Finite element analysis of buckling behaviour of stiffened cylindrical

shells
Mohd Shahrom Ismaila, 1, Azrol Jailanib and Syed Mahathir Muhammad al-Attasc
a
Jabatan Kejuruteraan Mekanikal, Politeknik Sultan Salahuddin Abdul Aziz Shah, Shah Alam 40150, Selangor,
Malaysia
b
School of Mechanical Engineering, College of Engineering, Universiti Teknologi MARA
Shah Alam, Selangor, Malaysia
c
Quadrant 2 Technologies Sdn. Bhd., Kuala Lumpur, Malaysia

Abstract
Stiffened cylindrical shell structures are commonly used in the marine industry, particularly offshore
structures and are often associated with initial geometric imperfections. A lack of knowledge in evaluating
geometric imperfections jeopardises the safety of shells. In the present study, the influence of
imperfections in axially compressed/externally pressurised stiffened cylindrical shells for different
geometries was investigated using finite element analysis (FE). This study focuses on the imperfection
techniques of Single Perturbation Load Approach (SPLA) and Smooth Inward Dimple (SID) for a
validation analysis with the published experimental data. The effect of the initial imperfection in terms
of its depth, magnitude, size, and location on the corresponding buckling strength of the shell was also
examined. Reasonable agreement with the published experimental data was obtained using both
techniques with the range of 2% – 13%. The presence of initial geometric imperfection has an adverse
effect on the load-carrying capacity of the shells with detrimental factors being its depth, magnitude, size
and location. In general, the results of this study contribute to the body of knowledge on the effects of
structural imperfections in the preliminary stages of the design, fabrication, and analysis of axially
compressed or externally pressurised stiffened cylindrical shell structures.

Keywords: Isotropic material; Buckling; Stiffened cylindrical shell; Finite element analysis; Structural
imperfection

1 Corresponding author:
E-mail address: [email protected] (Mohd Shahrom Ismail)
a Senior Lecturer, Jabatan Kejuruteraan Mekanikal, Politeknik Sultan Salahuddin Abdul Aziz Shah

Manuscript submitted to Material Research Express

1
Abbreviations
FEM Finite element method
SPLA Single perturbation load approach
SID Smooth inward dimple
∆U Load-controlled displacement
R Radius of cylinder
t Thickness of cylinder
L Length of cylinder
b Stiffeners width
h Stiffeners height
PEigen Eigenvalue buckling load
P0 lateral perturbation load
a Half-span of imperfection in the hoop direction of the corresponding cylinder
b Half-span of imperfection in the slant/axial directions of the corresponding cylinder
∆𝑧 /δ Maximum inward deflection is measured in the middle of the dimple
a/t Normalized imperfection amplitude to the thickness of the cylinder
ρ Cylinder’s knockdown factor
ρSPLA Cylinder’s knockdown factor obtained from the single perturbation load approach
ρSID Cylinder’s knockdown factor obtained from the smooth inward dimple technique
ρExp Cylinder’s knockdown factor obtained from the available experiment
E Young’s modulus
σyield Yield stress
υ Poisson’s ratio

1 Introduction
The buckling problem of shell structures has continuously attracted many engineers to find an
appropriate solution from the foundation of the theory to the experimental view. Certainly, there are a
few aspects that need to be accounted for, including the types of load, shell geometry, initial
imperfections, material characteristics, and boundary conditions. On the other hand, adopting stiffeners
may increase the stability and buckling resistance of shell structures while simultaneously reducing their
weight. Stiffened cylindrical shell structures are common panels that are widely used in the marine
industry, predominantly in offshore structures. Traditionally, offshore structures are made of steel and

2
fixed to the seabed; thus, their weight is not a major consideration in the design period. During service,
these structures are exposed to various loads, primarily compression, external pressure, or a combination
of both. Recently, the use of far lighter and slender cylindrical structural members has become more
apparent, as their weight is more important in satisfying the buoyancy requirement and the influence of
payload, that is, dead weight in total. Therefore, it is necessary to conduct a detailed analysis of the
buckling occurrence of slender structures. Hence, a more rational design that incorporates weight savings
and economic fabrication of particular structures is required. The unavoidable geometric imperfection
that could be introduced during the manufacturing process or shell exploitation can also adversely affect
safety. Therefore, a better understanding of the imperfection phenomena is required for updates to
anticipate the unexpected catastrophic failure of structures, thus producing a balanced design in terms of
weight, stiffness, and stability.
Details of previous studies on the sensitivity of the cylinder buckling load and knockdown factors
associated with initial imperfections can be found in Refs. [1] – [4]. It is known that all aerospace or
subsea regulatory agencies have adopted a safe and conservative approach to implementing the lower
bound of the buckling knockdown factor curve. This is necessary to keep the safety factor of the structure
at the highest level. Several approaches have been deployed to address geometric imperfection
uncertainties, although the eigenmode affine is usually used. The eigenmode affine is described by
imposing the buckling mode into the perfect shape of the shell and is computed with the imperfection
amplitude. The computed buckling mode and imperfection amplitude play vital roles in estimating the
buckling load. For example, the buckling analysis reported in [5] – [6] showed the need to perform linear
and nonlinear buckling analyses with different initial geometric imperfections on cylindrical shells. Ref.
[7] mentioned that imperfections in the shape of the buckling mode do not appear in actual structures.
Moreover, several researchers have suggested that the eigenmode-affine imperfection technique may not
serve as a worst-case geometrical imperfection [8], [9]. Considering the limitations highlighted above,
the approach seems to produce inconsistent estimations of the buckling load and permits uncertainties in
examining structural buckling resistance.
Over the last few decades, buckling and post-buckling analyses using the nonlinear Finite Element
Method (FE) have been widely accepted. Numerous nonlinear codes, including ANSYS, ABAQUS,
NASTRAN, and STAGS, are among the established FE software with the advancement of modern
computer technology, with a strong ability to offer accurate numerical predictions for a substantial
number of buckling problems. For instance, Paik et al. [10] used a nonlinear FE model with the ANSYS
software package to assess the ultimate limit state of a stiffened plate in ship structures. Several similar
analyses using ANSYS were conducted by Cao et al. [11] and Ghorbanpour et al. [12]. Arbocz and

3
Starnes [13] used a powerful nonlinear FE code, STAGS, to predict the stable and unstable post-buckling
responses of axially compressed cylindrical shells. Tao et al. [14], developed a three-dimensional
nonlinear finite element model using ABAQUS, that accounts for the nonlinear material behaviour, and
initial geometric imperfections, to predict the buckling behaviour of concrete-filled steel tubes, where
close agreement is achieved between the test and FE results in terms of load-deformation response and
ultimate strength.
The single perturbation load approach (SPLA) proposed by Hühne et al. [15] is a relatively new method
for designing imperfections in axially compressed shell structures. This method uses the lateral load
imposed on the outer surface of the model to simulate the worst-case geometrical imperfections of typical
structural models such as cylindrical shells. Ismail et al. [16] utilised the SPLA to study the influence of
imperfections on the critical buckling of axially compressed cylindrical shells for different geometries
and composite materials and aluminium using finite element analysis. Ref. [1], [3] employed the smooth
inward dimple (SID) technique to analyse cylindrical, toroidal, barrelled, and torispherical shells
subjected to uniform external pressure. More recently, Ifayefunmi and Błachut [7] used a similar
technique to examine steel cones under combined loads (axial compression and external pressure). A
study on the stability behaviour of stiffened and unstiffened cone cylinder shells using a similar
imperfection technique was reported in [17], [18]. Ismail et al. [19] also employed the imperfection
technique in the case of externally pressurised spherical shells.
However, there is still a lack of information and effective guidelines reported in the literature for
introducing imperfections in the case of stiffened cylindrical shells under axial compression and
pressurised loads. Thus, relevant updates are always sought to address new design challenges that suit
current technology demands. The present study aims to evaluate the response of imperfection-sensitive
stiffened cylindrical shells subjected to axially compressed or externally pressurised loads using the
SPLA and SID methods via finite element analysis. The current work is purely numerical and is verified
with published experimental data given in [20] – [22]. The present study also examines the influence of
initial imperfections in terms of depth, magnitude, size, and location on the corresponding buckling
strength of the shell.
The results of this study are highly useful and contribute to the body of knowledge on the effects of
structural imperfections in the preliminary stages of the design, fabrication, and analysis of axially
compressed or externally pressurised stiffened cylindrical shell structures.

4
2 Materials and methods
The present investigation considered different types of stiffened cylindrical shells being its geometry
or stiffener arrangement with the support of nonlinear static FE analyses. The knockdown factors of the
stiffened cylindrical shells are determined by normalising the collapse load over the eigenvalue load.

2.1 Finite element modelling and material properties

The arrangement of the axially compressed stiffened cylindrical shell under the SPLA is shown in Fig.
1 (a). Fig. 1 (a) also shows that the displacement and rotation constraints are applied to one end of the
shell, and the load-controlled displacement ∆U with slow quasi-static compression is applied to the other
end. Nevertheless, the allowable displacement ∆U is constrained for an externally pressurised cylinder
shell because the pressure is applied around the cylinder circumference. The analyses were performed
using the finite element software package ABAQUS.
The selection of S4R, a 4-node shell element with reduced integration from the ABAQUS element
library is considered in this study. Indeed, this is necessary as the element uses three translational and
three rotational degrees of freedom at each node. The element accounts for finite membrane strains and
arbitrarily large rotations. Subsequently, this element is suitable for large-strain analyses and
geometrically nonlinear problems. In particular, reduced integration was carried out in accordance with
the Simpson rule. Since S4R is a linear element, then the hourglass control settings are necessarily
required.
As expected, from the convergence analysis, the buckling load yields to be slightly lower and reaches
an almost constant state. For example, the mesh with the number of elements 10,206 found to be suitable
for the analysis as shown in model AR-1a (see Table 1). The denser the mesh, the more time it takes for
a solution to complete. The tested simulation models employed more than 10,000 elements. The analysis
is governed by a nonlinear static analysis based on a setting of Nlgeom and increment of 1,000. In
addition, the ABAQUS solver based on Newton-Raphson is used for the analysis.
The artificial damping factor was carefully selected based on sensitivity studies. Prior to performing
response analyses on the cylindrical shells, the sensitivity study was conducted to determine a suitable
value for the artificial damping factor. It is known that too small a value of the artificial damping would
possibly result in a singularity of the tangent stiffness matrix. Meanwhile, a large artificial damping value
would lead to over-damped results. Therefore, the best approach prior to the selection of artificial
damping factor is considered by choosing the lowest value of artificial damping, c = 4e- 9 first. For each
run, if the result seems to be unsatisfactory (i.e., unconverged) due to the singularity, then the value is
increased accordingly until a quality outcome is achieved.

5
 The geometries of the shells used in this study refer to those experimentally performed by Weller et
al. [20] for longitudinally stiffened panels for the case of axially compression and for the case of external
pressure is refer to those by Weller et al. [21] and Walker and McCall [22] for ring-stiffened panels. The
shell is designed to have a radius of R, thickness of t, and length of L. Stiffeners have a width of b and a
height of h. The detailed configurations are presented in Figs. 1 (b) – (c) and Table 2. The stiffeners of
the verification models were located outside the cylinder, except for model R2. The elastoplastic material
model and its corresponding properties are presented in Table 3 and Fig. 2.

2.2 Imperfection approach

This section describes the imperfection approach taken in the study. The approach is based on
validation studies comprising previous experiment data that are available in literature.

2.2.1 Single perturbation load approach

In general, as highlighted by Hühne et al. [15], the SPLA procedure in finite element analysis has three
steps. In the first step, a lateral perturbation load P0 was applied at the midsection of the stiffened cylinder,
as illustrated in Fig. 1 (a). It is purposely used to produce a single buckle or local dent as the worst
imperfection. The lateral perturbation load P0 was then increased linearly until a consistent buckling load
was achieved in the second step. In the second step, the cylindrical shell is driven by axial compression
or external pressure to reach the state of the buckling load, known as local snap-through. Finally, in the
third step, the load was sustained until the shell accomplished its post-buckled nature. The Newton-
Raphson algorithm was chosen for both steps.
It is worth noting that once the dent has formed, the residual stress is also built up. This situation
indicates the natural occurrence of the dent caused by a defect during operation or the manufacturing
process. The presence of residual stress also shows that the structure was manufactured without stress
relief having been carried out.

2.2.2 Smooth inward dimple

Alternately, along with the SPLA, the effect of the initial geometric imperfection was also evaluated
as a localised inward dimple. In practice, the localised inward dimple profile appears to be more realistic,
although it was not considered to be the worst imperfection [3]. Fig. 3 shows the smooth inward dimple
(SID) imperfection modelled using Eq. (1):

6
 𝜋𝑥 𝜋𝑦 (1)
∆𝑧 = 0.25𝛿 [1 +𝑐𝑜𝑠 𝑐𝑜𝑠 ( ) ] [1 +𝑐𝑜𝑠 𝑐𝑜𝑠 ( ) ]
𝑎 𝑏

The terms ‘a’ and ‘b’ indicate the half-span of the imperfection in the hoop and slant/axial directions
of the corresponding cylinder. On the other hand, the maximum inward deflection ∆z is measured in the
middle of the dimple and denoted as, δ. Nonetheless, δ is also considered as the imperfection amplitude.
The range of imperfection amplitude was computed between 0.0 - 2.5. This magnitude is employed to
simulate the worst-case scenario for a tested shell. The SID is converted to ABAQUS by rearranging the
selected nodes accordingly on the basis of equation (1). This produces the desired dent formation with
zero stress. This state is identical to the dent formation after the manufacturing process, which runs
through the stress relief or annealing process.

3 Results and Discussion

3.1 Sensitivity analysis and verification study
In the early stage of the analysis, a sensitivity study was conducted to determine a suitable value for
the artificial damping factor. It is known that an artificial damping value that is too small may result in
the singularity of the tangent stiffness matrix. Meanwhile, a large artificial damping value leads to over-
damped results. The responses of the different artificial damping values demonstrated by the AS-1L
model are presented in Fig. 4 (a). It can be observed that the use of the damping factor c = 4e-9 fails to
reach the first local snap-through. This situation is linked to the singularity of the tangent stiffness matrix,
resulting in a non convergence problem. Next, a larger artificial damping value of c = 4e-6 seems to
produce over-damped results. Thus, an artificial damping value ranging between c = 4e-7 – 2e-8, which
produces a convergence result, was chosen in this study.
Figs. 4 (b) – (d) show the responses of the collapse load for each stiffened cylindrical shell estimated
using the SPLA and SID techniques. A minimum of ten different lateral perturbation loads Po and
amplitudes were tested for each shell using the SPLA and SID, respectively. The perturbation load Po
was obtained from the nonlinear FE analysis on the perfect cylindrical shell. The imperfection sensitivities
can be noticed from the significant drop of the knockdown factors over the normalised values of a/t.
Noticeable reductions of the knockdown factor are observed in both models for small imperfection
magnitudes (up to a/t ≥ 0.3). Meanwhile, the knockdown factors become relatively stable towards the
lowest value before continuing with a stable state. The lowest knockdown factors are found to be around
0.570 to 0.765, respectively.

7
 For verification purposes, the estimated collapse loads were compared against published experimental
data [12-14] and are presented in Table 4. In general, more conservative results estimated by the SPLA
were observed in comparison with the SID technique, except for model R2. Notably, both approaches
overestimated the experimental data and did not include the AS-1L model. In comparison, the AS-1L and
AR-1a models indicated large discrepancies in the collapse load by nearly 10% - 13% through the SPLA
and SID, respectively. The SPLA signifies a minimal difference of less than 10% for the R2 and AR-1a
models, respectively. The SID technique yielded the closest estimation of collapse loads in the range of
2% - 6% for the AS-1L and R2 models, respectively. Apparently, a more conservative estimation of
knockdown factor was found to be produced by the SPLA (see Figs. 4 (b) – (d)). This is clearly seen
through lowest plotted data against the range of imperfection amplitude. The required lateral perturbation
load Po in estimating the knockdown factors are found to be between 3 kN - 9 kN thus permitting a/t wall
to be more than 0.75 in return. Meanwhile, AR-1a (ring-stiffened cylinder) model gives the smallest a/t
wall which is about 0.35. The model’s thinner wall may suggest the occurrences. Table 5 summarised the
knockdown factor of each tested model.
The imperfections caused by inaccurate measurements and load conditions during the experimental
setup are unavoidable. It can be seen that most of the initial imperfections obtained are significantly
higher than the experimental value except model R2. In contrast, the SID method also calculates the
closest estimate of the initial imperfection and the knockdown factor for model R2. In the case of the
externally pressurised cylinder (model R2), which is demonstrated by the SPLA shown in Fig. 4 (d), a
significant decrease in the collapse load can be observed. This occurrence is probably related to the large
bay which spans about 40 mm between the stiffeners (see Ref. [22]). It is also hypothesised that the lateral
load generated by the SPLA may trigger the drastic loss of buckling strength of the shell. In this case, the
shell would be more susceptible to catastrophic buckling failure.

3.2 Buckling load with various dimple size

Finite element analysis considers a model to be a perfect structure with presumed boundary conditions.
Thus, the differences between the results are acceptable. Overall, the SPLA and SID estimates are in good
agreement with the experimental data. Once the confidence in the FE models was established, further
investigation on the effect of dimple size on the structural collapse load was performed using the smooth
inward dimple (SID) technique. The investigation was conducted by increasing the size of the local
dimple with a constant imperfection amplitude in the midsection of the cylinder. Three different dimple
sizes were selected, as listed in Table 6.

8
 The results reveal that the collapse loads and initial structural imperfections are reduced to an average
of 7% to 30%, respectively, when the dimple size increases. The AS-1L and R2 models exhibited the
largest losses in collapse loads of approximately 9% to 10%, respectively. In contrast, the AR-1a model
experiences a large drop in imperfection, a/t, by 46%, despite being insensitive to the collapse load. The
present study suggests that a greater local dimple size has a strong influence on the structural collapse
load and initial structural imperfection. The results are summarised in Table 6.

3.3 Buckling load with different dimple location

The dimple location across the cylinder is assumed to influence the estimation of the structural
buckling resistance. The new location of the local imperfection was then selected at the top and bottom
of the shell sections. The study was performed by taking the AS-1L model with a constant dimple size
and imperfection of a/t = 2 as an example. As shown in Table 7, the SPLA still produced the lowest
collapse loads compared to the SID technique, with an average of 26 %. The results also implied that the
location of the dimples had a substantial influence on the structural buckling strength. This can be seen
as the collapse loads being reduced by 30% and 5% in comparison with the imperfection located at the
midsection of the shells, as demonstrated by the SID and SPLA.
It might be argued that simultaneous loading by a denting force of constant amplitude is very unlikely
to be seen in practice. But this approach, from a numerical standpoint, is computationally more efficient
than the SID that is based on the increased-radius imperfection (i.e., constant (a/t)-amplitude). In
particular, the SPLA appears to be highly effective in considering the worst structural geometrical
imperfections. However, the SPLA is limited by its approach, that is, the depth of the amplitude/dent
produced by the lateral load and position of the considered imperfection. Conversely, the smooth inward
dimple (SID) possesses most of the SPLA abilities with additional manipulation of dimple size.
The current findings also underline that the structural buckling strength is affected by the initial
imperfections, such as depth, magnitude, size, and location. Overall, both techniques were found to be
effective in the preliminary design stage, providing sufficient information for evaluating the uncertainties
in the stiffened cylindrical shell structure. From a design point of view, the inconsistencies in estimating
the structural collapse loads by relocating the imperfection through a single inward dimple seem to be
unfavourable owing to safety concerns. On the other hand, the SPLA proved to be insensitive to structural
collapse loads, even by repositioning the dimple locations.
Fig. 5 shows the summary of knockdown factor against radius over thickness ratio. The design
guideline based on NASA SP-8007 [23] was employed for a comparison analysis with the experiment
and numerically calculated knockdown factors. This is critical in order to evaluate the results with

9
established design guidelines. From observation, it suggests that NASA SP-8007 guidelines prevail to be
significantly conservative for being the lowest estimation of knockdown factor. On the other hand, the
knockdown factor of R/t = 267 was found to be much closer to the design guideline lower-bound curve
in comparison with R/t < 500.

4 Conclusion
The numerical results for axially compressed or externally pressurised stiffened cylindrical shell
structures are presented in this paper. Although the current study is merely numerical, it supports the
critical claim on the importance of structural imperfections and highlights the ability of each imperfection
technique, namely the SPLA and SID. The collapse load estimations made using both techniques were
shown to be in reasonable agreement with published experimental data. From a design point of view, the
inconsistencies in estimating the structural collapse loads by relocating the imperfection using the smooth
inward dimple (SID) seem to be unfavourable to the safety aspect of the shell. The SPLA demonstrated
attractive ability for designing imperfection in stiffened cylindrical shells and promoted a desirable
technique in a preliminary design process or an evaluation of a buckling failure.
The presence of initial geometric imperfections has an adverse effect on the load-carrying capacity of
the shells, with detrimental factors such as depth, magnitude, size, and location. These numerical results
are based on the FE calculations and it would be desirable to benchmark them against experimental data.
In general, the results of this study are highly useful and contribute to the body of knowledge on the
effects of structural imperfections in the preliminary stages of the design, fabrication, and analysis of
axially compressed or externally pressurised stiffened cylindrical shell structures.

Acknowledgements
The authors would like to express sincere gratitude to Politeknik Sultan Salahuddin Abdul Aziz Shah and
thank the Ministry of Higher Education, Malaysia, for supporting the research.

References

[1] J. Błachut and P. Wang, “Buckling of Barreled Shells Subjected to External Hydrostatic Pressure,”
J. Press. Vessel Technol., vol. 123, no. 2, p. 232, 2001.

[2] B. C. Cerik, “Ultimate strength of locally damaged steel stiffened cylinders under axial
compression,” Thin-Walled Struct., vol. 95, pp. 138–151, 2015.

[3] J. Błachut, “Imperfection sensitivity of externally pressurised shells,” Rev. Eur. des éléments finis,

10
 vol. 13, no. 8, pp. 787–810, 2004.

[4] P. K. Das, A. Thavalingam, and Y. Bai, “Buckling and ultimate strength criteria of stiffened shells
under combined loading for reliability analysis,” Thin-Walled Struct., vol. 41, no. 1, pp. 69–88,
2003.

[5] P. Iwicki, M. Wójcik, and J. Tejchman, “Failure of cylindrical steel silos composed of corrugated
sheets and columns and repair methods using a sensitivity analysis,” Eng. Fail. Anal., vol. 18, no.
8, pp. 2064–2083, Dec. 2011.

[6] S. G. P. Castro, R. Zimmermann, M. A. Arbelo, R. Khakimova, M. W. Hilburger, and R.

Degenhardt, “Geometric imperfections and lower-bound methods used to calculate knock-down
factors for axially compressed composite cylindrical shells,” Thin-Walled Struct., vol. 74, pp. 118–
132, 2014.

[7] O. Ifayefunmi and J. Błachut, “Instabilities in imperfect thick cones subjected to axial compression
and external pressure,” Mar. Struct., vol. 33, pp. 297–307, 2013.

[8] R. Degenhardt et al., “Investigations on imperfection sensitivity and deduction of improved knock-
down factors for unstiffened CFRP cylindrical shells,” Compos. Struct., vol. 92, no. 8, pp. 1939–
1946, Jul. 2010.

[9] R. Khakimova, C. J. Warren, R. Zimmermann, S. G. P. Castro, M. A. Arbelo, and R. Degenhardt,

“The single perturbation load approach applied to imperfection sensitive conical composite
structures,” Thin-Walled Struct., vol. 84, pp. 369–377, Nov. 2014.

[10] J. K. Paik, B. J. Kim, and J. K. Seo, “Methods for ultimate limit state assessment of ships and ship-
shaped offshore structures: Part I-Unstiffened plates,” Ocean Eng., vol. 35, no. 2, pp. 261–270,
2008.

[11] Q. shuai Cao and Y. Zhao, “Buckling strength of cylindrical steel tanks under harmonic
settlement,” Thin-Walled Struct., vol. 48, no. 6, pp. 391–400, 2010.

[12] A. Ghorbanpour Arani, R. Rahmani, and A. Arefmanesh, “Elastic buckling analysis of single-
walled carbon nanotube under combined loading by using the ANSYS software,” Phys. E Low-
Dimensional Syst. Nanostructures, vol. 40, no. 7, pp. 2390–2395, 2008.

[13] J. Arbocz and J. H. Starnes, “Future directions and challenges in shell stability analysis,” Thin-
Walled Struct., vol. 40, no. 9, pp. 729–754, Sep. 2002.

[14] Z. Tao, B. Uy, F. Y. Liao, and L. H. Han, “Nonlinear analysis of concrete-filled square stainless

11
 steel stub columns under axial compression,” J. Constr. Steel Res., vol. 67, no. 11, pp. 1719–1732,
2011.

[15] C. Hühne, Rolfes, E. Breitbach, and J. Teßmer, “Robust design of composite cylindrical shells
under axial compression — Simulation and validation,” Thin-Walled Struct., vol. 46, no. 7–9, pp.
947–962, Jul. 2008.

[16] M. S. Ismail, J. Purbolaksono, A. Andriyana, C. J. Tan, N. Muhammad, and H. L. Liew, “The use
of initial imperfection approach in design process and buckling failure evaluation of axially
compressed composite cylindrical shells,” Eng. Fail. Anal., vol. 51, pp. 20–28, 2015.

[17] O. Ifayefunmi, M. S. Ismail, and M. Z. A. Othman, “Buckling of unstiffened cone-cylinder shells

subjected to axial compression and thermal loading,” Ocean Eng., vol. 225, no. March, p. 108601,
2021.

[18] M. S. Ismail, O. Ifayefunmi, and S. H. Sheikh Md Fadzullah, “Buckling Analysis of Stiffened

Cone-Cylinder Intersection Subjected to External Pressure,” Key Eng. Mater., vol. 833, no. i, pp.
223–227, 2020.

[19] M. S. Ismail, J. Mahmud, and A. Jailani, “Buckling of an imperfect spherical shell subjected to
external pressure,” Ocean Eng., vol. 275, p. 114118, 2023.

[20] T. Weller and J. Singer, “Experimental studies on buckling of 7075-T6 Aluminium alloy integrally
stringer-stiffened shells,” TAE Rep., no. 135, 1971.

[21] T. Weller and J. Singer, “Further experimental studies on buckling of integrally ring-stiffened
cylindrical shells under axial compression,” Exp. Mech., pp. 267–273, 1974.

[22] A. C. Walker and S. McCall, “Strength of damaged ring and orthogonally stiffened shells - part I
Plain ring stiffened shells,” Thin-Walled Struct., vol. 5, pp. 425–453, 1987.

[23] Weingarten VI, Seide P, Peterson JP. Buckling of thin-walled circular cylinders. NASA SP-8007
Monograph, 1968, Volume 1.

12
Tables

Table 1: Convergence studies on the AR-1a model.

Number of Elements
4536 6366 7776 10,206 13,986
Buckling Load [kN]
19.02 19.01 19.02 17.55 17.55

Table 2: Dimension and geometrical properties of the models.

Shell R [mm] t [mm] L [mm] R/t L/R b [mm] h [mm]
AS-1L [12] 120.6 0.241 182 500 1.41 0.751 1.06
AR-1a [13] 120.43 0.238 108.03 506 0.897 0.75 0.262
R2 [14] 160 0.6 200 267 1.25 0.6 4.8

Table 3: Material properties used in the models.

Material E [GPa] σyield [MPa] υ
Aluminium-alloy 7075-T6 [12, 13] 70E3 540 0.3
Steel [14] 208E3 387 0.29

Table 4: The collapse loads by different imperfection techniques in comparison with the published
experimental data.
Model PEigen PExp PCollapse, SPLA PCollapse, SID
AS-1L [20] 32.84 kN 22.27 kN 19.11 kN 20.90 kN
AR-1a [21] 17.55 kN 11.67 kN 12.90 kN 13.42 kN
R2 [22] 0.965 MPa 0.550 MPa 0.618 MPa 0.563 MPa

Table 5: The knockdown factors of each tested model.

PExp PCollapse, PCollapse,
Model
/PEigen SPLA /PEigen SID /PEigen
AS-1L [20] 0.678 0.582 0.636
AR-1a [21] 0.665 0.735 0.765
R2 [22] 0.570 0.640 0.583

13
Table 6: The summarised results of increased imperfection dimple size for each model.
Dimple size PCollapse, SID
Model PCollapse, SID a/t
a [mm] b [mm] /PEigen
9 11.4 20.90 kN 0.636 1.24
AS-1L 13.5 11.4 20.21 kN 0.615 0.82
18 11.4 18.75 kN 0.571 0.88
9 3 13.42 kN 0.765 0.63
AR-1a 9 9 13.26 kN 0.756 0.34
12 12 13.12 kN 0.748 0.34
8 8 0.563 MPa 0.583 0.5
R2 12 12 0.531 MPa 0.550 0.42
16 16 0.510 MPa 0.528 0.42

Table 7: Model AS-1L with constant imperfection a/t = 2 at various locations.

Dimple size Collapse load [kN]
Location of imperfection
a [mm] b [mm] SID SPLA
Top-section 18 11.4 26.57 19.28
Mid-section 18 11.4 18.51 18.38
Lower-section 18 11.4 26.50 19.24

14
Figures

Fig. 1: (a) Boundary condition of FE modelling by SPLA (b) Details of ring stiffener and (c) stringer
stiffener profile and its variables.

15
Fig 2: Stress vs. strain response for Steel and Aluminium-Alloy 7075-T% materials.

16
Fig 3: (a) Initial imperfection located at the midsection of a stiffened cylinder with (b) details
modelling of SID represented by the R2 model.

17
Fig. 4: (a) Sensitivity studies of AS-1L and the estimation of collapse loads by (b) AS-1L (c) AR-1a
and (d) R2 models.

18
Fig. 5: Summary of knockdown factor against radius over thickness ratio.

19