Sei sulla pagina 1di 26

SPE-185941-MS

New Tubular Design Ellipse With Backup Pressure

Malcolm A. Goodman, Issa A. Kalil, Albert R. McSpadden, and Oliver D. Coker, III, Altus Well Experts, Inc.

Copyright 2017, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Bergen One Day Seminar held in Bergen, Norway, 5 April 2017.

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

Summary
This paper develops a new von Mises ellipse based on backup pressure (internal pressure for collapse and
external pressure for burst), so the new API collapse equations with dependence on internal pressure can
be shown simultaneously on the same plot with the ellipse. Unlike the current ellipse used for casing and
tubing design which is approximate, the new ellipse is exact for collapse, burst, and axial loads. Without
the new ellipse, the approximate ellipse will continue to be used, and the new API collapse equations must
be plotted separately. Example cases demonstrate the advantages of the new ellipse, including increased
accuracy. Results help clarify many questions about the new API collapse equations, namely where they
come from and how to use them.

Introduction: API Collapse History and Link to Von Mises Ellipse


The conventional tubular design ellipse introduced by Johnson et al. (1987) is constructed from the von
Mises relationship by assuming that external pressure is zero for the top burst half of the ellipse and internal
pressure is zero for the bottom collapse half of the ellipse. The resultant ellipse is not only approximate, but
also inaccurate; however, it has the advantage that the API pipe ratings (before 2015) can be displayed on the
same plot with the ellipse to show both the uniaxial and triaxial limits for tubular design. With the new API
collapse equations (API TR 5C3, October 2015), the old ellipse is no longer compatible for simultaneous
display.
The history of the API collapse equations and their link to the von Mises ellipse is interesting and
somewhat intriguing. The origins of the concepts are not obvious. In particular, it has been difficult to
determine where the biaxial and latest triaxial (with dependence on pi)collapse equations came from and
who are the original authors. This paper asserts that the earliest biaxial relationship can be attributed to
Holmquist and Nadai (1939), who developed the "Ellipse of Plasticity" by setting σr = 0 in the von Mises
relationship. This paper also asserts that the triaxial dependence in the new API collapse equations can be
attributed to Lubinski (1975) who developed the "Circle of Plasticity" and derived the triaxial ellipse for
collapse with dependence on (σz +pi). The Lubinski work occurred long before the date on the reference to
Clinedinst (1980) in the addendum release (API TR 5C3, October 2015).
By the time of the publication of API Bulletin 5C3 (1st Edition, 1970), several iterations had already
occurred in the evolution of the collapse formulation. Namely, theoretical work presented by several authors
2 SPE-185941-MS

at the twelfth annual meeting of the API in 1939 indicated that collapse was categorized between elastic
instability of pipe with larger D/t ratio and yield of pipe with smaller D/t ratio.
By the mid 1960's data was gathered on 2488 specimens of K55, N80 and P110 tubulars. This led to the
incorporation of an average collapse pressure represented by one of three formulae: 1) an analytical ‘yield’
limit; 2) an empirical ‘plastic’ limit with use of constants A and B in the Stewart-type collapse form (1906);
and 3) an analytical ‘elastic’ limit with 95% of the collapse limit proposed by the Clinedinst formula (1939).
A minimum plastic limit and a minimum elastic limit were implemented by reducing the average value
by 25%. Subsequently, a statistical approach was implemented to arrive at the minimum plastic collapse
values in the form of the equation presently available with the constants A, B and C. However, this approach
led to a discontinuity between the minimum plastic limit and the minimum elastic limit. Consequently, a
plastic/elastic transition collapse limit formula was developed to overcome this anomaly by incorporation
of constants F and G in the Stewart-type equations.
Initially, these collapse pressure formulae assumed zero axial force and zero internal pressure. Corrections
were proposed for the influence of axial force (stress). Originally, this correction took the form of the
Holmquist and Nadai (1939) quadratic as a multiplier to the originally-determined collapse pressure, namely

(1)

where Pc is the original collapse performance and Pca is the reduced collapse performance under the influence
of an axial (tensile) stress Sa and yield stress Y. Figure 1 shows a pipe vendor's proposal for use of this
technique.

Figure 1—Pipe vendor plot for biaxial correction of collapse

Subsequently, this quadratic multiplier was used to determine the "equivalent yield strength in the
presence of axial stress". That is, instead of multiplying the original collapse rating by this biaxial correction
factor, the original yield strength was multiplied by this biaxial correction factor,

(2)
SPE-185941-MS 3

This "equivalent yield strength" Ya was then used in each of the API collapse formulae to arrive at the
corrected collapse resistance under the influence of axial force. This particular correction was attributed to
Clinedinst (1980) in the API Bulletin 5C3 (1985).
Up to this point, industry proceeded to assume that collapse occurred with no effect of internal pressure, so
the equivalent pressure acting to collapse the pipe was simply the differential pressure in the well. However,
in 1989, the API Bulletin 5C3 (1989) provided a correction to this idea in the form of an effective external
pressure under the heading "Effect of Internal Pressure on Collapse",

(3)

This equation suggested that the effective external pressure pe was equal to the specific Δp = (po − pi) plus
some D/t function of pi. Reference is made in the API document that this was also attributed to Clinedinst
(1985). This particular view on the correction for pi to get the effective external pressure was employed by
the API until the ISO 10400 (2007) / API TR 5C3 (2008) release. At that time, Section 8.4.7 in API TR
5C3 modified the former equation above under the same heading "Effect of Internal Pressure on Collapse"
as below,

(4)

Note that the form of the two equations above is the same. The exception is that equation (3) dealt with an
effective collapse ‘load’ whereas equation (4) determined an effective collapse resistance due to the effect
of internal pressure. The API document references Clinedinst (1985) as the contributor of this change on
handling the internal pressure effect. The API TR 5C3 (2008) equation can be rearranged as follows to
define the differential pressure resulting in collapse,

(5)

In its most recent revision, the API TR 5C3 Addendum with Annex M (2015) once again references
Clinedinst work (1980) as the source for the triaxial representation of "the combined loading equivalent
grade, the equivalent yield strength in the presence of axial stress and internal pressure". As noted above, the
authors of this paper question this source and believe Lubinski (1975) is the original source. The combined
equivalent grade is

(6)

This combined "equivalent yield strength" fycom with minimum yield strength fymn(API started using f
instead of Y for yield strength and σ instead of S for stress) is then used in each of the API collapse formulae
to arrive at the corrected collapse resistance under the influence of axial stress and internal pressure.
However this final calculation (based on the "equivalent yield strength", the D/t ratio, and the proper collapse
regime – Yield, Plastic, Transition or Elastic) actually determines the differential pressure causing collapse,
(7)
Note the following two points: 1) there is no need for a correction for internal pressure, since it is directly
accounted for in the equivalent yield strength; and 2) this was the first time that the API referred to a
"differential pressure" instead of an absolute pressure as the collapse resistance. This latest API addendum
for collapse leads to and gives rise to the new ellipse with the new coordinate axes proposed in this paper.
4 SPE-185941-MS

To set the stage for the new ellipse with backup pressure, the derivations of the biaxial ellipse, the Lubinski
circle, and triaxial ellipse are presented in the next sections. The similarity in form of the biaxial and triaxial
relationships provides insight into the new collapse equations and helps with understanding the details and
use of them. Also, the triaxial collapse derivation with backup pressure pi points the way for the burst
derivation with backup pressure po. The top half of the burst ellipse and the bottom half of the collapse
ellipse are used to construct the new ellipse with backup pressure.

Ellipse of Plasticity (Biaxial Collapse)


As noted in the Introduction, the Ellipse of Plasticity was developed by Holmquist and Nadai (1939). To
obtain this biaxial ellipse, the starting point is the full triaxial relationship of the von Mises stress σVM

(8)

where σr, σθ, and σz are the principal stresses in cylindrical coordinates and Y is the tensile yield stress.
Holmquist and Nadai (1939) note that "in a moderately thick-walled tube the radial stress σr can in first
approximation be neglected", which means σθ and σz are much greater than σr. Assuming σr = 0, then
equation (8) becomes
(9)
This defines the biaxial ellipse on the axes σθ and σz, see Figure 2 which is reproduced from Holmquist
and Nadai (1939).

Figure 2—Biaxial Ellipse (based on Holmquist and Nadai, 1939)

Solving for σθ in the quadratic equation (9),

(10)
SPE-185941-MS 5

This is beginning to look a lot like the "axial stress equivalent grade" in the API biaxial collapse equations.
Equation (10) defines the equivalent grade concept. To complete this derivation to get the collapse pressure,
the Lame equations are used (tension positive and compression negative),

(11)

where r is the pipe radius and A and B are constants defined by

(12)

with a, b and pi, po being the internal and external radii and pressures respectively. From (8) and (11), the
von Mises stress σVM is maximum at the inner radius a, so first yield occurs at the inner wall. For the biaxial
case, σr = -pi = 0 at the inner wall, so from (10), (11) and (12) with po = pc for collapse

(13)

Since pc must be positive, the negative sign is required on the square root and, therefore,

(14)

The following important observations are made:


∘ Equation (14) is precisely the API biaxial equation for yield collapse.
∘ The quantity in the curly bracket on the right hand side of equation (14) is identically the "equivalent
yield strength in the presence of axial stress" used in the biaxial API collapse equations.
∘ The quantity on the left is the external collapse pressure for zero internal pressure, so it must
be adjusted for internal pressure according to the API correction formula. As mentioned in the
Introduction and shown later in the section for the triaxial case, the form of the triaxial ellipse is
the same as in the biaxial case, but the quantity on the left is (pc – pi) so no correction equation for
internal pressure is required.
∘ In the Holmquist and Nadai (1939) paper, a footnote states: "A more complete theory without
disregarding the stress in equation (37) has been worked out by one of the writers, and will be reported
elsewhere." Their equation (37) is the same as equation (8) in this paper. Presumably, the ‘more
complete theory’ is along the lines of the triaxial theory published by Lubinski (1975). However, this
later work by Holmquist or Nadai was never completed or not published, because no paper has been
found and no reference to such a paper was made by Lubinski (1975).

Circle of Plasticity (Triaxial Yield)


Lubinski (1975) developed the exact representation for the von Mises stress σVM using the Lame equations
for σr and σθ without any simplifying assumptions. Interestingly, the result is a circle and not an ellipse, see
Figure 3. Lubinski named it the Circle of Plasticity, apparently following the lead and naming convention
of Holmquist and Nadai (1939) with their Ellipse of Plasticity. The procedure to get the circle is described
in the Lubinski (1975) paper, but the details are sketchy and the derivation is not obvious. A complete
description of the circle derivation is presented in the Appendix of this paper.
6 SPE-185941-MS

Figure 3—Von Mises Circle (Lubinski 1975)

The equation of the circle is important, because it is needed to develop the burst and collapse halves
of the new ellipse based on backup pressure proposed in this paper. The circle has not been accepted or
used for tubular design in the petroleum industry, probably because the circle axes are not simply pressure
differential and axial force, and the API ratings cannot be superimposed. Goodman (2017) uses the circle,
which he calls the "von Mises Circle" analogous to the von Mises ellipse, for his new leak criterion for
threaded connections.
The y-axis in Figure 3 is

(15)

and the x-axis is

(16)

The equation of the circle is


(17)
The x-axis given by equation (16) is identically the stress due to the Lubinski fictitious force for buckling.
Goodman (2017) shows that this fictitious stress is associated with the shear part of the axial stress, i.e.
it is independent of the hydrostatic part, which explains the mystery or fiction about the fictitious force.
The reason that the buckling fictitious stress defined by (16) appears in the yield relation defined by
(17) is that both buckling and yield are independent of hydrostatic pressure, see Goodman (2017). Both
buckling and yield are pure shear phenomena, and the buckling stress in (16) represents the shear part of the
axial component in (17). Equation (17) means that von Mises yield is governed by only two independent
quantities, ΔP and Δσz, even though pi, po, and σz are three independent variables.
SPE-185941-MS 7

Ellipse with Backup Internal Pressure (Triaxial Collapse)


To investigate collapse, Lubinski (1975) started with the circle equation in (17), made a substitution, and
developed the full triaxial ellipse based on backup internal pressure. As with the circle, the ellipse derivation
by Lubinski is sketchy, so this section reproduces the collapse derivation with more detail. Understanding
this derivation is essential, since the authors of this paper use a similar technique in the next section to
develop the ellipse for burst based on backup external pressure.
For collapse, since
(18)
then
(19)
Using (18) in (15) and substituting (19) in (16) to eliminate po, then equation (17) becomes

(20)

Expanding the square and combining terms, then


(21)
where

(22)

Solving the quadratic equation (21) as before and again using the negative sign on the square root to
ensure that the collapse differential pressure (pc - pi) is positive,

(23)

This triaxial result in comparison to the previous biaxial result demonstrates:


∘ Equation (21) is identical to equation (9), so the two ellipses have the same skewed shape as in Figure
2.
∘ The only difference between the biaxial ellipse and triaxial ellipse is the value on each axis.
Specifically, for the triaxial plot, (pc - pi) replaces pc on the y-axis and (σz + pi) replaces σz on the
x-axis.
∘ The triaxial case is exact. There are no assumptions on σr as in the biaxial case, nor is pi or po set
equal to zero to get the top burst and bottom collapse regions as in the standard ellipse used currently
for tubular design, see Johnson et al. (1987).
∘ Equation (23) is identical to the new 2015 API equation for yield collapse.
∘ The quantity in the curly brackets is the same as the "combined loading equivalent grade" in the
addendum API TR 5C3 (October 2015), which defines "the equivalent yield strength in the presence
of axial stress and internal pressure". The derivation above shows where this comes from, and that
it is attributed to Lubinski (1975).
∘ To repeat an important but very subtle point, the left hand side of equation (23) shows that the
differential pressure governs the behavior, so pi is simply added to the result to get the collapse
8 SPE-185941-MS

pressure pc. No internal pressure correction is needed. This is different than the previous biaxial
relationship (14) which resulted in pc only, so the correction for pi was required.

Ellipse with Backup External Pressure (Triaxial Burst)


This section follows the same procedure as in the previous section but eliminates pi instead of po to get the
burst relationship. From (18),
(24)
so using (24) and (18) in (15) and (16), then (17) becomes

(25)

After expanding and combining terms,


(26)
where S is the same as in (22) and

(27)

(28)
The results in (26) - (28) are interesting. First, the stress quantity in (28) has the same form as (22),
namely it is simply the axial stress plus the backup pressure. Second, equation (26) is similar to equation
(21). In fact, for k = 1, i.e. in the limit when the radii a and b are close as in thin wall, then j = 1 and the
two equations (21) and (26) are the same. Solving the quadratic equation (26) for S,

(29)

Just as tension derates collapse, compression derates burst. Since tension is positive and compression is
negative, then To is negative in the upper left quadrant of the ellipse, so the plus sign on the square root is
selected to ensure that S is positive for burst. Setting Δp = (pb – po) in (22) where pb is the burst pressure,
then (29) becomes

(30)

Equation (30) for burst is analogous to equation (23) for collapse. These two equations are exact and are
used in the next section to construct the new ellipse with backup pressure. The y-axis of the new ellipse is
Δp = (pi – po) and the x-axis is (σz + pB) where the backup pressure pB is pi for the collapse bottom half of
the ellipse and po for the burst top half of the ellipse. The two halves are continuous across the x-axis, since
pi = po when Δp = 0 at the x-axis, so Ti = To. However, there is a slight kink at the x-axis, since the slopes are
different for the burst and collapse halves due to the radii ratio (a/b) in the constants j and k defined by (27).
SPE-185941-MS 9

Full Ellipse with Backup Pressure and Superimposed API Ratings


Given the form of exact ellipses for both collapse and burst developed above, the equations (23) and (30) are
used to determine the performance of the 7 in. 26 lbf/ft P110 pipe routinely referenced in the API Technical
Report 5C3 (2008 and Addendum 2015). Figure 4 shows the exact representation for collapse and burst
using each half of the required ellipse. The y-axis is Δp = (pi – po) and the x-axis is (σz + pB) where the
backup pressure pB is pi for the collapse bottom half of the ellipse and po for the burst top half of the ellipse.
Note the slight kink at the x-axis where Δp = 0. As stated previously, this is due to the difference in slopes
between the collapse and burst halves.

Figure 4—Exact ellipse for 7in. 26.00 lbf/ft P110

This 7 in. 26.00 lbf/ft P110 pipe was used to demonstrate the use of the original API equation with biaxial
equivalent yield strength and subsequent API equation for the "effect of internal pressure on collapse" as
given in the API TR 5C3 (2008). Note that for the API example case with axial stress of 11,000 psi and an
internal pressure of 1000 psi, the calculated pci – pi was 6010 psi. However, use of the ‘modified’ equation
in the Addendum of 2015 that included both axial stress and internal pressure within the triaxial equivalent
yield strength showed that the calculated pc – pi was now 6100 psi. The collapse resistance of this same
pipe as reported by API with no axial stress or internal pressure is 6230 psi. These results are demonstrated
in Figure 5.
10 SPE-185941-MS

Figure 5—Collapse for 7 in. 26.00 lbf/ft P110 with 11,000 psi axial stress and 1000 psi internal pressure

Although not obvious in Figure 5, it should be noted that what is referred to as API Collapse (2008)
indicates that the actual collapse limit (the differential pressure) varies with increasing internal pressure.
That is, the API Collapse (2008) values are shifted upwards on the ordinate while also shifted to the right
on the abscissa as pi increases. This is evident as indicated by the oval located at the value 12,000 psi for
σz + pi on the abscissa. Note that there is a 90 psi difference on the ordinate axis between the API Collapse
(2015) and the API Collapse (2008) [6100 psi – 6010 psi = 90 psi].
This difference is increased considerably if it is assumed that the calculation of the differential collapse (pc
– pi) is to be determined at 11,000 psi axial stress and 10,000 psi internal pressure as shown in Figure 6. For
this case, the API Collapse (2008) using the former API TR 5C3 equations for "equivalent yield strength"
due to axial stress only with the "effect of internal pressure on collapse" would result in a differential pressure
of 5080 psi. In contrast, the improved equation for the "triaxial equivalent yield strength" that includes both
the axial stress and the internal pressure results in a differential pressure of 5970 psi. There is now an 890
psi difference on the ordinate axis between the API Collapse (2015) and the API Collapse (2008). Figure
6 indicates the disparity in these two calculation methods.
SPE-185941-MS 11

Figure 6—Collapse for 7 in. 26.00 lbf/ft P110: 11,000 psi axial stress & 10,000 psi internal pressure

There are three important conclusions from Figure 6. First, the API Collapse (2008) limit must be ‘drawn’
uniquely for every internal pressure desired. Second, the shifting of the API Collapse (2008) limit as pi
increases indicates a greater reduction in collapse resistance than is calculated by the API Collapse (2015)
formulation. Third, an advantage of the API Collapse (2015) formulation is that the differential pressure
rating can be read directly from the plot, i.e. the rating curve for the "new" formulation does not shift with
internal pressure when plotted on the abscissa σz + pi.
A peculiarity of the API Collapse (2008) limit is demonstrated in this specific case of 10,000 psi internal
pressure. The shift to the right on the abscissa up to the 24,000 psi "equivalent yield strength" minimum
imposed by the API would result in a collapse value outside of the exact collapse yield ellipse. This suggests
some anomaly not supported by theory. Note that the API Collapse (2015) limit appears to "truncate" at or
near the exact collapse yield ellipse for the condition where σz+pi in the updated Addendum 2015 results
in the 24,000 psi "equivalent yield strength" minimum.
Of particular interest, is the behavior of both the exact collapse and burst yield ellipse in comparison to the
new API Collapse (2015) limit for API pipes with increasing D/t. Five pipes were selected to demonstrate
this phenomenon: 3-1/2 in. 12.70 lbf/ft L80; 7 in. 38.00 lbf/ft T95; 9-5/8 in. 53.50 lbf/ft N80; 13-3/8 in.
72.00 lbf/ft N80; and, 20 in. 94.00 lbf/ft K55. These five have D/t values that range from 9.333 for the 3-1/2
in. to 45.662 for the 20 in. pipe. Consequently, the API collapse resistance without axial stress or internal
pressure spans the ‘Yield’ regime for the 3-1/2 in. pipe to ‘Elastic' regime for the 20 in. pipe.
Table 1 below provides the D/t value, API collapse rating and appropriate regime for each of the five pipes
evaluated. Additionally, Figures 7a-7e show the exact yield ellipse superimposed with the API Collapse
(2015) limit and the API uniaxial limits for Burst, Tension and Compression for these same five pipes.
12 SPE-185941-MS

Figure 7a—Exact ellipse with API ratings: 3-1/2 in. 12.70 lbf/ft L80

Figure 7b—Exact ellipse with API ratings: 7 in. 38.00 lbf/ft T95
SPE-185941-MS 13

Figure 7c—Exact ellipse with API ratings: 9-5/8 in. 53.50 lbf/ft N80

Figure 7d—Exact ellipse with API ratings: 13-3/8 in. 72.00 lbf/ft N80
14 SPE-185941-MS

Figure 7e—Exact ellipse with API ratings: 20 in. 94.00 lbf/ft K55

Table 1—API Collapse of Five API Pipe

Pipe D/t Collapse (psi) Regime

3-1/2 in. 12.7 lbf/ft L80 9.333 15,310 Yield

7 in. 38 lbf/ft T95 12.963 13,430 Plastic

9-5/8 in. 53.5 lbf/ft N80 17.661 6620 Plastic

13-3/8 in. 72 lbf/ft N80 26.021 2670 Transition

20 in. 94 lbf/ft K55 45.662 520 Elastic

Figures 7a-7e lead to the following important observations and results:


∘ The σz+ pi required in the Addendum API TR 5C3 (2015) to reach the minimum 24,000 psi equivalent
yield strength is 38,922 psi for the K55 pipe, 65,253 psi for the L80/N80 pipe, and 80,698 psi for
the T95 pipe.
∘ The 3-1/2 in. pipe (yield collapse regime) tracks the exact collapse ellipse as would be expected for
the yield collapse criterion.
∘ The 7 in. pipe (plastic collapse regime) tends to track the exact collapse ellipse once yield becomes
the collapse criterion.
∘ The 9-5/8 in. pipe begins in the plastic collapse regime, but with increasing σz + pi the equivalent
yield strength decreases and the regime moves from plastic to yield collapse. At the extreme end of
the positive abscissa, the 9-5/8 in. pipe begins to track the exact collapse ellipse.
∘ The 13-3/8 in. pipe begins in the transition collapse regime, but with increasing σz + pi the equivalent
yield strength decreases and the regime moves from transition to plastic collapse. The 24,000 psi
minimum equivalent yield strength is reached before the pipe is able to attain the yield collapse
regime.
SPE-185941-MS 15

∘ The 20 in. pipe begins in the elastic collapse regime and moves towards the transition collapse with
increasing σz + pi until reaching the 24,000 psi limit. Consequently, the API Collapse (2015) curve
appears to be flat (horizontal) for the entire range of σz + pi values.
∘ In investigating these five pipes (with the 7 in. 26 lbf/ft P110 previously described above), an odd
occurrence was detected as the equivalent yield strength fell below the 24,000 psi limit. That is, the
API equations used to determine the D/t ranges for the various collapse regimes as a function of
yield strength tend to "act up". The equations generate inconsistent and sometimes non-logical ranges
where D/t for yield to plastic is greater than the required D/t for plastic to transition. Additionally, the
D/t for plastic to transition tends to decrease at equivalent yield strengths below 12,000 psi. Lastly,
the D/t for yield to plastic approaches negative values for equivalent yield strengths below 7500
psi. These odd occurrences explain or at least justify the API-imposed limit of 24,000 psi minimum
equivalent yield strength with use of the modified equation.
∘ Of particular interest is the "kink" that occurs on the abscissa (Δp = 0) for the ellipse. As described
in the derivation section, the burst and collapse halves of the ellipse are identical when k = j = 1 in
equations (27) and (29), which is approximated for thin wall, when the outer radius b nearly equals
the inner radius a. Whereas the 3-1/2 in. pipe in Figure 7a shows a noticeable kink at the abscissa,
the 20 in. pipe in Figure 7e appears to indicate a continuous ellipse from the plot of the two halves.
Note that for the 20 in. pipe with a D/t of 45.662, the calculated values for k and j are 0.914 and
0.959, respectively.
All the ellipses constructed from a top burst half and bottom collapse half have the same skewed shape
with a kink on the abscissa. In fact, even the old ellipse with pi = 0 for collapse and po = 0 for burst is
identical to the new exact ellipse. Only the axes are different between the old and new ellipses, which means
any plotted points (loads) are different and hence accuracy is improved with the new exact presentation.
This is demonstrated in the next section.

Implementation of New Ellipse: Well Design Example Case


To demonstrate the utility of the proposed exact ellipse formulation coupled with the API collapse (2015)
formulation, an example case is presented. The following example compares the new methodology directly
with results from industry software for design of casing and tubing. Minor modifications, including a vertical
well, were made to the example for ease of presentation and the benefit of hand-calculations to confirm
results. Well configuration is described in Table 2 below. For hand-calculations, additional information is
required. Table 3 below provides data needed to replicate the stress analysis for the initial conditions as-
landed.

Table 2—Well Configuration for Example Problem


16 SPE-185941-MS

Table 3—Additional Input Data for Example Problem

Two load conditions were evaluated for each of the five strings listed in Table 3. One burst load and one
collapse load were created to compare the existing graphical approach with the new approach. A description
of the two load cases for each of the five strings is provided in Tables 4 – 8.

Table 4—Burst and Collapse Loads for 3-1/2 in. 12.70 lbf/ft L80

Load Description
Load Type Load Name
Internal Pressure External Pressure Thermal Profile

7218 psi @ 40 ft. and 50 °F from 40 to 400 ft.


Burst Acid Job 10 lbf/gal brine to 17000 ft.
14700 psi @ 17000 ft. and 111 °F @ 17000 ft.

2500 psi on top 50 °F from 40 to 400 ft.


Collapse Collapse with APB 10 lbf/gal brine to 17000 ft.
of 10 lbf/gal brine and 111 °F @ 17000 ft.

Table 5—Burst and Collapse Loads for 7 in. 38.00 lbf/ft T9

Load Description
Load Type Load Name
Internal Pressure External Pressure Thermal Profile

7218 psi on top 49 °F from 40 to 400 ft.


Burst Tubing Leak – Acid Job 17.5 lbf/gal mud to 14800 ft.
of 10 lbf/gal brine and 108 °F @ 14800 ft.

Collapse Evacuated Casing Evacuated (0 psi) 17.5 lbf/gal mud to 14800 ft. Geothermal

Table 6—Burst and Collapse Loads for 9-5/8 in. 53.50 lbf/ft N80

Load Description
Load Type Load Name
Internal Pressure External Pressure Thermal Profile

14.5 lbf/gal mud to


Displace to Gas CH4 inside with 14.33 lbf/ 48 °F from 40 to 400 ft.
Burst 9500 ft. and 8.33 lbf/gal
Frac @ Shoe gal frac gradient at 15000 ft. and 132 °F @ 15000 ft.
mix-water to 15000 ft.

Mud level at 2000


Mud Drop due 48 °F from 40 to 400 ft.
Collapse ft. with 8.33 lbf/gal 14.5 lbf/gal mud to 15000 ft.
to Lost Returns and 132 °F @ 15000 ft.
below to 15000 ft.
SPE-185941-MS 17

Table 7—Burst and Collapse Loads for 13-3/8 in. 72.00 lbf/ft N80

Load Description
Load Type Load Name External
Internal Pressure Thermal Profile
Pressure

CH4 inside with 10 lbf/gal mud


46 °F from 40 to
Displace to Gas 15.20 lbf/gal to 6100 ft. and
Burst 400 ft. and 109
Frac @ Shoe frac gradient 8.33 lbf/gal mix-
°F @ 9700 ft.
at 9700 ft. water to 9700 ft.

Mud level at
46 °F from 40 to
Mud Drop due 5385 ft. with 10 lbf/gal mud
Collapse 400 ft. and 109
to Lost Returns 14.5 lbf/gal mud to 9700 ft.
°F @ 9700 ft.
below to 9700 ft.

Table 8—Burst and Collapse Loads for 20 in. 94.00 lbf/ft K55

Load Description
Load Type Load Name External
Internal Pressure Thermal Profile
Pressure

8.6 lbf/gal 152 °F @ 40 ft.


seawater to 450 181 °F @ 100 ft.
800 psi on top of
Burst Produce 1 Year ft. and 8.33 53 °F @ 400 ft.
10 lbf/gal mud
lbf/gal mix- 153 °F @ 450 ft.
water to 2000 ft. 272 °F @ 2000 ft.

Mud level at
8.6 lbf/gal 44 °F from 40
1200 ft. with
Collapse Mud Drop of 60% seawater to 400 ft. and 61
10 lbf/gal mud
to 2000 ft. °F @ 2000 ft.
below to 2000 ft.

Given the landing conditions for each of the five strings in Tables 2 and 3, the load cases of Tables 4-8
can be calculated and plotted. Figures 8a-12a show the burst and collapse load cases for each of these five
strings against the plot format presented previously. In contrast, these same load cases were captured as
results from commercial software in Figures 8b-12b.

Figure 8a—Load cases and exact ellipse: 3-1/2 in. 12.70 lbf/ft L80
18 SPE-185941-MS

Figure 8b—Commercial software results: 3-1/2 in. 12.70 lbf/ft L80

Figure 9a—Load cases and exact ellipse: 7 in. 38.00 lbf/ft T95
SPE-185941-MS 19

Figure 9b—Commercial software results for 7 in. 38.00 lbf/ft T95

Figure 10a—Load cases and exact ellipse: 9-5/8 in. 53.50 lbf/ft N80
20 SPE-185941-MS

Figure 10b—Commercial software results for 9-5/8 in. 53.50 lbf/ft N80

Figure 11a—Load cases and exact ellipse: 13-3/8 in. 72.00 lbf/ft N80
SPE-185941-MS 21

Figure 11b—Commercial software results for 13-3/8 in. 72.00 lbf/ft N80

Figure 12a—Load cases and exact ellipse: 20 in. 94.00 lbf/ft K55
22 SPE-185941-MS

Figure 12b—Commercial software results for 20 in. 94.00 lbf/ft K55

A quick look at the ‘a’ and ‘b’ pairs of Figures 8a-12b indicates a similarity in the overall shape of the
design limits and actual load profiles. However, there are some key differences in the results, including
the definitions of the abscissa and ordinate axes. In the figures based on this paper, both axes are in units
of pressure (psi). The abscissa is the sum of stress and either internal pressure or external pressure. The
ordinate is the actual differential pressure defined by Δp = pi – po. The commercial software uses force
units for the abscissa and pressure units for the ordinate. However, the ordinate is defined by an "effective
differential pressure" which is not the same as the simple Δp = pi – po. Note that the abscissa in the present
work could also be defined in force units by the simple multiplication of pipe cross-sectional area, but the
authors prefer the stress terminology to be consistent with API and to maintain the same units for both axes.
The main accuracy difference between the pairs of figures is in the use of the additional pressure term in
the load cases (pi for collapse loads and po for burst loads) of Figures 8a-12a. Plotting of the loads on these
figures with the new ellipse is exact relative to the limits (axial, burst, collapse and triaxial). The difference
is most notable between the pairs in Figures 8 and 10.
In Figure 8a for the 3-1/2 in. 12.7 lbf/ft L80 tubing, the collapse load is a constant 2500 psi and therefore
is a horizontal line. In Figure 8b, the commercial software implements the internal pressure correction and
the collapse load line is not constant or flat but angled upward.
In Figure 10a for the 9-5/8 in. 53.5 lbf/ft N80 casing, the collapse load line is positioned within the
collapse rating. However, Figure 10b indicates that the load exceeds the collapse rating. Again, this is due
to the use of the prior API TR 5C3 (2008) equation in the commercial software.
Lastly, another difference between the ‘a’ and ‘b’ pairs of these figures is in the definition of the API
collapse limit curve. In the commercial software, all five ‘b’ figures have the collapse curve extending to
the point on the von Mises ellipse where it crosses the abscissa on the tension side. In all five ‘a' figures,
the collapse curve truncates at a σz + pB tensile value in which the "combined loading equivalent grade"
equals 24,000 psi. This is shown as an abrupt terminus on the API collapse curve. The 24,000 psi limit is a
specification in all prior versions of API 5C3 involving the equivalent grade concept.
SPE-185941-MS 23

Conclusions
The API collapse history spans over 75 years. The latest API collapse equations released in 2015 link back
to work done 40 years ago by Lubinski (1975). It is this 1975 technology that drives the new tubular design
ellipse presented in this paper. The main conclusions are:
1. The new ellipse is exact for collapse, burst, and axial loads, unlike all prior ellipses that assume either
the radial stress is zero or the internal/external pressure is zero for collapse/burst respectively.
2. The skewed shape of the new and old ellipse is identical. The difference is the value on each axis,
which causes the load cases to be plotted differently.
3. With the new ellipse, the load cases are more accurate as demonstrated by the results in the example
well design presented in the last section. One casing exceeded the collapse rating using the old
methodology but did not with the new methodology.
4. Differential pressure is explicit in the new formulation, and the effect of internal pressure on collapse
is not an issue as it is with the prior API collapse equations before 2015.
5. How should collapse safety factors be calculated? The conventional method uses the value on the
ordinate of the collapse plot to determine the rating (differential pressure), which is divided by the load
to get the safety factor. Some industry experts prefer the "vector" method instead which calculates the
rating and load based on the magnitude of the vector from the ellipse origin through the load point
to where it intersects the collapse plot. The authors of this paper support the conventional method
for two reasons:
a. The API specifies the collapse rating as the ordinate value.
b. The vector approach could intersect the collapse plot in a different collapse region.
6. Should the industry begin to consider use of the von Mises circle in Figure 3 instead of the ellipse?
The advantage of the circle is simplicity, D/t is explicit, and the axes have physical meaning.
The disadvantage is the standard API ratings cannot be superimposed. With further experimental
investigation recognizing that yield, including collapse, is independent of hydrostatic pressure, it may
be possible to generate updated collapse curves (as well as burst and axial ratings) compatible with the
circle. Such an approach could motivate and justify the vector calculation for the collapse safety factor.

Acknowledgement
The authors thank Frans Klever for his help in sorting through several issues related to the API collapse
history and for providing certain references. Appreciation is extended to Marko Divljanovic for help with
figures. The authors thank Altus Well Expert, Inc. for permission to publish this paper.

Nomenclature
Pca = Reduced collapse performance in API equation
Pc = Collapse performance in API equation
Pci = Collapse rating after internal pressure correction
Sa σa = Axial (tensile) stress in API collapse equations
Y = Yield stress
Ya = Equivalent yield strength for collapse due to axial tension
Pe = Collapse effective external pressure due to internal pressure correction
D = Pipe outside diameter
t = Pipe wall thickness
fycom = Combined equivalent yield strength due to axial force and internal pressure
fymn = Minimum yield strength
24 SPE-185941-MS

σr σθ = Radial and hoop principal stresses in cylindrical coordinates


σz = Axial principal stress in cylindrical coordinates
σVM = Von Mises stress
r = Pipe radius
A, B = Constants in the Lame equations defined by (12)
σn = Neutral axial stress, introduced by Lubinski (1975)
Δσz = Excess axial stress above neutral value, abscissa of von Mises circle (Lubinski, 1975)
ΔP = Differential pressure quantity, ordinate of von Mises circle (Lubinski, 1975)
a, b = Internal and external radii
pi, po = Internal and external pressures
Δp = Pressure differential = (pi - po)
S = Ellipse pressure variable defined by (22)
Ti, To = Ellipse stress variables defined by (22) and (28)
j, k = Pipe dimensional constants defined by (27)
pB = Backup pressure, equals pi for collapse and po for burst

References
API Bull 5C3 (1970), API Bulletin on Formulas and Calculations for Casing, Tubing, Drill Pipe, andLine Pipe Properties,
First Edition, 1970, American Petroleum Institute, Washington, DC.
API Bull 5C3 (1985), API Bulletin on Formulas and Calculations for Casing, Tubing, Drill Pipe, andLine Pipe Properties,
Fourth Edition, 1985, American Petroleum Institute, Washington, DC.
API Bull 5C3 (1989), API Bulletin on Formulas and Calculations for Casing, Tubing, Drill Pipe, andLine Pipe Properties,
Fifth Edition, 1989, American Petroleum Institute, Washington, DC.
API TR 5C3 (2008), ANSI/API Technical Report 5C3/ISO 10400:2007, 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, First Edition, December 2008, American Petroleum Institute, Washington, DC.
API TR 5C3 (October, 2015), Addendum to: ANSI/API Technical Report 5C3/ISO 10400:2007, 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, First Edition, December 2008, Washington, DC.
Clinedinst, W. O (1939), A Rational Expression for the Critical Collapsing Pressure of Pipe under External Pressure,
Drilling and Production Practice, American Petroleum Institute, 1939, pp. 383 — 391.
Clinedinst, W. O. (1980), Calculating Collapse Resistance under Axial Stress using Existing API Collapse Formulas and
the Strain Energy of Distortion Theory of Yielding, Report prepared for the American Petroleum Institute, December
1, 1980.
Clinedinst, W. O. (1985), Collapse Resistance of Pipe, Dissertation presented to Century University, Los Angeles, CA,
1985.
Goodman, M. A. (2017), A Unified Approach to Yield, Buckling and Leak in Well Tubulars, Paper SPE-185855-MS, To
be presented at SPE Europec featured at 79th EAGE Annual Conference & Exhibition, Paris, France, 12-15 June 2017.
Holmquist, J. L. and Nadai, A. (1939), A Theoretical Approach to the Problem of Collapse of Deep-Well Casing, Drilling
and Production Practice, American Petroleum Institute, 1939, pp. 392 — 420.
Johnson, R., Jellison, M. J., and Klementich, E. F. (1987), Triaxial-Load-Capacity Diagrams Provide a New Approach to
Casing and Tubing Design Analysis, SPE Drilling Engineering (September).
Landmark (2015), WELLCATTM R5000.1.13.0 Build 6382, Halliburton Landmark, Houston, TX.
Lubinski, A. (1975), Influence of Neutral Axial Stress on Yield and Collapse of Pipe, (Transactions of the ASME), Journal
of Engineering for Industry, pp. 400 – 407 (May).
Stewart, R. T. (1906), Collapsing Pressure of Bessemer-Steel Lap-welded Tubes 3 In to 10 In in Diameter, Transactions
of the ASME — Volume 27, 1906, pp. 730 – 822.
SPE-185941-MS 25

Appendix: Derivation of Von Mises Circle


The circle equation (17) is constructed by using the Lame equations (11) in the von Mises equation (8).
Since first yield occurs at the inner wall r = a,

(31)

Squaring terms,

(σr - σθ)2 = 4B2/a4


(σθ - σz)2 = B2/a4 + A2 + 2AB/a2 - 2Aσz - 2Bσz/a2 + σz2

(σz – σr)2 = B2/a4 + A2 - 2AB/a2 - 2Aσz + 2Bσz/a2 + σz2

Total 6B2/a4 + 2A2 0 - 4Aσz 0 + 2σz 2

Using these quantities in equation (8),

(32)

From the definitions of A and B in (12) and ΔP and Δσz in (15) and (16), then the circle equation in (17)
is obtained. Note from (11) and (12)

(33)

which Lubinski (1975) calls the "neutral axial stress". The quantity (σr + σθ)/2 is the mean normal stress
in the cross-section and hence A represents the hydrostatic component. So (σz – A) in (32), which is the
buckling fictitious stress, reflects the independence of hydrostatic pressure. See Goodman (2017) for further
analysis and discussion on this topic of hydrostatic independence of yield and buckling.
26 SPE-185941-MS

Conversion Factors

inch X 2.54 E-0 = cm


psi X 6895 E-0 = Pa
lbf X 4.448 E-0 = N

Potrebbero piacerti anche