Sunteți pe pagina 1din 44

no-e-tic

Noetic Engineering Inc. Phone: 780.414.6241


Box 49036 - RPO Strathcona Fax: 780.469.1250
Edmonton AB T6E 6H4
engineering inc. Canada

Project Memorandum
Date: April 28, 2003 Fax Number

To: Per Angman, TESCO Corporation +1 (403) 252-3362

Cc:

From: Maurice Slack

File: 33.101.071 Pages: 44

Re: CDS Connections: Marketing Support Draft Report


Torque Ring Design for Enhanced API Connection Performance in Casing Drilling
Operations
Document Rev2
Urgent For review Original to follow by mail Please reply

This report was written to an internal Tesco audience as both a means to teach the fundamentals of
torque transfer in threaded connections and to transfer the essential outcomes of the tasks conducted
to develop the muli-lobe torque ring design. Various side issues have been deliberately excluded to
both avoid confusing the main thrust of understanding how to use the connections while drilling and
to protect sensitive aspects of the design basis from prying eyes. Similarly, details of the ring
geometry specification have not been made explicit since these are still evolving and do not need to
be largely decimated.
We trust that this report largely meets your requirements for a white paper but look forward to your
input prior to finalizing.
Best regards,

Maurice Slack
Note: This Torque values in Table 7 are revised in this revision

no-e-tic

WhitePaper.Rev2.2003Apr28.dda.doc
Notice: This document is directed to the person(s) named above, but occasionally computers make mistakes and phones get
misdialed. We ask you to respect the fact that this fax may contain privileged and confidential information that is exempt from
disclosure. If you are not the person(s) named above, we will be very upset and even distraught if you send copies of this
document to your friends and relatives. If weve made a mistake and you receive this transmission in error, please call us collect
to let us know, then kindly return the original fax by mail. Thanks for giving us the same consideration wed extend you!
Torque Ring Design for Enhanced API
Connection Performance in Casing
Drilling Operations
Prepared by Noetic Engineering Inc.
for
Tesco
28-April-2003
Rev 2: 2003-Apr-28

no-e-tic
Torque Ring Design for Enhanced API
Connection Performance in Casing
Drilling Operations

Prepared by Noetic Engineering Inc.


for
TESCO Corporation
28-April-2003

Prepared by:

no-e-tic

Trent Kaiser, PhD, PEng Maurice Slack, PEng

Under APEGGA Permit Number: P 6807

ii
Executive Summary
This report draws together the salient outcomes of a number of design and experimental tasks
conducted to develop a torque ring design enhancing the torque capacity of standard API connections
for use in casing drilling applications. While most premium connection designs incorporating torque
shoulders provide suitable torque capacity, these designs also carry a premium price and often are
special-order items manufactured on demand, rather than inventoried for immediate delivery.
Similarly, existing methods providing rings to convert or upgrade API connections to premium or
semi-premium seal performance still require modified or otherwise controlled coupling geometries.
These methods, thus only partially achieve the design goal of enhancing the torque capacity of
standard or off the rack API connections. Many of the economic and logistical advantages being
sought are therefore not obtained using such available approaches.
It is therefore economically and logistically attractive to find a means to press fully standard API
buttress and 8-round connections into service for casing drilling. Large inventories of standard API
connections are available world-wide, and the cost of deploying these designs is typically much lower
than premium solutions. The challenge to using these non-shouldered designs in casing drilling is the
limited torque capacity: inadequate for 8-round or barely adequate for buttress.
Although innocently simple on the surface, implementation of torque rings to optimally increase the
torque capacity of stock or off the shelf API buttress and round threaded connections requires
consideration of a large number of variables. This report explains how these variables influence
installation, operational performance and pin to box torque transfer in a complex interactive fashion,
and particularly how the relatively loose tolerancing allowed by API specifications is managed.
The design basis for a solution which implements a novel multi-lobe torque ring design with stock API
connections is presented. This ring architecture enables installation into already bucked collars at the
rig site or elsewhere. The mechanics of how this approach enhances torque is developed, the
tolerance issues related to the implementation are addressed, and a method for dealing with handling
issues is described. The result is an effective solution for reliably providing torque enhancement for
standard API connections that is cost-effective and tolerant of dimensional variation and field
practices.
This elegantly simple approach to torque ring design and implementation was proven in full scale shop
trials to provide increases in torque ring capacity for both BTC and LTC thread forms of practical use
in casing drilling. These tests also provided data to calibrate a simple analytic model allowing straight
forward prediction of torque capacity to many other sizes.

no-e-tic

iii
Table of Contents
Executive Summary .....................................................................................................................iii
Table of Contents....................................................................................................................... iv
Table of Figures .......................................................................................................................... v
1 Introduction ......................................................................................................................... 1
2 Mechanics of Torque Transfer in Tapered Threaded Connections .............................................. 2
2.1 Interfacial Torque Transfer ............................................................................................. 2
2.2 Thread Helix Torque Transfer Considerations.................................................................... 2
2.3 Bearing Loads ................................................................................................................ 5
2.3.1 Bearing Load Basics ................................................................................................. 6
2.3.2 Bearing Load Capacity Calculations ........................................................................... 6
2.3.3 Thread details and bearing loads .............................................................................. 8
2.3.4 Sensitivity to Axial Stress........................................................................................ 10
2.4 Additional factors in Torque Transfer ............................................................................. 11
2.4.1 Friction Factor ....................................................................................................... 12
2.4.2 Remobilized friction ............................................................................................... 12
3 API Thread Tolerances ........................................................................................................ 13
4 Torque Ring Design ............................................................................................................ 14
4.1 Design Requirements.................................................................................................... 15
4.1.1 Functional Requirements ........................................................................................ 15
4.1.2 Installation Requirements....................................................................................... 15
4.2 Design Concept............................................................................................................ 15
4.3 Lobe Design Basis ........................................................................................................ 16
4.3.1 Multi-lobe forming and installation........................................................................... 16
4.3.2 Multi-lobe Design................................................................................................... 17
4.3.3 Multi-lobe deformation characterization ................................................................... 18
4.4 Ring Installation Analysis .............................................................................................. 18
4.5 Installation and Removal Tests...................................................................................... 19
5 Torque Ring Performance Enhancement ............................................................................... 21
5.1 Analytical Performance Assessment ............................................................................... 22
5.1.1 Buttress Thread..................................................................................................... 22
5.1.2 Round Thread ....................................................................................................... 25
5.2 Full Scale Tests ............................................................................................................ 27
5.2.1 Equipment Configuration ........................................................................................ 27
5.2.2 Specimen Preparation & Procedures ........................................................................ 27
5.2.3 Data Reduction and Results .................................................................................... 28
5.2.4 Conclusions from Full Scale Tests............................................................................ 35
6 Summary ........................................................................................................................... 35

no-e-tic

iv
Table of Figures
Figure 1 Simplified free body diagram of torque transfer for incipient slip on axisymmetric pin element.
........................................................................................................................................... 2
Figure 2 Bearing load distribution examples: Axial bearing and stress (left), radial bearing induced
hoop stress (right) ................................................................................................................ 6
Figure 3. Biaxial yield envelop for tubular components.................................................................... 8
Figure 4. Load transfer on 8-Round Thread ................................................................................... 9
Figure 5. Load transfer on Buttress thread................................................................................... 10
Figure 6 General nature of bearing stress effect on friction coefficient............................................ 12
Figure 7. Illustration of taper variation combinations .................................................................... 14
Figure 8. Ring and plug gauge measurement error due to taper variation....................................... 14
Figure 9 Multi-lobe torque ring deformations ............................................................................... 16
Figure 10. Forming load results for 114.3mm torque ring. 3.175mm wall. 550MPa yield strength ..... 18
Figure 11. Installation response of torque ring ............................................................................. 19
Figure 12. Sketch of fixture used to install and remove rings in a coupling. .................................... 20
Figure 13 Summary of prototype ring installation and removal load maximums............................... 20
Figure 14 Photograph of four lobe fine pitch ring contact patch after removal showing yielding of
ratchet teeth. ..................................................................................................................... 21
Figure 15 Fast pin slow box nominal diameter interference 20 ppf buttress connection made up to
target torque...................................................................................................................... 24
Figure 16 Base case for 7 23 ppf L80 LTC connection.................................................................. 26
Figure 17 Ring installation behaviour predicted by finite element analysis (FEA).............................. 29
Figure 18 Typical torque turns graph (7 inch BTC) showing shoulder and yield plateau.................... 30
Figure 19 Summary of all LTC smooth ring results compared to model predictions. ......................... 33
Figure 20 Summary of all BTC smooth ring results compared to model predictions and expected
torque limits....................................................................................................................... 34

no-e-tic

v
1 Introduction
Torque requirements for connections used in the casing drilling system (CDS) are generally higher
than those required to simply run and case conventionally drilled wells. In addition to providing seal
and axial load capacity, CDS connections must also deliver torsional load to the bottom of the hole
while undergoing cyclic bending as the string is rotated through intervals with curvature. While most
premium connection designs incorporating torque shoulders provide suitable torque capacity, these
designs also carry a premium price and often are special-order items manufactured on demand, rather
than inventoried for immediate delivery.
It is therefore economically and logistically attractive to find means to press standard API buttress and
8-round connections into service for casing drilling. Large inventories of standard API connections are
available world-wide, and the cost of deploying these designs is typically much lower than premium
solutions. The challenge to using these non-shouldered designs in casing drilling is the limited torque
capacity: inadequate for 8-round or barely adequate for buttress. Furthermore, the torque
specifications for these designs are developed to reflect make-up operations, where curvature (and
cyclic curvature) are not part of the load environment. There is concern that the API designs could
suffer incremental make-up in a ratcheting fashion as a bending load rotates around the pipe axis in a
deviating interval.
One method available for improving the torque capacity of API connections is to provide a ring in the
J-section (between the pin noses) of a modified collar to develop additional torque. The couplings
using this strategy, such as Hunting Interlocks Convertible Buttress Connections, are not
manufactured to standard API tolerances, but are specially designed to facilitate the use of such rings,
particularly in terms of tolerancing. In addition, the pins are typically screened to further limit the
allowable tolerances. These available methods, thus only partially achieve the design goal of
enhancing the torque capacity of standard or off the rack API connections. Many of the economic and
logistical advantages being sought are therefore not obtained using such available approaches.
This report presents the design basis for a solution which implements a novel multi-lobe torque ring
design in stock API connections, enabling installation into already bucked collars at the rig site or
elsewhere. The mechanics of how this approach enhances torque is developed, the tolerance issues
related to the implementation are addressed, and a method for dealing with handling issues is
described. The result is an effective solution for reliably providing a torque enhancement for standard
API connections that is cost-effective and tolerant of dimensional variation and field practices.

no-e-tic

WhitePaper.Rev2.2003Apr28.dda.doc 1
28-April-2003

2 Mechanics of Torque Transfer in Tapered Threaded Connections


The ability to transfer torque between the pin and box of threaded connections is largely dependent
on the frictional stresses acting on the contacting surfaces at a distance from the centre of rotation.
The frictional stress in turn is the product of the friction coefficient and the bearing load acting normal
to the contact surfaces in the threads or on the shoulder as shown in Figure 1.
N
Contact force vector

r N

Surface of pin

Direction of
torque

Torque = Nr

Figure 1 Simplified free body diagram of torque transfer for incipient slip on axisymmetric
pin element.
While this simple relationship is generally true, the detailed mechanics involve a number of subtleties,
which deserve clarification. In addition, the friction coefficient operative between the mating surfaces
is not a simple constant for the thread compounds used with threaded connections. This section
therefore provides a detailed review of the mechanics, showing how the thread form and helix
geometries contribute to the contact stress, and hence torque, and discusses the frictional behaviour
of thread compound materials.
2.1 Interfacial Torque Transfer
The contact stress includes two components: the normal stress and the frictional stress as shown in
Figure 1. The frictional stress is related to normal stress by the friction coefficient, which is usually
assumed to be constant in the range of 0.08 to 0.15 for typical unconfined thread compounds, but in
fact includes some nonlinear characteristics. For surfaces axisymmetric to the pipe axis, e.g. shoulder
and seal surfaces, only the frictional component contributes to the torque transfer. On the other hand,
helical surfaces, i.e. in the threads, are skewed relative to the pipe axis. Consequently, normal
stresses on helical flank surfaces contribute a small component to the torque (not shown in Figure 1)
and the frictional stress contribution is diminished slightly, as demonstrated in the following section.
2.2 Thread Helix Torque Transfer Considerations
The magnitude of the relative contributions is controlled by the helix angle, which is small for virtually
all OCTG threaded connections. The relative contributions of normal thread contact stress relative to
tangential stress can be considered in terms of the helix angle. The helix angle is calculated from the
no-e-tic

following:
Pitch
tan ( ) =
Diameter Equation 1
For buttress and 8-round thread forms, the pitch is constant over the entire range of pipe sizes.
Therefore, smaller diameter products have higher helix pitches. Consequently, of the casing drilling
applications under consideration, the largest helix influence would be at the smallest casing size of

2 of 38
28-April-2003

114mm (4 ) with buttress threads. At this size the helix angle for a 5.08mm buttress pitch (0.2
pitch for 5 tpi) is only 0.81, and for larger sizes and 8-round threads the angle is even smaller.

Thread
angle ()

2.2.1 Effect of axial load transfer on torque


Comparing the relative torque contributions of normal and frictional stresses requires the tangential
components be calculated. The direct contribution to torque by normal stress on radial surfaces is
defined by the sine of the pitch angle. The contribution to torque by tangential stress is defined by the
cosine of the pitch angle. However, because it is related by the friction coefficient to the normal
stress, the frictional contribution can also be expressed in terms of the normal stress. The tangential
and axial contact stresses on a helical radial surface can therefore be given in terms of the normal
stress by:
t = n sin + f cos
z = f sin n cos
f = n
t = n (sin + cos )
z = n ( sin cos ) Equation 2
where,

t is the tangential stress,


z is the axial stress,
n is the normal stress due to axial load,
f is the friction stress, and
no-e-tic

is the helix angle.


Solving these two expressions to eliminate the normal force gives:

3 of 38
28-April-2003

sin + cos
t = z
cos sin
Assuming sin << cos ,
t = z (tan + ) Equation 3

This is the basis for determining torque in a shouldered connection, as given in API Specification
RP7G. Assuming the axial stress comes from a shoulder, and expressing the helix angle in terms of
thread pitch and radius, torque is given by:

p R
T = Fa + t t + s Rs
2 cos Equation 4
Where,
T is the torque,
Fa is the axial force on the shoulder,
p is the pitch,
Rt, Rs are the radii at the thread and shoulder respectively,
t, s are the friction coefficients on the thread and shoulder interfaces respectively, and,
is the thread flank angle.
In this expression the friction coefficients are assumed positive for make up torque and negative for
break out. Applying typical values for the 114mm (4 ) with buttress thread form considered earlier
( = 0.1), shows that if the friction coefficient magnitude remains unchanged between make up and
break out, break out shoulder torque will be approximately 86% of makeup.
2.2.2 Effect of radial load transfer on torque
The helix effect on torque arising from radial interference in 8-round or V-threads may be most simply
described in terms of virtual work. For an increment in rotation, the path length followed by a
contacting element on the helix is longer than that of the applied torque that follows a circular, not a
helical path. Since the work done must be the same, the torque that would otherwise arise from
axisymmetric rotation due to radial stress will be boosted by helix angle or pitch according to the
following relation:
2
T p
= 1+
T0 D Equation 5

T0 is the torque from radial stress with zero pitch.


no-e-tic

Substituting values for 114mm (4 ) with 8-round, gives:


2
T 0.125
= 1+ = 1.00004
T0 4.5 Equation 6
Clearly the effect of pitch may be neglected when considering the radial stress contribution to torque
in 8-round threads. Although the relationship for Buttress threads is somewhat more complex because
radial load is borne by crest to root contact separate from axial load reacted by the flanks (see

4 of 38
28-April-2003

discussion on bearing load in next section), it can be shown that the effect of pitch on torque induced
by radial load is also negligible1.0015T0 for the 114mm (4 ) example case.
A similar argument can be made for the relative contribution of taper to torque. One interesting
observation is that the normal contact stress component on cylindrical roots and crests of larger
buttress sizes contributes a mechanical torque, even though there is no obvious tangential component
of the surface in a section view of the connection geometry. This effect may be visualized by
imagining the zero friction case. In this case a made up taper threaded connection would tend to
back spin on its own when the make up torque is removed. Rather than developing this contribution
geometrically, a variational approach provides a more elegant solution.
First, we assume the radial stress is proportional to the interference. Second, we note that the
interference is proportional to the relative rotation between the pin and box. The variational torsional
work done to rotate the pin an infinitesimal amount relative to the box must be equal to the sum of
the frictional work done to slide the surfaces and the radial work done to increase the interference.
Mathematically, this is expressed:
td = r (dI + Rd )
Taper Pitch
dI = d
2 2
Taper Pitch
td = r d + R
4 Equation 7
where,

r is the radial stress,


is an angle of rotation,
I is the radial interference, and
R is the radius of the contact surface.
Comparing the relative contributions of the normal and frictional components of the bearing stress for
a buttress thread geometry reveals that the normal component is just 0.4% as large as the frictional
component for the 114mm (4 ) casing, and it becomes proportionally less for larger sizes.
These relationships demonstrate that the helix effect on torque transfer due to radial interference can
be neglected with only minor consequence on accuracy. In fact, uncertainty in other controlling
variables will have a much greater impact on the solution. Furthermore, ignoring the helix effect and
assuming the tangential stress is equal to the friction stress generates a conservative solution by
under-predicting the torque transfer in the threads by less than 1%. The remaining analytical
description of torque transfer will therefore use this assumption to simplify the relationships
developed.
2.3 Bearing Loads
The tangential stress transferring torque is assumed to be equal to the frictional stress. However, the
frictional stress is related to the normal stress by the friction coefficient. Thus, developing significant
no-e-tic

bearing loads between the mating parts is key to transferring torque. And, understanding the normal
stress distributions therefore allows one to understand how the torque is transferred in a connection.
The normal stress distribution can be evaluated in a variety of ways by considering the elastic
behaviour of the mating components. Numerical methods such as the Finite Element Method (FEM)
can provide very precise predictions of stress distributions for evaluating torque transfer capacity.
However, such techniques can be expensive and time consuming, and their application usually
benefits from a more fundamental parametric description of the processes governing the system to be
modelled. For this report, the fundamental parametric description is developed and implemented in a

5 of 38
28-April-2003

simple engineering numerical model (described in 5.1) to provide basic torque transfer predictions
that can be used for design purposes. Verification and calibration of this model was performed
experimentally, and the model was subsequently employed to develop the torque enhancement
system being implemented for standard API connections.
2.3.1 Bearing Load Basics
Bearing stresses on the connection components must be reacted by internal stresses in those
components: radial loads are carried by hoop stresses, and axial bearing loads are carried by axial
stresses. Figure 2 shows a connection cross-section illustrating both types of stress and bearing load
distributions. Premium connections typically have a torque shoulder that engages at make-up to
develop the axial internal stress field, typically at the end of the pin as shown in the left half of the
figure. This is where a metal-to-metal seal is usually located to contain internal pressure. Standard
API casing connections develop torque only through bearing loads generated by hoop stress, as
illustrated in the right half of the picture, because their designs do not include a torque shoulder.
These connections rely on high radial stresses in the threads and a long flow path plugged with a
solids-containing pipe dope for their seal.
axial hoop
tension tension

axial hoop
compression compression

Shoulder-induced torque Interference-induced torque

pipe axis

Figure 2 Bearing load distribution examples: Axial bearing and stress (left), radial bearing
induced hoop stress (right)
The figure illustrates the two types of bearing load distribution and torque transfer independently. In
fact, most premium designs generate some radial interference before shouldering, although it is
typically less than API connection designs because the threads are not required to provide a sealing
function, and some applications benefit from keeping hoop stresses lower for corrosion resistance and
other reasons. Much of the additional cost of premium connections is associated with managing the
tolerances required to maintain the hoop strain, particularly in the seal region, inside a target window.
Fortunately, from a torque transfer standpoint, the requirement for high bearing stresses to provide a
seal in both types of design has necessitated relatively high make-up torques. The mechanical
interlock between threads is very efficient, with most threads delivering the majority of their potential
axial strength with a hand-tight make-up. Thus, in many ways, high pressure capacity is synonymous
with high torque capacity.
no-e-tic

2.3.2 Bearing Load Capacity Calculations


The nominal radial stress developed due to interference is commonly calculated using shrink-fit
pressure vessel equations. Thick-wall pressure vessel equations provide better accuracy, and they
provide a basis for calculating a pressure rating for thread-based seals, but are more complicated to
deal with. For the purpose here of understanding how the radial interference develops torque, the
thin-wall pressure vessel equations are used.

6 of 38
28-April-2003

The average radial stress in the connection can be approximated by that created in the middle of the
thread. The maximum value will be the lesser of that generated by the box and by the pin:

r =
(
2 max S yb t b , S yp t p )
Dp Equation 8
, where
Dp is the pitch line diameter in the middle of the thread,
Sy is the yield strength of the material, with values specified independently for
the pin and box, and
t is the wall thickness to the pitch line at the middle of the thread, with values
specified independently for the pin and box.
This relationship demonstrates how the maximum radial bearing stress is reduced relative to the
material strength by the D/t ratio. However, the length over which the radial bearing stress is
developed offsets the associated reduction in torque. Thus, one mechanism for increasing the torque
capacity of the thread is to increase the threaded length. API connections have a relatively long
threaded length, but this length is not proportional to the casing size. Therefore, for similar D/t ratios,
the torque capacity of the thread form tends to increase linearly with the pipe diameterhence linear
rules of thumb such as 1,000 ftlb of torque per inch of diameter for buttress connections work
reasonably well. But this means API connection torque does not increase in proportion to the torque
capacity of the pipe body. (For constant strength and D/t ratio, pipe body torque capacity actually
increases with diameter cubed.)
The total radial bearing load can be calculated by integrating the radial stress over the threaded
length. This mathematical operation also allows variations in thread interference to be accounted for
in calculations of torque loads below the maximum level. There is also an added component
associated with an elastic bending wave that extends a modest distance beyond the interval with
imposed interference. This can be accounted for with a correction factor that can be determined
analytically or numerically.
The axial bearing load available from shouldering can be calculated directly using the material
strength in conjunction with the section area bearing the load. This is usually the pin section area,
which is some portion of the total pipe wall section area, because threads must consume some of the
section area. Since there is no D/t scaling of the axial load, as there is for the radial load, a
substantial load can be developed from a relatively small area. Furthermore threads that develop
additional torque with axial load effectively double the torque enhancement (if the friction coefficient
is constant everywhere), since the axial load is internally equilibrated or reacted by the threads.
(Thread shape and other effects on torque transfer are discussed in the next section)
In connections where both radial and axial stresses are developed, there is actually an increase in the
maximum value of both stresses, and an optimal ratio exists between the two. Consider the Mises
yield envelop for steel illustrated in Figure 3. The yield surface is given in terms of the axial and hoop
stresses (assuming the radial stress component and stress gradients through the tubular wall
no-e-tic

contribute insignificantly to the stress state). The equation for the yield surface is given by:

S y = h2 + a2 h a Equation 9

Considering the shouldering region of the connection, the pin will be in biaxial compression,
corresponding to the lower left quadrant of the diagram, and the box will be in biaxial tension,

7 of 38
28-April-2003

corresponding to the upper right quadrant of the diagram. The uniaxial yield strength is marked on
the axes at four locations, in compression and tension, in both directions.
a

Sy
Box
(tensile)

h
Pin
(compressive)

Sy
Optimal Torque
Stress

Figure 3. Biaxial yield envelop for tubular components


Stress states for connections with only radial bearing loads fall on the horizontal axis, inside the
ellipse, and for connections with only axial bearing loads the stresses fall on the vertical axis.
However, under biaxial stress, the individual stress components can actually exceed the uniaxial yield
stress of the material because of biaxial effects.
In fact, an optimum ratio between radial and axial bearing stresses can be found mathematically,
assuming an effective radius for both components. If we assume the radius of contact is the same for
both components, the optimum ratio is 1:1, giving a maximum torque contribution when the axial and
hoop stresses are both equal to the uniaxial material yield strength. Each stress component thus
supports the other, allowing the full torque contribution of both components to be combined.
The biaxial loading develops in the shoulder region, where the pin is typically smaller than the box and
therefore governs the total torque that can be transferred in that region. Figure 3 shows the point on
the yield surface where the maximum torque transfer will be developed. While it is not practical to
design a connection to reliably engage the optimal stress ratios to reach this location directly, this
solution does indicate where the stress state will evolve under plastic deformation to reach the
ultimate load. It also gives a direction for shouldering connection designs to be optimized for torque
transfer.
2.3.3 Thread details and bearing loads
Thread characteristics have some implications for bearing loads and how externally applied loads may
impact the torque capacity. Threads that bear radial interference-induced loads on the thread flanks
(8-round) behave different from those that carry those loads on the thread roots and crests
(buttress).
no-e-tic

Threads like the 8-round that bear radial interference loads on the thread flank tend to amplify the
radial bearing stress because of the large angle of the thread from the pipe axis. Also, depending on
how symmetric the load and stab flanks are, the incremental torque generated on the thread by axial
load can be small, and even zero, until one of the thread flanks us unloaded. Consider the illustration
of the 8-round thread in Figure 4, subject to three levels of axial load. There is no axial load in the
first case, which shows symmetric loading on the load and stab flanks because of the symmetric

8 of 38
28-April-2003

geometry of the thread form. It shows the average radial stress generated by radial interference
amplified according to:

r
b =

sin
2 Equation 10
where b is the bearing load, normalized in terms of the thread length. This effect is analogous the
that employed in V-belt designs for transmitting mechanical power between rotating shafts.

a) zero axial load transfer

n n

F F
r r

b) moderate axial load transfer c) high axial load transfer

Figure 4. Load transfer on 8-Round Thread


The diagram also illustrates how load is transferred from one flank to the other as axial load is
applied. However, because the flanks are symmetric the net bearing load generated to develop torque
does not change until the load transfer is sufficient to unload one flank. Until that point, the load
increase on one flank is offset by an equal load on the opposite flank. This also means that when the
no-e-tic

load transfer is large enough to unload one flank, one or both of two things must happen: incremental
load above the flank-unloading level must be transferred by another thread, or additional interference
must be generated in the thread to increase the equivalent bearing stress associated with the load
flank stress required to react the load being transferred.
Buttress threads behave quite different, because the axial load transfer across the thread is nearly
decoupled from the radial load. Figure 5 shows the load transfer distributions on the buttress thread
with no load transfer and with axial load transfer in each direction. The normalized radial bearing
stress is nearly equal to the radial stress generated by hoop deformation, because the contact surface
9 of 38
28-April-2003

on the roots and crests is nearly parallel to the pipe axis. There is slight magnification factor due to
the taper on the crests (for sizes smaller than 13-3/8), and a corresponding load flank bearing stress
that is small and therefore ignored at zero external load. Also note that the radial bearing load is
shown on the major thread diameter.
When axial load transfer is carried by the buttress thread, there is an increase in the total bearing load
on the thread, corresponding to the flank load developed. The radial load on the crest suffers a
negligible change, because the crest is parallel to the axial load direction. Consequently, axial load
transfer produces a corresponding increase in torque capability. This effect provides some
compensation for the lack of a load magnification like that generated by the 8-round thread form.

r
a) zero axial load transfer

S L

Ft Ft

r r
b) Stab flank load transfer c) Load flank load transfer

Figure 5. Load transfer on Buttress thread.


2.3.4 Sensitivity to Axial Stress
Bearing loads in the connection will change in response to axial load. In conventional applications this
no-e-tic

behaviour is of little consequence because the connections are not usually subjected to high torque
loads after make-up. However, in casing drilling applications the connections are subjected to both
tensile and compressive axial load (depending on where they are located in the string) while
transferring high torque loads to the drilling assembly.
For example, consider a non-shouldering buttress connection subjected to axial load, and assume the
interference-related stresses are in the elastic regime. At a neutral load, nearly the entire torque
capacity is generated by radial stresses between the thread roots and crests. When axial load is added
to the connection load, some torque capacity is added as a result of the bearing stress developed on

10 of 38
28-April-2003

the load (or stab) flanks. Some radial bearing stress is lost at the face of the coupling because of
Poisons ratio contraction of the pin relative to the box, because of the axial stress difference at that
location. However, the same effect increases the radial bearing stress near the centre of the
connection. Consequently, for a buttress connection the net change in radial bearing stress
approximately nets out to zero. Thus the net change in torque capacity contributed by axial load,
results from the bearing stress it creates on the thread flanks. This torque is equal to the product of
axial load, friction coefficient and mean radius of the bearing stress, which is roughly the radius of the
pitch line at the middle thread (see Figure 1 where N is equivalent to axial load).
On the other hand, a non-shouldering 8-round connection provides little change in torque capacity
with axial load, because of the way it transfers load from one flank to the next, as described in
Section 2.3.3. Only once all threads have unloaded one flank will the torque capacity increase, for a
given interference level. However, at such loads, seal integrity is compromised and tensile failure is
also likely. Therefore, additional torque capacity in 8-round threads must be provided through some
other mechanism.
The impact of tensile axial load on torque capacity in shouldering connections may be beneficial,
neutral, or detrimental, depending on the shoulder and thread design. Shoulder stresses are
compressive, generated by interference between the shoulder and thread load flanks. At the reference
neutral load, the shoulder load and net thread flank load are equal. Axial tension shifts the balance of
these loads, reducing the shoulder load and increasing the thread flank load. The load lost on the
torque shoulder is in proportion to the area of the torque shoulder relative to the total connection
cross section. The load added to the net flank bearing load is equal to the total applied load, less the
shoulder load removed. Thus, for large shoulders, more load will be lost on the shoulder than gained
on the thread. However, such configurations are rare, so generally, the load increase on the threads
will exceed that lost on the shoulder.
Connections based on a buttress-style thread will therefore have torque lost on the shoulder offset by
an increase in load transferred by the threads, because of the direct torque response of buttress
thread forms to axial load. There is an added bias in the load transfer increase torque capacity
because the thread flanks are at a larger diameter in most casing connection designs and the same
load therefore develops higher torque. Under compression, the torque capacity will drop until the load
shifts from the load flank to the stab flank, at which point torque capacity will increase as both flank
and shoulder loads increase with compression. However, a problem may develop during the transfer
between flanks that the axial backlash displacement will mobilize friction in the threads as discussed in
Section 2.4.2 below, relieving the torque generated by radial bearing loads and allowing incremental
make-up.
In designs using a V-type thread like the 8-round design, the torque response of the threads to axial
load is small, if there is any at all. Consequently, torque lost on the shoulder of these designs with
tension is not offset by an increase in the threads. On the other hand, compressive loads on these
designs increase the torque capacity by adding bearing load to the torque shoulder. Furthermore,
because there is no axial backlash in the thread design, there is no risk of remobilizing friction
developed by radial interference.
2.4 Additional factors in Torque Transfer
Most of the discussion here relates to bearing loads, how they are calculated and how they can be
no-e-tic

optimized to generate torque. The primary objective is to enhance the torque transfer characteristics
by modifying the bearing loads and distributions with a torque ring in the J-section of the connection.
However, there are other factors that control the torque generated. While these factors are not
considered as design variables in this exercise, understanding their role is important for characterizing
torque transfer, and such understanding may be valuable if other connection designs are pursued to
provide enhanced casing drilling performance. Following are the other primary factors controlling
torsional performance of casing connections.

11 of 38
28-April-2003

2.4.1 Friction Factor


The frictional characteristic of the interface carrying the bearing load is the most important factor
governing the conversion of bearing load into torque. While this characteristic is usually assumed to
be a constant value, in fact there are many sources of variation in this property.
Threads are always protected with a lubricant of some sort, usually thread compound (pipe dope),
that includes suspended solid particles. These particles, referred to as the solids pack, play an
important role in providing the seal function, preventing galling and controlling friction coefficient. The
pipe dope behaves like a fluid during make-up, until it is trapped in the interfacial regions carrying
bearing loads where it can dramatically reduce the effective friction factor during make-up. This
tendency to entrap all or a portion of the grease phase pressure is an inevitable consequence of
creating sealing threads: threads that cannot entrap thread compound will also not entrap well bore.
Once the connection is made up, and given sufficient time, the entrapped grease pressure may
further decrease or bleed off, causing the portion of normal surface contact stress born by the solids
pack to increase, which coincidentally increases the frictional capacity of the film. For typical oil field
thread compounds, the unconfined friction coefficients are in the range of 0.1 to 0.15 but the effective
friction factor can drop to one tenth of this range where dope entrapment is allowed to develop on
make-up. This effect explains why break-out torques are usually higher, and sometimes very much
higher, than make-up torques. The overall effect of dope entrapment over the entire thread length is
not as severe as might appear from the discussion here, because not all bearing surfaces are prone to
dope entrapment, particularly the imperfect threads. A typical result for an API buttress connection
test demonstrates torque values 65% higher for threads that disable dope entrapment.
Another source of variation in friction factor is the magnitude of the bearing stress. The friction factor
is usually assumed to be constant, but in fact it is a function of the contact stress born by the film as
illustrated in Figure 6 representing the behaviour on say a torque shoulder surface undergoing sliding
during make up. The variation with contact stress depends on factors such as grease and solids pack
composition, surface finish and shear displacement magnitude.

0.1

0.08

0.06
Friction
Coefficient
0.04

0.02

0
Bearing Stress
no-e-tic

Figure 6 General nature of bearing stress effect on friction coefficient.


2.4.2 Remobilized friction
The direction of the frictional stress is perpendicular to the normal stress, in the same direction as the
relative slip between the two surfaces. When connections are made up, the direction of friction is in

12 of 38
28-April-2003

the circumferential (or tangential) directions, thereby developing the torque to react that applied by
the make-up device.
If the direction of slippage changes, the direction of friction stress changes with it, so that slippage in
one direction can occur under low loads if the major component of the friction vector is occurs in one
of the orthogonal directions. This behaviour is referred to as compound sliding and can have negative
implications for incremental make-up. Designs that include any axial backlash in the threads are thus
prone to incremental make-up if cyclic loading cycles the backlash. For example, the standard buttress
connection design includes tolerances that provide some clearance between the flanks to
accommodate thread form variations. When load is cycled (either bending or axial) through the
neutral point the load transfer shifts from the load flank to the stab flank and vice versa as the
threads shift axially. During that shift, the surfaces slide axially and the friction stress rotates to the
axial direction. If torque is applied during this process, there must also be some tangential motion to
maintain a tangential component in the friction vector for torque transfer.
The best way to prevent incremental make-up is therefore to eliminate backlash movements under
normal operating conditions. This is an advantage in shouldering designs, because the shoulder load
creates a pre-stress in the connection that generally exceeds axial compressiv stresses due to bending
and cyclic axial loads, thereby preventing backlash movements from occurring in the threads,
preserving the torque capacity. Also, thread designs that carry radial loads on the thread flanks, such
as the API 8-round thread, prevent backlash because of the tight axial engagement created in that
type of thread form.

3 API Thread Tolerances


Thread tolerances specified by API were developed long before modern threading equipment was
available to provide the precise geometric control that is possible today. While threads can be
manufactured to tighter tolerances, there are a few compelling reasons why design solutions to
increase their torque capacity need to accommodate the API tolerances:
1. Some manufacturers still use older equipment for non-premium (i.e. API) connections,
2. There are large quantities of pipe in stock around the world that have been built to these
tolerances,
3. Quality control costs to maintain API tolerances are much lower than those to maintain tighter
tolerances.
The primary API tolerance controlling connection make-up and performance is on the thread taper.
The axial position of the thread relative to the coupling face or pin nose is relatively unimportant,
because interference is controlled by the taper thread taper and applied torque. The made-up position
of the thread is therefore quite variable. Furthermore, the variation in interference over the thread can
be significant because of the large range of allowable tapers. Figure 7 shows the possible
combinations of fast (high-taper) and slow (low-taper) threads between pin and box thread pairs at
tolerance extremes. If both threads are at the same extreme tolerance (fast or slow), then the
interference is still uniform, because the taper tolerance is the same for both parts. On the other
hand, if the two parts are at opposite extremes, then there can be a large interference variation, with
more interference at the mouth of the coupling if the pin is fast and the box slow, or at the tip of the
no-e-tic

pin if the pin is slow and the box fast.

13 of 38
28-April-2003

Slow Box Fast Box


Fast Pin Slow Pin

Fast Box Slow Box


Fast Pin Slow Pin

Figure 7. Illustration of taper variation combinations


In the limit, torque capacity will be governed by the yield strength of the pin and box. While taper
mismatch produces initial yield sooner in terms of applied torque, plastic deformation in the pin and/or
box will redistribute the interference after the yield strength is reached locally, and the ultimate torque
will not be dramatically affected by the taper mismatch.
The torque-turns relationship could be estimated by the average interference, near the middle of the
thread, for example at the pitch plane. Thus thread diameter can be used to interpret the position to
which a connection will be made-up. However, thread diameter on API connections is gauged with a
ring and plug system. These gauges have carefully machined thread forms and are screwed onto the
pin and box to indicate the hand-tight location or stand off, which is used to interpret the thread
diameter. Taper variation introduces an error into the pitch diameter measurement, because diameter
is measured at one end of the thread form as illustrated in Figure 8. Consequently, taper variation will
cause the pin to be smaller and the box to be larger than indicated by the gauges, and therefore the
average interference for a given pin-in-box penetration will always be equal to or less than inferred
from the measurement.

Ring gauge Ring gauge

Fast Pin Slow Pin

Slow Box Fast Box

Plug gauge Plug gauge

Figure 8. Ring and plug gauge measurement error due to taper variation
The gauging method therefore allows a wide range of relative pin-in-box positions for a given torque
load. The actual values are not presented here but require a careful assessment of the API
specifications for both geometry and gauging practice to infer the actual allowable part geometries.
This greatly complicates the process of providing torque enhancement with a device to induce an axial
load component on the pin nose while maintaining adequate thread sealing and avoiding over torque
or galling risk. This is also why many premium connection designs require tighter tolerances on their
geometry and employ alternative gauging methods to measure thread form and seal dimensions.
Quantitative management of these tolerances thus becomes the major design challenge to successful
implementation of torque rings.
no-e-tic

4 Torque Ring Design


Many premium connection designs incorporating a shoulder have demonstrated the superior torque
characteristics such configurations afford. Converting the standard API casing and tubing connection
configurations to employ similar torque transfer mechanisms is therefore an appealing option for the
casing drilling system. A successful conversion would allow inexpensive connections that are widely
stocked to be used in this moderately demanding operation.

14 of 38
28-April-2003

Various types of rings, placed between the pin noses have been used in some premium or semi-
premium applications. These rings have primarily been designed to provide enhanced sealing and
incidentally improve torque functionality since axial load and torque are transferred directly between
pins by the pin nose ring.
The couplings using this strategy, such as Hunting Interlocks Convertible Buttress Connections, are
not manufactured to standard API tolerances, but are specially designed to facilitate the use of such
rings, particularly in terms of tolerancing. In addition, the pins are screened to further limit the
allowable tolerances. These available methods, thus only partially achieve the design goal of
enhancing the torque capacity of standard API connections.
This section initially discusses the design requirements for converting stock API couplings to a
shouldering design with a pin nose ring or what is referred to here as a torque ring. A novel solution
that meets the requirements of casing driling while accomodating the broad tolerance band allowed
by API specifications is then presented.
4.1 Design Requirements
4.1.1 Functional Requirements
For optimal performance the torque ring implementation for API connections needs to accommodate
the wide tolerance variation allowed by the API specifications. The primary concern is that the solution
be able to accommodate the wide range of stand-off that may exist between the two pins in the
optimal make-up location for transferring torque. Avenues for providing this latitude could include
mechanical design options or field measurement methods for determining an optimal ring length. The
ring solution must also accommodate variations in diameter while meeting the installation
requirements discussed in the following section.
The design must carry the maximum axial and torsional loads that can be carried by the tip of the pin.
Stronger material can be used to offset smaller section characteristics in the ring relative to the pin
nose. The rings must not infringe on the drift diameter specified for the pipe being joined.
Surface characteristics of the ring should be chosen to maximize frictional load transfer. This will help
both installation requirements for ring retention and for torque transfer between the two pin noses.
4.1.2 Installation Requirements
The ring must be field-installable with simple hand-held equipment. Installation should be performed
prior to moving the pipe onto the rig floor, preferably on a pipe rack on-site or in a pipe yard. Once
installed, the ring should be secure to prevent it from falling out during handling and from being
knocked out by equipment being passed in and out of the casing before the field-end make-up is
complete. While the ring must be secure in the coupling prior to make-up, it should also be
removable, in case there is a problem with make-up and the ring must be removed to accommodate a
different thread. Furthermore, the installation must not risk damaging the threads by deforming their
edges or galling.
4.2 Design Concept
The design concept for meeting all of the necessary requirements is a torque ring deformed in a multi-
lobe pattern to provide a more compliant radial deformation characteristic. The radial compliance is
no-e-tic

used to accommodate the wide range of thread diameters that is allowed by API specifications at the
centre of the coupling. The outside surface of the ring is serrated in a unidirectional pattern to allow
for easy installation and good retention characteristics.
The API specifications allow a wide range of relative make-up positions between pin noses. This range
is broken into zones based on the analytical basis described below that was developed to estimate the
sensitivity of torque transfer characteristics to connection and ring tolerances. A measurement system
is described for determining the ideal position of the field-end pint relative to the mill-end pin location,
and the ring most closely matching the desired length is selected for each connection and installed.

15 of 38
28-April-2003

4.3 Lobe Design Basis


4.3.1 Multi-lobe forming and installation
The radial or hoop stiffness of a cylindrical ring is relatively high because of the uniaxial hoop strain
associated with the axisymmetric stiffness mode. Higher modes offer much lower stiffness because
most of the deformation is associated with flexural deformation of the ring. The elastic range of
deformation is also much greater as a result.
The next mode is the two-lobe or ovalization mode. An example of this mode being utilized is in
some childproof caps for pharmaceutical products. Squeezing the cap in two locations pushes the rim
in at those locations. This ovalizes the rim, causing it to displace outward at the points 90 degrees
from the locations pressed, releasing a catch and allowing the cap to be turned.
The two-lobe shape is the lowest stiffness mode with the largest range because it has the longest
distance to accumulate flexural deformation for a given diameter. It therefore offers the greatest
range for accommodating diameter variations in the collar due to tolerance specifications. However, it
is not the best choice for the application for two reasons:
1. It contacts at two locations, allowing the ring to rock and rotate when it is first inserted,
potentially leading to installation problems, and
2. It generates the lowest maximum elastic radial force, and therefore the lowest retention
potential of the multi-lobe shapes.
Therefore, depending on the size of the application, three or more lobes are best suited to provide the
preload and ring retention function. Figure 9 illustrates the multi-lobe behaviour when the ring is
formed by external loading and when it is subsequently installed in the connection. The heavier line
shows the shape of the ring at the start of the process and the moderate size line shows the
configuration at the end of the process being illustrated.
Fr
Ff
uf

Fr
Ff

Ff Fr
no-e-tic

a) Form multi-lobe ring b) Retained multi-lobe ring

Figure 9 Multi-lobe torque ring deformations


Part a) of the figure shows how the forming causes the ring to move outward at the points between
where the forming load is applied. The ring moves inward with the forming load, and with sufficient
load, plastic deformation in the ring causes a permanent multi-lobe shape to be set. The magnitude of
the outward displacement between load points is similar to the inward displacement occuring at the
16 of 38
28-April-2003

load points. The effective outside diameter (defined by the outside surface at locations between load
points) must be bigger than the maximum coupling ID allowed by API tolerances. The effective inside
diameter (defined by the inside surface at the load point locations) of the as-formed ring can be
smaller than the drift diameter specified for the pipe product, as long as the installed diameter
exceeds the drift after installation.
Part b) of Figure 9 shows the deformation of the ring when it is installed. The installation load points
develop between the forming load points, where the effective diameter is largest. The deformation
that results tends to reverse the multi-lobe shape created by the forming process, and the ring is
pushed back to a nearly circular shape. In fact, if the installation diameter approaches the original ring
diameter, the ring will come into conformance around the complete circumference, and the stiffness
increases dramatically as the deformation switches from the multi-lobe characteristic to the
axisymmetric deformation characteristic. If this happens the installation force required increases
dramatically, but this may be tolerable in extreme cases if an appropriate installation tool is available.
Generally, however, the design objective will avoid complete conformance, and some multi-lobe
characteristic will remain after the ring is installed.
The behaviour characteristics of the multi-lobe shape are markedly different from axisymmetric
behaviour. Deformation of an axisymmetric ring produces pure radial displacement of the annular
space occupied by the ring, where both the effective OD and ID of the ring move inward radially with
interference. In contrast to that, the multi-lobe shape produces an annular displacement of the
annular space, where the effective OD moves inward, but the effective ID moves outward. Thus, the
annular area consumed by the multi-lobe shape decreases with interference.
4.3.2 Multi-lobe Design
Reviewing the configurations shown in Figure 9 and considering the characteristics described in the
previous section, the ring geometry should be characterized by the following features:
1. The source material for the ring should have an outside diameter matching the minimum
coupling diameter allowed by thread tolerances, to avoid full conformance when the ring is
installed.
2. The source material for the ring should have an inside diameter that is larger than the drift
diameter of specified for the pipe size by an amount that will clear the drift diameter when the
formed ring is installed in the largest possible coupling thread diameter.
3. The lobe amplitude must be large enough to provide adequate interference with the largest
possible coupling diameter.
4. The number of lobes used for a given ring design is governed by the required stiffness. API
tolerances are the same over the full range of pipe sizes, so as pipe size goes up, lobe
numbers also go up to provide the same range of lobe amplitudes required to accommodate
the tolerance. The retention load is thus maximized of the required tolerance range for a
given size.
A significant portion of the annular space must be left empty to accommodate the range of
displacement required by API tolerances. This requirement reduces the amount of steel that can be
used in that annulus, thereby limiting the axial load available for torque enhancement. Where
additional torque is required, there are a few options that can be used to offset this limitation without
no-e-tic

increasing material strength:


1. Break the diameter range into sub-groups, and measure the thread to select the proper size
of multi-lobe ring. The disadvantage to this approach is that the number of ring options
increases dramatically with the possible combinations of ring diameter and length.
2. Design for a common sub-range of tolerance values, corresponding to the typical geometry
range produced by modern manufacturing methods, and up to the maximum possible
connection diameter. Allow for the minimum possible diameter by providing a tool that is

17 of 38
28-April-2003

capable of pressing the ring into full circumferential conformance in rare under-size cases.
This allows one multi-lobe design to be implemented that first increases the ring ID with
interference until conformance is reached, then decreases the ring ID when more interference
for the under-size cases is imposed.
Implementing either of these options allows the torque enhancement to be increased. This is
particularly valuable for light-wall sizes and for 8-round thread forms.
4.3.3 Multi-lobe deformation characterization
Multi-lobe deformation characteristics were determined by using Finite Element Analysis (FEA) to
model the behaviour of the ring under forming and installation conditions. The analysis accounted for
nonlinear behaviour associated with material yield, geometric/large-displacement, and contact
interaction. The ring was modelled with two-dimensional solid elements and the box/forming surface
was modelled as a rigid surface.
Figure 10 shows an example of the forming loads required to produce the multi-lobe shape, and the
stiffness of the ring to the deformation loads. The two-lobe shape has the lowest stiffness and the
largest range of deformation. Stiffness and load increases, and range decreases rapidly as more lobes
are deployed in the ring. The characteristics are essentially the same when the deformation is
reversed on installation, although there will be some influence from residual stresses. More details on
installation loads are presented in subsequent sections.

1200 2 Lobe Ring


(Plane strain results) 3 Lobe Ring
1000 4 Lobe Ring

800
Total radial
reaction force 600
(N/mm ring depth)

400

200

0
0 0.5 1 1.5 2 2.5 3 3.5
Radial forming displacement at ring OD (mm)

Figure 10. Forming load results for 114.3mm torque ring. 3.175mm wall. 550MPa yield
strength
The load-displacement characteristic required to accommodate a range of diameters can be selected
from a series of response calculations like this for a given size of product. A certain minimum load will
be required to provide a minimum required retention force on the ring, and the maximum load should
be, at most, slightly above the apparent yield (nonlinear) point on the response curve. Some
no-e-tic

additional compliance is needed to avoid having the ring come into full conformance on installation,
although this may be specified over a sub-range of the full tolerance band if such a design approach is
adopted, as described earlier.
4.4 Ring Installation Analysis
Analysis runs were conducted in forming and installing multi-lobe rings in API connections. Figure 11
shows the installation response of a torque ring as it is pressed into the taper of a coupling thread. In
this case, the ring was formed into a muli-lobe shape with a 1.95mm amplitude. The figure also marks
the original outside radius of the ring, to compare with the response of the effective inside and
18 of 38
28-April-2003

outside radii of the torque ring as it is pressed into place. Note that the inward forming displacement
is slightly higher than the outward displacement, because the entire forming load is compressive,
generating slight hoop contraction of the ring as it is formed.
When the ring is pressed into the thread cone it initially exhibits the elastic stiffness expected from the
three-lobe deformation mode, the effective outside diameter is reduced and the effective inside
diameter increases. After about 0.5mm of radial displacement, the load response curve softens as the
ring develops some yielding in the material, but the displacement response of the effective ID and OD
remains linear. At about 1.4mm of radial displacement, or 72% of the lobe amplitude, the ring begins
to develop more complete conformance, as the contact points move around the circumference. The
load response curve stiffens dramatically when this occurs, and the rate of growth of the inside radius
starts to decline, becoming almost flat by 1.6mm of radial displacement. (Unloading response is also
shown in this figure.)

350 56
3 LOBE
CASE
FORMING 300 55
DISP
1.95 mm STARTING OUTSIDE RADIUS
250 54

Radial load
200 53
per half lobe Radial Installation Load Radius
(N/mm Smallest Inside Radius (mm)
ring length) 150 52
Largest Contact Radius

100 51

50 50
MIN. DRIFT RADIUS
0 49
0 0.4 0.8 1.2 1.6
Radial installation displacement at largest radius (mm)

Figure 11. Installation response of torque ring


The results demonstrate the multiple characteristics of the installation response of the rings, when
they are pressed into a cone spanning the full range of the ring amplitude. While the minimum
diameter may be smaller than the drift diameter, installation loads cause the minimum diameter to
increase enough to clear the drift diameter, provided a minimum level of interference is assured.
Interference beyond the yield point causes the response to soften, which may lead to localization of
the deformation if the stiffness becomes too small. However, once most of the displacement range is
consumed, the stiffness climbs rapidly toward the axisymmetric characteristic, even before full
conformance is reached.
4.5 Installation and Removal Tests
Two options were proposed for enhancing the retention characteristics of the torque ring. Surface saw
tooth and ratchet profiles were considered to help retain the rings. The frictional sliding response
during installation, and the gripping efficiency of the external surface profiles both required physical
no-e-tic

testing of axial load and displacement. Load application and load/displacement monitoring were
provided by the relatively simple fixture shown in Figure 12.

19 of 38
28-April-2003

Push Plate
Stop Ring Threaded Plate

Ram Pressure Port

Figure 12. Sketch of fixture used to install and remove rings in a coupling.
With this fixture, load is applied to a torque ring through the push plate attached to the ram piston.
This load is reacted back into the coupling from the ram body through the threaded load plate
engaging the coupling threads. Load is inferred from pressure applied to the ram and measured by a
pressure transducer (not shown). Ring displacement relative to the coupling body was measured by
an LVDT held in a frame resting on the open end of the coupling and a rod attached to its core in
contact with the piston end (also not shown). These measurements were simultaneously acquired by
a digital data acquisition system. Prior to testing, the relationship between ram load and pressure was
validated by applying the ram load to a calibrated load cell.
Using this equipment configuration, a series of tests were conducted where formed rings were first
pushed to the coupling mid-plane while recording load and displacement. Following this installation
test the ability of the rings to resist a removal or dislodging force was similarly measured by reversing
the coupling ends and again pushing against the installed ring while recording load and displacement.
Figure 13 provides an overview of ring responses from a test program that, combined with test
observations, allows several tentative conclusions to be reached.

5000
Installation Peak Load
4500
Removal Peak Load
4000

3500

3000
Load (lbs) 2500

2000

1500

1000

500

0
1Fine2 2Fine 3Fine 4Fine 5Fine 1Coarse 2Coarse 3Coarse 4Coarse
Test Specimen
no-e-tic

Figure 13 Summary of prototype ring installation and removal load maximums.


Comparing results between three- and four-lobe configurations for the fine tooth ring tests, shows
that although the contact forces for the four-lobe configuration are much higher and the removal
gripping force is lower. Post-test inspection of the rings shows significantly greater plastic yielding of
the ratchet teeth occurred during installation of the four-lobe compared to the three-lobe rings. Figure
14 shows this yielding which is particularly noticeable on the leading edge of the ring shown in the

20 of 38
28-April-2003

bottom of the photograph. This resulted in little to no ratcheting action during unloading as seen in
the data of Figure 13. This appears to be as much a function of the radius of curvature in the contact
zone, as the higher load for the four lobe cases. The smaller radius of curvature for the four lobe rings
resulted in a zone of contact about that of the three lobe rings. Combined with the higher contact
load per lobe, the stress on the teeth edges was high enough to fold the teeth edges over during
installation and consequently the reverse load gripping or ratcheting action was disabled.

Figure 14 Photograph of four lobe fine pitch ring contact patch after removal showing
yielding of ratchet teeth.
The influence of lubrication on removal loads is not dramatic, particularly for the coarse ratchet
configuration. Instead, variation for the coarse rings appeared to be more a function of how well the
two coupling box helixes mesh. On the couplings where the helixes tended to mesh, the ring teeth
could more fully drop into and engage the coupling threads and therefore develop more removal load
capacity.

5 Torque Ring Performance Enhancement


The preceding sections have described how threaded connections transfer torque, how API connection
geometry is allowed to vary by their tolerance specifications and the challenges that variation presents
for enhancing torque transfer. These sections have also described a novel and practical torque ring
architecture that addresses those challenges to provide the additional bearing load required for
enhanced torque transfer. The question remains: How much of a difference will the torque ring
solution really make?
This section develops the specific torque transfer relationships for API buttress and 8-round
connections and the torque enhancement provided by the torque ring. An analytical basis is developed
no-e-tic

that provides the fundamental parametric description of torque transfer, which is most useful for
design purposes. An experimental program is also discussed, the results of which are used to verify
the analytical basis and calibrate the physical factors that are not accounted for in the analytical
description.

21 of 38
28-April-2003

5.1 Analytical Performance Assessment


5.1.1 Buttress Thread
The make-up torque/turns relationship, and several other parameters of importance to the CDS
application, can be estimated analytically by making a few reasonable assumptions:
1. bearing loads can be calculated using thin wall pressure vessel or shrink fit equations
2. friction factor is constant (note that this assumption can be relaxed if simple numerical
techniques are employed)
3. bending waves characterize end-effects in the thread contact zone
4. no lead interference is generated by manufacturing tolerances or Poisson's ratio effects.
A simple numerical model incorporating these assumptions was created and provides a measure of
the following performance parameters relevant to the CDS application:
1. Make-up torque,
2. Imperfect thread contact stress (limits leakage capacity),
3. Thread backlash load.
4. Load transfer in imperfect threads (controls axial and flexural strength or strain capacity), and
5. Galling tendency.
Torque is obtained by integrating the product of the element contact load with the friction coefficient
and contact surface radius. Similarly, thread backlash load is computed by integrating the product of
the element contact load with the friction factor. This simplification ignores the feedback of axial load
on contact stress through Poissons effect. Leakage pressure is computed as the internal pressure at
which the contact stress of the mid-length perfect thread element is equal to the applied pressure.
Interpretation of load transfer in the imperfect threads and galling tendencies are derived by inference
from the contact stress plots over the engaged thread length.
Figure 15 shows a sample output from the model for a 178mm (7 20 ppf) connection made up to a
target torque. The four plots provide a visual indication of the overall connection stress state and the
tabular values extract parameters relevant to design.
As initially configured the model assumed a constant friction coefficient and make-up to the base of
the triangle. The model was subsequently modified to allow the friction coefficient to be specified
independently for individual elements and for make-up to any arbitrary position. The more general
frictional description of this model allows more complex frictional behaviour to be modelled, such as
hydrodynamic or dope entrapment effects in the perfect thread interval. Similarly, the more
generalized make-up description allows the model to iteratively calculate the position for a given
torque.
Although the ultimate torque limit from radial interference is governed by the material strength and
D/t ratio of the mating parts, a practical limit is defined by the make-up specification for the
connection. This is defined by the collar face location relative to the triangle stamped on the pin. The
relative improvement afforded by the torque ring can be evaluated by comparing the delta torque to
the make-up or shoulder torque to reach the triangle without the torque ring. Table 1 gives a
no-e-tic

summary of the make-up results for 177.8mm (7in) BT&C casing for a range of tolerance conditions.
Calculations assume the pin is advanced to align the triangle base with the collar face. The tolerance
nomenclature uses SP/FB to refer to a slow pin taper, fast box taper, and Loose/Nom/Tight refers to
the pin/box pitch diameter size combination relative to the triangle position, giving the relative
interference at the defined make-up location.

22 of 38
28-April-2003

Table 1 Summary of 7in, 20 ppf buttress make up to triangle base.


Case Tolerance Profile Net torque Seal pressure Sliding load
Taper Dia ft lb Psi lb
13 SP/FB Loose 3230 9943 11257
17 FP/FB Loose 3497 6454 12130
11 SP/FB Nom 5226 9943 18166
15 FP/FB Nom 5617 9751 19478
9 FP/SB Loose 5903 3072 20395
7 FP/SB Nom 8016 6369 27716
5 Nom Loose 8467 9943 29341
1 Nom Nom 10309 9943 35714
3 Nom Tight 10841 9943 37544

These results illustrate the large range of input torque required to reach the nominal make-up
position, with a 3:1 ratio between the maximum and minimum make-up torques.

Another way to demonstrate the make-up variability is by comparing final make-up positions for a
given input torque. This method tends to be used for mill end make up and the associated variability
therefore represents the range of mill end axial positions the torque ring design must accommodate.
Table 2 shows the range of make-up positions that can be reached for an input torque of 7,000 ft-lb.
The range calculated for this level of torque is within the API range of 2.64 turns, but it starts slightly
below the API minimum of 1 turn.

Table 2 Summary of final positions for 7 20 ppf buttress made up to 7,000 ft lb.
Case Tolerance Profile Make-up past base of triangle
Taper Dia turns inches
3 Nom Tight -1.2 -0.240
1 Nom Nom -0.76 -0.152
7 FP/SB Nom -0.22 -0.044
17 FP/FB Loose 0.74 0.148
13 SP/FB Loose 0.84 0.168

Table 2 thus also shows what range of torque ring length would need to be available to provide good
thread sealability and thread-torque transfer. With a range of 0.408 on each side of the connection,
eight sizes would be required for 0.1 increments, presumably distributed in a statistical manner about
the mean.
The total torque capacity of buttress connections is obtained by adding the torque obtained from
radial interference, as shown in Table 1 and referred to as shoulder torque, to the delta torque
obtained from end load provided by the torque ring. The maximum available delta torque or torque
boost from the torque rings may be calculated assuming a material yield strength for the ring and
friction coefficients applicable on the shoulder and load flank surfaces in the connection. Two
coefficients are required because the torque derives from two sources in the buttress connection as
no-e-tic

described in 2.3.2: friction at the pin tip or shoulder, and friction on the load flanks from the stress
developed to react the axial load on the ring. These friction coefficients are best determined
empirically as described in the following Section 5.2 describing testing and analysis of torque capacity
results for both buttress and 8-round connections.

23 of 38
28-April-2003

Case Number 7 Case Number 7


0.010000
0.025 Box
0.008000 Pin

0.006000 Yield box


0.020 Yield pin
0.004000

0.002000
0.015
Interference Hoop
0.000000
(in) Strains
0.010 -0.002000

-0.004000

0.005 -0.006000

-0.008000

0.000 -0.010000
0 1 2 3 4 5 0 1 2 3 4 5
Axial Position from Cplg Face (in) Axial Position from Cplg Face (in)

Case Number 7 Case Number 7


10000 100000
Box
9000 80000
Pin
8000 60000

7000 40000

6000 20000
Contact Hoop
Stress 5000 Stress 0
(psi) (psi)
4000 -20000

3000 -40000

2000 -60000

1000 -80000

0 -100000
0 1 2 3 4 5 0 1 2 3 4 5
Axial Position from Cplg Face (in) Axial Position from Cplg Face (in)

Case Number 7 Make up past base of triangle turns -0.22


Yield stress ksi 80 Net torque ft lb 7000
Casing outside diameter in 7 Leakage pressure psi 4989
Casing weight ppf 20 Axial load to slide @ Pcontact lb 24195
Interference at cplg face in 0.0151 Friction coef in imperfect thrds 0.1
Int. gradient on radius in/in -0.003 Friction coef in perfect thrds 0.03
no-e-tic

Figure 15 Fast pin slow box nominal diameter interference 20 ppf buttress connection
made up to target torque.

24 of 38
28-April-2003

5.1.2 Round Thread


The analytic model developed to evaluated the make up response of buttress connections to radial
interference, thus predicting shoulder torque, may be adapted to consider the behaviour of API STC
and LTC connections. However, as discussed in Section 2.3.4, unlike the buttress thread form, the 8
round thread form of STC and LTC connections results in much greater interaction between axial,
radial and pressure loads. The model was therefore adapted to accommodate this interaction in the
terms described there, where the axial loads induced by make up against a torque ringcompression
in the pin, tension in the boxtend to unload the stab flanks.
Where this occurs in the presence of contained pressure, the pressure will invade the stab flank
interfacial region when the contact stress is reduced to equal the pressure. This behaviour was
therefore modelled to obtain a relationship between axial load applied at the pin tip and the length of
thread invasion. The non-invaded zone length, if any, thus governs the effective seal width. The bi-
axial stress effect shown in Figure 3 significantly affects the radial interference; therefore the model
also included this yield criterion and associated Poissons effects. This model thus relates the
connection geometry variables to both the make up torque-turns behaviour and hence end gap
controlled by torque ring length as well as torque ring delta torque, and thread seal length under
internal pressure.
The general effect of API taper and diameter tolerancing of STC and LTC connections is similar to that
already presented for buttress connections, therefore these tolerancing effects are not again
presented here for the STC and LTC threads. Instead, a sample output is presented to illustrate the
predicted axial load or torque ring effects.
Figure 16 presents results for a nominal geometry, 7 23 ppf L80 LTC connection. For this analysis
case, the connection is first made up to a target torque of 4,000 ftlb as shown, then an end load is
applied to simulate shouldering against a torque ring, and finally an assumed pressure is applied. The
lower right hand graph shows the axial load transfer occurring between the pin and box in terms of
axial stress. The stab flank stress is equal to the applied pressure over the interval where the axial
stress is greater than zerocorresponding to the leak path length. The interval between this point,
where the axial stresses equal zero, and the limit of the perfect threads is the active seal width.
The degree of make up shown (1.7 turns past hand tight) is only about one half the nominal API
power tight of 3 turns. Thus the case shown demonstrates that even for relatively low radial
interference the remaining seal length is likely adequate to seal the modest pressure shown. However
the model shows that at 2,000 psi, the effective seal length is nearly zero.
The model may also be used to illustrate how torque capacity is a function of internal pressure, since
the total torque calculated by the model takes this into account. For the case shown, the indicated
torque of 12,490 ftlb at 1000 psi increases to 12,881 ftlb at 2,000 psi and drops to 11,949 with zero
internal pressure. This demonstrates the relatively modest impact of internal pressure yet serves to
show that under extreme conditions changes in internal pressure may contribute to over torque.
The bearing load in the threads in the interval over which the axial load is reacted, is independent of
the radial interference, (see Section 2.3.4) so that the radial interference only contributes to torque
where the axial stress is zero. For the case shown, this only contributes 1,163 ftlb of torque, out of
the initial 4,000 ftlb of shoulder torque. This demonstrates the fact that the delta torque (torque
increase obtained by contact with the torque ring) is not additive with the shoulder torque. Therefore
no-e-tic

for high axial load on the torque ring, the torque capacity becomes increasingly independent of radial
interference. This fact is used in the evaluation of full scale data presented in the next section to
utilize a simplified basis for predicting total torque capacity of 8-round connections with torque rings.

25 of 38
28-April-2003

Case Number 1 Case Number 1


0.010000
0.0250 Box
0.008000 Pin

0.006000 Yield box


0.0200 Yield pin
0.004000

0.002000
0.0150
Interference Hoop
0.000000
(in) Strains

0.0100 -0.002000

-0.004000

0.0050 -0.006000

-0.008000

0.0000 -0.010000
0 1 2 3 4 0 1 2 3 4
Axial Position from Cplg Face (in) Axial Position from Cplg Face (in)

Case Number 1 Case Number 1


10000 100000
Box Hoop
9000 80000 Pin Hoop
Pin Axial
8000 60000
Box Axial
7000 40000 Limit Perf Thrds

6000 20000
Contact
Stress
Stress 5000 0
(psi)
(psi)
4000 -20000

3000 -40000

2000 -60000
Seal Length
1000 -80000

0 -100000
0 1 2 3 4 0 1 2 3 4
Axial Position from Cplg Face (in) Axial Position from Cplg Face (in)

Case Number 1 Make up torque ft lb 4000


Interference gradient of selected case in/in 0 Total torque ft lb 12490
Interference at cplg face in 0.0067 Turns past HT turns 1.70
Friction coefficient in perfect threads 0.04 Internal pressure psi 1000
Friction coefficient unconfined 0.05 Pin tip axial stress psi -54858
no-e-tic

Friction coefficient pin face 0.12


Figure 16 Base case for 7 23 ppf L80 LTC connection.

26 of 38
28-April-2003

5.2 Full Scale Tests


As introduced at the beginning of this Section, the remaining question to be answered is: How much
of a difference will the torque ring solution really make? The immediately preceding sections have
described the relatively complete analytic models developed to answer this question. However, the
friction coefficients required by these models can only be obtained by physical testing. Controlled full-
scale tests of the incremental torque capacity obtained using multi-lobe torque rings were therefore
conducted.
Unlike field trials, controlled shop tests permit destructive testing, which provide valuable physical
understandings of the ring failure process, and allow measurements to both verify and calibrate
predictions of connection performance. To ensure the effects of size and connection type are correctly
accounted for, tests were conducted on both BTC and LTC connections for each of three sizes:
140mm (5), 178mm (7) & 244mm (9 5/8). Testing also provided the opportunity to evaluate the
ring installation behaviour and test the tendency of the torque ring end load to propagate a leak path
into the 8-round threads.
The following sections summarise the full scale testing procedures and results.
5.2.1 Equipment Configuration
Data allowing these objectives to be met was obtained by incremental make up of coupled and
threaded specimens with strain gauged torque rings installed. Tests were conducted using commercial
power tongs to apply torque and incrementally make up specimens while monitoring torque, turns at
the coupling face and axial strain with an automated data acquisition system. To evaluate the effect
that axial end-load (applied by the ring) might have on sealing response of 8-round threads, pressure
ports were installed at the mid-plane of the LTC couplings. This allowed injection of relatively low
hydraulic pressure at incremental points of make-up after shouldering, to test for initiation of shoulder
seal and potential loss of thread seal. After initial testing demonstrated no leakage at maximum
torque, pressure ports were placed in the field end threaded region approximately 12 mm (0.5) from
the pin tip. Drift reduction was checked by ID measurement when significant yielding was indicated.
5.2.2 Specimen Preparation & Procedures
Using this equipment tests were planned on 18 specimens where two pins and one box constitute one
specimen. Geometry and material properties for the specimens are as shown in Table 3.

no-e-tic

27 of 38
28-April-2003

Table 3 Description of specimens and rings


Type LTC LTC LTC BTC BTC BTC
Size mm (inches) 140 178 244 140 178 244
(5 ) (7) (9 5/8) (5 ) (7) (9 5/8)
Weight kg/m (ppf) 23.1 34.2 53.6 25.3 34.2 53.6
(15.5) (23) (36) (17) (23) (36)
Pin grade K55 K55 K55 K55 K55 K55
Drift mm (in) 121.08 158.52 222.63 121.08 158.52 222.63
(4.767) (6.241) (8.765) (4.767) (6.241) (8.765)
Ring material grade L80 L80 L80 L80 L80 L80
Ring yield stress MPa (ksi) 93,270 92,300 87,893 92,300 92,300 87,893
Nominal Cplg ID mm (inches) 130.20 167.51 232.89 132.44 169.95 236.12
(5.126) (6.595) (9.169) (5.214) (6.691) (9.296)
Ring OD mm (inches) 130.15 167.46 232.89 132.61 170.13 236.32
(5.124) (6.593) (9.169) (5.221) (6.698) (9.304)
Nom pin & ring ID mm (inches) 125.7 161.70 226.6 125.7 161.70 226.6
(4.950) (6.366) (8.921) (4.950) (6.366) (8.921)
Torque ring thickness (in) (0.087) (0.1135) (0.124) (0.1355) (0.166) (0.192)
Number of lobes 3 5 6 3 4 5

Specimens were mill end made up to the API power tight position to provide a defined mill end make
up position. Strain gauged rings were then installed, while visually monitoring the peak pressure
applied to the installation tool hydraulic ram. This pressure was used to indicate installation load. With
the rings installed and strain gauges connected to the data acquisition system, field make up was
performed until ring yield was detected by plateau of the torque curve except for the 9 5/8 BTC
specimens where the maximum torque was limited by the power tong capacity and the end load
inferred from the measured ring strain.
For the LTC connections, make ups were performed incrementally to allow application of pressure to
the ports through the connection to test for the presence of a thread leak path, either past the
shoulder seal, or past all or a portion of the threaded interval depending on which port configuration
was available.
5.2.3 Data Reduction and Results
The test results were interpreted to answer three basic questions:
1. What force is required to install the rings?
2. What are the friction coefficients for the LTC and Buttress threads and shoulder torque reactions?
3. When do the LTC threads begin to leak?
The following sections therefore present the data in this framework.
Ring Installation Force
The ring installation forces were predicted from the finite element analysis in terms of the radial forces
no-e-tic

generated in the lobe contact region. The rings were formed an amount sufficient to develop most of
the available spring force of the rings when installed in couplings of nominal diameter. Measurements
on the couplings showed they were all nearly nominal. The frictional interaction of the thread crests
and the saw-tooth external finish of the rings relate the radial force to the sliding installation force.

28 of 38
28-April-2003

12000

10000 5.5 in. BTC (3 lobes)


5.5 in. LTC (3 lobes)
8000 7 in. BTC (4 lobes)
Total 7 in. LTC (5 lobes)
radial 9.625 in. BTC (5 lobes)
installation 6000
9.625 in. LTC (6 lobes)
load
(lbf)
4000
Note:
To obtain axial sliding
2000 force multiply by
effective friction
coefficient of say 0.25
0
0 0.5 1 1.5 2
Radial installation displacement at largest outside
radius (mm)

Figure 17 Ring installation behaviour predicted by finite element analysis (FEA).

Table 4 Axial installation force from pressure gauge visual peak detection.
Coupling Peak Load
(lb)
7-B1 1789
7-B2 2012
7-B3 1789
7-R1 2683
7-R2 2236
7-R3 2236
9.6-R1 2683
9.6-R2 2571
9.6-B1 2683
9.6-B2 2683
9.6-B3 3578
5.5-R1 671
5.5-R3 2236
5.5-B1 1118
5.5-B2 894
Note: Coupling identification sequence <size><type: B = buttress; R = round><number>
no-e-tic

Figure 17 shows the FEA predicted response curves up to the point where the ring valleys are forced
into contact with the coupling and the radial forces increase in accordance with the ring hoop stiffness
mode. Table 4 shows the axial installation loads. By comparison an effective friction coefficient of
about 0.25 to 0.3 applies to the BTC rings and 0.4 to 0.5 to the LTC. The higher value for the LTC
connections may be attributed to interlocking of the relatively sharp thread crests into the ring
serrated surface. The anomalously high data point for ring 5.5-R3 may be attributed to the ring
coming into conformable contact with the threads generating hoop stress. Nevertheless, observations
and the recorded data demonstrate that in all cases the ring installation loads met the objective of
29 of 38
28-April-2003

ease of installation with reasonable gripping force available to counteract inadvertent or premature
dislodging. Gripping on the 8-round thread peaks was especially positive.
Connection Torque Capacity and Friction Model
An example of a typical graph showing torque turns response for a field end make up is shown in
Figure 18. Turns zero was set with the pin threaded into the box an arbitrary number of full one-turn
increments of the measured API power make up position. Mill end make-ups were controlled to land
nearly at the API power tight position. Therefore since the rings were all controlled to 0.9 inches in
length, the field shoulder point lands 0.1 inches beyond the API power tight position plus an integer
increment for the zero position. For buttress, API make up is 2.5 turns and for LTC 3 turns.

FMU7.0BPin1B-Cplg1B.MU2

25000

20000

15000 Yield plateau


Delta torque
Torque
(ftlb)
Shoulder point
10000

5000
Shoulder torque

0
1 1.5 2 2.5 3 3.5 4
Turns

Figure 18 Typical torque turns graph (7 inch BTC) showing shoulder and yield plateau.
The shoulder point and peak torque values were graphically interpreted, as shown in Figure 18, to
obtain the torque and turn co-ordinates for these points. Interpretation used the strain and time data
in addition to torque and turns and sought to filter out anomalous behaviour such as torque spiking
after dump valve trigger. Axial stress at peak torque for the 9 5/8 BTC was calculated using strain
data obtained from the axial strain gauges installed on the rings since the tong had insufficient torque
capacity to yield the rings in these connections.
Pre-test geometry measurements taken on the couplings and pins showed that interference on the 5
BTC was high while the 7 & 9 5/8 LTC was at or below the API minimum. The low interference
for the LTC provides a worst case configuration for sealing and peak torque and were therefore
accepted for the test program. In general, the wide range of interferences provided an extra degree
no-e-tic

of randomisation to test the validity of cross size and geometry predictions of torque capacity.
Observations of the torque ring shoulder and pin end surfaces did not show significant amounts of
galling on the surfaces rotated up to and even beyond the onset of yield. Coupled with a characteristic
flat yield plateau as shown in Figure 18, it was thus concluded that similar frictional response could be
expected from all the end surfaces up to the onset of global yield.

30 of 38
28-April-2003

Torque Model Description


As described in earlier sections, the torque models for buttress and 8-round threads differ. In buttress
threads, the torque ring end-load is reacted by load flanks that are nearly perpendicular. The radial
make up interference loads are reacted on the root to crest contact, therefore the torque contribution
obtained from the threads or shoulder torque is additive to the delta torque obtained from
interference with the torque ring.
In 8-round threads, the radial and axial loads are both reacted on the angled thread surfaces (30
relative to radial direction or 60 included angle), so that for normally made-up connections, the only
contribution to delta torque is from the pin end. Another way of stating this, is that the end load is
reacted in the threads by load transfer from the stab to the load flank, and therefore does not
contribute to the delta torque. This condition applies unless the initial shoulder torque is low. In which
case, the maximum total torque obtained with the torque ring for a connection that shouldered at the
hand tight position would be very similar for a connection with the API minimum radial interference. If
the selection of ring length and make up procedures does not allow such a mill end connection to
incrementally make up, then the drilling torque will be limited by this worst case connection, and the
radial or shoulder torque contribution can be ignored at the limit condition when high end-load exists
on the torque ring.
On this basis the following relatively simple relationships may be used to describe the delta torque
response of Buttress and total torque response of 8-round connections, when the rings are at or near
yield, and the ring axial yield load is equal to or less than the pin tip yield load. The friction
coefficients are of course specific to the compound type and all the other applicable vagaries of nature
and man under field conditions. These relationships do not capture the full range of torque response
and do not account for variations induced by radial interference, but nonetheless are expected to
provide an effective means to estimate available torque capacity over the full range of API sizes under
zero external load conditions.

p Equation 11
but = Fend + bt rpitch + shld rring + shld
2
p rt rpitch Equation 12
rnd min = Fend + + shld rring
2 cos
p Equation 13
rnd = Fend + shld rring + shld
2
y (ODring
2
IDring
2
)
Fend = y Aend =
4
Where,
but = torque from buttress thread form
Fend = end force from ring at yield
no-e-tic

p = pitch
shld = friction coefficient on torque ring shoulder (same for buttress or 8-round)
rring = radius of ring centre
bt = friction coefficient on buttress thread load flanks
rpitch = thread pitch radius at centre (half E1 diameter of 8-round)
shld = shoulder torque (buttress only)
rndmin = lower bound torque from 8-round thread form
31 of 38
28-April-2003

rnd = torque from 8-round thread form with normal mill end makeup
rt = friction coefficient on 8-round thread flanks
= 8-round thread flank angle relative to radial direction (30)
y = ring yield stress (assumes ring matl similar to pin)
end = ring end area
ODring = outside diameter of ring
IDring = inside diameter of ring
As shown, the friction coefficient between the pin end and torque ring is expected to be the same for
either thread form. This occurs because the dope entrapment conditions on the ring shoulder are the
same for either connection type therefore only three friction coefficients need be determined. The
remaining variables may be determined from geometry.
It should be noted that this model only applies to zero axial load conditions. However the effect of
axial load can be readily taken into account by considering it is reacted in the threads through the
same friction coefficients as the torque ring axial load in the manner described in Section 2.3.2.
Based on this model the torque shoulder delta torque values as shown in Appendix A were identified
and averaged for each connection size and type. The pre-shouldering make up friction response was
also considered to obtain best fit values, shown in Table 5, for the three friction coefficients based on
the known geometry and strength of the rings and connections (given in Table 3).

Table 5 Friction coefficients for API Modified thread compound (Topco) correlated to all
sizes
Location of interfacial region Symbol Value
Torque ring shoulder (same for buttress or 8-round) shld 0.12
Buttress thread load flanks (combined perfect & imperfect threads) bt .07
8-round thread flanks (very nearly the same as required to model radial rt .05
interference)
Note these are essentially short-term dynamic friction coefficients. For drilling, static long-term
coefficients would be more representative; however other factors such as micro movements or
fretting with rotating bending loads must also be considered.
Based on these parameters and the torque model described, Table 6 summarises the torque capacity
data in terms of measured and predicted or calculated torque values. The values shown do not have
any conservatism built in, but reflect the ability of the model to consistently predict response
independent of size and type. As shown, the calculated torque reasonably predicts the measured
response with the greatest variance occurring on the relatively thin 140mm (5.5) specimens.
no-e-tic

32 of 38
28-April-2003

Table 6 Summary of torque capacity results


LTC Connection Size & Weight 5 LTC 7.0 LTC 9 5/8 LTC
15.5 ppf 23 ppf 36 ppf
Average measured total torque (ftlb) @ control stress * 4500 10562 20550
Calculated total torque (ftlb) 4889 10469 21062
BTC Connection Size & Weight 5 BTC 7.0 BTC 9 5/8 BTC
17 ppf 23 ppf 36 ppf
Average measured delta torque (ftlb) @ control stress* 7605 17000 17246
Predicted delta torque (ftlb) 9654 16259 15300
"Typical" Shoulder torque (ftlb) 5000 7000 9000
Projected maximum total torque (ftlb) 14654 23259 43748
*Note: Control stress for measured averages taken at initiation of global yield except for 9 5/8 BTC
where the control stress it as indicated by axial strain gauge readings at tong stall torque.
Observations made of the post make up pin tip and deformation patterns showed that in general
significant drift reduction only occurred for rotation well beyond the onset of ring axial yield. The post
yield deformation only tended to produce significant drift reduction in the 5.5 inch connections where
the rings were both relatively thin and had IDs greater than the corresponding made up pin tips. (The
rings were all machined to the nominal pipe ID, however the combination of: smaller initial pipe ID,
diameter reduction as a result of make up strain and increased radius at lobe peaks all conspire to
make the pin tip ID fall inside the ring end face over all or a portion of the circumference.) This
combined with the often excessively large conical chamfer on the pin ends tended to allow the pin tips
to bend inward. This effect can be remedied by providing rings with slightly reduced ID but still well
within drift allowance.

LTC Stop Ring Torque Capacities

25000

20000

15000
Torque Ave measured
(ftlb) Calculated
10000

5000

0
5.5 LTC 7.0 LTC 9.625 LTC
Connection Size
no-e-tic

Figure 19 Summary of all LTC smooth ring results compared to model predictions.

33 of 38
28-April-2003

BTC Stop Ring Delta Torque Response

50000
9.625 BTC projected total based on ring yield
45000
stress.
40000

35000 9.625 BTC measured values limited by power


tong. Ring stress therefore less than yield as
30000 indicated by strain gauge.
Delta Ave meas Delta-torq
Torque 25000 Calculated Delta-torq
(ftlb) Projected Total Torque
20000

15000

10000

5000

0
5.5 BTC 7.0 BTC 9.625 BTC
Connection Size

Figure 20 Summary of all BTC smooth ring results compared to model predictions and
expected torque limits.
Leakage of LTC Connections
Pressure testing on the first few connections demonstrated that the shoulder seal was formed almost
immediately after contact. This was evident even where make up was typically halted at less than
10% of the available delta torque (delta torque = yield torque shoulder torque) and a pressure test
conducted. The pressure tests indicated conformable contact had been established even at these
modest contact stress levels (also ~10% of yield).
As mentioned above, leakage path tests were initially conducted with ports placed into the connection
J-section (the mid-plane of the coupling, between the two pin tips). Initial pressure tests on the LTC
connections indicated the thread leakage path did not propagate across the complete threaded
interval even at full make up. It was postulated that the excess thread compound might have been
plugging the leakage path. Therefore for all subsequent tests, a second set of pressure injection ports
were placed in the LTC couplings to intersect the pin threads approximately 12 mm (0.5) from the
pin tip and pressure applied to this port. (The port was filled with thread compound and plugged
during initial make up to prevent loss of dope pressure.)
With this configuration it was possible to test for the development of a leak path over two intervals of
the threads: from the port into the J-section and from the port to the coupling face. It was found that
leakage was initiated over the port-to-tip interval for make up levels at or near yield but in no case
was a leak path formed across the port-to-face interval. (If port-to-tip leakage did occur, the port in
the J-section was plugged to allow pressure application across the port-to-face interval.)
no-e-tic

This response was generally consistent with model predictions, as described in Section 5.1.2, for the
level of interference in the test samples. The model predicts early unloading of the threads near the
pin tip propagating back to nearly unload all the threads at full yield. However various simplifying
assumptions make the model leakage prediction somewhat pessimistic therefore the test results
demonstrate that designs staying within the envelop prescribed by the model will be conservative.
Furthermore, it appears the shoulder may often also provide a satisfactory seal, especially because
the increased end load tending to diminish the radial seal tends to promote the axial shoulder seal.

34 of 38
28-April-2003

Therefore the need to carefully manage torque ring length in the interest of preserving high seal
capacity in the threads is diminished.
5.2.4 Conclusions from Full Scale Tests
Full scale make up tests conducted using power tongs to apply torque demonstrated that significant
and relatively predicable torque increases can be achieved, for not only buttress, but 8-round
connections as well. However somewhat premature deformation occurred in pins and rings of the
140mm (5.5) specimens. This occurred because these very thin rings, manufactured to equal the
pipe body nominal ID, had a larger ID than the actual made up pipe. Combined with the API
specification allowing excessive chamfering to create almost conical pin tips on these sizes, portions of
the pin tip tended to slip off the ring. Therefore particularly for the lighter casing sizes, the ring
design basis may be modified, in accordance with Section 4.3.2, to allow the ring ID to be somewhat
less than the pipe nominal ID, perhaps up to 50% of the API drift allowance of 0.125 inches. (The
effect of this diameter change on running the bottom hole assembly must be carefully considered to
ensure nothing will hang up or be accidentally set.)
Torque yield data fit to a relatively simple predictive model, using only three friction coefficients,
demonstrated a more than satisfactory degree of accuracy. The ability to thus satisfactorily model the
results across 6 different connection types (3 sizes each of BTC & LTC) portends well for predicting
torque capacity for other sizes and weights. The friction response is of course a function of the thread
compound used (API Modified) and therefore needs to be monitored. This could take the form of a
relatively simple field usable fixture to obtain more accurate estimates of maximum torque for
situations demanding operation nearer the connection capacity.
The tendency toward reduced thread leakage in 8-round threads was somewhat less than expected
from predictions obtained using the relatively simple numerical model developed to assess
interference behaviour. Finite element analysis would provide a sound basis to develop a more precise
measure of this effect, especially in combination with external loads; however the excellent seal
performance of the shoulder suggests that the thread seal capacity is of secondary importance. In
fact, there may be an advantage to allowing some seepage as a means to bleed off trapped gas
pressure that might otherwise tend to collapse the torque ring under conditions of pressure cycling.
Installation of the multi-lobe rings was straightforward and appears to provide a satisfactory level of
gripping in either buttress or 8-round threads. Prototype commercial means to form the rings were
used and subsequent improvements of this process demonstrate that satisfactory quality and
efficiency are achievable.

6 Summary
Although innocently simple on the surface, implementation of torque rings to optimally increase the
torque capacity of stock or off the shelf API buttress and round threaded connections requires
consideration of a large number of variables. This report has considered how these variables influence
installation, operational performance and pin to box torque transfer in a complex interactive fashion.
The relatively loose tolerancing allowed by API specifications presents a particularly challenging design
problem. This wide range of dimensions makes it impossible to install straight cylindrical rings simply,
without running the risk of some rings being loose and others requiring excess interference. Similarly,
no-e-tic

the wide range of tolerance allows for a wide variation in radial interference in connections made up
to a fixed shoulder position with commensurate variation in galling and thread sealing performance.
The installation dilemma presented by the wide tolerance was solved by forming the torque rings into
multi-lobe shapes. This novel and unique approach effectively increases the circumferential or hoop
compliance, easily allowing the rings to tightly grip the coupling inside surface over the full allowable
variation in coupling diameter. At the same time, this architecture avoids requiring high installation
forces and risk of galling, jamming and buckling associated with a press-fit forces. The multi-lobe
rings are thus eminently suited to simple and rapid field installation requirements.

35 of 38
28-April-2003

The effect of tolerancing on interference variation is accommodated by a combination of biasing the


API field make up power tight position toward increased interference and providing a selection of ring
lengths to adjust for variations in mill end make up position. Furthermore, the shoulder seal is
expected to perform at least as well if not much better than the API thread seal, mitigating the
negative effect of possible low radial interference on seal quality.
This elegantly simple approach to torque ring design and implementation was proven in full scale shop
trials to provide increases in torque ring capacity for both BTC and LTC thread forms of practical use
in casing drilling. These tests also provided data to calibrate a simple analytic model allowing straight
forward prediction of torque capacity to other sizes.
The following Tables provide an estimate of expected torque capacities for several casing sizes. These
torques are calculated based on the set of parameters shown at the bottom of each table including a
safety factor. The ring ID value shown is for the unformed ring. Only the delta torque values are
shown for the buttress connections.

no-e-tic

36 of 38
28-April-2003

Table 7 Estimated delta torque for buttress connections.

Casing
ID Pipe Ring Delta Thread
Nominal Weight Pipe Wall ID Ring1
Body Wall Torque2 Type
OD

(Inches) ppf in in in in ftlb

4-1/2 9.50 0.205 4.090


10.50 0.224 4.052 4.006 0.116 3339 BT & C
11.60 0.250 4.000 3.970 0.135 3836 BT & C
13.50 0.290 3.920 3.914 0.163 4585 BT & C
15.10 0.337 3.826 3.826 0.207 5730 BT & C

5-1/2 14.00 0.244 5.012 4.978 0.121


15.50 0.275 4.950 4.935 0.143 6115 BT & C
17.00 0.304 4.892 4.892 0.165 6988 BT & C
20.00 0.361 4.778 4.778 0.222 9245 BT & C
23.00 0.415 4.670 4.670 0.276 11307 BT & C

7 17.00 0.231 6.538 6.497


20.00 0.272 6.456 6.439 0.129 9061 BT & C
23.00 0.317 6.366 6.366 0.166 11523 BT & C
26.00 0.362 6.276 6.276 0.211 14489 BT & C
29.00 0.408 6.184 6.184 0.257 17452 BT & C
32.00 0.453 6.094 6.094 0.302 20285 BT & C
35.00 0.498 6.004 6.004 0.347 23052 BT & C
38.00 0.540 5.920 5.920 0.389 25576 BT & C

9-5/8 29.30 0.281 9.063 9.052 0.126 16903 BT & C


32.30 0.312 9.001 9.001 0.152 20203 BT & C
36.00 0.352 8.921 8.921 0.192 25361 BT & C
40.00 0.395 8.835 8.835 0.235 30825 BT & C
43.50 0.435 8.755 8.755 0.275 35833 BT & C
47.00 0.472 8.681 8.681 0.312 40400 BT & C
53.50 0.545 8.535 8.535 0.385 49233 BT & C

Note 1: Ring ID is prior to forming of lobes, i.e., diameter equivalent to ring circumference.
Note 2: Total Torque = Delta Torque + Shoulder Torque

Parameter Units Value


no-e-tic

Friction coef on pin end 0.11


Friction coef in threads 0.07
Safety factor 1.25
Ring yield strength psi 80,000
Pipe body yield strength psi 80,000

37 of 38
28-April-2003

Table 8 Estimated total torque for LTC connections.


Casing Wght Pipe ID ID Shld API Delta Limit Min Limit Thrd
Nom Wall Pipe Ring1 Yield Opt Torq Min M/U Comb Type
OD Body Torq M/U Torq Torque Load
w/o Torq for Joint
SF Thread Pullout
Seal
(Inches) ppf in in in ftlb ftlb ftlb ftlb kips

4-1/2 9.5 0.205 4.090 4.033 2220 629 909 280 265 ST & C
10.5 0.224 4.052 4.006 2539 827 1197 370 303 LT & C
11.6 0.250 4.000 3.970 2974 2230 1095 1586 491 355 LT & C
13.5 0.290 3.920 3.914 3644 2710 1496 2172 676 435 LT & C
15.1 0.337 3.826 3.826 4432 2105 3066 962 529 LT & C

5-1/2 14 0.244 5.012 4.978 4095 1715 2471 756 399 LT & C
15.5 0.275 4.950 4.935 4871 2207 3185 978 474 LT & C
17 0.304 4.892 4.892 5596 3410 2686 3881 1195 545 LT & C
20 0.361 4.778 4.778 7022 4200 3916 5678 1762 683 LT & C
23 0.415 4.670 4.670 8373 5027 7313 2286 815 LT & C

7 17 0.231 6.538 6.497 5385 ST & C


20 0.272 6.456 6.439 6930 3014 4332 1318 528 LT & C
23 0.317 6.366 6.366 8626 4350 4398 6333 1935 657 LT & C
26 0.362 6.276 6.276 10322 5110 6057 8740 2684 786 LT & C
29 0.408 6.184 6.184 12055 5870 7703 11141 3438 918 LT & C
32 0.453 6.094 6.094 13751 6610 9267 13431 4165 1047 LT & C
35 0.498 6.004 6.004 15447 7340 10783 15664 4881 1176 LT & C
38 0.540 5.920 5.920 17030 8010 12158 17697 5540 1297 LT & C

9-5/8 29.3 0.281 9.063 9.052 11559 4495 6440 1945 637 LT & C
32.3 0.312 9.001 9.001 13538 6396 9172 2776 746 LT & C
36 0.352 8.921 8.921 16091 9359 13439 4080 887 LT & C
40 0.395 8.835 8.835 18835 7270 12485 17954 5469 1038 LT & C
43.5 0.435 8.755 8.755 21388 8130 15338 22087 6749 1179 LT & C
47 0.472 8.681 8.681 23750 8930 17931 25853 7922 1309 LT & C
53.5 0.545 8.535 8.535 28409 10470 22916 33124 10208 1566 LT & C

Note 1: Ring ID is prior to forming of lobes, i.e., diameter equivalent to ring circumference.

Parameter Units Value


Friction coef on pin end 0.11
no-e-tic

Friction coef in threads 0.04


Safety factor 1.25
Ring yield strength psi 80,000
Pipe body yield strength psi 80,000

38 of 38

S-ar putea să vă placă și