Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

SPE/IADC-184689-MS

Maximizing Tubular Capacity and Well Integrity Under Near-Yield Loading

J. Nowinka, G. Meijer, and S. Miller, Noetic Engineering 2008 Inc.

Copyright 2017, SPE/IADC Drilling Conference and Exhibition

This paper was prepared for presentation at the SPE/IADC Drilling Conference and Exhibition held in The Hague, The Netherlands, 14–16 March 2017.

This paper was selected for presentation by an SPE/IADC program committee following review of information contained in an abstract submitted by the author(s).
Contents of the paper have not been reviewed by the Society of Petroleum Engineers or the International Association of Drilling Contractors and are subject to correction
by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers or the International Association of Drilling Contractors,
its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers or
the International Association of Drilling Contractors is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations
may not be copied. The abstract must contain conspicuous acknowledgment of SPE/IADC copyright.

Abstract
Achieving well integrity demands knowing how much loading your tubulars can withstand, but what
happens when near-yield loading causes unexpected deformations that are not predicted by code design
and conventional elastic design theory? It begs the question, are harsh operating conditions and narrowing
design safety margins increasing the potential for well failures? Alternatively, can tubulars be safely pushed
even harder without compromising well integrity?
This paper raises industry awareness, demonstrates material test results, and proposes steps to account
for possible near-yield load effects so that well designs can fully use the available tubular material capacity
and be cost effective without compromising the long-term integrity of the well. Examples of near-yield load
scenarios leading to substantial plastic deformation observed in laboratory testing are presented, and ways
to determine when those effects need to be considered in elastic design methods are proposed.
The paper illustrates the effects of near-yield loading on tubular deformations using results of mechanical
testing on a P110 material sample under cyclic uniaxial and multiaxial load. Uniaxial load cycles with stress
amplitude below the rated yield strength showed substantial growth of plastic strain in only a few cycles.
Combined-load tests on small-scale tubular P110 specimens loaded with axial force and internal pressure
cycles also demonstrated pronounced near-yield load effects.
We discuss the significance of material behaviors that trigger plasticity at stresses below the measured
yield strength and contribute to severity of near-yield load effects; such as changes in the stress-strain
response due to elevated temperature exposure, directional dependence of material properties, and cyclic
softening. We draw conclusions on how tubular material properties and operational loading extremes need
to be assessed and balanced to maintain structural integrity of wells in anticipated operational scenarios.
To maximize capacity of completion tubulars, the complexity of near-yield load effects needs to be
characterized with technically rigorous yet practical means. This paper points to relatively simple material
tests for determining the propensity for near-yield loading effects of common OCTG materials, and suggests
initial considerations for incorporating near-yield load considerations in tubular design.

Tubular Design – Peace of Mind or Challenge?


Well designers have been successfully using elastic design bases to evaluate and select Oil-Country-Tubular-
Goods (OCTG) materials and pipe diameter-to-thickness (D/t) ratios for casing and tubing pipe-connection
2 SPE/IADC-184689-MS

systems. Fundamentally, elastic tubular design bases assume a linear (proportional) relationship between
the applied stress and corresponding strain (deformation) in a structure subjected to either monotonic or
cyclic loads up to the material yield strength. In an elastic design framework, the concepts of load-path-
independence and superposition are frequently employed, and can safely be applied through the use of
appropriate safety factors even when small local plastic deformations exist. This greatly simplifies casing
body design and integrity considerations. Connection performance must also be assessed to ensure the
complete tubular system performs according to the design intent.
The yield strength of a tubular material is conventionally defined as the stress corresponding to a small
strain offset relative to the linear stress-strain curve (as described in ASTM E 8 – 16A), or the stress measured
at a prescribed strain value in a pull-to-failure test (as described in API Specification 5CT). Yield strengths
are typically characterized through uniaxial material tests, either in the longitudinal (axial) or transverse
(circumferential) direction of the pipe. API 5CT defines basic OCTG material characterization requirements
and mechanical response specifications for typical grades. In certain situations, casing designers also
consider more in-depth material characterization to ensure materials satisfy application-specific structural
and environmental performance requirements.
Casing and tubing loads will fluctuate during installation, completion, and operation of wells. The cyclic
response of the material influences how the tubular structure will respond, but API specifications do not
require characterization of these cyclic properties. Fortunately, the concept of superposition can be applied
in most cases because the assumption of linear elastic material response is valid. An example of linear
elastic material response under applied stress-controlled cyclic loading is illustrated in Figure 1(a). While
casing and tubing strings subjected to low and moderate cyclic load amplitudes (relative to the material
yield strength) will respond elastically, there are fundamental aspects of cyclic material response that must
be considered in the well design process if repeated cyclic loading to high stress levels is anticipated during
the life of the well.

Figure 1—Examples of three different types of fundamental cyclic material responses under fully-
reversed stress-controlled loading at 90% of measured yield strength (5 cycles): (a) Linear elastic
response (in virgin P110 material at 20°C), (b) Stable cyclic plastic response (in pre-cycled P110
material at 180°C), and (c) Increasing plasticity with cycling (in pre-cycled P110 material at 20°C).
SPE/IADC-184689-MS 3

While cyclic tubular mechanical design processes traditionally include some consideration of fatigue
resistance, this paper focuses on a separate facet of cyclic material response that can lead to large-scale
tubular deformations even when tubulars are never loaded to levels above the yield strength.
Figure 1(b) and Figure 1(c) illustrate two different material responses to such cyclic loading. Figure 1(b)
shows the response of a material that has a slightly non-linear monotonic stress-strain characteristic. Under
cyclic loads, a small amount of plastic strain is accumulated with each stress reversal in the initial cycles, but
the strain range increment is diminishing in each cycle, and thus the material stress-strain response becomes
stable after a few cycles – in the case illustrated in Figure 1(b), after approximately five load cycles. The
cyclic response example in Figure 1(c) is entirely different from Figure 1(b) because the plastic strain range
grows at an approximately constant rate with each loading reversal (in this example, under stress amplitude
of 90% of the yield strength) and does not stabilize. The cyclic material behavior displayed in Figure 1(c)
demonstrates cyclic softening of the tested material. This characteristic feature of the material behaviour
gives rise to our definition of near-yield loading effects.
In this paper, we define near-yield load effects as cyclic growth of plastic strain range in a cyclically-
softening material at stresses below the material yield strength. The paper objective is to raise industry
awareness of near-yield load effects and their possible implication on well integrity. We consider this
not in the context of questioning the validity of traditional design bases, but rather as a mechanism for
understanding the limits of elastic casing and tubing design and where more complex material behavior
should be anticipated. Specifically, where pressures and temperatures are high, cyclic loads may occur, and
practical considerations may be driving a tendency for lower safety factors, we should confirm whether the
elastic design assumptions are still valid considering the relevant material mechanical response.
Subsequent sections of this paper describe:

• Background and discussion of fundamental material properties and loading conditions that trigger
and contribute to near-yield load effects;
• Preliminary experimental studies in which near-yield load effects were explored in P110 material;
and
• Strategies to account for near-yield load effects at the tubular design stage.

Background on Material Behaviors Contributing to Near-Yield Load Effects


Individually, the fundamental material behaviors leading to the linear elastic, stable cyclic plastic, and cyclic
softening stress-strain responses shown in Figure 1 are generally well understood (Skelton et al, 1997;
Broom and Ham, 1957), but are typically not discussed as part of the elastic design method employed in
casing and tubing design. In the following discussion, we describe our experimental set-up and draw on the
results of physical tests to illustrate facets of material response that contribute to near-yield load effects.

Rounded Yielding Behavior


For this discussion of near-yield load effects, it is important to revisit the assumption of a linear elastic
stress-strain relationship at stresses up to the yield strength. While this assumption is valid for most
OCTG materials at ambient temperature, it is not universally true. For instance, some materials exhibit a
rounded transition between the elastic and plastic regions in their stress-strain response (i.e., as applied
stress approaches the yield strength) at ambient temperature. Similarly, the yield character of most OCTG
materials becomes rounded at elevated temperatures. For these cases, the linear elastic stress range, up to a
stress known as the proportional limit or elastic limit, can be considerably smaller than the yield strength.
Frequently, room temperature tensile stress-strain responses of low-alloy quenched-and-tempered
carbon-steel tubular materials demonstrate a linear-elastic increase in stress followed by a stress plateau, as
shown in the 20°C P110 stress-strain curve in Figure 2, and then strain hardening at larger strain magnitudes.
4 SPE/IADC-184689-MS

The elastic (proportional) limit is very similar to the yield strength measured at 0.6% strain for this shape of
stress-strain curve. When the elastic limit is equal to or greater than the yield strength, the material behavior
is anticipated to be elastic for applied stresses up to the yield strength.

Figure 2—Example stress-strain curves for virgin P110 material samples illustrating
sharp and rounded yielding characteristics at 20°C and 180°C, respectively.

At 180°C, the same material demonstrates a completely different transition between the linear elastic
and plastic ranges: the transition is not sharp, but instead rounded, as also shown in Figure 2. Elevated
temperature causes rounding of the yield character because increased molecular motion promotes the
progress of some dislocations within the metal crystal structure leading to early onset of plastic deformation.
The key facet to be noted here is that at 180°C, the measured elastic limit is approximately 640 MPa and
the yield strength (as measured at 0.6% elongation) is 796 MPa. Thus, the elastic limit is only 80% of the
yield strength.
Once stresses exceed the elastic limit in a material testing sample or a tubular structure, plastic strain
(permanent deformation) occurs. At 180°C, the P110 accumulates 0.188% of plastic strain as the stress is
increased from the elastic limit to the yield strength. Table 1 shows that even at stresses equal to 90% and
95% of the yield strength, appreciable plastic strain will accumulate at 180°C. By contrast, the plastic strain
will be negligible if the material is loaded to yield strength at 20°C.

Table 1—Plastic strain estimates for P110 material sample under


uniaxial loading to stress levels from 90% to 100% of API yield strength.

20°C 180°C

% of measured
Stress (MPa) Plastic strain (%) Stress (MPa) Plastic strain (%)
yield strength

90 803 0.001 716 0.012

95 848 0.002 756 0.047

100 893 0.005 796 0.188

From a tubular design perspective, small amounts of plastic strain alone are generally not a concern, but
their implication on subsequent tubular response must be clearly understood before such behaviors can be
SPE/IADC-184689-MS 5

factored into casing and tubing design. For instance, strain-based design practices are generally employed
to manage plastic deformations in thermal well tubulars (Nowinka et al., 2007). For most designs using
an elastic basis, the early onset of plasticity observed in rounded material responses, such as the one in
Figure 2, is avoided by introducing an appropriate safety factor on the applied well construction and service
loads, thus keeping the material in its elastic response range. Typical tubular materials also exhibit strain
hardening behavior and considerable ductility, which will augment the ability of the material to initially
withstand overload conditions and limit deformation.

Cyclic Material Behavior


Plastic deformation, and particularly reversed plastic loading, has a number of effects on subsequent material
response due to the accumulation and motion of dislocations within the metal crystal structure. OCTG
materials typically strain harden under monotonically increasing stress, and strain hardening will limit the
strain achieved for a given applied stress. Under cyclic (or reversed) plastic loading, the effective yield
strength drops (i.e., the Bauschinger effect) due to back stresses in the material and the cyclic stress-strain
curve becomes more rounded than the initial monotonic response (Ellyin, 1997; Kaiser et al., 2008). Plastic
deformation and the resulting change in material response continue to influence material performance
under subsequent loads unless the material is subject to processing, consisting of a combination of elevated
temperature and time, to recover the original material behavior.
Referring again to Figure 1, cyclic plasticity observed during 180°C fully-reversed cyclic loading to 90%
of the yield strength (shown in Figure 1(b)) is characterized by hysteresis loops, where the width of the
loop is the plastic strain range and the area of the hysteresis loop is the irrecoverable strain energy or work
per load cycle driving the plastic deformation. In contrast, in Figure 1(a), the elastic energy is recovered
when the specimen is unloaded. Compared to the relatively stable (i.e., constant plastic strain range) 180°C
hysteresis loops shown in Figure 1(b), the 20°C plastic cyclic loading in Figure 1(c) shows a growing plastic
strain range that occurred after plasticity was triggered in the material through pre-cycling. This behavior
is significantly different than the cyclic behavior at the same temperature to the same loading level shown
in Figure 1(a), which was not pre-cycled.
To evaluate the cyclic softening behavior of the sample P110 material, specimens were cyclically loaded
for 30 cycles at a constant axial strain range of 1.2% and the resulting stress range was monitored. Given
that cyclic softening behaviour can be demonstrated in either a cyclic constant-stress-range test when the
corresponding strain range increases; or in a constant-strain-range test when the corresponding stress range
decreases, we chose to conduct these 30-cycle tests in strain control, instead of stress control, to avoid
accumulation of a substantial amount of plastic strain and possible specimen failure.
Figure 3 compares the evolution of stress amplitude during cycling for tests completed at 20°C and 180°C.
During the 20°C test, the stress amplitude decreased rapidly in the first five cycles and then continued to
decrease at a moderate pace to the end of the 30 cycle test. In contrast, during the 180°C test, the stress range
decreased moderately in the first few cycles and then displayed a stable stress range for the remainder of
the test. These results are consistent with the relatively stable hysteresis loops observed at 180°C in Figure
1(b) and the rapidly growing 20°C hysteresis loops of Figure 1(c). At 20°C, this material exhibits what is
known as cyclic softening behavior, and at 180°C it exhibits stable cyclic behavior. Materials may also show
cyclic hardening behavior for which cyclic stress range increases with number of load cycles at constant
strain range.
6 SPE/IADC-184689-MS

Figure 3—Cyclic softening in 1.2% constant-strain range tests at 20°C and 180°C.

From a casing and tubing design perspective, the use of materials that exhibit cyclic softening behavior
should be managed with care when substantial exposure to cyclic operating loads is anticipated. If stresses
are partially or completely governed by load-controlled parameters such as pressure, a cyclic softening
material is likely to exhibit incremental plastic deformation with every load reversal (near-yield load effect).
This is clearly outside of the intent of an elastic design basis, and invalidates many of the underlying
assumptions for using the principle of superposition for predicting pipe response. In such situations, the
material behavior should be understood, the well conditions predicted, and an appropriate tubular design
and well operating strategy for mitigating near-yield load effects created.

Experimental Specimen and Apparatus


Our choice of the test specimen architecture and experimental set-up configuration was dictated by our
goal to broadly explore OCTG near-yield effects under uniaxial and multiaxial loading. We chose a tubular
specimen because it enabled application of combined loading resultant in a multiaxial stress state. The
experimental specimen and associated testing and control system were specifically designed to enable
tensile and compressive axial stresses to be combined with known tensile hoop stresses, and for any stress-
controlled testing sequence to be followed. The results of such tests are hoop and axial strains measured by
surface strain gages while hoop and axial stress are generated through the application of internal pressure
and axial force to the specimen. We recognize that a more complete multiaxial program would involve
application of both external and internal pressure, and we may extend the test capability in the future as
required.
All the results shown in this paper are from tests on full-body tubular specimens machined from
small-diameter P110 tubing. The full-body specimen was designed to have a substantially uniform strain
distribution and resist compressive buckling in its reduced ("gage") section. The D/t ratio for this specimen
(in its gage section) is 36. A photo of an instrumented specimen is shown in Figure 4.
SPE/IADC-184689-MS 7

Figure 4—Full-body tubular specimens made of P110 tubing for combined-load evaluation.

Testing was done in a custom load frame configured to enable loading to selected stress targets under
axial tension-compression and internal pressure, with specimen temperature controlled at set points of 20°C
(ambient) and 180°C. The system incorporated precise feedback control of applied load, strain, and internal
pressure and custom-designed grips that eliminated backlash under load reversals. Multiple strain gages
were used to ensure strain uniformity in the specimen gage section and to obtain redundant measurements of
local axial and hoop strains. Specimen heating was applied using an electrical resistance heater, which was
controlled through feedback from resistance temperature sensors on the specimen to maintain a uniform
temperature distribution within the reduced section.

Multiaxial Yield Characterization


As a first step, we characterized the load response of the P110 tube material that would subsequently be used
to evaluate near-yield load effects. API Technical Report 5C3 design equations assume OCTG mechanical
response is isotropic and the multiaxial yield state is adequately defined by the elliptical von Mises envelope.
To validate this assumption, P110 monotonic stress-strain behavior was characterized at 20°C and 180°C
under a total of five ratios of axial stress to hoop stress (achieved by axial tension/compression combined
with internal pressure). Such characterizations are difficult to do with full-scale pipes, but the load frame
and internal pressure control systems and small samples employed for this study permitted such tests with
a high degree of control.
In traditional mechanical engineering theory, we refer to the von Mises yield criterion (also known as
the maximum distortion energy criterion) as defining a combined stress "envelope" within which a material
will remain elastic (Boresi and Sidebottom, 1985). The applied axial-to-hoop stress ratios and measured
P110 von Mises yield strengths (σYield) and elastic limits (σElast) are listed in Table 2.

Table 2—Measured von Mises yield strengths and elastic limits for tubular
P110 specimens at five ratios of axial to hoop stress and two temperatures.

20°C 180°C

von Mises von Mises


Axial Stress to von Mises Yield von Mises Yield
Elastic σElast/σYield (%) Elastic σElast/σYield (%)
Hoop Stress Ratio Strength (MPa) Strength (MPa)
Limit (MPa) Limit (MPa)

1:0 (Axial Tension) 893 893 100 796 640 80.4

1:1 913 825 90.9 842 715 84.9

0:1 (Hoop Tension) 914 820 89.7 818 670 81.9

−1:1 880 825 93.8 771 675 87.5

−1:0 (Axial Compression) 899 825 91.8 808 625 77.4


8 SPE/IADC-184689-MS

The von Mises stress-strain curves for axial tension, axial compression, and hoop tension are shown in
Figure 5. The equivalent strain and von Mises stress are used for the comparison basis because a multiaxial
stress state consisting of hoop stress and radial stress is produced in the pipe wall when internal pressure
is applied.

Figure 5—Measured axial tension, axial compression and hoop tension


stress-strain curves for tubular P110 specimens at 20°C and 180°C.

The 20°C stress-strain curves show that the yield strengths under axial tension, axial compression and
hoop tension are all similar, but the axial compression and hoop tension curves have a rounded yield as
compared to the sharp yield point under axial tension. The 180°C results are consistent in yield strength and
yield character. At 180°C, the yield character is round and all three stress ratios have similarly low elastic
limits. Where elastic limits at 20°C are as low as 90% of the yield strength, the elastic limit drops to as little
as 77% of the yield strength at 180°C.
Note that for the hoop tension stress-strain curve shown in Figure 5, the hoop stress on the specimen
ID calculated from applied internal pressure is shown as a function of strain measured on the OD of the
specimen. Even for the large ratio of diameter to wall thickness of the tubular specimen (D/t = 36), there is a
small variation in hoop stress and strain through the wall thickness. Elastic hoop stress on the specimen OD
was calculated using Lamé equations and was found to be 94% of the hoop stress on the specimen ID. The
hoop stress and strain gradients lead to initial plasticity at the ID and the zone of plasticity grows through
the wall as the stress is increased. This non-uniform yielding leads to a rounded stress-strain response even
if the base material, tested under a uniform state of stress, produces a sharp yield point.
The 20°C and 180°C measured yield strengths and elastic limits are compared to best-fit yield and elastic
limit envelopes in Figure 6. The performance envelopes are based on the von Mises elliptical envelope and
were calculated following the assumption that the envelopes are centered in the axial-hoop stress space.
SPE/IADC-184689-MS 9

Figure 6—Yield and proportional limit test data and corresponding von Mises envelopes for tubular P110 material specimens.

The results listed in Table 2 and shown in Figure 5 and Figure 6 indicate the following behavior from
the tested P110 material sample:
1. The 180°C stress-strain responses have much rounder yield behavior than 20°C responses at all stress
ratios, and therefore considerably lower elastic limit than yield strength at each stress ratio. Elastic
limits at 180°C are 77% to 88% of the yield strength.
2. Axial tension at 20°C is the only material response that has a distinct yield point with equal yield
strength and elastic limit. The elastic limit of the other 20°C tests ranges from 90% to 94% of the
yield strength.
3. The elliptical von Mises envelope is a reasonable fit to the yield strength data.
4. The von Mises envelope is a good fit to multiaxial elastic limit data except under axial tension, at
which the elastic limit and yield strength are equal. For this P110 material the elastic limit from an
axial tensile test is a poor representation of the elastic limit at other stress ratios under a combination
of axial load and internal pressure.
From a well design perspective, the results shown in Figure 6 are informative in that they validate the
use of a uniaxial yield strength as the basis for drawing a von Mises yield ellipse for this specific tubular
sample. They do, however, also point out a substantial difference in the yield character (shape of the stress
strain curve) and the elastic range between uniaxial tension response and all of the other loading modes
(including uniaxial compression). Where the performance of casing designs may be tied to the material
response in the near-yield region, more comprehensive material characterization than that demanded by
API specifications may be warranted.

Near-Yield Load Effects under Uniaxial Load


Given the material elastic limit and yield strength properties identified in the preceding discussion, we
conducted an initial uniaxial evaluation of near-yield load effects in the same P110 material. This initial
testing simulated a load path with a combination of fully-reversed axial load cycles at 20°C and 180°C. The
first test, with stress-strain results shown in Figure 7, consisted of the following load path steps:
1. Five load cycles at 20°C with constant stress amplitude equal to 90% of actual 20°C yield strength,
2. Five load cycles at 180°C with constant stress amplitude equal to 90% of actual 180°C yield strength,
and
3. Five load cycles at 20°C with constant stress amplitude equal to 90% of actual 20°C yield strength.
10 SPE/IADC-184689-MS

Figure 7—Measured stress-strain response of tubular P110 specimen subjected to cyclic loading at 90% of yield strength.

The applied stress amplitudes used for this testing were equal to the values listed in Table 1 corresponding
to 90% of the measured yield strength at the respective temperatures.
Key facets of the results shown in Figure 7 include:

• Elastic cyclic stress response during the first set of five cycles at 20°C. The response remains elastic
because the maximum applied 803 MPa axial stress, equal to 90% of actual 20°C yield strength,
does not exceed the elastic limit of 893 MPa.
• Stable cyclic stress response during the set of five cycles at 180°C, where the applied stress
amplitude of 716 MPa does exceed the 640 MPa elastic limit, introducing a small amount of plastic
deformation. The hysteresis loops became nearly constant within the five applied load cycles,
indicating that the tested P110 sample was cyclically stable at 180°C.
• Cyclic softening behavior in the final set of cycles at 20°C, which was obvious by the immediate
increase in strain range and noticeable incremental strain with each subsequent cycle, indicating
that the tested P110 sample cyclically softens at 20°C.
These initial results are very interesting because they highlight two completely different responses for
the same material at 20°C, with the difference simply being the (sub-yield strength) loading history of the
specimen. The virgin material exhibited fully elastic response (as anticipated), but unstable plasticity was
observed at the same load level and same temperature after generating a small amount of cyclic plasticity
during cycling at elevated temperature.
Given the surprising results of cyclic testing at 90% of yield strength, we proceeded to assess the influence
of the applied load magnitude, expecting even more pronounced near-yield load effects when we increased
the stress amplitude to 95% of actual yield strength. The applied stress amplitudes are again listed in Table
1 and the results are shown graphically in Figure 8. The increase in stress amplitude from 90% to 95% of
yield strength led to the introduction of a slight amount of hysteresis in the initial 20°C cycles, a greater
plastic strain range for the 180°C hysteresis loops, and a much more rapid increase in strain range in the
final 20°C load cycles.
SPE/IADC-184689-MS 11

Figure 8—Measured stress-strain response of tubular P110 specimen subjected to cyclic loading at 95% of yield strength.

The comparison of the plastic strain range during the two cyclic tests (with stress amplitudes of 90%
and 95% of yield strength) in Figure 9 shows the increase in the plastic strain range during the 180°C load
cycles and a particularly large increase during the final five 20°C load cycles. In both cases, the plastic
strain range is still increasing at the end of the test and the incremental increase in plastic strain with each
cycle is relatively constant. These results suggest that the rate of growth in plastic strain range is dependent
on both the plastic strain accumulated during the 180°C cycling (indicated by the width of the hysteresis
loops) and the incremental plastic strain during the load cycles showing cyclic softening. They also indicate
that lower loading levels would reduce the rate at which the plastic strain range will grow, and point to a
stress level where cyclic incremental plasticity is completely avoided for the entire loading sequence, even
if the material cyclically softens.

Figure 9—Increase in measured strain range as a function of loading history at fully-reversed


stress levels of 90% and 95% of yield for tubular P110 specimen under uniaxial loading conditions.
12 SPE/IADC-184689-MS

To confirm that the fundamental material behavior was similar in tubing, casing and plate materials, we
performed supplementary tests on strip specimens cut from material samples of P110 casing and QT100
plate, and observed virtually identical responses to near-yield loading.
If the cyclic material behavior described in the preceding paragraphs and figures occurred in any single
direction in an in-service tubular, it would manifest as a gradual change in the shape of the tubular
(pipe ballooning or ovalization, elongation/contraction, and any associated instability effects), or as a
greatly reduced fatigue resistance relative to what might be anticipated from an elastically-loaded tubular.
Fortunately, the test results shown here likely represent an extreme loading situation, and we are now in a
position to refine our tubular performance prediction based on more field-representative loading paths.

Near-Yield Load Effects under Combined Axial and Internal Pressure Load
As a next logical step to this investigation, we employed the same tubular specimen design and material
sample to conduct preliminary cyclic multiaxial testing to learn about how near-yield load effects would
manifest under a complex stress state. Our starting expectation was that the effects would be similar to the
uniaxial tests. At a minimum, we hoped to demonstrate the types of responses that might be anticipated in
subsequent testing efforts and considered in casing design.
For the first tests we conducted, we noted that near-yield load effects have previously been observed as
test-string buckling and pipe body ballooning during connection qualification tests to the ISO 13679/API
5C5 CAL IV specifications. For this reason, we founded our multiaxial test loading path on this connection
qualification basis, modifying the sequence to remove all external pressure steps because our test apparatus
did not have external pressure capability. The load paths corresponding to the CAL IV test (according to
2014 draft of API 5C5) and the performed multiaxial near-yield load test are summarized in Figure 10.
Differences between the two load paths are highlighted in bold text. For the purposes of this discussion,
we align with relevant connection testing specifications on terminology, and reference tensile axial stress
combined with tensile hoop stress as "quadrant 1" of the von Mises envelope, and compressive axial stress
combined with tensile hoop stress as "quadrant 2". Omission of external pressure load points reduced the
severity of cyclic loading by eliminating part of Series A cycles in quadrants 3 and 4 (i.e., hoop stress is
not fully reversed), but the near-yield load effects in the two-quadrant test were expected to have similar
trends as those in a four-quadrant test. We recognized that the relative complexity of this loading path would
make deduction of multiaxial parameters governing near-yield load effects challenging, but expected that
an initial demonstration would provide high value in defining next steps in our investigation.
The CAL IV connection test load path described in ISO 13679/API 5C5 is based on a connection
evaluation envelope (CEE) that is typically provided by the connection supplier, and the test load envelope
(TLE) that results from prescribed percentages of the CEE and axial stress limitations. In this study, and
in many connection tests, the CEE coincides with the pipe body performance envelope defined by the von
Mises envelope for first yield on the pipe body ID in quadrants 1 and 2 and by pipe body collapse in quadrants
3 and 4. CEE calculations are based on API Technical Report 5C3, using the actual measured material yield
strength. For the tests reported here, we calculated the load points according to the 2014 draft of API 5C5.
SPE/IADC-184689-MS 13

Figure 10—Comparison of load steps from ISO 13679/API 5C5 CAL IV connection qualification
specification and multiaxial cyclic conditions applied to tubular P110 near-yield load specimen. Load steps
occurring in connection qualification path not replicated in near-yield load testing are show in bold text.

The first load cycle applied follows a path around quadrants 1 and 2 of the 20°C (ambient-temperature)
80% TLE where von Mises stress never exceeds 80% of the measured 20°C yield strength. Figure 11(a)
shows the load path (i.e., applied axial and hoop stresses and their sequence) within the 80% TLE and the
CEE; the path starts in quadrant 1 and moves counterclockwise to quadrant 2, and then returns to a zero
stress state. The resulting strains are compared to the calculated elastic 80% TLE strains and CEE strains
in Figure 11(b), where these strains are measured on the specimen OD. The strain results show that when
applied axial and hoop stress remain on or within the 80% TLE, the strains remain elastic and are accurately
predicted using Lamé equations and Hooke's law.

Figure 11—Applied stress load path (a) and corresponding measured elastic strain response (b) in tubular P110 sample
during initial 20°C loading step corresponding to initial low-stress API 5C5 CAL IV load step. (First load cycle of Series B)
14 SPE/IADC-184689-MS

The second load cycle follows a counterclockwise (CCW) path around the 20°C 95% TLE and returns
clockwise (CW) along the same path, again at ambient temperature. The stress path is shown in Figure
12(a) and the measured strains are compared to the calculated elastic 95% TLE and CEE strain envelopes
in Figure 12(b). In comparison with the first cycle (when stresses remained on and within the 80% TLE and
strains remained elastic), plastic deformation is observed when stresses follow the 95% TLE. In the first
(CCW) part of the cycle, measured strains (shown in purple) began to exceed the calculated elastic 95%
envelope (dashed blue) at high combined axial and hoop stress in quadrant 1, and clearly grew outside of
the elastic envelope as combined hoop and compressive axial stresses were applied in quadrant 2. When
the load path reverses, strains (shown in green) continue to increase, particularly as high hoop stresses are
applied. At the end of the cycle, a net plastic deformation of the specimen is observed, with 0.13% plastic
axial strain and 0.05% plastic hoop strain.

Figure 12—Applied stress load path (a) and corresponding elastic-plastic strain response (b) in tubular P110 sample
during second 20°C loading cycle corresponding to 95% API 5C5 CAL IV load step. (Second load cycle of Series B)

During the second load cycle, plastic deformation likely initiated due to the difference in yield properties
in different loading directions. Recall from Figure 6 that at applied stress ratios other than 1:0 (axial tension),
the measured elastic limit for this material sample is lower than the measured axial tensile yield strength.
In this part of the combined load test, we expect plasticity arose from the stress-ratio-dependence of the
elastic limit envelope.
While not described fully here, the strains that developed in the tubular specimen as the test progressed
behaved in a similar manner as shown in Figure 12. Plasticity continued to increase in subsequent 20°C
and elevated-temperature cycles. Given the small number of cycles undertaken at individual conditions, it
was difficult to isolate cyclic softening behavior from plastic strain that would be anticipated in a material
responding in a cyclically stable manner. We suggest that future investigation consider more cycles at
individual test conditions to enable the stability of the cyclic response to be assessed.
To illustrate the net effect of the testing sequence applied to the test specimen, we introduce Figure 13,
which indicates the measured specimen strains incurred during the final Series A load cycle. This cycle
follows a counterclockwise and subsequent clockwise path, with the applied stress on and within the 95%
SPE/IADC-184689-MS 15

TLE at 20°C. The cycle starts with no applied axial or hoop stress, and the specimen initially has plastic
axial strain of 0.29% and hoop strain of 0.30%. At one point in the cycle, the measured total hoop strain
achieves a maximum of nearly 1.5%. Similarly, the axial strain range traversed during the cycle has grown
by approximately 100% relative to the elastic strain prediction. The final plastic deformation corresponding
to this load path when the specimen has been unloaded results in plastic strains of 0.28% and 0.59% in the
axial and hoop directions, respectively.

Figure 13—Substantial plastic deformation observed in final ambient-temperature loading sequence on tubular
P110 samples after completing full loading history shown in Figure 10. (Results for last load cycle of Series A)

The multiaxial testing results described here are both interesting and significant. At a fundamental level,
we loaded a material specimen to within 5% of its yield strength at ambient temperature and within 10% of
yield at elevated temperature, and observed plasticity; this in itself is important but not surprising, given the
material has a rounded stress-strain response at elevated temperature and in certain loading directions. Of
greater significance is that the material began to yield in a 20°C load cycle in which von Mises stress did not
exceed 95% of the yield strength, and when API material testing protocols for standard grades would not
have identified a lower elastic limit. Furthermore, significant growth of plastic strain is observed as the test
progresses, particularly in the hoop direction, even though the severity of the loading relative to the material
yield strength does not increase. The observed plastic hoop deformation, which leads to a permanent increase
in pipe diameter, agrees with observations of ballooning during connection tests. Given the magnitude of
the hoop strain increase observed in the final cycle shown in Figure 13 and the demonstrated tendency of
the material to cyclically soften at ambient temperature, we would expect a rapid progression of plasticity
leading to specimen failure if the test had continued with additional loading cycles at this stress level.
During the multiaxial test, we also observed substantial increases in strain during stress holds at specific
load points; this is commonly referred to as creep, and was particularly prominent at 180°C. An example of
creep during a hold of a 180°C load point in quadrant 2 is shown in Figure 14. Creep is typically expected
to occur in low-alloy steel at temperatures greater than about 370°C (Campbell, 2008), but rate-dependent
response has been observed and documented over a wide range of temperatures in tensile tests of low and
medium grade OCTG (Kaiser, 2009).
16 SPE/IADC-184689-MS

Figure 14—Observed creep behavior during ten-minute stress hold in quadrant 2 at 180°C.

The results of these multiaxial near-yield material tests are both surprising and informative from a
tubulars design perspective. To summarize, we observe a tendency for commonly-used OCTG materials to
demonstrate highly non-linear behavior when subjected to near-yield stress levels, and we see that the extent
of plastic deformation is a strong function of the loading path to which the material is subjected. For the
P110 (and similar) materials we have examined, we see a tendency for cyclic softening behavior at ambient
temperature once plasticity is triggered, and a more stable cyclic response at elevated temperature. We
also observed a plasticity trigger associated with variation of elastic limit around the material performance
envelope, suggesting there are possible triggers for near-yield load effects even if the tubular is not exposed
to elevated temperature. Finally, we see that the extent of plasticity at a constant applied stress is rate-
dependent, which is consistent with traditional expectations of creep behavior, and infer that rate-dependent
material behavior is likely to influence near-yield load effects if casing or tubing loads are sustained for
long periods of time in service.

Addressing Near-Yield Load Effects in Tubular Design


The illustrative testing results presented in the previous sections of this paper indicate that subjecting
commonly used OCTG materials to certain load sequences at applied stresses below the material API yield
strength can lead to substantial, and in some cases growing, plastic deformation that is not anticipated in
traditional tubular design. Similar behavior has been observed in full-scale laboratory tests where pipe-
connection samples are subjected to combined cyclic loads at near-yield stress levels, and it is understood
that this behavior has led to tubular failure in at least one test. This information should be factored into the
casing and tubing design and evaluation process in a practical way, as considered in the following sections.

Conditions for Near-Yield Load Effects


For every case in which we have observed near-yield load effects, we see that four key elements must be
present:
SPE/IADC-184689-MS 17

• High loads on tubulars. Data we have gathered thus far indicates near-yield load effects will not
occur if loads (and associated stresses) are low enough that a tubular remains completely elastic in
its structural response. For the P110 material testing described herein, we measured elastic limits
in quadrants 1 and 2 of the von Mises yield envelope as low as 77% of the yield strength (axial
compression at 180°C). This suggests that a casing design safety factor in excess of 1.3 for this
material sample would generally prevent the onset of plasticity and avoid near-yield load effects.
Rate-dependence of material properties may also influence the elastic limit.
• A "trigger" for plasticity. It is apparent that even if a tubular is heavily loaded in a cyclic loading
mode, near-yield load effects will not invalidate elastic design assumptions unless plastic strain is
initiated in a sufficient volume of the material. An obvious trigger takes place when initial plasticity
arises from local stresses that are higher than the material elastic limit. Other factors might also
contribute – such as residual stress effects from pipe manufacturing, yielding through part of the
wall of a thick-walled tubular, or stress-strain concentrations.
• Cyclic loads. A mechanism must be present to cause incremental cyclic plasticity in the tubular.
This does not mean the loading needs to be fully reversed; our current hypothesis is that the stresses
must repeatedly breach the tubular material elastic envelope in multiple locations (i.e. stress ratios),
causing plasticity to evolve as loading is applied.
• Material propensity to cyclically soften. We do not expect near-yield load effects to occur unless
this key facet of material response exists at some point in the anticipated operating temperature
range. We clearly observed cyclic softening behavior in a common OCTG material at ambient
temperature in fully-reversed uniaxial testing, but have not fully explored this for the range of
available OCTG materials.
Well designers should consider the potential for near-yield load effects in well service, and be proactive
in minimizing well failure risks. In those well designs where conservative safety factors are assumed and
rigorously obeyed, the likelihood of near-yield load effects is minimal. If safety factors are reduced, more
rigour should be employed in evaluating candidate pipe and material combinations to ensure long-term
integrity in the context of anticipated well conditions.

Suggested Evaluation Approach


In situations where near-yield load effects may occur, the following steps should be considered by well
designers to assess the potential for near-yield load effects to occur under the anticipated loading conditions
and to mitigate their impact:

• The tendency for a material to cyclically soften over the planned operating temperature range
must be established. This can be done with a cyclic coupon test in load- or strain-controlled mode,
as has been demonstrated herein with a P110 material. Our suggestion is to conduct this type of
test at increasing levels of load or strain until initial plasticity is initiated, and then to measure the
cyclic strain per incremental cycle over a ten to 20 cycle period to establish whether stabilization
is imminent. Alternatively, a material's tendency to soften can be measured using appropriately-
placed strain gages on the pipe body if connection qualification testing is being undertaken to the
ISO 13679/API 5C5 evaluation protocol. In either case, if it is evident that the material does not
cyclically soften, the near-yield load effects described herein are unlikely to manifest in casing and
tubing during well operation, and need not be factored into the design.
• A sufficiently thorough characterization of the monotonic (uni-directional) response of the
candidate tubular material at representative temperatures is needed. In its most basic form, this
will include instrumented longitudinal tensile testing to characterize the elastic limit in addition
to the yield strength. Ideally, the designer will also obtain an understanding of how the measured
monotonic response will change in different loading modes. Compressive longitudinal testing
18 SPE/IADC-184689-MS

would be valuable, and any multiaxial characterization would be even more beneficial, but may
be difficult to obtain for full-scale pipe samples. Tests to characterize the rate-dependence of the
material may also add value; the most practical approach may be creep tests at increasing levels of
stress to determine the minimum stress level at which plasticity initiates.
• If a material is known to cyclically soften, and once monotonic properties are established, the
designer can choose to either avoid near-yield load effects entirely (by selecting sufficiently
large safety factors), or can embark on a more rigorous evaluation of the behavior of the
material under field-representative loading conditions. Small tubular specimens such as the
ones described in this paper are effective and efficient for such testing, but may not be fully
representative of larger pipes. Full-scale testing work on representative commercial pipe may be
effective, provided the sample is sufficiently uniform and stable to enable accurate measurement,
and provided the test control system is adequately precise.
• While not explored here, numerical modelling of pipe body (and connection) performance might
be undertaken to predict the stress states that will exist in casing service. However, such modelling
is not yet a reasonable alternative to testing for establishing the magnitude of deformation
associated with near-yield load effects. Explicit modelling of near-yield load effects will be very
difficult, because the constitutive model describing the tubular material will need to accurately
capture the transient cyclic material response.
• Ultimately, all steps should lead to selection of operational loads and design safety factors that
minimize the impact of near-yield load effects. Selections must account for the known near-yield
load material behavior and possible operational scenarios, including contingencies for more severe
loading than intended due to unplanned events.
• Where practical, the designer could consider close monitoring of tubular loads during well
construction and operations and adjust operational load limits according to the load history of the
pipe.

Future Investigations
The discussion in this paper is intended to raise industry's awareness of a facet of cyclic material response
that might limit functionality of tubulars operating in extreme service well environments. If such effects
have the potential to affect well tubulars in the same way as they have impacted tubular samples in the
described material tests and connection qualification programs, we believe there is considerable value in
continuing studies of near-yield loading behaviour.
Key facets of near-yield load effects that would benefit from further investigations include:

• Metallurgical fundamentals. Characterization of OCTG material response would benefit from


metallurgical insights to complement mechanical evaluations, ultimately enabling a better
understanding of the mechanisms at work and prediction of when near-yield effects occur.
• Product-specific material grades. While near-yield load tests performed to-date indicate
repeatable near-yield load effects for quenched-and-tempered P110 materials, it is uncertain if
near-yield load effects occur in all API material grades or proprietary materials (e.g., corrosion
resistant alloys). Similarly, consistency of this response across the range of products that satisfy a
grade designation could be evaluated, as could differences in response associated with variations
in commercial manufacturing processes.
• Multiaxial performance envelopes. Multiaxial material performance envelopes should account
for all combinations of material properties and external loading that lead to near-yield load effects.
The studies to-date identified that the elastic limit and the transient cyclic response should be
considered in assessing near-yield load effects in tubular design. While assessment of those
SPE/IADC-184689-MS 19

material properties is relatively simple under uniaxial loading, it is a more challenging task under
multiaxial loading when several stress and strain components change simultaneously.
• Representative field loading. Mitigation of near-yield load effects in well designs will be
facilitated by simulating various field-representative load paths in laboratory conditions, where the
material and/or structural response to applied loading can be precisely measured, and initiation
of the near-yield load effects can be detected. Specifically, it will be beneficial to observe
multiple loading cycles to establish whether strains are stabilizing or growing as cycling occurs.
Application-specific load paths of various severities (e.g. hydraulic fracturing wells designed with
various safety factors) can be examined to determine the likelihood of significant near-yield loading
effects for different tubular configurations and operational scenarios.
• Stress concentrations. Actual local stresses might be greater than design stresses in some regions
and components within a tubular string. Examples are stress concentrations in a connection due
to complex geometry, and/or increased stress adjacent to a connection due to the stiffness contrast
between the pipe body and the connection. Impacts of stress concentrations on propagation of near-
yield loading effects within the tubular string should be investigated.
• Engineering basis. Further investigation will be needed if industry sees the need to develop a more
complete engineering basis that will address near-yield load effects in well designs. For example,
a systematic comparison of conventionally-measured yield strength and elastic limit for OCTG
materials will facilitate derivation of appropriate de-rating factors for in-service allowable stresses.

Conclusions
The following conclusions are drawn from the information presented in this paper:

• Despite the conventional linear-elastic material assumption traditionally used for well design,
laboratory testing of a commonly-used OCTG tubular sample (P110 grade) demonstrates that
under some combinations of cyclic loading substantial plastic strain can accumulate in a material
that has not been loaded beyond its yield strength. Plastic deformation in the tested tubular
specimens increased with each loading cycle, suggesting an unstable response that would
ultimately compromise the specimen integrity. This response, not traditionally accounted for in
material evaluation or casing and tubing design practices, is referenced as near-yield load effects.
• Plasticity in the test specimens is believed to arise from the gradual nature of the yield response
(rounded transition) of the P110 material at 180°C, and from multiaxial-stress dependence of yield
strength and elastic limits. The cyclic softening character of the tested P110 material at 20°C was
associated with rapid progression of plastic strain under repeated cyclic load.
• Near-yield load effects were observed in fully-reversed uniaxial tests and in multiaxial loading
cycles combining axial tension-compression and internal pressure.
• Four conditions are believed to be necessary for near-yield load effects to manifest in well tubulars:
load levels that create high stresses in the tubular material; cyclic loading that causes evolution
of plastic strain over time; a plasticity trigger; and a cyclic softening character of the material at
temperatures representative of field conditions.
• If near-yield load effects are considered likely in an application, well designers should consider
evaluation steps (such as those presented herein) that will enable determination of the severity of
those effects under the anticipated operational loading.
• Well design processes will benefit from continued investigations of near-yield load effects, which
will enhance understanding of relevant material features and facilitate predictions of near-yield
load effects in various OCTG materials under field-representative loading scenarios, and in areas
of stress-strain concentrations such as connection thread roots.
20 SPE/IADC-184689-MS

• Development of an efficient near-yield load test methodology will be useful to conduct near-yield
load test programs on statistically-representative populations of material samples under a variety
of loading scenarios.

Acknowledgements
The authors acknowledge the participants of the multi-client project in which near-yield loading was
investigated and the testing reported herein was conducted: Evraz Inc. North America, Tenaris Connections
Limited, TMK IPSCO, United States Steel Tubular Products Inc., and Vallourec and Mannesman Oil and
Gas France.
The technical contributions and reviews of Dan Dall'Acqua, Trent Kaiser, and Victor Yung are greatly
appreciated.

References
API Technical Report 5C3, First Edition, December 2008. Technical Report on Equations and Calculations for Casing,
Tubing, and Line Pipe Used as Casing or Tubing; and Performance Properties Tables for Casing and Tubing. American
Petroleum Institute
API Specification 5CT, Ninth Edition, July 2011. Specification for Casing and Tubing. American Petroleum Institute
API RP 5C5, 2014 draft, Procedures for testing casing and tubing connections
ASTM E 8 – 16A Standard Test Methods for Tension Testing of Metallic Materials. ASTM International
Boresi, A.P. and Sidebottom, O.M. 1985. Advanced Mechanics of Materials, Fourth Ed., John Wiley & Sons, New York
pp. 93–95
Broom, T. and Ham, R.K. 1957. The Hardening and Softening of Metals by Cyclic Stressing, Proceedings of the Royal
Society of London. Series A, Mathematical and Physical Sciences, Vol. 242, No. 1229 pp. 166–179
Campbell, F.C. 2008. Elements of Metallurgy and Engineering Alloys, Campbell, F.C. Ed., ASM International, Materials
Park, Ohio p. 265
Dall'Acqua, D., Smith, D.T., Kaiser, T.M.V. 2005. Thermo-plastic Properties of OCTG in a SAGD Application, SPE
97776, SPE/PS-CIM/CHOA International Thermal Operations and Heavy Oil Symposium, Calgary, Alberta, Canada,
Nov 1–3, 2005
Ellyin, F. 1997. Fatigue Damage, Crack Growth and Life Prediction, Chapman & Hall, New York, p. 35
Kaiser, T.M.V., Yung, V.Y.B., Bacon, R. M. 2008. Cyclic Mechanical and Fatigue Properties for OCTG Materials, SPE
97775, SPE Journal Volume 13 Issue 4, pp. 480–486, 2008
Kaiser, T.M.V. 2009, Post Yield Material Characterization for Strain-Based Design, SPE 97730, SPE Journal Volume 14
Issue 1, pp. 128–134, 2009
Nowinka, J., Kaiser, T., and Lepper, B. 2007. Strain Based Design of Tubulars for Extreme Service Wells, SPE 105717,
SPE/IADC Drilling Conference Amsterdam, Holland, Fe 20-22, 2007
Skelton, R.P., Maier, H.J., Christ, H.-J. 1997. The Bauschinger Effect, Masing Model and the Ramberg-Osgood Relation
for Cyclic Deformation in Metals, Materials Science and Engineering Vol. A238 pp. 377–390

You might also like