Accepted Manuscript

Bifurcation, plastic and plastic instability loads for cones subjected to combined
pressure and axial force

J. Błachut

PII: S0308-0161(17)30317-4
DOI: 10.1016/j.ijpvp.2017.10.002
Reference: IPVP 3651

To appear in: International Journal of Pressure Vessels and Piping

Received Date: 12 September 2017

Revised Date: 19 October 2017
Accepted Date: 22 October 2017

Please cite this article as: Błachut J, Bifurcation, plastic and plastic instability loads for cones subjected
to combined pressure and axial force, International Journal of Pressure Vessels and Piping (2017), doi:
10.1016/j.ijpvp.2017.10.002.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
Bifurcation, plastic and plastic instability loads for cones subjected to combined pressure
and axial force

J. Błachut
The University of Liverpool, Mechanical Engineering, Liverpool, L69 3GH, U.K.

Abstract:

PT
The paper discusses load carrying capacity of steel cones subjected to combined loading by
simultaneously acting uniform pressure and axial load. The following combinations of loads are
considered: external/internal pressure plus axial tension. Depending on the load combination

RI
various modes of failure are studied. These includes bifurcation buckling, plastic load and plastic
instability load.
For the case of internal pressure/axial tension: envelopes of the first-yield, plastic load and plastic

SC
instability load are computed. All of these domains are convex and closed. This is not true for the
case of external pressure and axial tension where bifurcation buckling becomes an active mode of
failure.
Due to substantial plastic straining, numerical results are based on true stress-strain modelling of

U
material beyond necking and for up to fracture.
AN
1. Introduction
M

Structural integrity of conical shells has been the subject of intense research in the past due to
their usage in a wide range of applications. Large portion of the research effort has been confined
to their elastic behaviour, Ref. [1]. Thicker metallic shells, where plasticity is likely to be
D

important, on the other hand, have not been subject of a wide research effort. It is shown, for
example, in Ref. [2] that plastic bifurcation or other mechanisms of structural failure associated
TE

with plastic behaviour have been examined very rarely. Combined stability of cones within
elastic-plastic regime, until recently, has not been studied at all, Ref. [2]. The current paper
examines conical shells subject to different combinations of two simultaneously, and
independently, acting loads. These are pressure and axial load applied at the top of truncated
EP

cone. Hence there are four possible scenarios depending whether pressure is internal or external,
and axial load is compressive or tensile.
The case of externally pressurised cones has been explored over the years, with recent additions
C

covering post-yield behaviour. In the latter case, one of prime concerns under incremental
loading was the possibility of buckling/collapse, Refs [2-6]. The case of internally pressurised
AC

cones, on the other hand, is less obvious in terms of structural failure under incremental internal
pressure within the elastic-plastic regime. The choice of an appropriate failure criterion on which
a safe design could be based and the development of an efficient numerical algorithm for
predicting failure is not straightforward even for the case of a single load case, e.g., internally
pressurised domed ends onto cylindrical vessels. In the past, this subject related to conical shells
has not been addressed to any large extend. One obvious approach to a cone’s safe performance
would be to follow the strategy developed earlier for domes. This would mean finding the
magnitude of plastic instability and then to use a knock-down factor. Another approach would be
to adopt the concept of plastic load as an indicator of safe performance. However in the latter
case there have been different approaches to the definition of plastic loads, and they appear to
depend on the type of pressure vessel and on mechanisms of its loading. A brief discussion
regarding these issues is given later on in the paper.

1
 ACCEPTED MANUSCRIPT
2. Scope
The current paper studies the concept of plastic load for the case of combined loading of axial
load and uniform pressure applied to steel conical shell. The effect of various combinations of
these two loads on structural integrity of truncated conical shells is explored. This is entirely
numerical investigation and it aims to illustrate how difficult it can be to predict the failure
mechanism when two loads are applied simultaneously. This is particularly important because of
advocated design by analysis instead of design by rules. The paper starts with a brief summary of
experimental and numerical work into buckling/collapse of steel cones subjected to external
pressure and axial compression. It is followed by further two cases: (i) internal pressure/axial
tension, and (ii) external pressure/axial tension. In addition to plastic load envelopes associated

PT
with combined loading, the following performance measures are also explored: first-yield
envelopes, buckling/collapse envelopes, and plastic instability envelopes. As real material
properties of steel for up to fracture were available, the effect of taking this into account when

RI
computing plastic loads and plastic instability loads is also to be assessed.

SC
3. Brief background to past work on bifurcation, plastic and plastic instability loads
3.1 Bifurcation buckling for steel cones
Consider the case of steel cone of constant wall thickness, t, and its geometry defined by: r2/t =
55, h/r2 = 1.0, and β = 14o – see Fig. 1 for definition of geometry. Let the shell be made from steel

U
with material properties given by: Young’s modulus E = 198.56 GPa, the yield point of material
σyp = 205.85 MPa, and Poisson’s ratio υ = 0.28. Assume that this truncated cone is clamped at the
AN
bottom, and only axial movement is allowed at the top end. Computations show that under
incremental, uniform external pressure the shell fails by asymmetric bifurcation with a number of
hoop lobes, n = 9 – as illustrated in Fig. 2a. This result has been obtained under the assumption of
M

elastic perfectly plastic modelling of steel, and the follower nature of lateral pressure, p. Details
about related experimental programme can be found in Refs [5, 6]. One of failed models is
depicted in Fig. 2b. In this case the experimental value of failure pressure was p = 4.62 MPa with
D

n = 7 hoop lobes. Numerical prediction of failure pressure for this case was pnumerical = 4.57 MPa.
Details about buckling tests on all 10 steel, of laboratory size, models can be found in Ref. [6].
TE

The tests were carried out for pure lateral pressure, pure axial compression, hydrostatic pressure,
and combinations of simultaneously acting, axial compression plus external pressure. The
relevant numerical predictions of failure load, and estimates given by design codes are also
provided in Ref. [6].
EP

3.2 Concepts of plastic load for a single load

Plastic loads were introduced for internally pressurised domed ends onto cylindrical pressure
C

vessels nearly half century ago, e.g., Ref. [7]. The first definitions were based on deflections at a
specific point. Two cases frequently investigated were based: (i) on the maximum deflection at
AC

the knuckle, or (ii) on the deflection of the apex (see Fig. 3a). Under single incremental pressure,
the load-deflection curves were generated. Then using different definitions, the magnitude of
plastic load was calculated. For doubly curved shells most work has been done for torispherical
and ellipsoidal domes. A typical plot of internal pressure, p, versus the crown deflection, δ, of a
torispherical shell is shown in Fig. 3. As it may be seen, there is no pressure which could
definitely be called the collapse pressure. Hence two definitions of plastic loads, p1, and p2, based
on the crown deflection were introduced as follows: (i) p1, is the internal pressure at which the
crown deflection reaches twice the yield point deflection, i.e., δ = 2δyp, and (ii) p2, which is the
internal pressure at which a line drawn from the origin and having the Twice Elastic Slope (TES),
meets the p-δ curve (see Fig. 3b). But difficulty here was finding the location of the yield point
experimentally. An alternative definition of the yield point was defined as that pressure which
causes the crown displacement versus pressure curve to deviate from linearity. The corresponding

2
 ACCEPTED MANUSCRIPT
plastic pressures are denoted pc1, and pc2, and these definitions are sketched in Fig. 3c. Generally,
for loading by pressure the choice of the deformation parameter is not arbitrary. It has been
postulated that a work criterion for plastic loads represents a more unified approach, Ref. [7].
This makes the concept strongly embedded in physics, and it also gives an indication of a global
behaviour of a given structural component. Definition of the corresponding plastic loads pv1 and
pv2 are shown in Fig. 3d.
It has been argued that use of global criteria to estimate plastic design pressures carries a benefit
of better utilization of material in the post-yield range and that they provide better measure of
structural integrity, Refs. [7-9]. Use of such criteria for shape and weight optimisation would also
be advantageous as shown in Ref. [10]. The Twice-Elastic-Slope pressure, the Twice-End-of-

PT
Linearity pressure (TEL) based on apex deflection, and the volumetric expansion pressure (based
on work criterion) seem to be good measures of such global behaviour. Ref. [9] provides
numerical results based on the twice-elastic-slope approach for torispherical shapes. Calculations

RI
were based on spherical cap radius, Rs, to diameter, D, ratio, Rs/D = 0.8. Plastic loads based on
crown deflection and on volumetric expansion were also studied experimentally, Ref. [11]. The
plastic loads based on the above definitions of plastic loads were computed for a frequently used

SC
torispherical head known as Korbbogen (Rs/D = 0.8, r/D = 0.154), with the diameter-to-thickness
ratio, D/t = 33.33. It was found that the difference between computed p1 and pC1 amounted to
17% whilst the difference between p2 and pC2 was 0.5%. The difference between pC1 and pv1 was
13% whilst the difference between pC2 and pV2 was negligible. The results led to the conclusion

U
that the predictions of pC1 and pC2 could be made much easier during the experimentation, i.e.,
without the need of strain-gauging and/or measurements of volume change. In order to compare
AN
like with like – all tested torispheres had the same mass but different geometry. Tests were
carried out on ten mild steel torispheres of approximately 200 mm diameter and 6 mm wall
thickness (5 pairs of nominally identical geometry). The Rs/D ratios were 0.7, 0.8, 1.0, 1.1, and
M

1.3. The ratio r/D varied between 0.119 and 0.171 to satisfy the condition of the same mass.
Experimental magnitudes of pC1 varied from 13.79 MPa to 24.9 MPa (see Fig. 4 and Table 1 for a
set of selected results). The errors [(experimental-numerical)/experimental x 100%], varied
D

between -12.1% and +0.1%. Further details are available in Ref. [11].
As mentioned earlier, most work on plastic loads has been done for internally pressurised domed
TE

vessel heads, with a new criterion for plastic load based on plastic work dissipation being
proposed in Ref. [12]. Other single load cases include a nozzle in a torisphere subjected to
moment loading, and a piping branch junction subjected to out-of-plane moment loading, Ref.
[13]. The same reference also provides additional background to the adoption of plastic loads and
EP

a caution to its wider use in pressure components. It appears that studies into plastic loads under
the action of more than one, simultaneously acting, loading are very limited, Ref. [14]. From
practical point of view the ultimate pressure which a given pressure vessel component is capable
C

to withstand under incremental loading is also of interest. This leads the idea of plastic instability
loads – as it is briefly outlined next.
AC

3.3 Plastic instability for a single load

As mentioned earlier, plastic loads for 10 mild steel torispheres were benchmarked against
experimental results in Ref. [11]. Table 1 provides summary of results for 3 pairs of these domes.
Experimentation provided not only plastic loads but also the ultimate load carrying capacity, i.e.,
burst pressures. It is seen, in Table 1, that all numerical values of plastic loads (pC1), are higher
than experimental values by 0% to 12%. Fig. 4 shows that all tested heads were able to carry
much higher values of internal pressure than those associated with plastic loads. A question then
arises how to estimate the ultimate load carrying capacity of these pressure vessel components. In
this context Refs [9, 15] outlined a procedure for numerical calculations of pressure at tensile
plastic instability for internally pressurised axisymmetric vessels. The proposed pressure is to be
an upper bound to the burst pressures that could be achieved in a real vessel. Then the burst

3
 ACCEPTED MANUSCRIPT
pressure is to be modified through the use of an appropriate strength reduction factor. Followed
from this, failure pressure is to be the pressure for which the structures approach dimensional
instability, i.e., unbounded displacement for a small increment in pressure. This can be
characterized by the slope of numerically generated load-deflection curve approaching zero – see
also Refs [16, 17]. The latter approach was used to compute plastic instability load with the aid of
the FE code Abaqus, Ref. [18]. The results which were obtained for models tested in Ref. [11] are
given in the last column of Table 1.

4.0 Concept of plastic load and plastic instability load for combined loading

PT
The case of plastic load and plastic instability load under simultaneous action of two loads are
discussed next for a conical shell.

RI
4.1 Plastic load
Consider a truncated conical shell (frustrum) under the action of tensile force, F, and
independently acting internal pressure, p – as illustrated in Fig. 5. Assume that under a single

SC
force, F, the first yielding takes place at the magnitude of the force amounting to, Fypo . Similarly,
let p oyp , be the pressure at which the first yielding of the material takes place whilst F ≡ 0. When
both loads (F,p) are applied simultaneously then the first yield takes place at (Fyp, pyp). Varying

U
the axial force between 0 < F < FP − I and pressure between 0 < p < pP − I one obtains an envelope
AN
limiting the elastic behaviour of the shell. Any combination of (F,p) which falls inside the
envelope results in the elastic behaviour of the shell, and it is immaterial what is the loading path
as the domain remains energy non-dissipative. When one enters the elastic-plastic (non-potential)
domain then the loading path matters. In all of the ensuing computations the proportional loading
M

is to be assumed. In the search of plastic loads, the very first step would be the evaluation of
plastic loads under the action of a single load. Denote the plastic load associated with axial
tension acting alone as, Fplo , and the plastic load associated with internal pressure only as, p opl .
D

Here again, when both loads are applied simultaneously varying tension between Fyp ≤ F ≤ FP−I ,
TE

and pressure between p yp ≤ p ≤ p P− I one obtains the envelope of combined plastic load as a set of
points (Fpl , p pl ) . One obviously has to adopt a criterion for the evaluation of (Fpl , p pl ) , and this is
to be described later in the paper. When the shell reaches a point on combined plastic load
EP

envelope this does not mean an instant loss of its structural integrity. It is expected that some
reserve of strength exists beyond the (Fpl , p pl ) - envelope. As mentioned earlier one of measures of
ultimate loss of strength under incremental loading can be associated with plastic instability
C

mechanism. The next section outlines the procedure leading to computing plastic instability load
for the case of two, simultaneously acting load-components, i.e., axial tension and internal
AC

pressure.

4.2 Plastic instability load

Computing of plastic instability under the action of two, simultaneously acting loads begins with
numerical estimation of: (i) tensile plastic instability force, FPo− I , whilst internal pressure, p, is set
to zero, and (ii) pressure plastic instability, poP − I , whilst axial tension, F, is set to zero. The FE
code Abaqus is used here to generate the required load-deflection curves. As previously, plastic
instability load is to be the load for which the cone approaches dimensional instability, i.e.,
unbounded displacement for a small increment in load. This means the slope of numerically
generated load-deflection curve approaching zero.

4
 ACCEPTED MANUSCRIPT
The combined plastic instability envelope is computed next by applying both loads (F,p),
generating load-deflection curves, and finding magnitudes of (FP− I , p P− I ) at which the slope of
load-deflection curves approaches zero.

5. Material properties, geometry and details of modelling

The FE calculations in what follows are carried out for material properties established
experimentally in an earlier research for the case of steel cones subjected to external pressure and
axial compression – as detailed in Ref. [5]. The material used for manufacture of cones (and
toricones) was carbon steel with the yield point, σyp = 205.85 MPa (based on 0.2% proof stress),

PT
Young’s modulus, E = 198.56 GPa, Poisson’s ratio υ = 0.28, and the ultimate tensile strength,
UTS = 450.2 MPa. This material properties were obtained experimentally from a solid steel billet
of 252 mm diameter by testing a number of round test samples. The engineering stress-strain

RI
curve obtained from these uniaxial tensile tests is plotted in Fig. 6 as material model ‘A’. This
curve can be converted into true stress-strain curve in a usual way (as outlined below). Since the
onset of necking destroys the uniaxial state of stress relation, the above curve will remain valid

SC
only for up to UTS. Due to perceived large plastic straining it would be instructive to assess the
magnitudes of the loads for the full material property curve. In order to account for diffuse
necking, a corrective method known as weighted average has been adopted. The method, details

U
of which can be found in Ref. [19], allows the true stress-strain curve to be generated from
uniaxial stress-strain curve. From available uniaxial stress-strain test data, and following
AN
reasoning given in Ref. [20], it is assumed that true stress-strain curve beyond the UTS can be
approximated in two ways:
The first assumes a power law:
M

σ = Kσ n , (1)

where K and n are empirical constants determined from known, experimental, true stress-strain
D

data before necking takes place.

Let εu be the true strain at which necking starts, and σu be the corresponding true stress. They can
TE

be calculated from the ultimate tensile strength, su, and the corresponding uniform engineering
strain, eu :

ε u = ln(1 + e u )
EP

(2)

σ u = s u (1 + e u ) (3)
C

Both, eu and σu, are known from the experiment – in our case from Ref. [5]. At the onset of
AC

necking one has:

σ ε =ε u
=σu
(4)
and

dσ
= σu (5
dε ε =ε u

If the power law is used to extrapolate the true stress-strain relation beyond necking, the onset of
necking will allow the constants K and n to be established:

5
 ACCEPTED MANUSCRIPT

σu
K=
(ε u )ε
, and n=εu
u
(6)

The extrapolation equation, Eq. (1), then becomes:

 σ  εu
σ =  uε ε (7)
 (ε u )


u

PT
One can also assume a linear relationship for the true stress-strain after necking, i.e.:

σ = a o + a 1ε

RI
(8)

where ao and a1 are two constants determined from Eqns (4-5) as:

SC
ao =σ u (1 − ε u ) and a1 = σ u
(9)

U
Eq. (8) becomes then:

σ = σ u (1 + ε − ε u )
AN
(10)
M

It transpires that the exponential and linear approximations constitute the lower and upper bound
to stress values beyond the necking, respectively. One way forward for approximation of the true
stress-strain curve would be the use of the weighted average of the lower and upper bounds,
D

respectively. This gives the following:

TE

εu
 ε  
σ = σ u  w(1 + ε − ε u ) + (1 − w)   (11)
  εu  
EP

where w is an unknown constant with the range 0 ≤ w ≤ 1.0. The value w = 1 represents the linear
extrapolation, and w = 0 corresponds to the power law. The value, w = 0.5 gives the average of
both laws. The weight constant, w, was determined in Ref. [20] both experimentally and
C

numerically for round tensile specimens and similar grade of carbon steel to the one which was
used in Ref. [5]. In the latter approach, known engineering stress-strain curve was considered as a
AC

target in the FE based optimization. The cost function was taken as the difference between the FE
calculated true stress-strain curve and known engineering stress-strain curve, with the constant,
w, being the design variable. Reasoning here was that a true stress-strain curve that reproduces
experimentally measured engineering stress-strain curve should correctly represent the strain
hardening relationship of a material. Ref. [20] provides experimental verification of this
procedure. It was found that w = 0.45 reproduced the true stress-strain curve very well.
In our case, at the UTS one has su = 450.0 MPa which gives σu = 528.77 MPa and εu =
ln(1+0.175053)=0.161313. Eq. (11) gives for, w = 0.45, the following value of true stress, σ =
579.4 MPa. In the FE, structural analyses, the true stress-strain was represented by a set of 17
linear segments. Exact values of (σ, ε)-pairs can be found in Ref. [17].

6
 ACCEPTED MANUSCRIPT
As an illustration, in addition to engineering stress-strain curve obtained from experiments, Fig. 6
also shows true stress-strain equivalent which was obtained using the weighted-average approach
– as outlined above.
With material properties detailed above, consider the same conical shell as defined in
Section 3.1, i.e., a conical shell of constant wall thickness, t, with the geometry given by the
following parameters: r2/t = 55, h/r2 = 1.0, and semi-apex angle, β = 14o . Assume the bottom of
the cone to be fully clamped, and neither rotation nor radial displacement at the top end, to be
allowed. Let the cone be loaded by pressure, p, and independently acting axial force, F, applied at
the top end. Results are given next for: (i) internal pressure/axial tension, and (ii) external
pressure/axial tension.

PT
6.0 Results: internal pressure, p, plus axial tension, F

RI
6.1 Load-deflection curves - first-yield envelope
The first phase of calculation, aimed at establishing the first yield envelope under combined
action of internal pressure and axial tension, is closely associated with generation of load-

SC
deflection curves. Load-deflection curve for the cone is plotted in Fig. 7 as the equivalent
load, Feq , versus change of internal volume, ∆V/V. The equivalent force, Feq , is defined as

U
2 2
 F   p 
F eq =  o  +  o  (12)
F  p 
AN
 yp   yp 

where: Fypo is the magnitude of first yield under action of axial tension F, only; poyp is the
M

magnitude of first yield under internal pressure p, only.

The envelope of elastic behaviour is shown in Fig. 8. It is seen here that the elastic domain
remains convex, and it looks like being bounded by two more-or-less linear boundaries. Typical
D

deformed shapes at both boundaries are depicted in Fig. 9a and Fig. 9c for points C1 and C2,
respectively. Deformed profiles at points ‘C1’ and ‘C2’ are clearly different. The cone at ‘C1’
TE

undergoes large deformation at the top end. The shape at ‘C2’ indicates a dominant role of
internal pressure which manifests itself by the membrane behaviour of the cone nearly along the
full length of the slant with bending confined to top and bottom ends. A similar, bi-linear shape
EP

of first-yield envelope has been noted for externally pressurised cones subjected simultaneous
axial compression, Ref. [4].

6.2 Plastic load envelope

C

The procedure for obtaining plastic loads comprises of the following two steps: (i) applied load
versus change of internal volume is computed first, and (ii) applied load versus the change of
AC

volume is plotted. In the current situation, the load versus the (∆V/V)-ratio proved to be
nonlinear. This in turn required an objective measure of deflection to be adopted for computation
of plastic load. The twice slope associated with the first yield, σyp, was used for the extraction of
plastic load. This is illustrated in Fig. 7 for loading path ‘0 → C1 → C3’. The first yielding
occurs when the axial force reaches the value, Fyp = 145.22 kN, and pressure reaches, pyp = 0.503
MPa. It is only then the slope, α1, is found numerically. Next, line is drawn from the origin at the
angle, 2 α1, to the intersection with Feq versus ∆V/V curve. In Fig. 7 this happens at plastic load
Fpl = (233.23kN, 0.805MPa). The full plastic load envelope is plotted in Fig. 8 It was obtained
for a range of proportional loading pairs (F,p). Deformed shapes of cones at points ‘C3’ and ‘C4’
are shown in Fig. 9b and in Fig. 9d. They by-and-large resemble shapes at ‘C1’ and ‘C2’.
Deformations at ‘C1’ and ‘C3’ occur at the top. The dominant role of internal pressure seen at

7
 ACCEPTED MANUSCRIPT
point ‘C2’ also exists at the point ‘C4’. The domain of plastic load under combined loading
remains convex and it is much larger than the elastic envelope. The ratio of relevant surface area
magnitudes, Spl/Syp, is 3.30. A sample of values of plastic loads, Fpl ≡ (Fpl , p pl ) is given in Table
2. As mentioned earlier, the cone under a single incremental loading is unlikely to suffer the loss
of structural integrity at plastic load. Hence an additional reserve of strength associated with
plastic instability is computed next.

6.3 Plastic instability load envelope

The load-deflection curve shown in Fig. 7 corresponds to the loading path 0 → ‘C1’ → ‘C3’.

PT
Plastic instability in this case will occur when the load-deflection curve seen in Fig. 7 reaches a
horizontal tangent once higher values of (F,p) are applied. The magnitude of the corresponding
plastic instability load was found to be: F P − I = (445.63kN, 1.645MPa ) . This corresponds to

RI
the point ‘C5’ in Fig. 10. The latter Figure depicts the whole envelope of combined plastic
instability. The domain is convex as are the other two. The size of the domain is significantly
larger than the domain of plastic load. The ratio of surface area of corresponding domains is as

SC
follows: SP-I/Syp = 25.8; SP-I/Spl = 8.16; Spl/Syp = 3.30. Selected values of plastic instability loads
are given in Table 3. Spread of plastic strains has been recorded along the loading path 0 → ‘C1’
→ ’C3’ → ‘C5’. Their spread through the wall thickness at all three points is shown in Fig. 11. It
is interesting to see that at plastic instability level no plastic hinge has developed.

U
AN
7.0 Results: external pressure, p, plus axial tension, F
Domains of elastic behaviour, plastic load and instability load were computed here applying the
same computational framework as for the case of internal pressure/axial tension in Section 6.0.
M

Results are shown in Fig. 12. As previously, proportional loading was applied. It is seen from
Fig. 12 that bifurcation buckling became an active mode of failure for some loading conditions.
This puts the results into a new consideration. Bifurcation buckling results in a catastrophic
D

failure, i.e., it means the end of load carrying capacity. The rest (to the right of bifurcation line in
Fig. 12), of the response graphs becomes then irrelevant from a practical point of view. But, as
TE

seen in Fig. 12, bifurcation buckling does not form a closed envelope as there is no bifurcation
buckling for smaller values of external pressure, i.e., for p o ≤ 0.4. Hence, plastic load and
p P−I
plastic instability concepts can only apply to a fraction of combined stability domain.
EP

8.0 Discussion and Closure

C

The paper illustrates behaviour of steel conical shell subjected to two independently acting loads,
i.e., internal pressure/axial tension, and external pressure/axial tension. In the first case, the
AC

structural response of the component is measured through three domains/envelopes, and they are:
(i) envelope of pure elastic behaviour, (ii) envelope associated with plastic load, (iii) envelope of
plastic instability. Plastic and plastic instability loads were obtained from plots of equivalent load
versus change of internal volume. The latter provides a measure of global behaviour of this
component, and it has been argued that this approach is well embedded in physics. All three
domains are convex. Numerical results show a sizeable strength beyond purely elastic domain. In
terms of the area associated with each mode, the values are: plastic load domain is 3.2-times
larger than elastic domain, and plastic instability domain is 26-times larger than the elastic
domain. Past tests on internally pressurised torispheres showed that the FE estimates of plastic
instability load were, as exemplified in Table 2, between 17.6% and 42.8% above experimental
burst pressures. The FE-computed values of burst pressures (given in Ref. [16]), brought the

8
 ACCEPTED MANUSCRIPT
above over-estimate down to the range 12.1% - 32.0%. Hence it would be advisable to estimate
the relevant burst loads for conical shells and eventually compare them with experimental data.
The case of external pressure and axial tension introduces a new dimension, i.e., appearance of
bifurcation buckling as likely mode of failure. This illustrates how careful one has to be when
assessing structural integrity of the component.
Finally, it has to be said that the above remains true for the specific component, and for given
modelisation of material properties.

9.0 References

PT
[1] J. Singer, J. Arbocz, T. Weller, “Buckling Experiments – Experimental Methods in
Buckling of Thin-Walled Structures”, vol. 2, 2002, John Wiley & Sons Inc., NY, 1732 p.,
(ISBN 0-471-97450-1).

RI
[2] O. Ifayefunmi, “Combined Stability of Conical Shells”, PhD Thesis, 2011, The University
of Liverpool, U.K.
[3] O. Ifayefunmi, J. Błachut, “Combined stability of unstiffened cones - Theory, experiments

SC
and design codes”, Intl Journal of Pressure Vessels and Piping, vol. 93-94, 2012, 57-68.
[4] J. Błachut, O. Ifayefunmi, “Buckling of unstiffened steel cones subjected to axial
compression and external pressure”, Journal of Offshore Mechanics and Arctic
Engineering, Transactions of the ASME, vol. 134, 2012, 031603-1 – 031603-9.

U
[5] J. Błachut, “Interactive plastic buckling of cones subjected to axial compression and
external pressure”, Ocean Engineering, vol. 48, 2012, 10-16.
AN
[6] J. Błachut, A. Muc, J. Ryś, “Plastic buckling of cones subjected to axial compression and
externa; pressure”, Journal of Pressure Vessel Technology, Transactions of the ASME, vol.
135, 2013, 011205-1 – 011205-9.
M

[7] J.C. Gerdeen, “A critical evaluation of plastic behaviour data and a united definition of
plastic loads for pressure components”, WRC Bull., 1979, No. 254.
[8] J. Spence, A.S. Tooth, (eds), “Pressure Vessel Design: concepts and principles”, London,
D

Chapman and Hall, 1994, 491 p.

[9] D.P. Updike, A. Kalnins, “Tensile plastic instability of axisymmetric pressure vessels”,
TE

Journal of Pressure Vessel Technology, Transactions of the ASME, vol. 120, 1998, 6-11.
[10] J. Błachut, “Minimum weight of internally pressurised domes subject to plastic load
failure”, Thin-Walled Structures, vol. 27, 1997, 127-146.
[11] J. Błachut, “Plastic loads for internally pressurised torispheres”, Jnl Pressure Vessel
EP

Piping, vol. 64, 1995, 91-100.

[12] D. Camilleri, D. Mackenzie, R. Hamilton, “Evaluating plastic loads in torispherical heads
using a new criterion of collapse”, Journal of Pressure Vessel Technology, Transactions
C

of the ASME, vol. 130, 2008, 011202-1 – 011202-8.

[13] D.G. Moffat, M.F. Hsieh, M. Lynch, “An assessment of ASME III and CEN TC54
AC

methods for determining plastic and limit loads for pressure system components”, The
Journal Of Strain Analysis For Engineering Design, vol. 36, 2001, 301-312.
[14] D. Mackenzie, H.J. Li, “A plastic load criterion for inelastic design by analysis“, Journal
of Pressure Vessel Technology, Transactions of the ASME, vol. 128, 2006, 39-45.
[15] D.P. Updike, A. Kalnins, “Ultimate load analysis for design of pressure vessels”, ASME
PVP 353, Pressure Vessel and Piping Codes and Standards, Proceedings of the 1997
ASME Pressure Vessels and Piping Conference, ASME, New York, NY, pp. 289-293.
[16] J. Błachut, V.T. Vu, “Burst pressures for torispheres and shallow caps”, Strain, vol. 43,
2007, 26-36.
[17] J. Błachut, O. Ifayefunmi, “Burst pressures for toriconical shells – experimental and
numerical approach”, Journal of Pressure Vessel Technology, Transactions of the ASME,
vol. 139(5), 2017, 051203-1 – 051203-9.

9
 ACCEPTED MANUSCRIPT
[18] Hibbitt, Karlsson, and Sorensen Inc., ABAQUS – Theory and Standard User’s Manual
Version 6.3, 2006, Pawtucket, RI, 02860-4847, USA.
[19] Y. Ling, “Uniaxial true stress-strain after necking”, APM Journal of Technology, vol. 5
(June), 1996, 37-48.
[20] K.S. Zhang, Z.H. Li, “Numerical analysis of the stress-strain curve and fracture initiation
for ductile material”, Engineering Fracture Mechanics, vol. 49(2), 1994, 235-241.

PT
RI
U SC
AN
M
D
TE
C EP
AC

10
 ACCEPTED MANUSCRIPT
Tables

Model D Rs t m p yp pCExptl
1 pCFE1 p plFE
r/D
(mm) D (mm) (kg) (MPa)
K1 200 0.80 0.154 6.01 6.48 16.6 20.20 21.95 41.98
K1a 200 0.80 0.154 5.93 6.47 16.1 21.60 21.58 49.50
K5 200 1.10 0.169 5.845 6.45 11.2 14.20 15.41 48.00
K5a 200 1.10 0.169 5.953 6.45 11.4 14.00 15.69 48.90
K6 200 1.30 0.1714 6.012 6.49 9.2 12.80 13.29 49.26

PT
K6a 200 1.30 0.1714 6.034 6.45 9.4 12.66 13.35 49.26
Table 1. Sample of experimental and FE results for internally pressurized steel torispheres (from
Ref. [11]).

RI
Fpl p pl
o
F P−Ip oP − I

SC
0.53 0.0
0.53 0.032
0.52 0.069

U
0.50 0.106
0.40 0.157
AN
0.35 0.165
0.20 0.177
0.06 0.179
M

0.0 0.179
Table 2. A sample of values of plastic loads normalized by: (i) Plastic Instability Force, FPo− I =
440.06 kN, and by (ii) Plastic Instability Pressure, p oP − I = 25.16 MPa.
D
TE

FP − I p P− I
FPo− I p oP − I
1.0 0.0
EP

1.01 0.0682(1)
1.02 0.18
0.97 0.34(2)
C

0.77 0.54
0.51 0.72
AC

0.25 0.86
0.0 1.0
Table 3. A sample of Plastic Instability loads (FP − I , p P − I ) normalized by: Plastic Instability force,
FPo− I , and by Plastic Instability pressure, pPo − I . Note: (1) corresponds to point C5 on loading path
C1→C3→C5; (2) corresponds to point C6 on the path C2→C4→C6.

11
 ACCEPTED MANUSCRIPT
Figures

PT
RI
U SC
AN
M
D

Fig. 1 Various combinations of simultaneously acting loads, and adopted notation for geometry
TE

of cones.
C EP
AC

Fig. 2 FE-generated eigenshape and view of corresponding experimental mode of failure.

12
 ACCEPTED MANUSCRIPT

PT
RI
U SC
Fig. 3 Definitions of plastic loads in internally pressurised dome.
AN
M
D
TE
C EP
AC

Fig. 4 Experimental plastic and burst pressures (adapted from Ref. [11]).

13
 ACCEPTED MANUSCRIPT

PT
RI
SC
Fig. 5 Geometry of conical shell subjected to independent loading by internal pressure, p, and
axial tension, F.

U
AN
M
D
TE
C EP
AC

Fig. 6 Experimental stress-strain obtained from uniaxial tensile tests, and true stress-strain.

14
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M

2 2
 F   p  ∆V
Fig. 7 Plot of equivalent load, Feq =  o  +  o  , versus change of volume, , for
D

F   p  V
 yp   yp 
loading along C1 – C3 path. Note: Fyp = 128.75kN , and p oyp = 2.40MPa .
o
TE
C EP
AC

Fig. 8 Domains of purely elastic behaviour and of plastic load, Fpl = (Fpl , p pl ) . Note: Spl/Syp =
3.32; FPo− I ≡ Plastic Instability force; pPo − I ≡ Plastic Instability pressure.

15
 ACCEPTED MANUSCRIPT

PT
RI
U SC
Fig. 9 Initial and deformed shapes at C1 (First Yield), and at C3 (Plastic Load). Also, initial and
AN
deformed shapes at C2 (First Yield), and at C4 (Plastic Load).
M
D
TE
C EP
AC

Fig. 10 Domains of elastic, plastic load, and plastic instability behaviour.

16
 ACCEPTED MANUSCRIPT

PT
RI
SC
Fig. 11 Spread of plastic strains at different load levels: (a) C1: (Fyp = 145.22 kN, pyp =

U
0.503MPa), (b) C3: (Fpl = 233.23 kN, ppl = 0.805 MPa), (c) C5: (FP-I = 445.63 kN, pP-I = 1.465
MPa).
AN
M
D
TE
C EP
AC

Fig. 12 Domains of elastic, plastic load, bifurcation, and plastic instability behaviour.

17
 ACCEPTED MANUSCRIPT
1. Combined stability of cones
2. External/internal pressure
3. Axial tension/compression
4. Full stress-strain material data for steel

PT
RI
U SC
AN
M
D
TE
C EP
AC