18 views

Uploaded by h_a_mech

A Study on Effects of Initial Deflection on
Ultimate Strength of Ring-stiffened Cylindrical
Structure under External Hydrostatic Pressure”.

- 006 Leer Chapter9a
- LRFD Pre Standard - Revised FINAL - Nov 9 2010
- The Stability of a Facetted Glass Shell Structure
- compression-1.pdf
- NTU Composite Column
- Bruhn + Supplement.pdf
- Buckling Mode Interaction in ColdFormed Steel Columns and Beams
- Buckling of Struts
- Chapter 11
- Lecture Summary
- buckling load analysis
- EXACT THEORY OF BUCKLING OF A THICK SLAB
- Design of Columns and Struts in Structural Steel
- Nazli Azimikor_Composite Deck Design Report
- Groutand Hertz
- Centre of Mass
- Colum n Strut
- F09 CE470 Course Outline
- Struts
- LAB 5 Buckling and Structural Stability Under Compression

You are on page 1of 16

cylindrical shells subjected to axial compression

J.F. Jullien*, A. Limam

Research Unit of Civil Engineering, Institut National des Sciences Appliquees, Lyon, France

Abstract

Experimental and numerical methods are used to study the stability problem of cylindrical

shells with cut-outs. The paper presents parametric research of the shape (square, rectangular,

circular), the dimensions (axial and circumferential sizes, diameter) of the hole. The effect of

the location and the number of the holes are also studied. The analysis indicates that the

critical load is sensitive to the opening angle or circumferential size of the hole. The function

(critical load-opening angle) is linear for large openings and independent of the geometrical

imperfections of the shell. However for small openings, it is necessary to take into account

the coupling between the initial geometrical imperfections and the openings. The linear

approach does not fit because of the importance of the evolution of the displacements near

the openings. These results will be used for the development of European rules. 1998

Elsevier Science Ltd. All rights reserved.

Keywords: Opening; Buckling; Cylindrical shell; Axial compression

1. Introduction

In many industrial applications, shells are equipped with openings of various

shapes, sizes and locations within the lateral surface. The objective of the present

paper is to improve the understanding of the effect of cut-outs on the critical buckling

load of thin cylindrical shells.

Little work has been done in the literature on the subject. Early work dates back

to Tennyson [1], who performed an experimental study on the effect of small circular

* Corresponding author

0263-8231/98/$see front matter 1998 Elsevier Science Ltd. All rights reserved.

PII: S 0 2 6 3 - 8 2 3 1 ( 9 8 ) 0 0 0 0 3 - 2

188

openings on shells which buckle in the elastic range. This work was followed by

Starnes [2] and Toda [3], who tested shells with a larger range of opening diameters.

Knodel and Schulz [4] were the first to perform tests on steel cylinders with a large

spectrum of imperfections (geometry, loading, material). This gave the possibility for

a statistical evaluation of the buckling strength of shells. At the same time no physical

understanding of the phenomenon was gained.

The first numerical calculations on cylindrical shells with cut-outs were reported

by Almroth and Holmes [5]. Based on the available results, Samuelson and Eggwertz

[6] proposed a simple analytical description of the effect of the diameter of the hole

on the buckling strength of shells.

None of the above-mentioned researches have provided sufficient information to

develop a reliable physical model of the considered problem, or to deduce a general

understanding. Many tests had been carried out with specimens in Mylar, giving a

purely elastic buckling, and the studied parameters did not take into account the

meridional or circumferential size of rectangular openings, or the position of the

opening in relation to the boundaries. Furthermore, the analysis of coupling geometrical imperfections with openings has not been carried out.

In order to achieve this goal, a new study has been performed on steel shells

subjected to uniaxial compressive load. The non-uniform axial stress induced by the

opening induces an internal bending, and is included in this work.

A series of tests have been undertaken with different shapes (square, rectangular,

circular), sizes and positions of openings with respect to the boundaries of the shell.

All tests were of high quality with procedures reproducing all the parameters, like

the initial geometrical imperfections. In fact, it is important to know to what extent

the effect persists in the case of shells with openings. The present paper gives a

clear answer to that question.

In parallel to the physical testing, an extensive parametric numerical study was

performed, in order to develop the know-how for such a complex case, taking into

account the relative influences of all the parameters.

Furthermore, this study gives a guide on developing the design rules.

2. Experimental methods

2.1. Definitions of shells

The notation used is defined in Fig. 1. The upper and lower cross-sections (A and

B) were assumed to remain plane and refrained from warping and/or ovalisation.

The upper cross-section was free to translate and rotate around the axes of the

section, while the lower cross-section was fixed. The loading was applied at the

centre of the upper cross-section by means of a displacement uz.

In the circumferential direction, the opening was defined by the central angle ,

or the curvilinear length 2c r or by the dimensionless length r

r

c

rt

Fig. 1.

189

In the axial direction, the height of square and rectangular cut-outs was defined

by h0. The distance from the lower edge of the opening to the lower cross-section

B of the shell was denoted by hb.

hb f

l h0

2

The experimental study was limited to relatively short shells with approximately

/r 2, with an aspect ratio of r/t 280, diameter 2r 99 mm, thickness t

0.175 mm and length 104 mm.

Different types of square, rectangular and circular openings were made, as shown

in Table 1. Characteristic dimensions of openings and their positions are defined for

the different cases studied (cases 2 to 7), and for comparison the case of a shell

without a cut-out (case 1).

190

Table 1

Characteristic dimensions of openings and their positions

Case 1

Case 2

2c

(mm)

r

(degree)

Case 3

Case 4

Case 5

Case 6

Case 7

Without opening

Dimensions of centrally positioned square opening

Fixed parameter h0 = 2c Variable parameter 2c

8.64

17.28

25.92 38.88

1.47

10

2.94

20

4.40

30

6.61

45

51.84

8.81

60

Fixed parameter r = 4.40

f

1

0.75

0.50 0.25

0

Number of equidistanced square openingsnb: number of openings r: 1 opening

r

1.47

4.40

4.40

8.80

2

2

3

2

nb

Height of centrally positioned rectangular opening

Fixed opening r = 2.94

0.17

0.25

0.37 0.50 0.75

h0/

Dimensionless curvilinear width of centrally positioned rectangular opening

Fixed parameter h0/ = 0.17

r

2.94

4.40

6.61 8.81 13.23

Dimensionless diameter of centrally positioned circular opening

r

0.15 0.28

0.73

1.46 2.88

4.34 5.75

Over 100 specimens were manufactured. Cylindrical shells were made from sheet

metal by means of rolling and electric welding along a generator. These operations

were automated to ensure that the resulting geometric imperfections were the same

for all specimens, which is a necessary condition for the present comparative study.

The opening was always situated on the opposite side of the weld. Great care was

taken so that the process of cutting the opening did not introduce any additional

imperfections.

The tests were carried out on a special device, under displacement control according to the geometrical and boundary conditions defined above (Fig. 1). The boundary

conditions were introduced in the specimens to be tested by means of two rigid

plates, which were connected to cylinders by a sobering technique.

2.3. Initial geometrical imperfections

For each specimen tested the geometrical imperfections were measured by means

of a contactless sensor at 25,000 points. For that purpose a special automated scanning system was mounted on the testing machine.

The geometric imperfections were located mainly in the vicinity of the welded

generator. The imperfections were characterised in the circumferential direction by

the wavelength corresponding to n 7 or 8, with a maximum amplitude of

191

w

(n) 0.9.

t

There were no imperfections in the meridional direction (m 0) except along the

welded generator which shows two half waves (m 2) with an amplitude

w

2w(m) 2 98 m

(m) 0.56 (Fig. 2).

t

The results for different cases of openings are compared with those corresponding

to shells without cut-outs. This resulted in a large experimental and numerical study

of shells without cut-outs and with typical geometrical imperfections.

The reproductibility of the present test methodology is verified by comparing

results for eight specimens without an opening. The mean value of buckling load

was P0 11.78 KN, with a standard deviation of 58 N, which corresponds to 5%

of P0. The corresponding stress is cr 216 MPa representing 47% of the theoretical

buckling stress cr.

3. Effect of shape of cut-out

3.1. Square openings

For given dimensions of the square and centred opening under consideration, the

force-displacement characteristic of the shell remains linear up to the first local buckling (LB), which occurs in the vicinity of a contour of the opening. After this point

the behaviour is linear and stable with a reduced slope until a global buckling

(collapse buckling, CB) takes place which involves the entire shell, Fig. 3.

In the pre-buckling range, before the first local buckling, large radial displacements

develop near the edges of the openings, in particular around the rim of the hole.

Fig. 2.

192

Fig. 3.

Compression of load-deflection curves for cases 1 and 2 showing the effect of hole size.

These observations are visualised in Fig. 4, which shows radial displacements for

two types of square openings at 50 and 80% of the local buckling, at the first local

buckling and finally at the collapse buckling.

A square cut-out in a cylindrical shell subjected to axial compression reduces the

first local buckling load and the global collapse buckling load as compared to the

reference shell without a hole. The drop of critical loads with the dimensionless

geometrical parameter of the opening r is described by a linear function in the studied

range 1.47 r 8.81. The ratio between these two critical loads is a function of

r, with a mean value of approximately 1.1 (Fig. 5). Beyond the point of approximately r 10, the first local buckling does not develop.

For chosen dimensions of the square cut-out r 4.40 the position of the opening

along the meridional direction (axis of a cylinder) with respect to cross-sections A

or B does not change the critical load. This interesting result indicates that the critical

load is not linked to the direct distribution of stresses above or below the hole. The

above conclusion is valid as long as cross-sections A and B remain plane but could

rotate. In practice, this boundary condition could be satisfied by placing reinforcing

rings of sufficient rigidity at these cross-sections, Fig. 5.

Multiple and equidistanced openings over the circumferential direction produce a

similar type of instability (first local buckling followed by collapse buckling) with

a corresponding loss of rigidity. In the case of multiple openings the reduction in

the critical load is smaller when the comparison is made taking the sum of the

opening angle, as shown in Fig. 6. Because of symmetry, the appearance of critical

loads is not accompanied by the development of internal bending. The above

dependence can be converted to a different co-ordinate system in which the abscissa

193

Fig. 4. Evolution of buckling forms (prebuckling, local buckling (LB), global buckling (CB). White

represents peaks and blue represents valleys as viewed from inside.

is not the sum of openings but the width of a single opening. It is interesting to note

that in the above co-ordinate system, the first local buckling load is the same for a

shell with single and multiple openings.

194

Fig. 5. Reduction of critical loads with an opening angle, showing weak effect of the height and position

of a hole in the meridional direction.

Fig. 6.

Reduction of the first local buckling load with the size and number of openings (cases 2 and 6).

A cylindrical shell with a rectangular opening carries the same critical buckling

load as an identical shell with a square opening, provided the cut-outs are of an

identical length in the circumferential direction, Fig. 5. The height of the opening

does not intervene in the studied range of parameters. At the same time this conclusion cannot be extended to cut-outs in the form of a slot, without a separate study.

195

A cylindrical shell with a circular opening of a diameter equal to the width of the

square opening develops the same collapse buckling load (Fig. 7). At the same time

the first load buckling is not observed in this case. The above result indicates that

the occurrence of the first local buckling is preceeded by a redistribution of stresses

due to stress concentration at the corners of square or rectangular cut-outs.

For this type of opening the effect of a small hole dimension r 0.15; 0.28; has

also been analysed. It was found that very small holes do not cause any appreciable

change in the critical load.

4.1. Opening angle

The present experimental study has determined the effect of the shape of the opening on the critical load of cylinders subjected to axial compression. It was shown that

the key parameter controlling the buckling strength of the shell was the dimension of

a cut-out in the circumferential direction. In the case of multiple openings positioned

equidistantly around the circumference, the control parameter was the sum of hole

dimensions of all openings.

Fig. 7. Dimensionless critical load versus and opening angle for square and circular holes. Note that there

is no local buckling for a circular opening and a generalised buckling load is the same for both shapes.

196

The above experimental results are confirmed by finite element calculations. The

calculations were run using the three-dimensional code CASTEM 2000, developed

by the French Atomic Energy Commission. The code is capable of dealing with

various types of singularities. Half of the shell was modelled by 3500 triangular

elements DKT (three displacements and three rotations per node), as shown in Fig.

8. Contrary to previous analysis on buckling of shells, linear analysis (LA) was found

to give a smaller critical load than geometrically non-linear analysis (GNA), as

shown in Fig. 9. This can be explained by an inability of the geometrically linear

model to account for large radial displacements around the opening.

In fact, already at 20% of the critical load, radial displacements exceed by approximately two times the displacements predicted by linear theory. Further refinement

of the model by introducing the non-linear material behaviour (GMNA) brings the

level of stresses even lower.

The result of the linear (LA) and non-linear (GMNA) analyses mentioned above

corresponds to loading case (a) when the cross-sections A and B remain parallel.

This condition does not fully reflect reality because a single opening produces an

asymmetry and so-called internal bending. At the same time, the results of calculations with geometrical and material non-linearities referred to in Fig. 9 as case (b)

correspond to the boundary condition in which cross-section A is free to displace

and rotate in all three directions.

It should be noted that for design purposes one has to consider rather conservatively only the first critical buckling load, even though there could be a stable postcritical range until collapse buckling occurs. Remember that for circular holes there

is no first local buckling. Therefore the design should be made on the basis of global

collapse buckling.

Fig. 8.

Finite element mesh of a cylinder with an opening (3500 elements for one half of the shell).

197

Fig. 9. Finite element results for two types of boundary and loading conditions (a) and (b) and linear

(LA) and non-linear (GNA and GMNA) formulations.

As was mentioned in Section 2.1 all the results presented so far are valid only

for shells with /r 2.

The amount of bending deformation introduced by the presence of non-symmetrically positioned openings will depend on the distance between two cross-sections,

taking into account the second order effect and greater susceptibility for shape distortion in the critical cross-section in the case of longer cylinders. Fig. 10 indicates that

there is an asymptotic limit for the length dependence. A more detailed analysis is

necessary to quantify the above dependence. In the mean time, the existing results

can be used for shells reinforced by two heavy rings at cross-sections A and B

distanced 2r apart. This will ensure that cross-sections remain plane without

warping and ovalisation.

4.3. Geometrical imperfections

The experimental results for different forms of openings correspond to metal cylinders characterised by reproducible geometric imperfections encountered in industrial

practice. These results were compared with earlier results reported by Starnes [2], see

Fig. 11. While the shell material is different (Mylar versus steel) and the geometrical

imperfections in Starness experiments are much smaller compared with the present

ones, the overall behaviour of both types of shells is identical in a certain range of

198

Fig. 10.

Effect of dimensionless length /r on critical buckling load (numerical study GMNA (b)).

Fig. 11. Comparison of INSA tests with earlier results by Starnes [2].

the parameter r. The range of interest corresponds to angular openings larger than

a limiting value denoted by r.

In the first approximation, the geometrical imperfections are seen not to have any

significant effect compared with the effect of opening beyond the critical limit r.

It is possible to identify a reduction function for a given opening over the range

r > r. From Fig. 11 it transpires that for small openings up to this limiting value,

199

the effect of imperfections must be coupled with that of the cut-out. At the same

time, for very small openings, their size is not important compared with a significant

imperfection sensitivity of cylinders subjected to compressive loads.

As shown in Fig. 12 a comparison between calculations and experimental results

is in full agreement.

The obtained results are limited to the aperture size r 6.61 (i.e.

45) and 212 r/t 400. These limits could be extended in the continuation of

this research.

In Fig. 12, the limits of two distinct zones are clearly identified. In the case of

small openings, it is necessary to include in the analysis a coupling between the

geometrical imperfection and presence of an opening. In the range of moderately

large openings, the existence of geometrical imperfections of any size can be neglected.

A careful examination of post-critical geometry for all cases considered indicates

that the modal geometry is not effected by the presence of openings. This statement

is valid for different dimensions and shapes of shells tested. One can conclude that

the mode numbers n and m are invariants, because they depend on the geometry

of a shell rather than that of an opening.

It should be recalled that the critical load has an asymptote with increasing amplitude of imperfections. It is important to note that the dependence of the critical load

on the hole size has a similar form as the known imperfection sensitivity. There

Fig. 12. Comparison between experimental results and numerical calculations for perfect (GMNA) and

imperfect (GMNIA) shells. Note the existence of two ranges. For small r there is a coupling between

geometrical imperfection and hole size. For medium opening, there is no coupling.

200

the reduction factor reaches a maximum (asymptotic) value where the opening size

corresponds to the half-length of the buckling wave [7,8]. It is shown in Fig. 12 that

geometrical imperfections no longer have any effect on cr for r > 2.94, i.e. for an

opening equal or larger than the half-wave of the Yoshimura pattern.

4.4. Principle of the proposed rule

The proposed rule uses a reduction (knock-down) factor to the critical Donnell

stress cr in the same way as is done for a cylindrical shell without an opening. This

rule is developed for the most unfavourable case, i.e. for cut-outs with sharp corners

(squares, rectangles). For the purpose of evolving a European rule [9], the same

notation is used for the reduction factor , as in the classical buckling analysis of

imperfect shells. The reduction factor is a linear piece-wise function of the geometrical parameter of the openings r, (Fig. 13). In the design rule it is proposed that

three distinct regimes be distinguished, denoted respectively by (1), (2) and (3).

Regime (1) corresponds to very small openings in which the size of the hole has

practically no effect on the critical load. The shell response is dominated by the

geometrical imperfections. The reduction factor is constant with respect to the opening parameter r, and its magnitude is equal to the reduction factor of an imperfect

cylinder without an opening subjected to axial compression.

Responses of shells with relatively large openings are dominated by the presence

of cut-outs rather than geometrical imperfections. This function is described by a

linear dependency given by segment (3) in Fig. 13. For intermediate hole sizes there

is a strong coupling effect between the opening parameter r and geometrical imper-

201

fections. The reduction factor in this range is approximated by a straight line, designated in Fig. 13 as segment (2).

In order to define the position of the above three lines on the r plane it is

necessary to identify six parameters.

5. Conclusion

The above experimental and numerical work leads to the following conclusions,

while enhancing the European rules relating to shell buckling.

The opening parameter r characterizes the effect of a hole on the critical load of

a cylindrical shell with an opening under axial compression;

The behaviour shows two large zones:

for small openings in which there is a coupling between the geometrical imperfection and the presence of an opening;

for a moderately larger opening the existence of geometrical imperfections of any

size can be neglected;

The limit between these two zones is defined by an opening angle equivalent to

the length of the half-wave of the shell without any opening;

The critical mode is independent of an opening;

For a moderately larger opening the reduction of critical load is linear with the

angle of the opening;

The height and the position of the hole have little influence;

The linear bifurcation theory underestimates the carrying capacity because it predicts buckling of the vertical edge of the cut-out.

Acknowledgements

The authors wish to thank the CECA-ECSC, which has supported this research

(contract no. 7210-SA/208, part E Enhancement of ECCS design recommendation

and development of Eurocode 3, parts related to shell buckling).

References

[1] Tennyson RC. The effect of unreinforced circular cutouts on the buckling of circular cylindrical shells

under axial compression. Journal of Engineering for Industry 1968;November:5416.

[2] Starnes JH. The effects of cutouts on the buckling of thin shells. In: Fung YC, Sechler EE, editors.

Thin shell structures. Englewood Cliffs, New York: Prentice Hall, Inc., 1974:289304.

[3] Toda S. Buckling of cylinders with cutouts under axial compression. Journal of Experimental Mechanics 1983;3:4147.

[4] Knodel P, Schulz U. Stabilite de cheminees dacier a` ouvertures dans les tuyaux. Stahlbau (Der)

1988;57(1):1321.

202

[5] Almroth BO, Holmes AM. Buckling of shells with cutouts. Experimental and analysis. Journal of

Solids Structures 1972;8:1057566.

[6] Samuelson AL, Eggwertz S. Shell stability handbook. London: Elsevier Applied Science, 1992:278

pp.

[7] Jullien JF, Limam A. Effect of openings on the buckling of cylindrical shells subjected to axial

compression. International Conference on Advances in Steel Structures, Hong Kong, 1114

December 1996.

[8] Al Sarraj M. Effets des ouvertures sur la stabilite des coques cylindriques minces soumises a` compression axiale. These de Doctorat, INSA de Lyon (France), 5 Octobre 1995:260 pp.

[9] ECCS/TWG8.4 Buckling of steel shells. European recommendations, 4th edn, no. 56, 1988.

- 006 Leer Chapter9aUploaded bynelsonsainz
- LRFD Pre Standard - Revised FINAL - Nov 9 2010Uploaded bydakotaaa
- The Stability of a Facetted Glass Shell StructureUploaded byqatarstructz30
- compression-1.pdfUploaded byjademarielle812
- NTU Composite ColumnUploaded bybsitler
- Bruhn + Supplement.pdfUploaded byCraig Swenson
- Buckling Mode Interaction in ColdFormed Steel Columns and BeamsUploaded byLatoya Jones
- Buckling of StrutsUploaded byMohamed Zamri
- Chapter 11Uploaded bysharathr22
- Lecture SummaryUploaded bysamir_ssh7151
- buckling load analysisUploaded bySrinivas Cherukuri
- EXACT THEORY OF BUCKLING OF A THICK SLABUploaded byLeonardo de Araujo
- Design of Columns and Struts in Structural SteelUploaded byMaqsood
- Nazli Azimikor_Composite Deck Design ReportUploaded byDani Torz
- Groutand HertzUploaded byoe07m007
- Centre of MassUploaded byaditya2053
- Colum n StrutUploaded byAjit Bole
- F09 CE470 Course OutlineUploaded by2011kumar
- StrutsUploaded byAjit Bole
- LAB 5 Buckling and Structural Stability Under CompressionUploaded byYousif Said
- BucklingUploaded byAbdul Rahman
- Lecture -13 Plate GirdersUploaded bygundulp
- Design of ColumnsUploaded byVenkatesh Gangadhar
- Mercedes BenzUploaded by郑 凯伦 Tey Kai Loon
- Unit 15Uploaded byBiplab Sadhukhan
- Kashani Et Al Accepted Manuscript 1Uploaded byPuneet Kaura
- som15.pdfUploaded byjhalakdutta
- 2002-1 The Roman OvalUploaded byEDITORIAL STARBOOKS
- Blast Resistant Control BuildingUploaded byseena107
- Mechanics of Deformable Bodies IIUploaded bykarlajane03

- 20 en Poliflex-tUploaded byGatot Nugroho
- CEL ConsultingUploaded byHuy Tran
- Physical layer standards for ATMUploaded byoureducation.in
- Principles of AssessmentUploaded byDarren Cariño
- Emulator - Passenger Seat Occupant Detector EnUploaded byOliver García
- Caliber 115Uploaded byAroldo Ruiz
- 8254 MicroprocessorUploaded byFaiz Rahman
- 07500 RoofingUploaded byShiraz Khan
- 896803_Tiguan_Overview.pdfUploaded byAlmas Bogdan
- Calculating ARIEL Lube RatesUploaded byLenin Marcel Fajardo Almeida
- Portfolio ResumeUploaded byTyler Shumock
- Methodology for Liquid Dye Penetrant TestUploaded byBheema Bhargavi
- Physical and Financial Accomplishments 2017 for RMEA ONLY FINALUploaded byStrep Rerref
- NAIS--Lipan Apache Women's Defense--Context-NARRATIONUploaded byapi-885332
- SpeakingUploaded byjsayer622
- Ultrasonic Rangefinder Using 8051Uploaded byKaos Polos Nakira
- air distributionUploaded byRamon Chavez B
- 2 - Human Neuroplasticity and Education (Capítulo 1[Language and Literacy] Pág 19 a 70)Uploaded byGustavo Caribé
- docUploaded byBhavisha Singh
- msg00009Uploaded byAswath78
- 1-s2.0-S187704281502933X-mainUploaded byLupu Adriana
- A Heuristic Algorithm to Optimise Stope BoundariesUploaded bypaulogmello
- FAM Catalogos Processing Plants 150dpiUploaded byluisantonio2
- Part2Uploaded bypulilatha
- More "human" or simply less Neanderthal? "Classic" vs "Levantine" NeanderthalsUploaded byNoelle Tankard
- BOQ toilet.xlsxUploaded byAjit D Gandhi
- 5 Evaluating Performance- Measuring Results and BehaviourUploaded byDewesh Shukla
- Wind Turbine Blockset in Matlab SimulinkUploaded byArifujjaman Sumon
- Chapter 4Uploaded byAlia Al Zghoul
- Security Assessment Rfp Cheat SheetUploaded bys1x8