You are on page 1of 8

IADC/SPE 128931

Case Study: API Box Boreback Stress Relief With Truncated Threads Can
Cause Premature Connection Fatigue Failure
Michael B. Gerdes, P.E., Dr. Kang Lee, SPE, T H Hill Associates, Inc.
Copyright 2010, IADC/SPE Drilling Conference and Exhibition
This paper was prepared for presentation at the 2010 IADC/SPE Drilling Conference and Exhibition held in New Orleans, Louisiana, USA, 24 February 2010.
This paper was selected for presentation by an IADC/SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not
been reviewed by the International Association of Drilling Contractors or the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily
reflect any position of the International Association of Drilling Contractors or the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any
part of this paper without the written consent of the International Association of Drilling Contractors or the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is
restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of IADC/SPE copyright.

Abstract
A rotary shouldered connection (RSC) can be modified
with two stress relief features to improve its fatigue
performance: 1) A stress relief groove on the pin and/or 2)
A boreback feature on the box, commonly called a box
boreback. Both are API approved stress relief features,
whose only purpose is to mitigate stress concentration
caused by unengaged threads in the connection. However,
under certain operating conditions the box boreback
feature as traditionally machined can actually reduce the
fatigue life of the connection. Prompted by recognition of
unique failure locations in the box of numerous BHA
connection types, the authors performed finite element
analysis to better understand the fatigue mechanisms
driving the failures.
In a typical API box boreback, the crests of the last four or
five threads are cut short, resulting in truncated threads.
Regarding the box boreback feature, API Specification 7-2
(API, 2008) controls for three primary dimensional
variables by connection type & size: 1) Cylinder diameter
2) Depth to last thread scratch 3) Depth to end of cylinder.
A machined boreback will be technically acceptable
adhering to these dimensional requirements, which allow
for truncated threads, but this method is not optimal. The
approach does offer two commercial advantages for the
tool owner as it is both simpler to execute and leaves
greater material behind for future recuts. API 7-2 also
allows for a box stress relief groove, but this feature is
rarely used commercially for similar reasons. To quantify
the impact of the box boreback on fatigue life, the authors
modeled a failed NC46 drill collar connection with the
feature and compared with hypothetical configurations of
the same connection.
Introduction
Rotary shouldered connections typically fail from fatigue
in the last engaged thread root of either the box or pin,
determined by the member weaker in fatigue and the
operating conditions impacting the connection. To improve

the fatigue performance of connections, stress relief


features (SRF) are cut into the threads. The stress relief
features in the box member of the connection are designed
to reduce the stress concentration created by the unengaged
threads in the back of the box. Unengaged threads increase
the concentration of stress at the last engaged thread and
reduce the fatigue performance of the connection.
As mentioned, when machined, a box boreback feature
usually leaves truncated threads at the back of the box.
Figure 1 compares a box boreback machined with
truncated threads to one machined with full-height threads.
As seen in the detail of the illustration, the reduced height
of the rear box threads limits engagement with the pin
threads especially as the pin tapers toward the back of the
box. Operation exposes these truncated threads to greater
wear and deformation than fully engaged ones, magnifying
the engagement deficiency. The reduced thread
engagement increases the bearing stress between load
flanks and results in higher stress concentration.
Eventually, threads with insufficient height become worn
to a point where there is no contact between the pin and
box threads. As this occurs, the last engaged thread
becomes the last full-depth box thread, a location further
toward the front of the box.
At this point, the total stress in the connection is distributed
over a smaller area at a location in the box with less crosssectional area. The combination of these factors reduces
the fatigue life of the box, and consequently results in
premature connection failures. Failure analyses for NC38,
NC46, NC50, 6-5/8" FH, 6-5/8" Reg., and 8 5/8" Reg. size
connections have confirmed that the unique location of
failure is not specific to one connection type or size.
Additionally, the failures manifest in connections with the
box boreback stress relief feature, machined with truncated
threads. A study of one failure, in an NC46 drill collar
connection, was chosen as a case example to illustrate and
define the forces behind these unexpected failure results.

IADC/SPE 128931

BB with Truncated
Threads

BB with Full-Height
Threads

Figure 1 - The truncated threads version of the API box boreback is generally preferred by machinists because of the
ease of manufacture. The cutter is set at the boreback diameter, then pulled straight out of the box connection.

An NC46 Drill Collar Connection Failure


A 6 " OD x 2 1/4" ID drill collar displaying a classic
fatigue crack was examined to determine the cause of the
unique crack location. The crack was located at the root of
the fourth from the last thread of the box in the NC46
connection (Figure 2), and the box had been modified with
a truncated thread boreback feature. The threads from the
location of the crack to the back of the box all showed
visible wear on the thread crests.

The failed box was sectioned and dimensionally analyzed.


A portion of the connection that was not cracked was used
to evaluate the condition of the threads (Figure 3, 4). An
optical comparator was used to precisely measure the radii
of the box threads (Figure 5). The dimensions were used
to create two-dimensional models of the connection.

Crack Location

4"

Figure 3 Sectioned portion showing crack location.

Length to the Back of Box

Length to Crack
2 11/16"
Worn Threads

Figure 2 Failed NC 46 drill collar box

Figure 4 Sectioned portion of failed NC 46 connection


showing the thread wear on the threads at the back of the
connection.

IADC/SPE 128931

Figure 7 Standard machined threads with no stress relief


features on pin or box.
Figure 5 An optical comparator image used to measure the
worn threads.

The models showed that the last four box threads at the
back of the box did not engage the pin threads, and further,
that the wear on the threads was sufficient to cause a loss
of contact with the mating pin threads. CFI analysis was
used to evaluate the impact of this loss of contact on the
fatigue life of the failed connection. CFI is based on an
application of the Morrow strain-life model (Dowling,
1993), which considers both the crack initiation and
propagation phases in determining fatigue life. Connection
features and loading conditions are modeled using a finite
element analysis program, which gives a prediction of
fatigue life in the form of cycles to failure. Details of the
process used to calculate CFI can be found in Appendix A
and IADC/SPE 112105 (Ellis, 2008). Figure 6 displays the
FEA results for the example failure, and Figures 7-9
represent alternative hypothetical connection configuration
options holding inputs constant, except as noted.

Figure 6 Model with worn threads representative of case


example, stress relief groove on pin.

Figure 8 API Stress relief groove on pin and box.

Figure 9 Box boreback without truncated threads, pin stress


relief groove.

IADC/SPE 128931

Table 1 displays the CFI values of the example failure as


modeled from Figure 6 at varying dog leg severities
(highlighted row). The CFI values for the hypothetical
connection configurations are shown for comparative
purposes. These include an NC46 connection with no
stress relief features (Figure 7), a NC46 connection with a
stress relief groove on the pin and box (Figure 8), and a
connection with a full-thread height boreback box and pin
stress relief groove (Figure 9). In the table below, 17(2.0)
represents 2.0 x 1017 cycles.
CFI @ DLS(degree/100feet)
Connection

SRF

NC46

BB
th
4 thread
worn

17(2.0)

12(2.8)

5(4.8)

NC46

No SRF

22(7.2)

18(1.0)

10(3.0)

NC46

API SRG

25(1.0)

20(1.5)

12(3.5)

contact (Figure 11), and a BB thread in which four threads


have lost contact (Figure 6). As box threads lose contact
with pin threads, fatigue life shortens dramatically. In the
table below, 23(1.4) represents 1.4 x 1023 cycles.
CFI @ DLS(degree/100feet)
Connection

SRF

BB
NC46
no worn
23(1.4)
18(2.3)
11(1.1)
thread
BB
st
NC46
1 thread
21(1.4)
16(2.1)
8(7.9)
worn
BB
th
NC46
4 thread
17(2.0)
12(2.8)
5(4.8)
worn
Table 2 CFI fatigue performance values for different NC46
BB thread contact geometries.

BB w/ full
thread
24(4.7)
19(6.9)
12(1.7)
height
Table 1 CFI fatigue performance values for different NC46
stress relief feature geometries.
NC46

The modeled fatigue life of the failed connection is


significantly lower, at all dog-leg severities, than any of the
alternative geometries, including the hypothetical
connection with no stress relief features. To put the results
in perspective, at a DLS of 1/100 feet (the blue column),
the connection with no stress relief features has relative
fatigue life 360,000 times that of the case example.
Modifying the box boreback to eliminate truncated threads
(Figure 9), yields a connection with more than 23 million
times the relative fatigue life of the failed connection
model. The stress concentration created by the unengaged
threads, the reduced cross section of connection material,
and the fewer number of threads in contact all contribute to
shortened fatigue life. After the threads have worn to a loss
of contact, the unengaged threads become a stress
concentration much like the unengaged threads on a
connection without a SRF. The movement of the last
engaged thread also moves the high stress point in the
connection to a position down the thread taper where there
is less cross-sectional material in the box. Additionally, the
loss of engaged threads requires the connection to
distribute the loads placed on the connection across fewer
threads, contributing to larger point stresses.
A further analysis was conducted to evaluate the life of a
connection as the location of the last engaged thread
changed with usage. Table 2 shows the results of
examining the fatigue performance of a box boreback
thread with full thread contact up to the last cut box thread
(Figure 10), a BB thread in which the first thread has lost

Figure 10 BB thread modified such that there is thread


engagement up to the last machined thread.

Figure 11 BB thread modified such that the first box thread


is not engaged: (note: scale adjusted from other figures to
show the location of the last engaged thread stress).

In comparing Tables 1 and 2, notice that even with one


worn thread losing contact with the pin, a connection with
a traditional box boreback feature has a significantly lower
fatigue life than a connection with no stress relief features
at all. Further, fatigue performance declines rapidly as
more thread contact is lost.

IADC/SPE 128931

Box Boreback Thread Wear


The fatigue performance of a BB connection as predicted
by CFI modeling is dependent on the location of the last
engaged thread in the back of the box. Examination of
failed field samples has shown thread wear that removes
box and pin thread engagement up to four threads from the
original last engaged thread. This type of thread wear is
not typical of other types of common connection thread
modifications. The rate of thread wear will be dependent
on the contact stress acting between two threads. An
analysis was performed to compare the thread contact
stresses along the connection.
The values show that the contact stresses between the BB
type connection threads are much larger than other thread
stress relief modifications. At the same bending moment,
relatively equal forces will be distributed over a connection
regardless of the type of thread stress relief feature. The
truncated thread crests of the BB reduce the contact area
between the pin and box threads and thereby increase the
contact stress between threads caused by bending and
make up forces.
The contact stresses between the last engaged pin and box
threads were evaluated for different thread configurations
on the same NC46 connections. Figure 12 shows the stress
distribution for a connection with a BB feature at the last
engaged thread. The average principal normal stresses at
the load flanks between the box and pin last engaged
threads were used for this comparative analysis. The
comparison of thread stresses at the same MUT is shown in
Table 3.

Stress at last
engaged
thread

Figure 12 Contact stresses at the last engaged thread of a


NC 46 connection with a BB feature.

Connection

SRF

Ave. Last Engaged Thread


Contact Stress (ksi)
@ API MUT

NC46

BB
no worn thread

166.29

NC46

No SRF

19.11

NC46

API SRG

29.07

BB
NC46
w/ full thread
29.719
height
Table 3 Last engaged thread contact stress for different
NC46 thread geometries.

An evaluation of thread contact stress after contact was lost


with the first thread of a BB cut was also conducted. Table
4 shows the amount of contact stress at the last engaged
thread after all the worn threads have lost contact.

Connection

SRF

Ave. Last Engaged Thread


Contact Stress (ksi)
@ API MUT

NC46

BB
no worn thread

166.29

NC46

BB
st
1 thread worn

133.36

NC46

BB
th
4 thread worn

69.23

Table 4 Last engaged thread contact stress for different


NC46 BB thread geometries.

Thread Wear Dependence on Operational


Conditions
The change in fatigue performance due to the change in
last engaged thread location suggests a dependence of
fatigue life with thread wear and the operational factors
that increase thread wear. The accumulation of thread
wear will be dependent on the level of stress at the threads
and corrosion from wellbore fluids. The stress level in the
threads will be affected by the MUT and bending moment
placed on the connection as well as any downhole
vibrations. The MUT should be dictated by standard API
values, so the bending moment will be the primary
operational factor other than vibrations, that increases the
rate of thread wear and fatigue damage.
Corrosion can lead to further degradation of the box
threads which will reduce the thread contact. The reduced
height of the BB threads also allows greater exposure to
well bore fluids. Thread dope can be washed away with
less difficulty and there is a larger path to deeper threads
compared to threads with full thread height. Corrosion
impacts the fatigue life of any connection, but any
accelerated corrosion due to the BB geometry was not
quantified as a part of this study.

IADC/SPE 128931

The exact relationship between all operational parameters


and the resulting effects on fatigue life and thread wear are
elaborate and difficult to quantify. A study of failed
threads has shown that the box connection can fail at any
thread along the truncated threads in the back of the box.
An increase in dog leg severity and subsequent bending
moment has not always shown an increase in the number
of worn threads. The increase in bending moment will
greatly increase the fatigue damage to the connection
which can cause the connection to develop a crack and fail
before a large amount of thread wear has occurred. In
other cases with minimal service conditions BB
connections have been shown to generate significant thread
wear. Further research is necessary to determine the
operational parameters and conditions that can lead to
increased wear and reduced fatigue life caused by BB
connections thread wear.
2-D Model Limitations
Another concern with the value of the predicted fatigue
performance of the box boreback is the ability to model the
entire feature. For simplification of FEA studies, the
threaded connection models are made in 2-D. The 2-D
models assume 360 symmetry of the model. This is not
an issue for typical threaded connections. However, the
BB thread will not have full contact circumferentially
around the connection. The thread contact in the back of
the box will change along the thread taper. Even with
thread contact at the cut threads, there will always be some
portion of exposed box thread root that does not contact the
pin threads. The models referenced in this paper were
created to match the dimensions at the location where a
cracked sample piece was cut. Analyzing the thread at a
rotationally different location may result in a different
fatigue performance value, but likely similar qualitative
results.
Summary

A typical API boreback (BB) stress relief feature is


machined with the crests of the last four or five box
threads cut short. This results in truncated threads at
the back of the box.
Operations expose the truncated threads to greater
wear and deformation and can lead to loss of contact
between the pin and box threads with subsequent
reduction in fatigue performance of the connection.
The wear rate and subsequent effect on fatigue
performance of the BB threads is dependent on
operational conditions.
FEA and comparative fatigue life studies of the BB
feature with increasing wear in the truncated threads
indicates that the fatigue life shortens dramatically as
the thread wear increases and as the box threads lose
contact with the pin threads. A typical machined API
BB feature is thus only effective as long as there is no
wear on the truncated threads.
Fatigue life comparison studies of the BB feature with

other connection configurations like API box stress


relief groove and boreback with full thread height
indicates that with wear on the truncated threads, the
other configurations provide vastly improved fatigue
performance.

IADC/SPE 128931

Appendix
For the analyses to be discussed herein (taking plastic
strain, stress, and fatigue into account), the Morrow strainlife model2 is used. The Morrow strain-life model was
chosen as it accounts for the fatigue crack initiation phase,
not just the crack propagation phase. This is important as
the connection geometry (thread design or stress relief
groove width) heavily influences fatigue performance.
Using a fatigue model such as the Forman model, which
assumes an initial crack size, would not capture the effects
of connection geometry. Equation 1 is the Morrow strainlife equation used to calculate fatigue life.

a =

f m
E

(2 N ) + ( )(2 N )
b

into discrete pieces or elements. The mesh is the lattice


that is the sum total of all of the elements. In order to
accurately describe the geometry and shape of a model, the
elements must be small enough to closely approximate the
contours present in the specific area(s) of interest. In the
case of a rotary shouldered connection, the threads and the
stress relief features are the prime areas of interest.
Therefore, it is important to have a dense mesh in these
regions. Regions of the connection away from these
critical areas may have a coarser mesh as the stresses and
strains here are generally not of much interest.

mp

...

(1)
Where:

f
m

= Strain amplitude
= Material fatigue constant determined
experimentally
= Mean stress (psi), when

m > mys

m = mys

E
Nf
b

f
mp

= Youngs Modulus (psi)


= Number of cycles to failure
= Material fatigue constant determined
experimentally
= Material fatigue constant determined
experimentally
= Mean plastic strain, when

mp

, set

m mys

, set

=0
= Material fatigue constant determined
experimentally

A quick study of this equation shows that if the mean stress


in the critical area is above the yield strength of the
material (100 KSI in the case of the drill collars used in
these studies), the variables that drive the fatigue life
calculation are strain amplitude and mean plastic strain.
Therefore, the model with the highest combination of
strain amplitude and mean plastic strain will have the
lowest fatigue life. If the mean stress in the critical area is
less than the yield strength of the material, the strain
amplitude and mean stress are the variables that drive the
fatigue life calculation. In these cases, the model with the
highest combination of strain amplitude and mean stress
will have the lowest fatigue life.
Mesh
An example of the type and size of mesh used is shown in
Figure A1. Each model that is evaluated is broken up

Figure A1 6-5/8 REG axis symmetric connection model.


Smaller elements are necessary in the regions of interest,
while coarser mesh is allowable in non-critical areas.

Application of MUT
Applying torque to a computer generated model of a rotary
shouldered connection is much more difficult than loading
an actual connection into an iron roughneck and pressing
the GO button.
Successfully modeling a threedimensional connection (getting a convergent solution) can
be a daunting task because of the amount of sliding contact
between the thread flanks, the helical nature of the thread
path, and the boundary conditions. As a result, rotary
shouldered connections are most easily modeled in two
dimensions. Two dimensional axis-symmetric models
with asymmetric (bending) loads must be created. MUT is
approximated by either assuming a certain amount of
thread and shoulder overlap, or by applying a thermal
gradient (to expand the material thereby creating thread
flank and shoulder interference) to the pin and/or box. The
bulk stresses induced from applying MUT to an actual
connection are known or calculable in three key areas

(Figure A2). Using these known or calculable values,


correctly approximating the MUT on the two dimensional
connection model can be achieved by adjusting the overlap
or thermal layer temperature, and comparing the resultant
stresses to those defined to be present in an actual
connection of the same type and dimensions.

IADC/SPE 128931

References
API Specification 7-2, Specification for Threading and
Gauging of Rotary Shouldered Thread Connections, first
edition, American Petroleum Institute (June 2008).
Dowling, Norman E., Mechanical Behavior of Materials:
Engineering Methods for Deformation, Fracture, and
Fatigue. Prentice Hall, 1993, ch. 9 & 14.
Ellis, S., Wadsworth, T., Lee, K., Gerdes, M., Altizer, S.:
Connection Fatigue Index (CFI): An Engineered Solution
for Connection Selection and a Replacement for BSR,
SPE/IADC 112105 (2008)

Figure A2 6-5/8 REG 8 x 2-13/16 rotary shouldered


connection critical stress areas: Seal area, 3/4 from shoulder
in pin, 3/8 from shoulder in box.

Bending Load
Axial loads and pressure are generally simple to apply.
Applying a bending load to a two dimensional axissymmetric asymmetric model generally requires the use of
a special element type. When modeling bending in a
rotary shouldered connection, it is important to bend the
connection multiple times in order to generate the best
approximation of the stress state at the thread roots.
During connection make up, the last few engaged pin
threads, any remaining unengaged pin threads, and some
areas of a stress relief groove (if present) are usually
plastically deformed. Since the stresses are plastic,
bending the connection once will not give accurate stress
and strain values. The connection must be bent multiple
times to allow for the effects of stress relieving and stress
redistribution. In an actual connection made up and run
down hole, this will occur once the connection is rotated
more than one revolution while bent or buckled.
CFI Calculation
CFI is expressed as a numeric representation of the relative
fatigue life. The relative fatigue life is determined by the
inputting the stress and strain values determined from FEA
into the Morrow Strain Life equation. These resulting
values are often expressed as a logarithmic value. CFI is
represented by a simplified expression of the logarithmic
value. The exponent of the base 10 multiplier is displayed
as the first number and the coefficient is displayed in
parenthesis. The exponent value affects the fatigue life far
more than the coefficient. By displaying the relative
fatigue life in this manner a quick and easy comparison can
be made.
25600000 = 2.56 x 107 = 7(2.56)

You might also like