the incomplete threads at the loaded face;
3 to 6 is that the general shape of the data is similar to that secondly, only the nominal axial stress has been used to norJournal of Offshore Mechanics and Arctic Engineering

AUGUST 1994, Vol. 116/165

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

out to values which would occur in reality. The values of R

which one would expect to achieve are normally between 0 and
1. Obviously, a value of zero indicates a purely axial load.
The range 0 to 0.5 applies to preloaded bolts in which the
prestress is sufficiently high to protect the bolt from high
bending loads (Junker, 1974). The results indicate that it is
preferable to be in the region of i?< 1.0. In this range, the
value of the maximum normalized is lower than that for a
concentrically loaded bolt. This might lead to the expectation
of a higher fatigue limit when 0<Ra< 1; however, the absolute
stress levels induced by bending will usually be higher than in
axially loaded connectors and will negate the benefit of a lower
maximum normalized stress. In these experiments, the maximum stress in the unthreaded portion of the bolt was approximately the same in all cases.

0.0
-14.0

-7.0

0.0

7.0

14.0

21.0

Distance from loaded face of nut, m m

Fig. 8 Normalized thread root stress distribution for bolt with Ft, = 0.5
normalizing stress is that in the plane of the slice (axial + bending stress)

malize the results, leading to a cyclic variation corresponding

to the variation of bending around the bolt circumference.
Ideally, the normalization would use the stress in the plane
of the bolt at which the slice was taken. This approach was
investigated, and an example is given in Fig. 8. The results
were found to be difficult to interpret and to compare with
those from other connectors. In addition, the scatter was accentuated and no conclusion could be drawn from them. The
problem of the appropriate normalization has to be addressed
further and will be dealt with in future work.
There seems to be a slight trend between peak normalized
stress and the stress ratio Ra. At values of Ra = QA and 1.0,
the maximum normalized stress is about 15 percent lower than
that found in concentric connectors (Ra = 0). At higher values
of Ra, the maximum normalized stress is consistently, but only
slightly, less than that found in the concentric connectors.
To put the results obtained into perspective, it is necessary
to compare the values of Ra for which the tests were carried

1 6 6 / V o l . 116, AUGUST 1994

Conclusions
There is a noticeable effect when bending is applied to loaded
bolts. The results show that this effect varies with the ratio of
bending stress to axial stress, and thus warrants further investigation at Ra ratios similar to those usually found in real
engineering situations, i.e., between 0 and 1. Using the nominal
axial stress for normalization, the maximum normalized stress
is lower in the presence of bending and appears to rise with
increasing Ra.

References
BS 3643 Part 1, 1981, ISO Metric Screw Threads-Principles and Basic Data.
Fessler, H., and Jobson, P. K., 1983, "Stresses in a Bottoming Stud Assembly
with Chamfers at the Ends of Threads," Journal of Strain Analysis, 18(1), pp.
15-22.
Fessler, H., and Wang, J. H., 1984, "Stress Analysis of Some Unsymmetric
Screwed Connections," Journal of Strain Analysis, 19(2), pp. 111-119.
Hollister, G. S., 1967, Experimental Stress AnalysisPrinciples and Method,
Cambridge University Press, Cambridge, U. K.
Junker, G. H., 1974, "Principles of the Calculation of High Duty Bolted
Connection Interpretation of Guideline VDI 2230, VDI-Berichte, No. 220, p.
11.
Kenny, B., 1965, "The Casting of a Low Exotherm Epoxy Resin," Journal
of Scientific Instruments, Vol. 42, pp. 719-720.
Patterson, E. A., 1990, "A Comparative Study of Methods for Estimating
Bolt Fatigue Limits," International Journal of Fatigue Fracture Engineering
and Material Structures, 13(1), pp. 59-81.
Patterson, E. A., and Kenny, B., 1987, "Stress Analysis of Some Nut-Bolt
Connections with Modifications to the External Shape of the Nut," Journal of
Strain Analysis, 22(4), pp. 187-193.

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