New method addresses
W.T. in offshore pipeline design 

Alexander Aynbinder Fluor Daniel Inc.
Houston
A new numerical iterative method for
analyzing and designing hightemperature/highpressure (HT/HP) pipelines
provides for considerably greater reduction of wall thickness for these
and for highpressure flow lines than is called for under current design
codes.
Such pipelinesoperating at up to 350° F. and pressures of more than
5,500 psigare currently envisioned for much nearfuture offshore
development (OGJ, Mar. 10, 1997, p. 27).
A joint industry project, supported by such major operating companies
as Amoco, BP, Shell, and others, was launched in April 1997 to address
certain problems that emerge during design, construction, and operation of
HT/HP pipelines. The project identified a major need for moreaccurate
methods for determining pipe wall thickness.
This study focuses only on a single problem in HT/HP pipeline design.
Additional investigations within the joint industry project initiative,
however, will address such other problems essential for developing HT/HP
technology as the effect of cyclic loading on the development of plastic
strains and the dependence of the mechanical properties of pipes on
temperature.^{1}
OverconservatismDetermining wall thickness for restrained
pipelines is important for advancing HT/HP technology. Existing
engineering codes for oil and gas pipelinesASME B31.4 and B31.8have no
limitations with respect to the values of temperature differential and
pressure.
At the same time, the simplified equations in these design codes for
determining hoop, longitudinal, and combined stresses in restrained
pipelines are based on the assumption of elastic properties of steel and
on the omission of radial stresses distributed across the pipe wall. These
assumptions induce significant conservatism into HT/HP pipeline analysis.
A growing number of pipeline engineers in the offshore industry are
concerned that these pipeline codes lead to overconservative designs for
highpressure flow lines^{2} and therefore to greater cost.
The present study shows these codes require a substantialand
unjustifiedincrease of the wall thickness of pipelines. Moreaccurate
methods of stress analysis can reduce wall thickness of HT/HP pipelines.
The purpose of this study has been to develop a moreprecise and
accurate numerical method for determining stresses and strains in HT/HP
pipelines. This goal is attained by considering the nonlinear property of
pipe materials and by properly handling essential radial and hoop stresses
with a specific distribution across the pipe wall.
Thus, the proposed method permits determining the wall thickness by
incorporating the conventional criterion for limiting combined stresses in
the pipeline or the criterion specifying strain limits.
The method considers the nonlinear behavior of pipe materials,
important for stress analysis of HT/HP pipelines because the allowable
stress may considerably exceed the limit of proportionality of the
stressstrain diagram.
The proposed method can be readily implemented on a personal computer
by use of common, commercially available spreadsheet software. And it can
be used with both stress and strain criteria, both of which are equally
suited for engineering design if an appropriate stressstrain diagram is
utilized. 

The proposed model of steel describes the corresponding stressstrain
relationship quite realistically for a complete range of stresses and thus
ensures adequate pipeline analysis for a large range of loading
conditions: up to the burst pressure, that is.
Furthermore, the proposed method ensures that the maximum hoop stress
for pipes due to internal pressure is located on the external surface of
the pipe if stresses across the pipe wall exceed the limit of
proportionality of the stressstrain diagram. This result agrees well with
experiments that show that bursting of pipes due to internal pressure
starts at the external surface of the pipe wall.
Finally, the method can also be applied for analysis of generalpurpose
pipelines with regular internal pressure and temperature differential. The
beneficial effect of the proposed method is considerably less, however:
usually on the order of 37%.
Solution methodThe developed method for evaluating the thickness
of HT/HP pipeline walls considers several essential factors that are not
included in the method of existing design codes.
Radial stresses are assumed to be significant, and their role in
determining pipewall thickness is not neglected. Also, the combined
stress is calculated by including all three components of the stress
tensor: hoop, longitudinal, and radial stresses.
Finally, at the calculation phase, the method considers the inelastic
behavior of pipe steel of the stress tensor components' values.
Note that the ASME B31.4 design code provides that for restrained
pipelines, the allowable combined stress be up to 90% of the yield stress.
At that level of loading, pipe materials exhibit strongly inelastic
behavior.
Toward this goal, the finiteelement method is used to obtain an
approximate distribution of radial and hoop stresses across the pipe wall.
The crosssection of the pipe wall is therefore modeled as a multilayered
annulus.
The cumulative thickness of the annulus layers is set equal to the
thickness of the pipe wall. Further, for every annulus layer, the hoop and
radial stresses are assumed to be constant and equal to the corresponding
average value.
An iterative computational algorithm has been developed to capture the
inelastic behavior of pipeline steel. The algorithm aims at matching the
nonlinear stressstrain constitutive relation.
In the first iteration, therefore, the effective modulus of all layers
is set equal to the elastic modulus. On subsequent iterations, this value
is modified for all layers by calculating the corresponding secant moduli
based on the evaluated level of strains associated with every layer.
Note that, for every iteration, the concept of the secant modulus
facilitates the analysis of the pipeline as a linear elastic system: On
every iteration, the secant moduli are computed based on the nonlinear
stressstrain diagram and are treated as elastic constants.^{3}
The total loadingthat is, pressure and temperature differentialis
completely accounted for on every iteration, and the calculation process
is repeated until displacements of modulus of elasticity and the effective
Poisson's ratio values at adjoining iterations converge.
Stressstrain diagramThe stressstrain diagram can be written in
the form of Equations 1 and 2. (See accompanying box for all equations.)
The effective elastic parameters are calculated by utilizing a
stressstrain diagram that contains three sections: 

 The first part of the stressstrain diagram is linear; it is in line
with engineering traditions and the methodology of the existing pipeline
codes. The corresponding slope is equal to the modulus of elasticity of
steel. This part of the diagram is valid up to the limit of
proportionality between stresses and strains.
 The second part of the diagram is described by a curve that
continuously and smoothly connects the first and the third linear parts
of the diagram.
 The third part of the diagram is linear. The corresponding slope is
equal to hardening modulus. This part of the diagram extends from the
yield strength to the ultimate strength.
A comparison of the
proposed steel stressstrain diagram with some other approximations and
experimental data has been described previously (OGJ, Feb. 20, 1995, p.
70).
The difference between the proposed nonlinear constitutive relation and
the corresponding linear model can be seen in Fig. 1
[59,649 bytes]. This figure presents the ratio of strains
corresponding to the proposed diagram and the corresponding linear model
vs. steel grade specified by API 5L.
The two curves plotted in this figure correspond to two different
values of stress equaling 90% and 100% of yield strength, respectively.
Clearly, analysis of the nonlinear behavior of HT/HP pipelines is
important for highloading regimes. It should be noted, therefore, that
HT/HP pipeline technology might consider strains as high as 90100% of
SMYS.
Stress determinationA straight restrained pipeline can be modeled
as a long cylindrical shell subjected to the axially symmetric loading in
the form of internal and/or external pressure and temperature
differential. The latter is a result of the change of the temperature
conditions during the installation period and the period of pipeline
operation.
Since geometry and loading do not change in the longitudinal direction,
the strain of the cylinder is symmetrical to its axis. Then, the
straindisplacement relations for a restrained long cylinder can be
written as shown in Equation 3.
As has been alluded to previously, every step of the iterative
procedure for determining the corresponding stresses can be considered as
an equivalent elastic problem. The corresponding linear constitutive
relation for axisymmetric elastic isotropic annulus is shown in Equation
4.
The equation^{4} from static condition of equilibrium in terms
of the components of stresses.is shown in Equation 5.
Substituting Equations 3 and 4 into Equation 5 yields the equation of
equilibrium with respect to the displacement (Equation 6).
The solution of Equation 6 with respect to radial displacement is
derived as Equation 7.
Finally, the stresses may be obtained by substituting Equation 7 into
Equations 3 and 4 and deriving Equation 8.
The obtained solution for the components of the stress tensor is valid
for an annulus of arbitrary thickness. It also shows that the radial and
hoop stresses and, therefore, the combined stress vary along the wall. The
effective modulus and Poisson's ratio depend on the combined stress.
Then, a moreaccurate solution may be derived by dividing the annulus
into layers so that each layer has such insignificant thickness that
corresponding stresses can be assumed constant for every layer.


Note that two constants of Equation 8 (C_{1} and C_{2})
must be determined for every layer from the boundary conditions and
conditions of continuity with respect to stress and displacement.
Specifically, the radial stresses at internal and external surfaces are
equal to the internal and external pressure for the first and last layers
(Equation 10).
Further, the radial stresses and displacement for adjoining layers are
equal on the boundary of the layers (Equation 11).
The system of linear algebraic equations can be obtained by combining
Equations 10, 11, 7, and the first equation of Equation 8. It can be
written in the general form as Equation 12.
The linear system of algebraic Equation 12 can be readily solved
yielding constant C_{i} of Equation 8 for all layers. Note that
the effective modulus and Poisson's ratio depend on the combined stress.
Evaluating these mechanical constants can assume that hoop and radial
stresses are constant for every layer; they are equal to the average value
across the wall of the layer (Equation 13).
Average values for radial, hoop, and longitudinal stresses can be
expressed by the equations shown in Equation 14.
The VanMises combined stress can be defined based on the maximum
distortion energy theory (Equation 15). Upon evaluation of the effective
modulus of elasticity by using the combined stress of Equation 15 in
conjunction with considering a simplified uniaxial tensile state of steel,
strains can be readily evaluated for every iteration (Equation 16).
Note that the modulus of elasticity corresponding to design codes is
used on the first iteration.
The iterative process proceeds by evaluating stresses s_{j,i}
by using the steel stress diagram of Equation 1. Strains obtained from
Equation 16 are used in this regard. Next, the effective modulus of
elasticity and the effective Poisson's ratio are computed for the next
iteration (Equation 17).
Note that on every step of the iterative process, all loading is
considered; the process is repeated until displacements of modulus of
elasticity and the effective Poisson's ratio values at adjoining
iterations converge.
It should be emphasized that the process converges to the exact
solution from above (that is, the result on every subsequent iteration is
closer to the exact solution with a reduced margin), providing for the
adequate level of conservatism in the obtained solution.
Numerical examplesThe proposed method for determining HT/HP
pipeline wall thickness can be readily implemented by using digital
computers in conjunction with such readily available software products as
Lotus or Excel.
Two numerical examples are considered here dealing with restrained
pipelines of different dimensions and under quite different loading
conditions.
The pipeline walls are partitioned into seven layers of equal thickness
for both cases. As the iterative algorithm of the proposed method
converges quite rapidly, the presented results were obtained after 6 and
12 iterations for the first and the second numerical example,
respectively.
The first numerical example addresses a restrained pipeline of 10.75
in. OD made of X60 steel. The thickness of its walls is selected to equal
0.58 in. The loading conditions involve the internal pressure of 5,000
psig and the temperature differential of 200° F.
Note that at this level of loading, the maximum value of combined
stresses is approximately equal to 90% of yield strength.


The proposed method of analysis of HT/HP pipelines is compared with the
code method of ASME B31.4 and with the method built upon the linear Lame's
solution.6 The code method is therefore examined in conjunction with the
Tresca equation for determination of combined stress considering only hoop
and longitudinal stresses.
Some pertinent results corresponding to this numerical example are
plotted in Fig. 2
[47,558 bytes], Fig 3
[55,225 bytes], and Fig. 4
[54,462 bytes]. Note that the difference between the maximum and
minimum values of the combined strain and the combined stress for linear
and nonlinear methods is approximately equal to 14 and 17%, respectively.
This discrepancy between the linear and nonlinear methods of HT/HP
pipeline analysis originates from the changes of the modulus of elasticity
and the Poisson's ratio (Fig. 4) associated with the nonlinear
stressstrain diagram. Also, the results of pipeline analyses
corresponding to the three methods of interest in HT/HP design are
summarized in Table 1
[10,907 bytes].
Therefore, the required wall thickness has been calculated based on the
existing B31.4 criterion; that is, the allowable combined stress is 90% of
yield strength. The comparison of results shown in Table 1 clearly
demonstrates the advantages of the proposed nonlinear method of HT/HP
pipeline analysis.
Thus, the proposed method yields much lower maximum combined stress.
Correspondingly, the thickness of pipeline walls can be reduced almost by
50% in comparison with the current ASME B31.4 code.
This method may be applied to pipelines subjected only to internal
and/or external pressure. A substantial number of theoretical and
experimental studies address such loading conditions.
The example of calculations was made for pipe with parameters listed in
Table 1 for which the actual internal burst pressure was obtained by tests
performed by Shell E&P Technology Co. in 19951996.^{2}
The pipe characteristics are shown in Fig. 5
[54,772 bytes] and Fig. 6
[52,703 bytes]. The calculations were provided for internal pressure
of 16,600 and 25,200 psig. The combined stress is calculated in accordance
with Tresca equation by considering hoop and radial stresses. That is, the
tensile longitudinal stresses due to pressure are not considered.
Plotted in Fig. 5 is one of the calculations results, the hoop stresses
distribution across the wall due to pressure of 16,600 psig. The maximum
hoop stress as a result of this pressure, by use of nonlinear solution, is
approximately 47 ksi; that is, the required wall thickness may be 0.78 in.
compared with 1.02 in. per B31.4 method to satisfy existing code
hoopstress criteria.
Note that the linear Lame's solution yields the maximum hoop stress
located on the internal surface of the pipe. Alternatively, the maximum
hoop stress is on the external surface for the nonlinear model of steel.


The same qualitative result of hoopstress distribution was obtained
previously using analytical solution for simplified bilinear approximation
of the stressstrain diagram (OGJ, Apr. 29, 1996, p. 57). This result,
which is obtained by using nonlinear models of steel, agrees with
experiments which show that bursting of pipe from internal pressure starts
at the external surface of the pipe wall.
The suggested method of calculation allows for predicting burst
pressure. The combined and hoop stresses distribution across the pipe wall
due to a burst pressure obtained by the test described by Langer2 are
shown in Fig. 6.
The maximum combined stress due to burst pressure of 25.2 ksi is
approximately 86 ksi. The ultimate tensile strength of tested pipes is
close to the value of combined stress.
A first stepIt should be strongly emphasized that if combined
stress is up to 90% of the yield strength, equal to the allowable combined
stress of B31.4 code, the actual strain that produces this stress may
significantly exceed the elastic strain.
Therefore, the nonlinear stress analysis should be applied for the
successful development of the HT/HP technology. The proposed method can be
considered as the first step in the direction of improving the design of
HT/HP pipelines.
References
 Beckman, J., "Erskine performance indicator for other high
temperature lines," Offshore, September 1997, pp. 10124.
 Langer, C.G., and Shah, B.C., "Code Conflicts for High Pressure
Flowlines and Steel Catenary Risers," OTC Paper 8494, May 1997.
 Desai, S., and Abel, J., Introduction to the Finite Element Method,
A Numerical Method for Engineering Analysis, Van Nostrand Reinhold Co.,
1972.
 Timoshenko, S., Strength of Materials, 3rd Edition, Part II,
Kriegas, Huntington, New York.
The Author
Alexander Aynbinder is a senior project engineer at
Fluor Daniel Co., Houston. Previously, he was in the civil engineering
department of Gulf Interstate Engineering, Houston. Before emigrating to
the U.S. in 1990, Aynbinder was a lead research scientist in the Russian
State Research Institute for Pipeline Construction. He is a graduate of
the Moscow Civil Engineering University and received a PhD in civil
engineering from the Central Research Institute of Civil Structures,
Moscow. Aynbinder is a member of ASME.
Copyright 1998 Oil & Gas Journal. All Rights Reserved.
