You are on page 1of 10

1 Copyright 2012 by ASME

ANALYSIS OF UNBURIED PIPELINE SPANS




Lawrence M. Matta
Stress Engineering Services, Inc.
Houston, Texas, USA
Rhett Dotson
Stress Engineering Services, Inc.
Houston, Texas, USA

ABSTRACT
Support requirements for unburied spans on existing
pipelines can be difficult to assess. An approach for
investigating the adequacy of support of unburied spans is
presented. It begins with a screening based on theoretical
longitudinal stress evaluation. An estimate of the allowable
free span lengths between supports is calculated, considering
the weight of the pipe and its contents, longitudinal stresses
from internal pressure, thermal stresses, and steady wind
loading. Existing span lengths are compared to the allowable
length and any segments exceeding the limit are flagged for
further study. Review by finite element analysis and
comparison with survey data for the identified spans is then
performed. Comparison of the FEA model deflections and
survey data can help determine whether pre-existing stresses
or plastic deformations are present, due, for example, to
installation procedures or ground movement. The FEA results
provide a factor of safety that can be used to help identify
potential safety issues and prioritize mitigation efforts.
INTRODUCTION
On many of the pipelines that crisscross the country, a
number of unburied spans occur at, for example, stream
crossings and gullies. Some of these spans have no
intermediate support structures, while longer spans often have
one or more supports along their length. Since these spans
often cross waterways, a failure of such a span would result in
environmental damage requiring costly cleanups. Many of
these unburied spans are aging, and regulators are bringing
their safety into question.
This paper describes an approach for evaluating the safety
of these spans. In this approach, the pipe spans are initially
screened based on an analysis of the theoretical longitudinal
stress magnitudes. Following the screening process, spans
identified for further review are analyzed with a finite element
analysis (FEA) approach.
In the screening method, theoretical evaluation of stresses
in the unsupported pipeline segments is performed by
estimating the longitudinal stresses in the pipe. The developed
method is intended as a tool to estimate allowable free span
lengths between supports. Longitudinal stresses in the pipe are
calculated considering the weight of the pipe and its contents,
longitudinal stresses from internal pressure due to the Poisson
effect, thermal stresses due to operation of the pipeline at a
temperature different from its installation temperature, and
steady wind loading due to potential sustained winds. The
available stress for bending loads is calculated and used to
determine the maximum allowable free span length. It should
be noted that the analysis can only account for defects and
flaws in the pipe or welds through efficiency factors. Caution
should be used when the presence of defects is known or
suspected.
Finite element analysis is applied to spans that do not pass
the initial screening. This can be considered a higher level
analysis to investigate whether these spans are sufficiently
strong to remain in service and to help prioritize maintenance
efforts.
This paper will first describe the analytical approach used
as a screening method for determining whether or not unburied
pipe spans have adequate support. The following section
provides a description of the FEA modeling approach taken in
this analysis. Two examples are provided showing typical
results of the FEA analysis. Finally, conclusions drawn from
the developed analysis method are discussed.
NOMENCLATURE
C
f
wind force coefficient
D pipe diameter
D
c
outside diameter w/ coatings & insulation
E pipe modulus of elasticity
F
D
stress design factor
G wind gust factor
I
f
wind load importance factor
L
S
allowable support spacing
S
p
longitudinal stress due to internal pressure
S
r
longitudinal stress due to thermal expansion
S
S
longitudinal stress due to elongation
SMYS specified minimum yield stress
T
i
installation temperature
T
op
operating temperature
Proceedings of the 2012 9th International Pipeline Conference
IPC2012
September 24-28, 2012, Calgary, Alberta, Canada
IPC2012-90579
2 Copyright 2012 by ASME
U
max
design wind speed
W combined distributed load
W
wind
distributed wind load
k
d
wind directionality factor
k
z
velocity exposure coefficient
k
zt
topography factor
p pipe internal pressure
t wall thickness
linear coefficient of thermal expansion
lateral pipe displacement
Poissons ratio
SCREENING ANALYSIS APPROACH
The developed approach for investigating the adequacy of
the support of unburied pipe spans begins with screening the
spans by evaluation of the longitudinal stresses in the pipe. This
is then followed by finite element analysis of spans requiring
further investigation. These tasks are detailed below.
Screening Assessment Methodology
The first task of the analysis was to determine a
conservative allowable length of an unsupported pipe span
based on steady-state loading. The analysis was primarily
based on the methodology of API RP-1117, Recommended
Practice for Movement in In-service Pipelines [1]. This
approach is based on evaluation of the longitudinal stresses in
the pipeline and restricting them to a value below an allowable
maximum stress based on the size and grade of the pipe. Figure
1 shows an idealized representation of a span with one support
and two unsupported segments.
An unsupported pipe segment will sag under the
distributed load of its own weight and the weight of its
contents. Steady wind loading also results in a distributed load,
which is assumed to be perpendicular to the weight of the pipe
and its contents. The distributed loads result in bending and
elongation of the pipe, contributing to longitudinal stresses.
Figure 2 shows a schematic of the idealized distributed loading
and supports for the two-segment span from Figure 1.
The pressure of the pipelines contents also generates
longitudinal stress due to the Poisson effect and the restraint of
the soil. The contribution of the internal pressure to the
longitudinal stress may be estimated as:


t
D p
S
p
2

=
(1)
where S
p
is the longitudinal stress due to internal pressure, is
the Poisson ratio of the pipe, p is the internal pressure, D is the
pipe diameter, and t is the wall thickness.
Because the pipe is restrained from longitudinal expansion
and contraction by the soil, temperature changes will generate
longitudinal stresses. The longitudinal tensile stress in the pipe
due to temperature change is estimated using the following
equation:


( )
op i r
T T E S =
(2)

where S
r
is the longitudinal stress due to temperature change,
is the linear coefficient of thermal expansion of the pipe, E is
the modulus of elasticity, T
i
is the installation temperature of
the pipe, and T
op
is the operating temperature of the pipe (which
may not be equal to the temperature of its contents, particularly
in gas service). Determination of temperature induced stresses
is usually difficult, in part because the installation temperature
of the pipe is not typically known. Estimation based on
anticipated temperature extremes may be helpful. Annex A
provides some information of how pipe operating temperatures
may be estimated when actual temperature data is not available.
In this analysis, static wind loading is determined based on
the methodology of ASCE 7 [2]. The Wind Velocity Pressure
is determined from:

Figure 1. Schematic of a single support, two segment span.



Figure 2. Idealized supports and distributed loading of a two
segment span.
3 Copyright 2012 by ASME


max
00256 . 0 U I k k k q
zt d z z
= (3)

where q
z
is the wind pressure, U
max
is the Basic Wind Speed, K
zt

is the Topography Factor, K
z
is the Velocity Pressure Exposure
Coefficient, K
d
is the Directionality Factor, and I is the
Importance Factor. The Basic Wind Speed is determined using
ASCE 7 6.5.4, and is typically equal to the 3-second gust wind
speed. A map of the 3-second gust wind speeds for the
continental United States is shown in Figure 3. The other
factors used in the equation are also determined using the
methodology in ASCE 7 as appropriate for the location of the
span.
Using the Wind Velocity Pressure, the loading on the pipe
is determined from the following equation:


c f z wind
D C G q W =
(4)

where W
wind
is the force of the wind per unit length of pipe, G is
the Gust Factor, C
f
is the force coefficient, and D
c
is the
outside pipe diameter (including coatings and insulation). The
wind loading is assumed to be in the horizontal direction,
perpendicular to the weight loading on the pipe. Note that
since the wind loading tends to be much smaller than the
weight of the pipe and its contents, making conservative
estimates in the wind data does not greatly affect the overall
results of the allowable span length calculations.
Following the methodology of API RP-1117, the allowable
stress remaining is determined from:


r p D A
S S SMYS F S =
(5)

where S
A
is the remaining stress available, F
D
is a design factor,
SMYS is the specified minimum yield stress for the respective
grade of pipe material, and S
p
and S
r
are defined in Eqs. (1) and
(2). Note that the pipe may have pre-existing bending stresses
from installation or ground movement. As the presence of any
pre-existing stresses is unknown, they cannot be included in the
screening analysis. Based on ASME B31.4 Table 403.3.1
(Allowable Values for Pipeline System Stresses) [3], the sum of
the longitudinal stresses should be less than 54% of SMYS, and
this is the design factor F
D
for Eq.(5).
The remaining available stress S
A
is the upper limit of
stress that can be applied by the bending and associated
elongation of the pipe under the steady loading. The axial
stress due to bending of the pipeline under the distributed
loading is estimated following the approach used in API RP-
1117. Calculation of bending stresses in pipe spans can be
calculated using basic beam theory from classic references
[e.g., 4 & 5]. For this analysis, however, we are interested in
developing a conservative method of estimating allowable
segment lengths for spans with a variable number and
placement of supports. In order to develop a general approach
for estimating bending stresses in pipe spans, some assumption
must be made regarding the geometry of the pipeline and
supports. The equation for the longitudinal stress due to

Figure 3. Basic wind speed, defined as 3-second gust speed at 10m.
4 Copyright 2012 by ASME
bending used here is derived from AISC beam diagram 39,
which is for a uniformly loaded beam with four equal length
simply supported segments [6]. This model overestimates the
bending stress (is conservative) for spans with less than 4
segments. The equation used to estimate the maximum free
span length is:


W
OD
ID OD
S
L
T
S
1071 . 0
098175 . 0
4 4
|
|

\
|
=
(6)

where L
s
is the maximum allowable support spacing, OD and
ID are the pipe outside and inside diameters, W is the combined
magnitude of the weight and steady wind loading. The
constants are unitless, and the equation should work for any
consistent units. S
T
is the stress available for bending after
elongation due to bending is accounted for, defined as:

S A T
S S S =
(7)

where S
S
is the longitudinal stress due to elongation of the pipe
due to bending. The longitudinal stress due to elongation of the
pipe can be estimated as:


|
|

\
|
=
S
S
L
E S 67 . 2
(8)

where is the maximum deflection of the pipe in feet, E is in
psi, and L
s
is in feet. In this analysis, the deflection is estimated
based on a uniformly loaded simply supported span. The
equation for the deflection is:


|
|

\
|
=
I E
W
L
S
4
875 . 1
(9)

where and L
s
are in feet, W is in lb/ft, E is in psi, and I is the
pipe moment of inertia in inches
4
. Equations ((6) through ((9)
can be solved simultaneously to find L
s
.
Screening Analysis Application
A specific pipeline may have a number of unburied spans
of different lengths, with and without supports present.
Typically the diameter of the pipeline will remain constant, but
individual spans may be constructed of different types of pipe,
with variations in wall thickness and grade. The screening
assessment described above results in an allowable span length
for pipes with specific characteristics. Therefore, the
calculations need only be performed for each type of span
occurring on the pipeline, not for each individual span.
For example, Table 1 shows the calculated maximum
allowable lengths for several types of pipes that vary in
diameter, wall thickness, and grade. For these examples, the
pipe was assumed to be transporting water. Note that the
results shown were determined using the calculated maximum
allowable working pressure (MAWP) of the pipe and are more
conservative than calculations based on the maximum operating
pressure of the pipeline as set by pressure relief devices. The
lengths of each segment of the analyzed spans are compared to
the allowable maximum length for that type of pipe. Spans
with segments that exceed the allowable length are identified
for further analysis.
FINITE ELEMENT MODELING
The above described screening analysis provides a
benchmark by which to identify spans that may be in need of
additional support. The screening approach is intended to be
conservative, and makes some generalizations about the pipe
spans that may be excessively so. Spans identified as over the
allowable length have, in some cases, been operating for years
without incident. Further analysis is required for pipe span
segments that exceed the maximum length calculated by the
screening tool in order to establish a quantitative factor of
safety that can be used to assess their fitness and possibly
prioritize remediation. FEA of these spans provides a higher
level analysis to investigate spans.
Model Description
Finite element modeling for this study was performed
using the general purpose finite element code ABAQUS.
Models are constructed using first order, two-dimensional pipe
elements (type PIPE21). Ground terminations and support
locations are determined for the pipe spans from survey data.
Typically, 500 feet of pipe on either side of the ground
terminations are added to the modeled spans to adequately
model the boundary conditions. Figure 4 shows how the
boundary conditions are applied for a pipe span with a single
support. For buried sections of the pipe, the vertical motion is
constrained, but axial motion is not. Axial motion is constrained
at one end of the modeled pipe. Pipe supports are modeled as
simply supported vertical boundary conditions. An elastic-
perfectly plastic material model is used in the analysis, with a
yield strength determined by the grade of the pipe.
Table 1. Maximum allowable lengths for 5 example pipe
types.

5 Copyright 2012 by ASME
An unburied pipe span is initially modeled as straight and
horizontal prior to the start of the analysis. The analysis is then
carried out in several steps, which are listed below. Non-linear
geometry is included in all of the analysis steps.

Step 1: Apply gravity to the undeformed pipe
Step 2: Deform pipe to correct ground and support
elevations
Step 3: Apply internal pressure
Step 4: Incrementally increase gravity until performance
limit is reached

In the first step, gravity is applied to the pipeline. In the
second step, vertical boundary conditions are applied to the
buried portions of the pipeline and pipe supports. Next, the
internal pressure of the line is increased to the desired level
typically the maximum allowable operating pressure (MAWP)
or the maximum operating pressure (MOP) of the pipeline at
the span location. Finally, the gravitational force is
incrementally increased until the specified performance limit is
reached or until analysis fails to converge, indicating failure.
While a much more detailed model of the soil and its
interaction with the pipe can be included if deemed necessary, it
requires knowledge of the soil properties. The simplified
modeling of the soil boundary conditions used in this analysis
provided a good match with measured pipe geometry,
indicating that the exposed pipe end conditions were modeled
adequately. Simulations were performed with soil friction
included, and it was found that with friction, the pipe could
withstand higher loads. Due to uncertainty with regard to the
soil friction, it was conservatively neglected.
In this study, steady wind loads at peak gust are generally
an order of magnitude smaller than the weight loads. Since
they are modeled as orthogonal to each other, the effect of wind
on the total distributed load is small. The combined loading
from weight and wind will not be vertical, but the resulting load
direction shifted was less than 10 degrees from vertical at peak
gust wind speeds. Since no friction information about pipe
supports was known, the pipe was assumed to be restrained
horizontally as well as vertically at the support. Since the loads
are modeled as uniform, a 2D analysis is sufficient, but be
aware that the direction of the deflection matches that of the
vector sum of the loads. For more complex load models,
switching the FEA elements to 3D is trivial.
FEA Example 1
Finite element analysis was applied to an unburied pipe
span with a length of about 96 feet and a single off-center
support. The pipe diameter was nominally 12 inches with a wall
thickness of 0.322 inches. The specific gravity of the pipeline
contents was assumed to be 1 (i.e., unit weight of 62.4 lb/ft
3
).
The pipeline contents were not modeled explicitly; instead, they
were accounted for by adjusting the density of the pipe
material.
Figure 5 shows a plot of the displacements for the analyzed
span. The dotted curve is actual pipe geometry as determined
by survey. The solid curve shows the FEA results at the end of
step 3 (with gravity and pressure loads applied to the model). It
can be seen that the shape of the red and blue curves match
very well. The good agreement between the model and the
survey data provides an indication that there are no hidden
pre-existing stresses or plastic deformations of the pipe
segment, since gravity and position of the supports alone
accounts for the shape of the span. The dashed curve shows the
displacements for the span when the distributed load has been
increased to the point of failure. Failure occurs when a plastic
hinge forms on the long segment of the pipe span. The green
curve shows that the long span sags significantly prior to
failure. The analysis showed that distributed loads on the span
could be increased by a factor of 6.5 before the model predicts
failure.
The plastic strain at the support as a function of
gravitational load is shown in Figure 6. The curve shows that
plastic deformation begins at 2.7gs and increases steadily until
6 gs. At 6 gs, the plastic deformation rapidly increases until
the analysis fails to converge at 6.5 gs. At this point, a plastic
hinge has formed at the support location and it is incapable of
supporting additional load.
The results of the analysis show that despite failing the
initial screening, this span has a high factor of safety against
failure caused by gravitational loads. The analysis showed that
the loads could be incrementally increased to 6.5 gs at which

Figure 4. Applied boundary conditions for finite element modeling
6 Copyright 2012 by ASME
point the system became numerically unstable. At these loads,
the maximum plastic strain in the pipe was approximately 10%
and located at the pipe support.
The actual ductility of the pipe material may not permit
plastic strains as high as 10%, and the failure limits cannot be
predicted due to unknown factors including material properties,
defects, and weld quality. Some reasonable limit must be
chosen, and for this analysis a 2% maximum plastic strain was
used as the allowable limit. This limit was chosen because it
governs piping systems with longitudinal stresses greater than
yield in ASME B31.4, Section 451.12 (non-cyclic displacement
of existing pipelines) and B31.8, Section 833.5 [3, 7]. Also,
ISO 13628-7 limits the principal structural strain in a limit-state
analysis to 2% [8]. A joint factor of 1 was used in this analysis
and no defects were considered.
The gravity scaling factor at which the maximum plastic
strain reaches 2% provides a factor of safety. In other words, if
the model indicates that multiplying the force of gravity by 3
results in a maximum plastic strain of 2%, then the load could
be increased by a factor of 3 before the strain limit is reached.
Based on a 2% strain limit, Figure 6 shows that the ultimate
capacity of the example span would be 3.5 gs, which is
substantially larger than typical operational factors of safety.
Based on this analysis, this span has sufficient factors of safety
against failure by gravitational loads to continue operating.
Note that caution should be exercised regarding the
potential for local buckling on the compressive side of the
bend. Following AISC, a circular section is considered compact
at a D/t ratio of 58.3 for a Class B pipe (or 50 for X42 pipe) [9].
In all of our cases, the pipes were compact, so failure by plastic

Figure 5. Example 1 pipe span configuration and deflections

Figure 6. Example 1 plastic strain distribution
7 Copyright 2012 by ASME
hinge is expected and local buckling shouldnt control. Also,
local stressing at supports was checked by ensuring that the
pipe didnt yield in shear. The FEA pipe elements utilized did
not account for local cross section deformation. If warranted in
a specific case, cross section deformation can be included in the
FEA.
FEA Example 2
Another analyzed pipe span has an exposed length of 75
feet and no supports. As in the previous example, the pipeline
was initially modeled as horizontal, and loading was applied in
several steps. In the first step, gravity was applied. In the
second step, a vertical displacement was applied to one side of
the pipeline to model the correct elevation change between the
ends of the span. Next, the internal pressure of the line was
increased to the MAWP, in this case 980 psi. It was observed,
however, that if the pipe was modeled as horizontal at the
locations of the ground penetrations, the sag of the modeled
pipe was much less than observed in the survey data. The
existing sag in the pipe span does not appear to be due to the
gravity load alone. Review of the survey data showed that the
pipeline exits the ground penetrations at an angle to the
horizontal. A correction to the model was made by matching the
slope of the pipe at the ends to that in the survey data (the pipe
was kept straight for 20 feet into the ground on each side then
turned to horizontal). Correcting for the burial angle of the pipe
resulted in a much better match between the model and the
survey data, as shown in Figure 7, indicating that the apparent
sag is the result of the angle at which the pipe enters the soil at
the ends of the span. The angle of ground penetration may be
an artifact of installation, but the history of the pipe is not
known. The good quality of the agreement between the
predicted span shape and the survey data again indicates that
the model is substantially capturing stresses in the span, and
that pre-existing stresses due to installation or ground
movement are accounted for.
After matching the model displacement to the survey data, the
gravitational force was incrementally increased (simulating
increased weight of the fluid carried) until the analysis failed to
converge, indicating failure of the span. In this case, the
example span reached the 2% plastic strain limit at 3.4g.
Figure 8 shows the survey data, the modeled displacement
of the pipe at 1g (prior to increasing the gravity load factor) and
the displacement when the span reaches the 2% plastic strain
limit. The span sags significantly below horizontal prior to
reaching the limit load.
CONCLUSIONS
While the majority of transmission pipelines are
underground, unburied spans occur at, for example, stream
crossings and gullies. Some of these spans have no support
structures, while longer spans may have one or more supports
along their length. The developed methods provide an approach
to define a maximum allowable unsupported length of unburied
pipeline segments and to further investigate the performance of
segments that exceed the determined limit.
The first part of this method is aimed at determining an
allowable length for segments of pipe between supports. This
allowable length provides a means of screening the pipe spans
to determine which are in need of further analysis. This
provides an economical approach to limiting the number of
spans requiring FEA and potentially limiting detailed surveys to
identified spans. Longitudinal stresses in the pipe were
calculated considering the weight of the pipe and its contents,
longitudinal stresses from internal pressure due to the Poisson
effect, thermal stresses due to operation of the pipeline at a
temperature different from its installation temperature, and
steady wind loading due to potential sustained winds. Actual
determination of thermal stresses can prove difficult, as it
requires, for example, some estimate of the temperature of the
pipe during installation. Wind loads can be estimated
conservatively, as they typically turn out to be much smaller
than the weight loads on the pipe. The remaining available
stress for bending is determined by subtracting the stresses due
to internal pressure and thermal effects from the maximum
allowable longitudinal stress for the pipe. In this study, the
maximum allowable stress levels are based on the limits
provided in ASME B31.4 and B31.8. This available stress is
then used to determine the maximum allowable free span length
based on the distributed load from the weight and wind force.
The developed method is not intended for calculation of design
stresses, especially with regards to hoop stress. This simple
analysis cannot account for individual span characteristics such
as pre-existing stresses in the pipe due to bending during
installation or ground movement unless these stresses are
known and added into the calculation.

Figure 7. Example 2 pipe shape with and without ground
angle correction
8 Copyright 2012 by ASME
The span length analysis is expected to be conservative,
and pipe spans that have been operating for years have been
found to exceed the determined allowable length. A
straightforward finite element analysis approach was used to
investigate pipe spans that exceeded the calculated length
limits. The FEA analysis can be used to help identify if a pipe
span is under significant pre-existing stress by comparing the
calculated pipe displacements under normal loading to survey
data of the existing pipe span. If the geometry is found to be in
good agreement, it is reasonable to assume that the FEA model
is adequately capturing the stress profile of the pipe. If,
however, the FEA predicted displacements do not match the
actual shape of the pipe, then the answer is not as simple. In
this case, some additional information about the span will be
required to adequately model it. Note that we have assumed
that the pipes were laid in a relatively unstressed state, and
thermal stresses may exist if the installation temperature differs
from the current pipe temperature. While a uniform tensile or
compressive load will not cause lateral deflection in a straight
pipe, it will affect the amplitude of deflection in a pipe with
side loads due to secondary moment effects. How much effect
compression or tension will have on the deflected shape of a
pipe span has not been investigated in this study.
If the FEA model proves to give good results at normal
loading, the model is then used to determine a factor of safety
for the span based on chosen failure criteria. In this study, we
have used a 2% strain limit based on current standards. The
factor of safety is determined by increasing the force of gravity
in the model until the plastic strain limit is reached. The factor
of safety is then defined as the gravity force in gs at the strain
limit. This factor of safety value can be used to compare spans
and prioritize remediation as deemed necessary.
This analysis is based on steady loading only. A complete
analysis should include an investigation of the potential for
vortex induced vibrations due to wind (or water, if flooding is
possible). If appropriate, an analysis of fatigue due to vortex
induced vibration should be performed.
It is notable that there is a large gap between the limit
applied for the analytical method (54% SMYS), and the limit
for FEA (2% strain). The FEA model is not intended to give an

Figure 8. Example 2 span analysis results
9 Copyright 2012 by ASME
exact loading of failure, or even to exactly identify the 2%
strain load, but rather to give a factor of safety. While this
analytical methodology described in this paper allows the user
to develop an allowable limit for the lengths of unburied spans,
and FEA permits the analysis of spans that exceed this length
by determining a safety factor for the span, the analysis cannot
provide guidance on what safety factors can be tolerated or for
how long. What factor of safety is ultimately deemed
acceptable must be based on the confidence level in the results,
and perhaps testing. Ultimately, the choice of what limits to
tolerate and how and when to remediate spans that exceed the
limits must be determined by the operators.
REFERENCES
1. API RP-1117, Recommended Practice for Movement in In-
service Pipelines, 3
rd
Edition, 2008.
2. ASCE 7, Minimum Design Loads for Buildings and Other
Structures, Chapter 6, American Society of Civil
Engineers, 2005.
3. ASME B31.4, Pipeline Transportation Systems for Liquid
Hydrocarbons and Other Liquids, 2009 Edition.
4. Young, W.C., Budynas, R.G., and Sadegh, A., Roark's
Formulas for Stress and Strain, 8
th
ed., Revised, McGraw-
Hill, Inc., 2012.
5. Shigley, J.E., Mechanical Engineering Design, 2
nd
ed.,
McGraw-Hill, Inc., 1972.
6. AISC Manual of Steel Construction, 2
nd
Ed., 1994.
7. ASME B31.8, Gas Transmission and Distribution Piping
Systems, 2009 Edition.
8. ISO 13628-7, Petroleum and natural gas industries --
Design and operation of subsea production systems -- Part
7: Completion/workover riser systems, 2005 Edition.
9. AISC Specification for Structural Steel Buildings 2010
(ANSI/AISC 360-10).
10. http://www.climate-zone.com/climate/united-
states/texas/houston/
11. Wu, J. and Nofziger, D.L., Incorporating Temperature
Effects on Pesticide Degradation into a Management
Model, J. Environmental Quality, 28:92-100, 1999.


10 Copyright 2012 by ASME
ANNEX A
THERMAL STRESSES IN PIPE SPANS



Long runs of pipeline tend to be axially restrained, or
anchored, by the surrounding soil at some effective length. The
length over which this occurs is often referred to as the virtual
anchor. Pipe branches and bends can locally affect the
anchoring. Thermal expansion of the pipe, resulting from
transportation of hot or cold fluids and from variations in soil
temperatures, can result in stresses in the pipe that are
dependent on the soil anchoring.
For long sections of straight buried pipe, the longitudinal
stress due to thermal expansion is


( )
0
T T E S
T
=
(A.1)

where

S
T
= longitudinal stress due to thermal expansion
E = pipe modulus of elasticity

= coefficient of thermal expansion


T = current pipe temperature
T
0
= pipe installation temperature

For unburied spans, the pipe is longitudinally restrained at
the ends based on the anchoring of the buried pipe. If the pipe
is straight, the longitudinal stress can be predicted using the
above equation. If the pipe has bends, or buckles under
compressive load, the thermal stress will be reduced (possibly
at the expense of bending loads or plastic deformation).
The thermal stress in the pipeline depends on the
temperature of the pipe when it was installed and the current
temperature of the pipeline. Installation temperatures depend
on the weather at the time of installation. As an example,
consider installation of a pipeline in Houston, TX. Climatic
information for Houston gives a typical January average daily
temperature of about 52F and a July average daily temperature
of about 83F [10]. For a long pipeline, it is likely that the fluid
reaches an equilibrium temperature with the surrounding soil at
the average burial depth of the pipe. Daily temperature
variations penetrate only a few inches into the soil, so annual
variations are of interest for buried pipelines. Note that the
fluid temperature will not change significantly when the pipe
emerges for a short unburied span, as long as the contents are
flowing. The annual variation of daily average soil temperature
at different depths can be estimated using a sinusoidal function
[11]. The model assumes a sinusoidal variation of surface
temperature, as


( )
(


+ =
days
t t
A T T
a S
365
2
sin
0
0

(A.2)
where

date start arbitrary an from time lag
amplitude re temperatu smoothed
re temperatu surface average
re temperatu surface smoothed
0
0
=
=
=
=
t
A
T
T
A
S


The graph in Figure A.1 shows an example of calculated
soil temperatures at various depths based on average
temperature information for Houston, TX based on estimated
soil properties. As burial depth increases, the seasonal
excursion from the annual mean temperatures becomes smaller,
and the phase lags the surface temperature cycle.





0 100 200 300 400
50
60
70
80
90
Surface
5 ft
10 ft
15 ft
20 ft
Day of Year
A
p
p
r
o
x
i
m
a
t
e

S
o
i
l

T
e
m
p
e
r
a
t
u
r
e
,

F

Figure A.1. Approximate soil temperatures for Houston, TX

You might also like