Journal of Pressure Equipment and Systems 5 (2007) 66-74

Elastic-plastic behaviours of pressurised pipes under cyclic thermal

stresses with temperature gradients
Chang-Sik Oh, Yun-Jae Kim *
Department of Mechanical Engineering, Korea University, 5 Ka, Anam-dong, Sungbuk-Ku, Seoul 136-701, Korea

Abstract
Elastic-plastic behaviours of pressurised pipes under cyclic thermal stresses are investigated in this paper, incorporating temperature
(and resulting the yield strength) gradient across the pipe thickness. Based on analytical investigations, explicit equations for various
stress regimes representing elastic-plastic behaviours are obtained. In the limiting case of no temperature (yield strength) gradient
across the thickness, proposed equations recover those for the Bree diagram. For the case when temperature (yield strength) gradient
exists across the thickness, overall shapes of stress regimes are slightly different from those in the Bree diagram. The main difference
is the shakedown and elastic regimes. Due to the yield strength variation across the thickness, shakedown should be checked at the
inner wall, at the outer wall, and at both walls. Accordingly elastic limit should be checked at the inner and at the outer walls. As results, boundaries between elastic and shakedown regimes in the proposed diagram are different from those in the Bree diagram.
Another difference is the plastic collapse regime which is not present in the Bree diagram. Boundaries between shakedown and ratchetting or reverse plastic in the proposed diagram are quite similar to those in the Bree diagram.
Keywords: Cyclic thermal stresses; Elastic-plastic behaviours; Pressurised pipes; Temperature and strength gradient.

1. Introduction
Pressurised pipes in power and chemical plants are
often subjected to repeated thermal loads. When the
loading of such components is severe, then cyclic plastic straining and/or incremental collapse or due to
gross distortion (ratcheting) could occur. Thus knowledge of the load conditions is required to avoid ratcheting in the design of such components. In obtaining
analytical solutions for elastic-plastic behaviours of
pressurised pipes under cyclic thermal stresses, Bree [1,
2] used a uniaxial model to simplify the problem, of
which the result is now well-known as the Bree diagram [3]. In his analysis, however, through-thickness
variations of radial and hoop stresses (due to the application of internal pressure) and of axial loading
(which may be caused by closed ends or axial restraint,
for example) were not fully considered. Later these
effects were systematically investigated via detailed
finite element analyses [4-8], including effects of
through-thickness variations of stress, of axial loading
and of the end restraint. Furthermore, the Bree analysis
was based on the assumption that the yield strength is
constant over the entire cycle. Noting that yield
strengths of materials are typically dependent on temperature, the effect of the temperature-dependent yield
strength was also incorporated into elastic-plastic behaviours of pressurised pipes under cyclic thermal
stresses [8, 9].
*

Corresponding author.
E-mail address: kimy0308@korea.ac.kr (Yun-Jae Kim).

Above analyses assume that temperatures in both

inner and outer walls are the same during thermal cycles, that is, during operating and shut-down conditions,
temperatures in both inner and outer walls increase to
the hot temperature and decrease to the cold temperature. Thus no temperature gradient exists across the
pipe thickness in the operating condition. In certain
applications, however, temperature at the inner wall of
the pipe can be different from that at the outer wall,
implying that temperature gradient can exist. Typical
examples include, for instance, heat exchangers and
co-axial pipes [10]. In such cases, yield strengths
through the thickness can be different due to temperature gradient, which makes elastic-plastic behaviours
of pressurised pipes different from existing ones.
This paper investigates elastic-plastic behaviours of
pressurised pipes under cyclic thermal stresses. An
emphasis is given on temperature gradient across the
pipe thickness and the resulting variations in yield
strengths, which is not fully considered in existing
works. Section 2 sets up a model for the present work.
Explicit equations for various stress regimes representing elastic-plastic behaviours are given in Section
3. Section 4 validates the proposed method against full
cyclic elastic-perfectly plastic FE analysis. The present
work is concluded in Section 5.
2. A model
This paper considers a pressurised pipe subject to
cyclic temperature gradients, as depicted in Fig. 1. The
mean radius and thickness of the pipe are denoted by

67

Journal of Pressure Equipment and Systems 5 (2007) 66-74

rm and 2w, respectively, and the pipe is subject to constant internal pressure P. The pipe is assumed to be
subject to cyclic thermal stresses as follows (see also
Fig. 1). Initially the pipe is subject to a constant temperature To both at the inner and outer walls. The
start-up and operating conditions are that the temperature at the inner wall is increases by To+T (T>0),
and remains constant. The temperature at the outer
wall, on the other hand, remains To. The shut-down
condition is the same as the initial condition of constant temperature To at both inner and outer walls. This
thermal loading cycle is repeated. The objective of
this problem is to quantify the effect of cyclic temperature gradients through the pipe thickness on elastic-plastic behaviours of pressurised tubes. However, it
should be pointed out that the problem is highly idealised in the sense that temperature at the outer wall is
assumed to be constant at start-up and shutdown. Consideration of temperature changes at the outer wall
adds much more complexities to the problem, and thus
is neglected in the present work.
As the yield strength of the material should depend
on temperature, the yield strength should vary across
the thickness under steady-state temperature conditions.
Let us denote the yield strength at temperature To as
YL and that at (To +T) as YH. Introducing the
non-dimensional variable defined by the ratio of the
yield strengths:

YL
YH

The value of =1 corresponds to the case of the

constant yield strength across the thickness, and thus to
the case of the temperature-independent yield strength.
As the yield strength typically decreases with increasing temperature, the value of should be greater than
unity, 1.0.
For simplicity, following assumptions are made in
this work to investigate elastic-plastic behaviours of
this problem:
(i) Material is assumed to be elastic-perfectly plastic.
(ii) The length of the pipe is sufficiently long so that
the end effect is negligible.
(iii) The pipe is assumed to be thin and open-ended,
and thus only the non-zero stress component is the
hoop stress (both axial and radial stresses are
zero).
(iv) Temperature distribution along the thickness is
assumed to be linear.
(v) Yield strength decreases linearly with increasing
temperature.
Note that the last assumption of linear dependence of
the yield strength on temperature is imposed to analytically develop diagrams for elastic-plastic behaviours. However, it will be shown that such an assumption is not central, and the proposed method can be
applied to any functional dependence of the yield
strength on temperature.
3. Stress regimes

(1)

Fig. 1. Schematic diagram of a pressurised pipe with linear

temperature gradient across the thickness.

In this section, boundaries of stress regimes for elastic-plastic behaviours of pressurised pipes under cyclic
temperature gradients across the thickness will be derived in dimensionless forms. For compact notation,
the following non-dimensional variables are introduced.
The pipe is subject to primary stress, p, (due to internal pressure) and secondary (thermal) stress, t (due to
temperature gradient), given by

Fig. 2. Stress Regimes for temperature-dependent yield strength

as =2.

Chang-Sik Oh and Yun-Jae Kim / Elastic-plastic behaviours of pressurised pipes under cyclic thermal stresses

Pr
2w
ET x
x

t
1 1
w
2 w

(2)

The total stress in the pipe can be found simply by

linear superposition:

p t

P r ET x

2w
2 w

Y t / YH

x
d 2 p
w
1

Y X 1

(6)

for the case when the compressive stress at the inner

wall reaches the yield strength, and

Y X

(7)

for the case when the tensile stress at the inner wall
reaches the yield strength. By equating these two
equations, the crossing point XE (shown in Fig. 2) is
given as

(4)

Note that stresses are normalized with respect to the

yield strength at temperature (To +T), YH.
The diagram with different stress regimes for the
case of =2 is shown in Fig. 2. The diagram consists
of five different stress regimes: (i) the elastic regime
(E), (ii) the shakedown regime (S1, S2 and S3), (iii) the
reversed plasticity regime (P), (iv) the ratcheting regime (R1 and R2) and (v) the plastic collapse regime
(F). In Fig. 2, the subscript 1 and 3 denote the case
when shakedown or ratcheting occurs at the outer wall
and at the inner wall, respectively. The subscript 2
indicates the case when shakedown occurs at both
walls. For instance, the regime S3 denotes the case
when shakedown occurs at the outer wall, and the regime R2 indicates the case when ratcheting occurs at
both walls. Noting that thermal stresses are
self-equilibrating, boundaries of stress regimes can be
derived from the condition that the average stress in
the thickness equals to twice the mechanical stress due
to internal pressure:
1

Fig. 3b to the case when the stress at the outer wall

yields. Based on the results in Fig. 3, boundaries for
the elastic regime are given by the following two equations:

(3)

Boundaries of stress regimes will be given in terms

of normalized primary and secondary stresses, defined
by

X p / YH

68

(a)

(5)

where denotes hoop stress in the pipe thickness. In

the subsequent sub-sections, explicit equations for
boundaries are given.
3.1. Boundaries for the elastic regime
Stress states for the boundary of the elastic regime
are schematically shown in Figs. 3a and 3b for the case
of =2. Noting that the yield strength at the inner wall
is different from that at the outer wall upon thermal
loading, the stress state should be checked not only at
the inner wall but also at the outer wall. The stress state
in Fig. 3a corresponds to the case when the stress at the
inner wall reaches the yield strength, whereas that in

(b)
Fig. 3. Stress distribution at the boundary (a) between the E and
S3 regimes and (b) between the E and S1 regimes.

Journal of Pressure Equipment and Systems 5 (2007) 66-74

69

XE

(8)

3.2. Boundaries between S1 and S2 (R1 and R2) regimes

Fig. 4 illustrates stress distributions for combined
pressure and thermal loading at initial start-up and
shutdown in the S1 regime, including stress distributions due to pressure and thermal loading. At start-up,
the stress in a portion of the outer wall exceeds the
yield strength and thus yielding takes place. At shutdown, the thermal stress is removed and the residual
stress distribution is also shown in Fig. 4. For the
re-start-up, addition of the thermal stress to the residual
stress gives the stress distribution that is exactly the
same as that for the first start-up. Thus the yield condition is nowhere violated, indicating that shakedown
condition is satisfied.
From the condition that the (compressive) stress in
the inner wall at the initial start-up equals the yield
strength, together with Eq. (5), the boundary between
S1 and S2 (and R1 and R2) can be expressed by

1 2 X 2Y 1 4

(9)

The extent of the plastic zone can be also estimated

from the location of the elastic-plastic interface, a (Fig.
4), which is given by

4 1
2
a

1
1 2 X

2
w
Y

2
Y

(10)

Fig. 5. Stress distribution at the boundary between the S2 and S3

regimes.

3.3. Boundary between S2 and S3 regimes

Fig. 5 illustrates stress distributions at initial start-up
and shutdown in the S3 regime. At start-up, yielding
takes place in a portion of the outer wall. The residual
stress at shutdown is also shown in Fig. 5. For the
re-start-up, addition of the thermal stress to the residual
stress gives the stress distribution that is the same as
that for the first start-up, which again indicates that
shakedown condition is satisfied.
From the condition that the (tensile) stress at the
outer wall at the initial start-up equals the yield
strength, together with Eq. (5), the boundary between
S2 and S3 is given by

2 X 1 2Y 1 4 2

(11)

The extent of the plastic zone can be estimated from

the location of the elastic-plastic interface, b (Fig. 5),
which is given by

4 1 X 2 1
b
1
2
w
2 1 Y 1

Fig. 4. Stress distribution at the boundary between the S1 and S2

regimes.

(12)

3.4. Boundary between S2 and P regimes

Note that, in the regime S2, the shakedown condition
is satisfied at both inner and outer walls. A stress distribution corresponding to such a condition is illustrated in Fig. 6. At the start-up, yielding takes place at
both inner and outer walls. The residual stress distribution upon removing thermal stress is shown in Fig. 6.
For the re-start-up, addition of the thermal stress to the
residual stress recovers the stress distribution that is
the same as the first start-up without violating yielding
condition everywhere, which in turn indicates that
shakedown condition is satisfied.

Chang-Sik Oh and Yun-Jae Kim / Elastic-plastic behaviours of pressurised pipes under cyclic thermal stresses

70

Extents of plastic zones at the inner and outer walls

can be estimated from the locations of elastic-plastic
interfaces, a (for the inner wall) and b (for outer wall)
(see Fig. 6), which are given by

1 a 2 2 1 a

2 1 Y 3 2 2 3 4 2Y 1 X

2Y 1

1 b 2 2 1 b

(15)

2 1 Y 3 2 2 3 4 2Y 1 X

2Y 1

Fig. 6. Stress distribution at the boundary between the S2 and P

regimes.

Fig. 8. Stress distribution at the boundary between the S1 and R1

regimes.

Fig. 7. Stress distribution at the boundary between the P and R2

regimes.

From the condition that the (tensile) stress in the inner wall for shutdown equals the yield strength (YL),
the boundary between S2 and P is given by

Y 1

(13)

The position of XS, shown in Fig. 2, can be found by

equating Eq. (9) and Eq. (13):

XS

2 4 1
2 3

(14)
Fig. 9. Stress distribution in the F regime.

Journal of Pressure Equipment and Systems 5 (2007) 66-74

71

3.5. Boundary between R2and P regimes

Fig. 7 shows stress distributions for the boundary
between P and R2. It shows that, at start-up, yielding
takes place both at the inner wall and at the outer wall.
Upon removing thermal stress, residual stresses in portions of both the inner wall and the outer wall equal the
yield strength. If the magnitude of the residual stress in
those portions is greater than the yield strength, ratcheting occurs at both walls in the subsequent thermal
cycles. From this condition together with Eq. (5), the
equation for the boundary between R2 and P can be
obtained:
4 X 1

8 1 1 5 2 2 1

2Y 1 2Y 1

(16)

3.6. Boundary between S1 and R1 regimes

The equation for the boundary between S1 and R1
can be obtained by stress distributions shown in Fig. 8.
At start-up, yielding takes place at the outer wall. At
shutdown, the residual stress in a portion of the inner
wall equals the yield strength. If the magnitude of the
residual stress is greater than the yield strength, ratcheting occurs at the outer wall in the subsequent thermal
cycles. From this condition together with Eq. (5), the
equation for the boundary between S1 and R1 can be
obtained:
2 X

Y2
2Y 1

(17)

3.7. Boundary for the F regime

The regime F corresponds to plastic collapse due to
the hoop stress resulting from internal pressure. Fig. 9
shows typical stress distributions for the regime F. It
shows that, if the hoop stress due to internal pressure is
greater than the average of YL and YH, and thermal
stress, then plasticity occurs in the entire thickness of
the tube. Based on this condition, the boundary for
the regime F is given by

Fig. 10. Effect of temperature dependent yield strength on stress

regimes and three cases considered for full-cyclic elastic-plastic
FE analyses to validate the proposed shakedown diagram.

XF

(18)

3.8. Comparison of stress regimes for temperature-independent yield strengths

In the previous sub-sections, closed-form equations
of boundaries for elastic-plastic behaviours of pressurised pipes under cyclic thermal stresses, incorporating
the temperature effect on the yield strength of the material. As the proposed equations explicitly include the
term, detailed shapes of the pressure-thermal stress
diagram can be constructed when the temperature difference (and thus the strength difference) in the operating condition is known.
It is worth comparing the proposed diagram with the
Bree diagram. It should be noted that, for the case of
=1.0, the proposed diagram recovers the well-known
Bree diagram [1]. For >1.0, the proposed diagram
differs from the Bree diagram. This is because the Bree
diagram assumes no temperature gradient across the
thickness during operating condition. To estimate elastic-plastic behaviours of the problem considered in the
present work, one could use the Bree diagram with the
thermal stress corresponding to the average temperature Tave=(T+T/2). Fig. 10 compares the proposed
diagram with the Bree diagram for the case of =2.0.
Due to the fact that strength properties corresponding
to the average temperature to use the Bree diagram, the
elastic and shakedown regimes in the Bree diagram are
smaller than those in the proposed diagram. However,
it should be noted that the original intention of the
Bree diagram was different from the present work, and
thus direct comparison is not really meaningful.
4. Validation against full cyclic elastic-plastic FE
analysis
To validate the proposed method to estimate elastic-plastic behaviours of pressurised pipes under temperature gradients, full cyclic elastic-perfectly plastic
FE analysis is performed using ABAQUS [11]. Resulting behaviours are compared with predictions using
the proposed diagrams. A thin strip across the pipe
thickness is meshed with twenty axi-symmetric elements (CAX4T in ABAQUS). Note that coupled temperature-displacement analysis can be performed using
the CAX4T element [11]. In the FE model, the pipe
has the mean radius r=52.0(mm) and the thickness
2w=4.0(mm). For the kinematic boundary condition,
the nodes on the bottom side of the thin strip are constrained in the y-direction, inducing plane stress condition. The internal pressure is applied on the pipe inner
wall. Note that axial stress due to internal pressure is
neglected. A linear temperature gradient is applied
across the pipe thickness. Material properties are worth
noting. For more realistic comparisons, actual
time-dependent material properties of 316 stainless

Chang-Sik Oh and Yun-Jae Kim / Elastic-plastic behaviours of pressurised pipes under cyclic thermal stresses

72

steels were taken from Ref. [12]. Fig. 11 shows relevant material properties (the thermal expansion coefficient, ; Youngs modulus, E; and the yield strength,
Y) as a function of temperatures (in the range of 0 to
800oC). For analyses, regression fits to temperature-dependence data were performed, leading to the
following expression:
15.24 8.88 103 T 4.63 106 T 2

(19)

(0 T 800o C )
E 197.964 0.059T 32.32 106 T 2

(20)

(0 T 800o C )

206.843

(0 T 150o C )

3 2
249.1565 0.3246T 0.2834 10 T
Y
o
o
(150 C T 565 C )

3 2
1522.3422 3.6965T 2.263 10 T

o
o
(565 C T 800 C )

(a)

(21)

Both the thermal expansion coefficient and Youngs

modulus can be fitted using a quadratic function in the
entire range of temperature. However, for the yield
strength, three distinct regions are fitted using constant
and quadratic functions.
Table 1. Three cases considered for full-cyclic elastic-plastic FE
analyses.
T
Case
y
p
t
X
Y
[]
155.1
Tout 566
0
155.7
0
1.5
1
Tin
648
77.25
2
3

Tout

632

90.05

Tin

698

44.73

Tout

400

164.7

Tin

642

81.92

53.68

90.35

1.2

65.53

352.2

0.8

4.3

(b)

Table 2. Expected elastic-plastic behaviours for three cases considered.

Case

Constant yield
strength (=1)

Temperature dependent
yield strength (=2)

Boundary between
E&S2
- No plastic strain

R1 Regime
- Uniaxial strain

S3 Regime
- Shakedown in inner
wall
S1 Regime
- Shakedown in outer
wall

R2 Regime
- Ratcheting

S2 Regime
- Reverse plasticity

(c)
Fig. 11. Variations of material properties of SS316 with temperatures [12]: (a) the thermal expansion, (b) Youngs modulus, and
(c) the yield strength.

Journal of Pressure Equipment and Systems 5 (2007) 66-74

73

For the analysis, three generic cases (as indicated in

Fig. 10 and summarized in Table 1) were chosen for
the following reasons. In view of comparing two different diagrams (one for temperature-independent yield
strength and the other for temperature-dependent yield
strength, proposed in this paper), the conventional diagram predicts that the case 1 corresponds to the elastic
regime, whereas the proposed one to the shakedown
regime at the inner wall. The case 2 locates in the
ratcheting regime (at the outer wall) according to the
conventional diagram but in the shakedown regime (at
the outer wall) according to the proposed diagram. For
the case 3, ratcheting should occur at both walls from
the conventional diagram, but reverse plasticity should
occur from the proposed diagram. Expected behaviours
for these three cases are summarized in Table 2.
Variations of equivalent plastic strains with a few
thermal cycles, resulting from full cyclic elastic-perfectly plastic FE analysis, are shown in Fig. 12,
for three different cases. As equivalent plastic strains
are plotted, they are positive. It shows that shakedown
occurs for the case 1 and case 2, as predicted from the
present method. For the case 3, the results in Fig. 12c
show that ratcheting occurs at both (inner and outer
walls), as also predicted from the present method.

(a)

(b)

(c)
Fig. 12. FE results of full-cyclic FE analyses in three different
cases; (a) the case 1: shakedown in inner wall; (b) the case 2:
shakedown in outer wall; and (c) the case 3: reverse plasticity in
both walls.

5. Conclusions
Elastic-plastic behaviours of pressurised pipes under
cyclic thermal stresses are investigated in this paper.
One interesting feature is that the present problem considers temperature gradient and the resulting gradient
of the yield strength across the pipe thickness to simulate cyclic elastic-plastic behaviours of heat exchangers or co-axial pipes. This gives a contrast to, for instance, the Bree problem [1] where cyclic thermal
stresses are applied without temperature gradient
across the thickness, of which application areas are
different. Based on analytical investigations, explicit
equations for various stress regimes representing elastic-plastic behaviours are obtained. In the limiting case
of no temperature (yield strength) gradient across the
thickness, proposed equations recover those for the
Bree diagram. For the case when temperature (yield
strength) gradient exists across the thickness, overall
shapes of stress regimes are slightly different from
those in the Bree diagram. The main difference is the
shakedown and elastic regimes. Due to the yield
strength variation across the thickness, shakedown
should be checked at the inner wall, at the outer wall,
and at both walls. Accordingly elastic limit should be
checked at the inner and at the outer walls. As results,
boundaries between elastic and shakedown regimes in
the proposed diagram are different from those in the
Bree diagram. Another difference is the plastic collapse regime which is not present in the Bree diagram.
Boundaries between shakedown and ratcheting or reverse plastic in the proposed diagram are quite similar
to those in the Bree diagram.
The present results are based on several assumptions.
One assumption is to neglect axial stresses due to internal pressure, and thus only the non-zero stress
component is the hoop stress. For completeness, axial
stresses not only from internal pressure but also from

Chang-Sik Oh and Yun-Jae Kim / Elastic-plastic behaviours of pressurised pipes under cyclic thermal stresses

axial forces should be considered, together with the

end condition. Due to complexities, numerical analyses
using finite element method would be useful.
References
[1] Bree J. Elastic-Plastic behaviour of thin pipes subjected to
internal pressure and intermittent high heat fluxes with application to fast nuclear reactor fuel elements. Journal of
Strain Analysis 1967; 2: 226-238.
[2] Bree J. Plastic deformation of a closed pipe due to interaction of pressure stresses and cyclic thermal stresses. International Journal of Mechanical Sciences 1989; 31:
865-892.
[3] ASME Section III Division 1 Subsection NH. Class 1
components in elevated temperature service. The American
Society of Mechanical Engineers, New York 2001.
[4] Hyde TH, Yahiaoui K. An experimental study of the ratcheting and creep of thick flanged pipes subjected to steady
axial mechanical loading and transient thermal loading. International Journal of Mechanical Sciences 1984; 26:
47-61.
[5] Dawson RJ, Fesseler H, Hyde TH, Webster JJ. Elasto-plastic and creep behaviour of axially loaded shouldered

74

tubes. J Strain Analysis 1980; 15: 21-29.

[6] Hyde TH, Webster JJ, Fesseler H. Thermal ratcheting of
axially loaded pipes operating in the creep range. Journal
of Strain Analysis 1982; 17: 243-251.
[7] Hyde TH, Sahari BB, Webster JJ. The effect of axial loading and axial restraint on the thermal ratcheting of thin
tubes. International Journal of Mechanical Sciences 1985;
27: 679-692.
[8] Gross-Weege J, Weichert D. Elastic-plastic shells under
variable mechanical and thermal loads. International Journal of Mechanical Sciences 1992; 34: 863-880.
[9] McGreevy TE, Leckie FA, Carter P, Marriott DL. 2006,
The effect of temperature dependent yield strength on upper bounds for creep ratcheting. In Proceeding of 2006
ASME Pressure Vessels and Piping, Vancouver.
[10] Shuyan H, Zhengming Z, Zhenya Q. The primary loop
confinement and pressure boundary system of the HTR-10.
Nuclear Engineering and Design 2002; 218: 111-122.
[11] ABAQUS Version 6.4 Users manual. Hibbitt, Karlsson
and Sorensen, Inc., R.I, USA 2003.
[12] ASME Section II Part D. Properties- maximum allowable
stress values, thermal expansion and moduli of elasticity
for ferrous materials, Table 1A, TE-1 and TM-1. The
American Society of Mechanical Engineers, New York
2001.