T.R.

GEBZE TECHNICAL UNIVERSITY

INSTITUTE OF ENERGY TECHNOLOGIES

BUCKLING OF THE ISOGRID STIFFENED CYLINDERS

UNDER COMPRESSIVE LOADING

ORHAN KEMAL CAN

A THESIS SUBMITTED FOR THE DEGREE OF
MASTER OF SCIENCE
DEPARTMENT OF APPLIED PROPULSION SYSTEM
DESIGN & ENGINEERING FOR AEROSPACE
TECHNOLOGIES

GEBZE
2022
 T.R.
GEBZE TECHNICAL UNIVERSITY
INSTITUTE OF ENERGY TECNOLOGIES

BUCKLING OF THE ISOGRID

STIFFENED CYLINDERS UNDER
COMPRESSIVE LOADING

ORHAN KEMAL CAN

A THESIS SUBMITTED FOR THE DEGREE OF
MASTER OF SCIENCE
DEPARTMENT OF APPLIED PROPULSION
SYSTEM DESIGN & ENGINEERING FOR AEROSPACE
TECHNOLOGIES

THESIS SUPERVISOR
PROF. DR. OSMAN SAİM DİNÇ

GEBZE
2022
 GEBZE TEKNİK
ÜNİVERSİTESİ YÜKSEK LİSANS JÜRİ ONAY FORMU

GTÜ Enerji Teknolojileri Enstitüsü Yönetim Kurulu’nun …30/05/2022….

tarih ve .2022…/…07… sayılı kararıyla oluşturulan jüri tarafından 08/06/2022
tarihinde tez savunma sınavı yapılan ORHAN KEMAL CAN’ın tez çalışması Enerji
Teknolojileri Anabilim Dalında YÜKSEK LİSANS tezi olarak kabul edilmiştir.

JÜRİ
ÜYE
(TEZ DANIŞMANI) : Prof. Dr. OSMAN SAİM DİNÇ

ÜYE : Prof. Dr. MAHMUT FARUK AKŞİT

ÜYE : Prof. Dr. MEHMET ALİ ARSLAN

ONAY
GTÜ Enerji Teknolojileri Enstitüsü Yönetim Kurulu’nun
……………………………. tarih ve ………/……... sayılı kararı.

İMZA/MÜHÜR
 SUMMARY
In this work, buckling capacity of isogrid-stiffened cylinders is examined with
changing parameters as R/t, isogrid triangle number on the circumference and
manufacturing tolerances. Firstly, hand calculations are performed based on NASA
Isogrid Design Handbook (NASA CR 124075) and NASA/SP-8007 (Buckling of
Thin-Walled Circular Cylinders). Secondly, a shell-beam FEA model is created using
ANSYS Workbench 2020 R1 where the skin and stiffeners are represented as shell
and beams respectively. This FEA model is employed to perform linear Eigenvalue
buckling analysis and nonlinear buckling analysis. Nonlinear analysis is run as large
deflection activated for with geometrical nonlinearity, nonlinear material properties
activated and with an initial geometrical imperfection applied to the FEA model.
Results of hand calculation and FEA are compared and interpreted considering
different R/t ratios and circumferentially isogrid triangle number.
Buckling capacity is determined and compared for isogrid-stiffened cylinder and
unstiffened cylinder which have same diameter and weight.
Knock down factors (KDF’s) which scale down the theoretical buckling load to
actual critical critical buckling are compared and interpreted for hand calculation and
FEA methodologies. Additionally, the effect of the geometrical imperfections caused
by manufacturing tolerances is showed on the KDF level.

Key Words: Isogrid Stiffened Structure, Non-Linear Buckling Analysis, NASA

Isogrid Design Handbook, Knock Down Factor (KDF) for buckling.

iv
 ÖZET
Bu çalışmada; R/t, çevredeki güçlendirici izo-kafes üçgen sayısı ve imalata
toleransları gibi parametrelerin değişimi ile izo-kafes güçlendirilmiş silindir
yapılarının burkulma kapasiteleri incelenmiştir.İlk olarak, NASA Isogrid Design
Handbook (NASA CR 124075) ve NASA/SP-8007 (Buckling of Thin-Walled Circular
Cylinders) dokumanlarına dayanarak el hesapları yapılmıştır. İkinci olarak, ANSYS
Workbench 2020 R1 programı kullanılarak, yapının ana-kabuk kısmının kabuki
elaman olarak, güçlendirici kafes yapısının ise kiriş eleman olarak modellendiği bir
kabuk-kiriş SEA(sonlu elemanlar analiz) modeli oluşturulmuştur. Bu SEA modeli ile
lineer özdeğer burkulma ve lineer olmayan burkulma analizleri gerçekleştirilmiştir.
Lineer olmayan analizde, malzeme ve geometrik lineer olmama durumları ile imalat
kaynaklı geometri hataları da hesaba katılmıştır.
Farklı R/t oranları ve çevredeki güçlendirici izo-kafes üçgen sayılarına göre
yapılan el hesaplamaları ve SEA sonuçları karşılaştırılmış ve yorumlanmıştır.
Aynı çapa ve ağırlığa sahip, izo-kafes güçlendirilmiş silindir ile
güçlendirilmemiş tek kabuk silindirin burkulma kapasitesi belirlenmiş ve
karşılaştırılmıştır.
Teorik burkulma kuvvetini, gerçek burkulma kuvvetine dönüştürmek için
kullanılan düşürme katsayısı (DK) değerleri, el hesapları ve SEA için karşılaştırılıp,
yorumlanmıştır. Ek olarak, imalat toleransları kaynaklı geometrik kusurların DK
üzerine olan etkisi incelenmiştir.

Anahtar Kelimeler: İzogrid Güçlendirilmiş Yapı, Doğrusal olmayan burkulma

Analizi, NASA Izogrid Tasarım Elkitabı, Burkulma Düşürme Katsayıları.
v
 ACKNOWLEDGEMENTS
I would like to express my deep and sincere gratitude to my supervisor, Prof.
Osman Saim Dinç, who is the architect of this Master of Science program collaboration
between TEI and GTU.
I wish to express my warm and sincere thanks to my friends and collogues Mithat
Murat Yalçın and Semih Babur Oğuz for their collaborations to this research and to
my team leader Bahtiyar Taylan Aksongur who encouraged me to attend the program.
Finally, I am grateful to my wife Merve Can, my son Tuna Can and my parents
Ayşe &Yaşar Can for their love and support.

vi
 TABLE of CONTENTS
Page
SUMMARY iv
ÖZET v
ACKNOWLEDMENTS vi
TABLE of CONTENTS vii
LIST of ABBREVIATIONS and ACRONYMS ix
LIST of FIGURES x
LIST of TABLES xii

1. INTRODUCTION 1
1.1. Grid Structures 1
1.2. Isogrid Structures and Usage in Aerospace 1
1.3. Buckling 3
1.4. Scope of the Thesis 4
2. ANALYTICAL CALCULATIONS 5
2.1. Buckling of Thin Walled Cylindrical Structure 5
2.2. Buckling of Isogrid Stiffened Thin Walled Cylindrical Structure 6
3. FINITE ELEMENT ANALYSIS 10
3.1. Introduction 10
3.2. Geometry 11
3.3. Material 12
3.4. Loading and Boundary Conditions 14
3.5. Meshing 15
3.6. Eigenvalue Buckling Analysis 16
3.7. Non-Linear Buckling Analysis 18
4. DISCUSSIONS AND CONCLUSION 22
4.1. Comparison of FEA and Hand Calculation 22
4.2. Isogrid Effect on Buckling Capacity and Weight 24
4.3. KDF by NASA Manual and FEA 25
4.4. KDF and Geometrical Imperfection Amplitude 26
4.5. CONCLUSION 28

vii
REFERENCES 29
BIOGRAPHY 30

viii
 LIST of ABBREVIATIONS and ACRONYMS
Abbreviations Explanations
and Acronyms
𝑘𝑥 : Buckling Coefficient
𝑁𝑥 : Critical Buckling Load
𝑚 : Axial Half-Waves of the buckling mode shape
𝛾 : Knockdown factor
𝐸 : Young Modulus
𝑡 : Thickness of the shell
𝑛 : Circumferential full waves of the buckling mode shape
𝑣 : Poisson’s Ratio
𝑏 : Rib thickness
𝑑 iso : Rib depth
hiso : Height of isogrid triangle
tiso : Thickness of skin
𝐸* : Equivalent Young Modulus
𝑡∗ : Equivalent Thickness of the Shell

ix
 LIST of FIGURES

Figure No: Page

1-1: Grid Structure Example. 1
1-2: Grid Types. 1
1-3: EJ200 Engine. 2
1-4: Fabrication Process of Stiffened Test Structure. 3
1-5: Buckling Test of a Cylindrical Shell. 3
2-1: Dimensions for a thin walled cylinder. 5
2-2: Isogrid Parameters. 7
2-3: Critical Buckling Loads per Triangle Number. 9
2-4: Buckling Load to Weight Ratio per Triangle Number. 9
3-1: Design Iterations per the circumferencially triangle number. 10
3-2: Geometry Modeler. 11
3-3: Geometry Model Parameters. 11
3-4: Titanium Alloy Bilinear Isotropic Hardening Curve. 12
3-5: Boundary Conditions. 13
3-6: Finite Element Types. 13
3-7: Buckling Load-Node Number-Mesh Size Plot. 15
3-8: Project Schematic for Eigenvalue Buckling Analysis. 15
3-9: Buckling Mode Shape for the option w/36 circumferentially 16
triangles.
3-10: Ansys Solver Controls. 16
3-11: Ansys Step Controls. 17
3-12: Project Schematic for Non-Linear Buckling Analysis. 17
3-13: KDF for Actual Buckling Load. 18
3-14: Stress Levels before Buckling Failure. 18
3-15: Non-linear Deformation – Buckling Mode Shape Comparison for 19
isogrid option w/36 circumferentially triangle number.
3-16: Non-linear Deformation – Buckling Mode Shape Comparison for 20
Unstiffened Cylinder.
3-17: Critical Buckling Loads per Triangle Number. 21

x
3-8: Buckling Load to Weight Ratio per Triangle Number. 21
4-1: Theoretical Buckling Loads per FEA and NASA Manual. 22
4-2: FEA-Hand Calculation Difference per Stiffener Density. 23
4-3: Theoretical Buckling Loads per FEA&NASA Manual for different 23
R/t ratios.
4-4: FEA-Hand Calculation Difference per R/t. 24
4-5: Theoretical and Actual Buckling Loads per FEA and Hand 26
Calculations.
4-6: Radial Deflection of First Eigenvalue Buckling Mode Shape. 27
4-7: KDF’s per Manufacturing Tolerances. 28

xi
 LIST of TABLES

Table No: Page

2-1: Design Iterations. 7
2-2: Material Properties. 7
2-3: Calculation Results. 8
3-1: Material Properties. 12
3-2: Mesh Sensitivity Results. 14
3-3: FEA Results. 20
4-1: Buckling Capacity of Unstiffened and Isogrid Stiffened Cylinder. 25
4-2: KDF’s per Manufacturing Tolerances-1 27
4-3: KDF’s per Manufacturing Tolerances-2 27

xii
 1. INTRODUCTION
1.1 Grid Structures
Grid structures means a thin outer skin and above, repetitive pattern of stiffening
ribs to support the whole structure as shown in Figure 1. In that way, structures
increases their stiffness while the weight addition is limited.

Figure 1-1 Grid Structure Example.

Structural efficiency and environmental robustness are major advantages of grid
structures. [1] By changing the direction of ribs, required stiffness is achieved in any
direction, which is needed per the loading condition of the structure. Other directions
that are not on a load path might be used for weight reduction opportunities.
Additionally, cellular structure provides higher damage tolerance capability.
According to the pattern of the repetitive ribs, grid structures are subdivided as
orthogrid, diamond grid, triangle rectangle grid and isogrid shown in Figure 1-2. [2]
The pattern affects the directional material behavior of the whole structure.

Figure 1-2 Grid Types.

1.2 Isogrid Strucuters and Usage in Aerospace

Isogrid structures has equilateral triangular pattern of stiffening ribs. Thus, tri-
directional grid provides isotropic material properties in all these directions. Isotropic

1
material properties ensure good compressive and bending combined load capabilities.
Isogrid structure has high stiffness to weight ratio, thus it is used in aerospace and
aviation applications. As an example, Eurojet EJ200 [3] which is a low-bypass
turbofan engine has isogrid stiffened casings as shown in Figure 1-3 and powers
Eurofighter Typhoon aircraft.

Figure 1-3 EJ200 Engine.

In turbojet engines of aircrafts, cold section casings which works under
compression loads are very sensitive to buckling failure. Due to weight restrictions,
design engineers aim to keep the casing thickness as thin as possible while the
hardware has still enough buckling margin.
Also, aerospace launch vehicle program are interested in grid stiffened
cylindrical structures. Hilburger Mark W. et al worked on the buckling of large-scale
, integrally-stiffened metallic cylinder. [4] In the scope of this study, eight different 8
ft diameter cylinders were fabricated and tested under various loading conditions and
test results were compared with finite element analysis results. Figure 1-4 is taken from
the article and shows the fabrication process of integrally stiffened, 2195-T3
Aluminum alloy cylinder with 96 inch diameter and 78 inch length. 2195 Al-Li flat
plates were machines to stiffened panels and bump-formed to curved ones. Finally,
curved stiffened panels were friction-stir welded together.

2
 Figure 1-4 Fabrication Process of Stiffened Test Structure.
1.3 Buckling
Buckling is an instability issue for structures under compressive loading
condition. Especially, it is very critical failure mode for long and slender structures
[5]. The shape of the structure changes suddenly at a critical level if a gradually
increasing load is applied on it. For thin and slender structure as shell cylinders, the
critical level is well below the yielding stress, so their strength is limited by buckling.
Figure 1-5 is taken from reference and shows buckling test of a cylindrical shell under
axial compression loading

Figure 1-5 Buckling Test of a Cylindrical Shell.

3
1.4 Scope of the Thesis
The aim of this study is to investigate the effect of added isogrid stiffeners on
the buckling capacity and the total weight of the cylindrical structure. While the critical
buckling load of the structure increases, also the total weight, which is one of the most
important key parameters for aircrafts, is increasing. Thus, the correlation between the
interval of the stiffeners and stiffness to weight ratio is worth to investigate.
Firstly, analytical calculations are performed per NASA Isogrid Design
Handbook. Theoretical, critical buckling loads are calculated and these values are
multiplied with KDF to get the real critical loads. Following, finite element analysis is
used to study the theoretical critical buckling loads and the nonlinear results
considering geometrical and material nonlinearities. This method is implemented by
the software ANSYS Workbench 2020 R1.
Titanium alloy is chosen as material because of its often usage in aero engine
cold section casings.

4
 2. ANALYTICAL CALCULATIONS
2.1 Buckling of Thin Walled Cylindrical Structure
Based on Donnell’s shell theory, critical buckling load for a isotropic,
moderately long thin walled cylinder under axial compression is given by [7]:
𝜋2 𝐷
𝑁𝑥 = 𝑘𝑥 (2-1)
𝐿2

Where
12 (𝛾𝑍)2
𝑘𝑥 = 𝑚2 (1 + 𝛽 2 )2 + 𝜋4 (2-2)
𝑚2 (1+𝛽 2 )2

𝐸𝑡 3
𝐷 = 12 (1−𝑣 2 ) (2-3)
𝐿2
𝑍 = 𝑅𝑡 √1 − 𝑣 2 (2-4)
𝑛𝐿
𝛽 = 𝑚𝜋𝑅 (2-5)

For moderately long cylinders, which have 𝛾𝑍 > 2,85 , equation (2-2) is
minimized to:
4√3
𝑘𝑥 = 𝛾𝑍 (2-6)
𝜋2

Using equations (2-1)-(2-6), theoretical critical buckling load is obtained as:

2𝜋𝐸𝑡 2
𝑁𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙,𝑡ℎ𝑒𝑜𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = (2-7)
𝑅√1−𝑣 2

As seen in equation (2-7), critical buckling load is dependent on the Young’s

Modulus and Poisson Ratio of the material and thickness and radius of the geometry.

Figure 2-1 Dimensions for a thin walled cylinder.

5
 In reality, actual critical buckling load is less than calculated theoretical due to
non-linearities and geometrical imperfections. To consider this reality, a knock down
factor (KDF) 𝛾 is applied
𝛾 = 1 − 0,901(1 − 𝑒 −𝜑 ) (2-8)
1 𝑅 𝑅
𝜑 = 16 √ 𝑡 (for < 1500) (2-9)
𝑡

2.2 Buckling of Isogrid Stiffened Thin Walled Cylindrical

Structure
As described in previous section, those formulas (2-1)-(2-9) are applicable for
isotropic unstiffened thin walled cylinders. However, isogrid stiffened cylinders have
ribs on the skin structure. Based on NASA Isogrid Design Handbook [8], an equivalent
Young’s Modulus E* and equivalent thickness t* can be calculated for isogrid
stiffened cylinders using formula (2-10) and (2-11) in order to be treated as a
monocoque, isotropic unstiffened cylinder. With those equivalent Young’s Modulus
E* and equivalent thickness t* values, equations (2-1)-(2-7) are applicable as stated in
Chapter 2.1.
𝛽
𝑡 ∗ = 𝑡𝑖𝑠𝑜 1+𝛼𝑖𝑠𝑜 (2-10)
𝑖𝑠𝑜

(1+𝛼𝑖𝑠𝑜 )2
𝐸∗ = 𝐸 (2-11)
𝛽𝑖𝑠𝑜

𝛽𝑖𝑠𝑜 , 𝛼𝑖𝑠𝑜 and 𝛿𝑖𝑠𝑜 are non-dimensional parameters and formulated as equations
(2-12)-(2-14) below using the values of the skin thickness, rib thickness, rib height and
the triangle height shown in Figure 7.

𝛽𝑖𝑠𝑜 = √[3𝛼𝑖𝑠𝑜 (1 + 𝛿𝑖𝑠𝑜 )2 + (1 + 𝛼𝑖𝑠𝑜 ) (1 + 𝛼𝑖𝑠𝑜 𝛿𝑖𝑠𝑜 2 )] (2-12)

𝑏𝑖𝑠𝑜 𝑑𝑖𝑠𝑜
𝛼𝑖𝑠𝑜 = (2-13)
𝑡𝑖𝑠𝑜 ℎ𝑖𝑠𝑜
𝑑𝑖𝑠𝑜
𝛿𝑖𝑠𝑜 = (2-14)
𝑡𝑖𝑠𝑜

6
 Figure 2-2 Isogrid Parameters.
Using the equivalent Young’s Modulus E* and equivalent thickness t*,
theoretical critical buckling load for the isogrid stiffened cylinder is calculated per the
formula (2-7).
NASA-SP8007 suggest KDF per formula (2-8) for lightly stiffened cylinders
and KDF of 0.65 for moderate or heavy stiffened cylinders based on NASA-TN-D-
5561[9].
As a initial design, an unstiffened, isotropic cylinder from Titanium Alloy is
taken with a radius of 150 mm, length of 450 mm and a shell thickness of 1 mm.
Following, six design iterations are studied by increasing the circumferential triangle
numbers with the same rib cross section of 1x2 mm as in Table 2-1.
Table 2-1 Design Iterations.

Skin Radius(mm) 150

Skin Thickness(mm) 1
Skin Length(mm) 450
Rib Height(mm) 0 2
Rib Width(mm) 0 1
Circumferential Triangle
0 12 16 24 36 48 72
Number

Table 2-2 Material Properties.

Young's Modulus(MPa) 9,60E+04

Poisson's Ratio 0,36
Density (kg/mm³) 4,62E-06

7
 Material properties of titanium alloy is taken from ANSYS Library [2] and listed
in Table 2-2.
Although the limits are not clear for a moderate or heavy stiffening definition,
KDF is taken as 0,65 for 6 iterations.
Table 2-1 Calculation Results.

Skin Radius(mm) 150

Skin
1
Thickness(mm)
Skin Length(mm) 450
Rib Height(mm) 0 2
Rib Width(mm) 0 1
Circumferential
Triangle 0 12 16 24 36 48 72
Number
Mass(kg) 1,966 2,142 2,198 2,319 2,486 2,663 3,007
t*(Eq. Thickness)
N/A 1,355 1,447 1,607 1,804 1,964 2,212
mm
E*(Eq. Youngs 7,295E 6,895E 6,325E 5,791E 5,462E 5,106E
N/A
Modulus) +04 +04 +04 +04 +04 +04
Theoretical
Critical 3,733E 5,205E 5,612E 6,352E 7,328E 8,194E 9,712E
Buckling Load +05 +05 +05 +05 +05 +05 +05
(N)
KDF 0,518 0,65 0,65 0,65 0,65 0,65 0,65
Actual Critical 1,934E 3,383E 3,648E 4,129E 4,763E 5,326E 6,313E
Buckling Load(N) +05 +05 +05 +05 +05 +05 +05
Buckling Load to
9,836E 1,579E 1,660E 1,780E 1,916E 2,000E 2,099E
Weight Ratio
+04 +05 +05 +05 +05 +05 +05
(N/kg)

Calculated equivalent thickness and Young’s Modulus, theoretical and actual

buckling load, buckling load to weight ratio are listed in Table 2-3.

8
 Figure 2-3 Critical Buckling Loads per Triangle Number.
As expected, increasing the number of triangles per circumference makes the
structure more capable to buckling. Changes in theoretical and actual critical buckling
load is seen in Figure 2-3. In addition, buckling capability to weight ratio is plotted in
Figure 2-4.

Figure 2-4 Buckling Load to Weight Ratio per Triangle Number.

9
 3. FINITE ELEMENT ANALYSIS
3.1 Introduction
In this section, a finite element model of isogrid stiffened cylinder is developed
using the software ANSYS Workbench 2020 R1. Geometrical model, meshing details,
material properties and analysis types are described in following sub-sections.
Because the cylinder thickness is too much smaller compared to the diameter
and length dimension, shell model is used for the skin [10].If the R/t ratio is above 25,
using solid elements become increasingly uneconomical considering the accuracy of
the results and computational time.
Stiffeners have a constant cross-section and the dimension of this cross-section
is too much smaller compared to the length dimension, thus they are modelled as beam
elements.
As described in section 2, one unstiffened cylinder and six different design
iterations as shown in Table 1 are studied using FEA. Six iterations differ in the
stiffener density as shown in Figure 3-1.

Figure 3-1 Design Iterations per the circumferencially triangle number.

10
3.2 Geometry
Using Geometry Modeler module available in ANSYS Workbench, a parametric
model consisting of a surface body and line bodies is developed as shown in Figure 3-
2.

Figure 3-2 Geometry Modeler.

The related parameters are the skin outer diameter, the skin length and the grid
number as shown in Figure 3-3. With the first two parameters creates the top surface
of the cylinder skin. Grid number parameter defines the number of the equilateral
triangle on the circumference. Therefore, it changes the density of the stiffeners. The
more the grid number is, the more the isogrid structure is stiffened. However, stiffeners
adds also weight to the structure.

Figure 3-3 Geometry Model Parameters.

11
3.3 Material
Titanium alloys are suitable materials for casings of aero-engine cold sections
modules as fan and compressor due to their low weight to stiffness ratio [11]. Due to
this fact, titanium alloy is chosen as material to investigate in this study.
ANSYS Workbench material library provides physical and material properties
as shown in Table 3-1.

Table 3-1 Material Properties.

Young's Modulus(MPa) 9,60E+04

Poisson's Ratio 0,36
Density (kg/mm³) 4,62E-06
Bulk Modulus(MPa) 1,1429E+05
Shear Modulus(MPa) 35294

In order to take the material non-linearity’s, bilinear isotropic hardening curve

of titanium alloy is also provided by ANSYS Material library as shown in Figure 3-4.

Figure 3-4 Titanium Alloy Bilinear Isotropic Hardening Curve.

12
3.4 Loading and Boundary Conditions

Figure 3-5 Boundary Conditions.

A remote unit force of 1 N is applied on the edge shown as A in Figure 3-5 in
negative Z- direction. The rotations of nodes on the same edge are also fixed.
The edge shown as B is fixed in Z-direction and free in radial direction. Finite
elements are SHELL281 and BEAM189 for surface body and line bodies respectively.

Figure 3-6 Finite Element Types.

13
 SHELL181 is suitable element with four nodes with six degrees of freedom at
each node: translations in the x, y, and z directions, and rotations about the x, y, and z-
axes for analyzing thin to moderately thick structures [12].
BEAM188 is proper element with two nodes, which have six degrees of freedom
at each node: translations in the x, y, and z directions, and rotations about the x, y, and
z-axes. It is suitable for analyzing slender to moderately stubby/thick beam structures
[12].

3.5 Meshing
FEA method is based on the concept of discretization of the interested system in
smaller simpler elements. This process is called as meshing [13]. Mesh density that is
determined by the element size is very critical for the accuracy of the analysis results.
Decreasing the mesh size may result in results that are more accurate; however, it
means also increase in element and node numbers. The more the element number is,
the more the computing time is. Optimum point should be determined between the
result accuracy and solution efficiency. A mesh sensitivity analysis is performed for
the design option with circumferential triangle number of 36. The resulted eigenvalue
buckling load, the created node number and their changes for each iteration is listed in
Table 5.
Table 3-2 Mesh Sensitivity Results.

Critical
Circumferential Mesh Change in
Eigenvalue Node Node Number
Triangle Element Buckling
Buckling Number Increase
Number Size(mm) Load
Load(N)
15 6,93E+05 REF 17064 REF
12 6,808E+05 -1,80% 17640 3,38%
9 6,811E+05 0,04% 30780 74,49%
36
8 6,746E+05 -0,95% 35620 15,72%
7 6,741E+05 -0,07% 58754 64,95%
5 6,729E+05 -0,18% 79325 35,01%

14
 Figure 3-7 Buckling Load-Node Number-Mesh Size Plot.
Element size of 8 mm is chosen as the accuracy-efficiency point where the
critical load result is 6,746E5 N and the node number is 35620 by viewing Figure 3-7.

3.6 Eigenvalue Buckling Analysis

The eigenvalue buckling analysis is performed using the static structural analysis
as pre-stressed model for isogrid stiffened model. Project schematic is seen in Figure
3-8.

Figure 3-8 Project Schematic for Eigenvalue Buckling Analysis.

Per the solution of the model, the first critical eigenvalue buckling load of the
isogrid cylinder with 36 triangles in the circumference is 6,746 E+5 N and the mode
shape which has 16x deformation scale is seen in Figure 3-9.

15
 Figure 3-9 Buckling Mode Shape for the option w/36 circumferentially triangles.

3.7 Non-Linear Buckling Analysis

Using the same geometry with the same mesh structure, a non-linear static
analysis is performed to take the geometrical and material nonlinearities and
geometrical imperfections into account. For geometrical nonlinearity effect, large
deflection is enabled in the ANSYS solver control options. (Figure 3-10)

Figure 3-10 Ansys Solver Controls.

16
 By linear elastic analysis, stiffness matrix is kept same. However in reality,
stiffness of the structure changes due to change in the shape of the structure [14].
In order to take the material non-linearities, bilinear isotropic hardening curve of
titanium alloy is also provided by ANSYS Material library as shown in Figure 3-4.
Titanium alloy has a yield strength of 930 MPa. It has a strain of 0.00969 at yield
stress 930 MPa. Up to this point, material behaves linear elastic. After that stress level
of 930 MPa, plasticity begins until the strain limit of 0.0484 at the stress level of 1010
MPa.
The critical buckling load of 6,746 E+5 N determined by the eigenvalue buckling
analysis is applied in Z-direction by pre-defined substeps in Analysis settings as shown
in Figure 3-11.

Figure 3-11 Ansys Step Controls.

Geometry used in the FEA model is perfect, however in reality; there are
imperfections due to manufacturing tolerances. To simulate this condition, the
buckling mode shape and deformations are transferred with a scale of 0,025 to the
nonlinear static analysis as shown in Figure 3-12.
Maximum total deformation for the first buckling mode is 1,02 mm as seen in
Figure 18. Considering inward and outward deformations, total relative deflection is 2
mm. So, scale factor of 0,025 means 0.05 mm profile tolerance, which is a reasonable
machining tolerance for a diameter 300 mm in aviation industry.

Figure 3-12 Project Schematic for Non-Linear Buckling Analysis.

17
 At the load level where the structure becomes unstable, solution will be
unconverged and the time step at that point is the knock down factor (KDF).

Figure 3-13 KDF for Actual Buckling Load.

As seen in Figure 3-13, KDF due to geometrical and material non-linearities and
geometrical imperfections is found as 0,895. This means structure buckles under 6,038
E+5 N compression load which is the 0,895 times of the theoretical critical buckling
load of 6,746 E+5 N.
Before buckling, maximum equivalent von-Mises stress is 874,5 MPa at the
upper edge and stress level is around 500 MPa overall per Figure 3-14. Both are well
below the yield stress of 930 MPa. Thus, cylinder buckles in elastic region. Therefore,
failure mechanism is buckling.

Figure 3-14 Stress Levels before Buckling Failure.

18
 Total deformation before buckling is maximum 2,8 mm at the upper edge of the
cylinder. As seen in Figure 3-15, total deformation shape before buckling failure
matches well with the linear buckling mode shape.

Figure 3-15 Non-linear Deformation – Buckling Mode Shape Comparison for

isogrid option w/36 circumferentially triangle number.

Unstiffened cylinder has a theoretical critical buckling load of 3,675 E+5 N and
KDF of 0,881. Thus, the actual critical buckling load is 3,238 E+5 N and total
deformation shape before buckling point matches well with the linear buckling mode
shape as seen in Figure 3-16.
The isogrid-stiffened cylinder is 26 % heavier than the unstiffened cylinder,
however it has 87 % higher actual critical buckling load.

19
 Figure 3-16 Non-linear Deformation – Buckling Mode Shape Comparison for
Unstiffened Cylinder.

The FEA results for the other design iterations with different circumferential
triangle number are listed in Table 3-3.

Table 3-3 FEA Results.

Skin Radius(mm) 150

Skin Thickness(mm) 1
Skin Length(mm) 450
Rib Height(mm) 0 2
Rib Width(mm) 0 1
Circumferential Triangle
0 12 16 24 36 48 72
Number
Mass(kg) 1,966 2,142 2,198 2,319 2,486 2,663 3,007
Linear Critical Buckling 3,675 3,962 4,335 5,497 6,746 7,665 9,068
Load (N) E+05 E+05 E+05 E+05 E+05 E+05 E+05
KDF for geometrical and
material non-linearity
0,881 0,808 0,875 0,896 0,895 0,896 0,923
w/geometrical
imperfections
Nonlinear Actual Critical 3,238 3,201 3,793 4,925 6,038 6,868 8,370
Buckling Load E+05 E+05 E+05 E+05 E+05 E+05 E+05
Actual Buckling Load to 1,647 1,495 1,726 2,124 2,429 2,579 2,783
Weight Ratio (N/kg) E+05 E+05 E+05 E+05 E+05 E+05 E+05

20
 Linear and actual critical buckling loads are graphed per circumferential triangle
number in Figure 3-17. As expected, increasing triangle number lead to a higher
buckling load.

Figure 3-17 Critical Buckling Loads per Triangle Number.

Figure 3-18 Buckling Load to Weight Ratio per Triangle Number.

21
 4. CONCLUSIONS AND DISCUSSIONS
4.1 Comparison of FEA and Hand Calculation
This section compares the FEA results from shell-beam model with the hand
calculation results per NASA design manual, concerning theoretical buckling load.
Figure 4-1 shows the FEA and calculation results of unstiffened cylinder and six
isogrid-stiffened cylinder with different triangle numbers. Regarding unstiffened
cylinder, FEA and hand calculations result in 3,675 E+5 N and 3,733 E+5 respectively.
The difference is only 1,6 %and it is acceptable for early design phase activities.

Figure 4-1 Theoretical Buckling Loads per FEA and NASA Manual.

As seen in the plot (Figure 4-1), hand calculation overestimate the theoretical buckling
load for isogrid designs. Difference between FEA and calculations reduces with
increasing number of triangles in the circumference. If the structure is lightly stiffened,
difference is up to 31%. However, it decreases to 7-8% if the cylinder is heavily
stiffened as minimum 36 triangle on the circumference.

22
 Figure 4-2 FEA-Hand Calculation Difference per Stiffener Density.
Above results are for the specific R/t ratio of three. Similar assessment is
performed for different R/t ratios from 50 to 750. Cylinders are moderate/heavily
stiffened as 36 triangle on the circumference. Figure 4-2 and Figure 4-3 shows the
results and difference between them again for FEA and hand calculation.

Figure 4-3 Theoretical Buckling Loads per FEA&NASA Manual for different
R/t ratios.
As seen in Figure 4-4, hand calculation is not suitable for R/t below 100.
Difference is 28% and 15% for R/t ratio of 50 and 15 respectively. Hand calculation
approach does not work properly if the slenderness is low.

23
 Figure 4-4 FEA-Hand Calculation Difference per R/t.
To sum up, hand calculation and FEA match very well if the cylinder is not
stiffened. For isogrid-stiffened cylinders, hand calculation approach overestimate the
buckling capability of the structure. The overestimate level is around 10% for slender,
moderate/heavily stiffened cylinders, which have minimum 36 stiffener triangle
number on the circumference and minimum slenderness level with R/t ratio of 100 to
800. If the slenderness is below R/t ratio of 100 or the structure is just lightly stiffened,
using hand calculation might mislead the design.

4.2 Isogrid Effect on Buckling Capacity and Weight

Design option with circumferential triangle number 36 has a weight of 2,486 kg.
Considering the same radius of 150 mm and the same weight and same material, 2,486
kg corresponds to an unstiffened cylinder with shell thickness of 1,269 mm. Nonlinear
buckling analysis of this shell results in a critical buckling value of 5,322E5 N while
the same weight isogrid stiffened cylinder has 6,038 E5 N. Isogrid design provides 13
%. improvement for the same weight as listed in Table 4-1.

24
 Table 4-1 Buckling Capacity of Unstiffened and Isogrid Stiffened Cylinder.

Unstiffened
Isogrid Stiffened
Cylinder w/same
Cyclinder
Weight
Skin Radius(mm) 150 150
Skin Thickness 1 1,269
Skin Length 450 450
Circumferential Rib
36 N/A
Number
Rib Height(mm) 2 N/A
Rib Width(mm) 1 N/A
Mass(kg) 2,486 2,486 Improvement
Linear Critical Buckling
6,746E+05 5,842E+05 15%
Load(N)
Nonlinear Critical
6,038E+05 5,322E+05 13%
Buckliing Load(N)

4.3 KDF by NASA Manual and FEA

In reality, structures buckle earlier than the expected, theoretical critical load due
the geometrical and material nonlinearities and geometrical imperfection caused by
manufacturing tolerances. In order to consider this phenomenon, theoretical buckling
load is multiplied with KDF and actual critical load is obtained. As discussed in 2.2,
NASA suggests a KDF of 0,65 for moderate/heavily stiffened cylinder. This number
is based on the test data obtained at NASA from the 1930s through the 1960s [15].
Considering the high accuracy of today’s modern manufacturing processes and
technology; this KDF looks conservative. Less conservative factors will help to gain
weight saving for aerospace industry.
On the other hand, non-linear FEA buckling analysis result in an average KDF
of around 0,85 depending on the stiffening level as discussed in 3.7.
Due to this fact, hand calculation ends with a lower actual critical buckling load
compared to FEA result, although it overestimate the theoretical, critical load by 7-16
% for moderate/heavily stiffened ones.

25
Figure 4-5 Theoretical and Actual Buckling Loads per FEA and Hand Calculations.
4.4 KDF and Geometrical Imperfection Amplitude
There are always initial imperfections due to nature of manufacturing processes
and this fact reduces the buckling strength of the structure. Obviously, imperfections
are random and could not predicted until we measure the actual, manufactured part.
Real 3D geometry of the part can be modelled by modern scanning technology and
can be transferred to FEA software. However, this methodology is piece part unique
and very costly process considering mass production. Instead of this technique, a pre-
deformed 3d model of the structure can be built up for initial geometry with
imperfections. As discussed xxx, first buckling mode shape from eigenvalue buckling
analysis is suitable for this technique. A scale factor is used to regulate the amplitude
of the imperfections. To remember, the structure that is studied in 3 is 300 mm
diameter cylinder with a shell thickness of 1 mm and it has 36 isogrid triangles on the
circumference with rib cross section 2x1 mm. If one checks the first eigenvalue
buckling mode shape, 300 mm diameter varies between 302 and 298,2 mm as seen in
Figure 4-6. Total deviation is 1,9 mm for 300 mm if the scale factor is assumed as 1.

26
 Figure 4-6 Radial Deflection of First Eigenvalue Buckling Mode Shape.
Scale factors and the corresponding manufacturing tolerances are listed in Table
4-2 and Table 4-3.
Table 4-2 KDF’s per Manufacturing Tolerances-1.

Scale Factor 0,01 0,02 0,03 0,04 0,05 0,06 0,07

Corresponding Manufacturing
0,02 0,04 0,06 0,08 0,1 0,12 0,14
Tolerance(±)(mm)
KDF 0,96 0,91 0,88 0,86 0,84 0,84 0,81

Table 4-3 KDF’s per Manufacturing Tolerances-2.

Scale Factor 0,08 0,09 0,1 0,2 0,5 0,75 1

Corresponding Manufacturing
0,16 0,18 0,2 0,4 1 1,5 2
Tolerance(±)(mm)
KDF 0,79 0,78 0,77 0,67 0,5 0,43 0,38

As expected, the more the structure have manufacturing tolerance, the lower
KDF is for actual critical buckling load. Figure 4-7 shows the change of KFD per
manufacturing tolerance. In section, ±0,05 mm tolerance is assumed for the isogrid
structure with a diameter of 300 mm. Considering the modern manufacturing
techniques used in aero industry, this tolerance level looks quite achievable.

27
 Figure 4-7 KDF’s per Manufacturing Tolerances.
This correlation (Figure 4-7) shows the importance of the manufacturing
tolerances for aero/aerospace industries considering thin walled cylinder structures.
Buckling capacity of those type structures is very sensitive to geometrical
imperfections. For instance, manufacturing tolerance of ± 0,1 mm results in a KDF of
0,84, while KDF is 7 % lower for ± 0,2 mm tolerance level. Just 0,1 mm change in
manufacturing accuracy reduces the buckling capacity by 7 %.

4.5 Conclusion
Regarding thin walled cylinders, results of hand calculation per Donnell’s shell
theory and FEA method are very close as 1,6 % for critical buckling load. However;
hand calculation per NASA’s manual overestimates theoretical critical buckling load
by 7-8 % compared to FEA results for moderately isogrid stiffened cylinders with R/t
ratio of 150. The difference increases up to 31 % if the structure is lightly stiffened. In
addition, if the R/t ratio is below 100, 7-8 % difference increases up to 28 %. Thus,
hand calculation is not proper for R/t ratios below 100.
One of the most important design criterions in aerospace technologies is the
weight. Using isogrid stiffeners increases buckling capability of the structure. This
study shows that isogrid-stiffened cylinder has 13% more buckling capability than the
monocoque cylinder with the same weight.
Manufacturing tolerance is very critical factor to determine the buckling capacity
of the structure, because it effects KDF, which allows calculating the critical buckling
load from theoretical buckling load. This study shows that the change in manufacturing
tolerance from ± 0,1 to ± 0,2 mm decreases the buckling capacity by 7 %.

28
 REFERENCES
[1] Steven M. Huybrechts, Grid Stiffened Structures: A Survey of Fabrication,
Analysis and Design Methods, 1999

[2] Web 1, (2021), https://www.satorsrl.it/grid-structures/, (Erişim Tarihi:

20/06/2021).

[3] Web 2, (2021), https://en.wikipedia.org/wiki/Eurojet_EJ200, (Erişim Tarihi:

20/06/2021).

[4] Hilburger Mark W., Test and Analysis of Buckling-Critical Stiffened Metallic
Launch Vehicle Cylinders,

[5] Hafiz Pouria and Baru A. Maikanti, An Investigation of Buckling Phenomenon in

Steel Elements, 2018

[6] Web 3, (2021), https://www.dlr.de/fa/en/desktopdefault.aspx/tabid-

10589/18840_read-43672/, (Erişim Tarihi: 20/06/2021).

[7] NASA/SP-8007, Buckling of Thin-Walled Circular Cylinders, 2019 Revision

[8] NASA CR-124075, Isogrid Design Handbook, 1973

[9] NASA TN D-5561, Buckling of Stiffened Cylinders in Axial Compression and

Bending- A Review of Test Data, 1969

[10] Sadowski Adam J. and Rotter J. Michael, Solid or shell finite elements to model
thick cylindrical tubes and shells under global bending, 2013

[11] Okura Takehiro, Materials for Aircraft Engines – ASEN 5063 Aircraft Propulsion
Final Report, 2015

[12] ANSYS Workbench Help Document

[13] Hassan Ahmed and Kurgan Naci, Modeling and Buckling Analysis of
Rectangular Plates in ANSYS, 2019

[14] Web 4, (2021), https://www.padtinc.com/blog/to-use-large-deflection-or-not-

that-is-the-question/, (Erişim Tarihi: 20/06/2021).

[15] Lovejoy Andrew E. and Hilburger Mark W., SBKF Modeling and Analysis Plan:
Buckling Analysis of Compression-Loaded Orthogrid and Isogrid Cylinders, 2013

29
 BIOGRAPHY
Orhan Kemal Can was born in Demirci/Manisa.
He worked as manufacturing and mechanical design engineer at different
aerospace and defense industry companies.

30