Sei sulla pagina 1di 120

Structural analysis of columns

with initial imperfections

Tjörvi Björnsson

Faculty
Faculty of
of Civil
Civil and
and Environmental
Environmental Engineering
Engineering
University
University of
of Iceland
Iceland
2017
2017
STRUCTURAL ANALYSIS OF COLUMNS WITH
INITIAL IMPERFECTIONS

Tjörvi Björnsson

30 ects. ECTS thesis submitted in partial fulllment of a


Magister Scientiarum degree in Structural Engineering

Advisors

Baldvin Einarsson

Egill Þorsteins

Faculty Representative

Guðmundur Valur Guðmundsson

Faculty of Civil and Environmental Engineering

School of Engineering and Natural Sciences

University of Iceland

Reykjavik, June 2017


Structural analysis of columns with initial imperfections

30 ects. ECTS thesis submitted in partial fulllment of a M.Sc. degree in Structural


Engineering

Copyright c 2017 Tjörvi Björnsson


All rights reserved

Faculty of Civil and Environmental Engineering


School of Engineering and Natural Sciences
University of Iceland
Sæmundargata 2
101, Reykjavik, Reykjavik
Iceland

Telephone: 525 4000

Bibliographic information:
Tjörvi Björnsson, 2017, Structural analysis of columns with initial imperfections, M.Sc. the-
sis, Faculty of Civil and Environmental Engineering, University of Iceland.

Printing: Háskólaprent, Fálkagata 2, 107 Reykjavík


Reykjavik, Iceland, June 2017
Abstract
This thesis presents a theoretical study of varied calculation models used to estimate
the stability of columns under combined axial load and bending moment. The
purpose of comparing these models is to study dierent methods of applying initial
imperfections to columns. Eurocode 3 is used as a basis for all models.

Presented rst are the reasons for why it is necessary to include initial imperfections
in the analysis of columns. Secondly the work provides mathematical background
to the buckling curves and equations presented in Eurocode 3. Thirdly, a series of
worked examples to show the application of the methods with the software chosen
and that it is compatible with the methods. Lastly the application of the same
methods to two types of guyed transmission towers. This is done to inspect the
eect of initial imperfections on the stability of the tower legs.

The work highlights the dierence between the methods. Using a linear analysis
method is simpler than dening the initial imperfections and using non-linear anal-
ysis. Modeling imperfections after the critical buckling shape is logical while model-
ing the imperfections with equivalent forces requires more assumptions, though the
results are similar.

The Statnett tower, with hundreds of elements, reacted well in analysis and gave
credible results. The Landsnet tower, with a much simpler geometry, was more
complex to analyze and interpret. Tower complexity is not as obvious as it seems.

v
Útdráttur
Verkefnið fjallar um fræðilega greiningu á mismunandi reiknilíkönum, sem notuð eru
til að meta stöðugleika súlna sem á verkar áslægur kraftur og beygjuvægi samtímis.
Tilgangur greiningarinnar er að bera saman mismunandi aðferðir sem notaðar eru
til að taka tillit til upphafsútbeygju í súlum.

Í uppha eru tilgreint hvers vegna er nauðsynlegt að taka tillit til upphafsútbeygju í
greiningu á súlum. Næst er farið í stærðfræðilegan bakgrunn kiknunarkúrfa og þeirra
jafna úr Eurocode 3 sem notaðar eru. Í þriðja lagi eru reiknuð sýnidæmi til að sýna
hvernig aðferðunum er beitt á einfaldari virki og að hugbúnaðurinn sem er valinn sé
hentugur. Að lokum er aðferðunum beitt á tvær tegundir stagaðra háspennumastra.
Það er gert til að skoða áhrif upphafsútbeygju á stöðugleika leggja mastranna.

Verkefnið gaf góða mynd af þeim mun sem er á aðferðunum. Línuleg greining er
einfaldari leið en að notast við upphafsútbeygjur og ólínulega greiningu. Að forma
upphafsútbeygjur eftir krítiska kiknunarforminu er virðist rökrétt, að forma þær með
jafngildum kröftum krefst meiri ágiskana.

Greining á Statnett mastrinu, sem inniheldur hundruðir stanga, gekk vel og skilaði
trúverðugum niðurstöðum. Landsnet mastrið, sem er mun einfaldara virki, var ók-
nara í greiningu og túlkun á niðurstöðum. Flækjustig mastra er ekki eins augljóst
og það virðist vera.

vi
Contents
List of Figures xi

List of Tables xv

Symbols and notations xix

Acknowledgments xxi

1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Theoretical background 5
2.1 Denitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Dierent analysis methods . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Column stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Elastic buckling (Euler buckling theory) . . . . . . . . . . . . . . . . 10
2.5 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.6 Column imperfections . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.6.1 Initial out-of-straightness . . . . . . . . . . . . . . . . . . . . . 14
2.6.2 Load eccentricity . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6.3 Residual stresses . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.6.4 Material eects . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.7 Structural analysis in EN1993-1-1:2005 . . . . . . . . . . . . . . . . . 18
2.8 Buckling resistance of members (EN1993-1-1:2005 Section 6.3) . . . . 20
2.8.1 Uniform members in compression . . . . . . . . . . . . . . . . 21
2.9 Uniform members in bending and axial compression (EN1993-1-1:2005
Section 6.3.3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.10 Sway and bow imperfections (EN1993-1-1:2005 Section 5.3.2(3)) . . . 24
2.11 EUGLI imperfection method (EN1993-1-1:2005 Section 5.3.2(11)) . . 26
2.11.1 EUGLI method history . . . . . . . . . . . . . . . . . . . . . . 27
2.11.2 EUGLI method formulation . . . . . . . . . . . . . . . . . . . 29
2.12 SAP2000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.12.1 Linear load case analysis . . . . . . . . . . . . . . . . . . . . . 33
2.12.2 Buckling analysis . . . . . . . . . . . . . . . . . . . . . . . . . 33

vii
Contents

2.12.3 Nonlinear static analysis . . . . . . . . . . . . . . . . . . . . . 34


2.12.4 Initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.12.5 Geometric non-linearity . . . . . . . . . . . . . . . . . . . . . 34
2.12.6 Guyed towers . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.12.7 Stress Output . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.12.8 Procedures for various methods of analysis . . . . . . . . . . . 36

3 Worked examples 39
3.1 Uniform column, pinned supports . . . . . . . . . . . . . . . . . . . . 39
3.2 Uniform beam, xed and pinned supports . . . . . . . . . . . . . . . 42
3.3 Simple frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3.1 Linear analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3.2 Non-linear analysis with sway and bow imperfection equiva-
lent forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3.3 Non-linear analysis with EUGLI imperfections . . . . . . . . . 47
3.3.4 Method comparison . . . . . . . . . . . . . . . . . . . . . . . . 50

4 Statnett tower in Norway 51


4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2 Statnett method of analysis . . . . . . . . . . . . . . . . . . . . . . . 52
4.3 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4 Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.5 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.6 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.6.1 Buckling analysis . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.6.2 Linear analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.6.3 Non-linear analysis with sway and bow imperfection equiva-
lent forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.6.4 Non-linear analysis with EUGLI imperfection . . . . . . . . . 66
4.6.5 Resistance according to Statnett method . . . . . . . . . . . . 67
4.7 Method comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

5 Guyed tubular tower 71


5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.2 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.3 Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.4 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.5 Buckling of a intermediately restrained leg . . . . . . . . . . . . . . . 79
5.6 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.6.1 Buckling analysis . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.6.2 Bracing stiness . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.6.3 Linear analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.6.4 Non-linear analysis with sway and bow imperfections . . . . . 86
5.6.5 Non-linear analysis with EUGLI imperfection . . . . . . . . . 89

viii
Contents

5.7 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.8 Tower test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

6 Concluding remarks 95

Bibliography 97

ix
List of Figures

1.1 a) Statnett tower, b) Landsnet tower . . . . . . . . . . . . . . . . . . 3

2.1 Stress-Strain relationship (Wikipedia, 2017a) . . . . . . . . . . . . . . 5

2.2 Stress-strain diagram (Wikipedia, 2017b) . . . . . . . . . . . . . . . . 6

2.3 Load-deformation curve for dierent analysis types (Leroy Gardner


& David A. Nethercot, 2011) . . . . . . . . . . . . . . . . . . . . . . . 8

2.4 Column classication (Ed Akin, 2009) . . . . . . . . . . . . . . . . . . 10

2.5 Simply supported beam (Jørgen Amdahl, 2014) . . . . . . . . . . . . 11

2.6 Deection curves showing the eective length Le for a column xed
at the base and free at the top (Vladislav F. Demenko, 2015) . . . . . 13

2.7 Recommended eective lengths for compression members (Leroy Gard-


ner & David A. Nethercot, 2011) . . . . . . . . . . . . . . . . . . . . 13

2.8 Simple supported column with initial deection (Rangachari Narayanan


et al., 2012) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.9 Column with an initial deection (Primoº Moºe, 2015) . . . . . . . . 15

2.10 Strength curve for eccentrically loaded columns (Rangachari Narayanan


et al., 2012) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.11 Typical distribution of residual stresses (Primoº Moºe, 2015) . . . . . 16

2.12 Stress-strain relationship with strain hardening (Rangachari Narayanan


et al., 2012) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

xi
LIST OF FIGURES

2.13 a) Lack of a clearly dened yield, b) Lack of a clearly dened yield


with strain hardening (Rangachari Narayanan et al., 2012) . . . . . . 18

2.14 Equivalent sway and bow imperfections (CEN, 2007) (d) System, load
and displacement (e) Initial local bow and buckling length for exu-
ral buckling (f ) Second order moment including moment from sway
imperfection (g) Initial local bow and buckling length for lateral-
torsional buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

2.15 Replacement of initial imperfections by equivalent horizontal forces


(CEN, 2005) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.16 Apollo bridge in Bratislava, Slovakia (Eurocodes, 2015) . . . . . . . . 28

2.17 Pentele bridge in Dunaújváros, Hungary (Magdi51, 2012) . . . . . . . 28

2.18 An illustrative example. Note the hidden location of e0d,m (Marian


Dallemule, 2015) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.1 Example 1 - Analysis of a simply supported IPE300 column, a) Col-


II
umn setup, b) ηcr (x), c) Mη,cr (x), d) NEd (x) . . . . . . . . . . . . . . 40

3.2 Example 2 - Beam from (Ivan BalẠ& Yvona Koleková, 2012) . . . . 43

3.3 Example 3 - Analyzed frame, a) scheme and loading, b) sway imper-


fection, c) and d) local bow imperfections . . . . . . . . . . . . . . . . 44

3.4 Example 3 - a) Shape of the rst buckling mode, b) Bending moments


necessary to bend the structure into the shape of the buckling mode . 48

3.5 Example 3 - Eects due to external loading using non-linear analysis:


II
a) Deformation, b) Bending moments MEd . . . . . . . . . . . . . . . 49

4.1 Stanett tower - Suspension tower, type BM (courtesy of Statnett) . . 52

4.2 Statnett tower - Geometry of tower (courtesy of Statnett) . . . . . . . 54

4.3 Statnett tower - Geometry of bridge (courtesy of Statnett) . . . . . . 56

4.4 Statnett tower - Load points sketch . . . . . . . . . . . . . . . . . . . 57

4.5 Statnett tower - SAP2000 3D model . . . . . . . . . . . . . . . . . . . 59

xii
LIST OF FIGURES

4.6 Statnett tower - Unbalanced ice critical buckling load . . . . . . . . . 61

4.7 Statnett tower - Wind and ice critical buckling shape . . . . . . . . . 61

4.8 Statnett tower - Buckling shapes, from left, Unbalanced ice mode 2,
Unbalanced ice mode 3, Wind and ice mode 2, Wind and ice mode 3 62

4.9 Statnett tower - Sway and bow imperfection equivalent forces. Upper
row: Load case 1-Unbalanced ice 1) Sway forces, 2) +Bow forces
Lower row: Load case 2-Wind and ice 1) Sway forces, 2) +Bow forces 64

4.10 Statnett tower - Comparison of methods . . . . . . . . . . . . . . . . 68

4.11 Statnett tower - Location of the critical cross section by method . . . 69

5.1 Landsnet tower - Guyed tubular M-tower (courtesy of Landsnet) . . . 72

5.2 Landsnet tower - Outline drawing of the suspension tower (courtesy


of Landsnet) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

5.3 Landsnet tower - Location of foundations (courtesy of Landsnet) . . . 74

5.4 Landsnet tower - Primary/secondary structure . . . . . . . . . . . . . 75

5.5 Landsnet tower - Location of load point attachments . . . . . . . . . 77

5.6 Landsnet tower - SAP2000 3D model . . . . . . . . . . . . . . . . . . 79

5.7 Column with intermediate elastic restraint . . . . . . . . . . . . . . . 80

5.8 Inuence of elastic spring stiness on buckling capacity . . . . . . . . 81

5.9 Three factors leading to section loading in the right leg . . . . . . . . 82

5.10 Landsnet tower - Buckling modes, in top row modes 1 and 2, in


bottom row 3 and 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

5.11 Landsnet tower - Buckling modes, in top row modes 5 and 6, in


bottom row 7 and 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

5.12 Landsnet tower - Secondary guy elastic modulus eect on the critical
buckling mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

xiii
LIST OF FIGURES

5.13 Landsnet tower - Linear analysis of load case 8a, a) Axial force NEd ,
b) Bending moment My,Ed . . . . . . . . . . . . . . . . . . . . . . . . 85

5.14 Landsnet tower - Modeling of sway and bow imperfection on the tower
leg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

5.15 Landsnet tower - Sway and bow imperfection form . . . . . . . . . . . 87

5.16 Landsnet tower - Maximum stress for load case 8, a) Sway, b) Sway+Bow,
c) Sway-Bow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

5.17 Landsnet tower - Elastic critical buckling mode shape number 1 for
load case 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

II
5.18 Landsnet tower - Load case 8, a) Axial force NEd , b) Bending moment
II
Mη,init . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

5.19 Landsnet tower - Usage comparison of dierent methods . . . . . . . 91

5.20 Landsnet tower - Buckling of the full-scale tower test (courtesy of


Landsnet) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

xiv
List of Tables

2.1 Dierent structural analysis methods . . . . . . . . . . . . . . . . . . 8

2.2 Buckling curves (CEN, 2005) . . . . . . . . . . . . . . . . . . . . . . . 22

2.3 Imperfection factors (CEN, 2005) . . . . . . . . . . . . . . . . . . . . 23

2.4 Values for NRk = fy Ai , Mi,Rk = fy Wi and ∆Mi,Ed (CEN, 2005) . . . . 24

2.5 Design values of initial local bow imperfection e0 /L (CEN, 2005) . . . 26

2.6 SAP2000 - Linear analysis . . . . . . . . . . . . . . . . . . . . . . . . 37

2.7 SAP2000 - Force calculation . . . . . . . . . . . . . . . . . . . . . . . 37

2.8 SAP2000 - Non-linear analysis with sway and bow imperfections . . . 37

2.9 SAP2000 - ηinit,m amplication calculation . . . . . . . . . . . . . . . 38

2.10 SAP2000 - Non-linear analysis with EUGLI imperfection . . . . . . . 38

3.1 Example 1 - Steel and column properties . . . . . . . . . . . . . . . . 39

3.2 Example 1 - Linear buckling analysis . . . . . . . . . . . . . . . . . . 40

3.3 Example 1 - Non-linear analysis with bow imperfection equivalent


horizontal forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.4 Example 1 - Non-linear analysis with EUGLI imperfection . . . . . . 41

3.5 Example 2 - Prole properties . . . . . . . . . . . . . . . . . . . . . . 42

3.6 Example 2 - SAP2000 analysis and calculation . . . . . . . . . . . . . 43

xv
LIST OF TABLES

3.7 Example 2 - Result comparison . . . . . . . . . . . . . . . . . . . . . 43

3.8 Example 3 - Section properties . . . . . . . . . . . . . . . . . . . . . . 44

3.9 Example 3 - Frame loading . . . . . . . . . . . . . . . . . . . . . . . . 45

3.10 Example 3 - Linear analysis . . . . . . . . . . . . . . . . . . . . . . . 46

3.11 Example 3 - Non-linear analysis, forces and moments due to loading


and sway and bow imperfections . . . . . . . . . . . . . . . . . . . . . 47

3.12 Example 3 - Buckling analysis node displacement . . . . . . . . . . . 48

3.13 Example 3 - Forces and moments from buckling analysis and non-
linear analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.14 Example 3 - ηinit,m imperfection calculation . . . . . . . . . . . . . . . 49

3.15 Example 3 - Non-linear analysis of deformed frame with ηinit,m im-


perfection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3.16 Example 3 - Column utilization according to sections 3.3.1-3.3.3 . . . 50

3.17 Example 3 - Utilization of frame columns analyzed, from (Eugen


Chladný & Magdalena ’tujberová, 2013) . . . . . . . . . . . . . . . . 50

4.1 Statnett tower - Buckling length factors for main body members . . . 53

4.2 Statnett tower - Suspension towers, buckling length options . . . . . . 53

4.3 Statnett tower - Prole specication . . . . . . . . . . . . . . . . . . . 55

4.4 Statnett tower - Guy specication . . . . . . . . . . . . . . . . . . . . 55

4.5 Statnett tower - Load cases . . . . . . . . . . . . . . . . . . . . . . . 57

4.6 Statnett tower - Load case buckling factors . . . . . . . . . . . . . . . 60

4.7 Statnett tower - Usage from linear analysis . . . . . . . . . . . . . . . 63

4.8 Statnett tower - Sway and bow imperfection parameters . . . . . . . 63

4.9 Statnett tower - Factors to calculate sway and bow equivalent forces . 65

xvi
LIST OF TABLES

4.10 Statnett tower - Axial forces necessary to calculate sway and bow
equivalent forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

4.11 Statnett tower - Usage from non-linear analysis with sway and bow
imperfection equivalent forces . . . . . . . . . . . . . . . . . . . . . . 65

4.12 Statnett tower - ηinit,m amplication factor calculation . . . . . . . . 66

4.13 Statnett tower - Usage from non-linear analysis with EUGLI imper-
fection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.14 Statnett tower - Guy eld buckling resistances . . . . . . . . . . . . . 67

4.15 Statnett tower - Usage according to Statnett method of analysis . . . 67

4.16 Statnett tower - Comparison of dierent analysis methods . . . . . . 68

5.1 Landsnet tower - Prole specication . . . . . . . . . . . . . . . . . . 73

5.2 Landsnet tower - Guy specication . . . . . . . . . . . . . . . . . . . 75

5.3 Landsnet tower - Load cases . . . . . . . . . . . . . . . . . . . . . . . 76

5.4 Landsnet tower - Load case 8 stepped increase . . . . . . . . . . . . . 77

5.5 Landsnet tower - Load case buckling factors . . . . . . . . . . . . . . 83

5.6 Landsnet tower - Linear analysis, critical section usage . . . . . . . . 86

5.7 Landsnet tower - Sway and bow imperfection parameters . . . . . . . 86

5.8 Landsnet tower - Non-linear analysis with sway and bow imperfec-
tions, critical section usage . . . . . . . . . . . . . . . . . . . . . . . . 88

5.9 Landsnet tower - ηinit,m amplication factor calculation . . . . . . . . 89

5.10 Landsnet tower - Non-linear analysis with EUGLI imperfection, crit-


ical section usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

5.11 Landsnet tower - Critical section usage from dierent analysis methods 91

xvii
Symbols and notations
Latin letters
A cross-sectional area
Aef f eective cross-sectional area
b width of a cross-section
e0 initial bow imperfection
e0,d maximum amplitude of a member imperfection
E Young's modulus of elasticity
fy yield stress
i radius of gyration
iy radius of gyration along major axis
iz radius of gyration along minor axis
Iy , Iz second moment of area about the y, z axis
k eective length factor
L member length
Lcr critical length
M applied moment
N axial force
P point load
q distributed load
x, y cross-section principal axes
z distance along a member

Greek letters
α imperfection factor for a buckling curve
αcr buckling factor
αh reduction factor for height h applicable to columns
αm number of columns in a row
χ reduction factor for relevant buckling curve
γM i particular partial factor
λ column modied slenderness
λ non dimensional slenderness
ηinit amplitude of elastic critical buckling mode
ηcr shape of elastic critical buckling mode
φ value to determine the reduction factor χ / global initial sway imperfection
σ stress

xix
LIST OF TABLES

Suxes
cr critical
Ed design value
el elastic
pl plastic
Rd design resistance
Rk characteristic resistance
u ultimate strength
x x-x axis
y y-y axis
z z-z axis

xx
Acknowledgments
I would like to thank my thesis advisors Egill Þorsteins and Baldivn Einarsson for
their ideas, guidance and wisdom in the writing of this thesis. I would also like to
thank my boss, Steinþór Gíslason, for his patience and EFLA Consulting Engineers
for providing me with the necessary facilities and equipment to do my work.

My thanks also go to my family, who have supported me throughout my studies and


my life, and to my girlfriend Lára Halla, for her understanding and encouragement.

Tjörvi Björnsson
Reykjavík, May, 2017

xxi
1 Introduction

1.1 Background
A column is a structural member that transmits the weight of the structure above
to other structural members or foundations below through compression.

The determination of column strength has been a subject of study for a long time.
Headway was made when Euler solved the dierential equation for elastic buckling
in 18th century. Navier found out that the method proposed by Euler described
the upper limit of experimental results. High safety factors were used in design
to address that problem. In the 19th century, researchers focused their eorts on
nding out why columns never reach their theoretical buckling strength. Young
pointed out the importance of including initial deformations and load eccentricity.
Perry found a convenient method to include the initial imperfections. Dierent
design methods were developed in dierent places, all with the same objective of
developing a simple method to estimate the eects of initial imperfections on the
strength of columns. In the mid-19th century organizations were founded, both in
America and in Europe, to direct steel research. These organizations did extensive
experimentation to establish common design methods for columns.

Structural members subjected to axial compression and bending moment are known
as beam columns. In principle, all members in a frame structure are beam-columns,
with the two extremes of beams (N=0) and columns (M=0).

The need to somehow combine the eects of axial load causing direct stress and
lateral load causing bending stress was apparent early. Design guidance for struc-
tural steel members under combined axial and transverse loads has evolved from
very simple assumptions to the present ultimate-strength approaches. The changes
in design recommendations have followed the state of knowledge of the behavior of
these structural elements. Early elastic models have given way to an understanding
of inelastic behavior.

The design of transmission towers is a broad subject. They come in many shapes
and sizes and dierent designs can serve dierent roles in a power-line. The two

1
1 Introduction

main types are suspension and tensions towers. In suspension towers the conduc-
tors are simply suspended from the tower, in tension towers the conductors can be
mechanically terminated.

Guyed towers and freestanding towers are similar but their behavior is very dierent.
The presence of pre-stressed cables allows the lattice frame of a guyed tower to be
signicantly smaller in both geometry and member size. The analysis of these struc-
tures is complex due to non-linear characteristics, large deections and variations in
axial load during extreme loading.

1.2 Objectives
The main objective of this study is to compare dierent methods used to apply
initial deection in structural analysis. This is done to gain a better understanding
of the eects that initial deections have on the structural stability and column
buckling resistance.

For the analysis part of this thesis, the focus will be on two types of transmission
towers, which are based on structural models and technical drawings of real towers
that were available to the author.

• Statnett tower: A lattice tower, internally guyed, with proles made mostly
of UNP and L sections. The focus will be on one of the tower legs. It is made
of dierent sized UNP elements. Its strong axis is and braced by two force
planes and its weak axis by multiple bracing members.

• Landsnet tower: A tubular tower, outward guyed, with proles made of CHS
and RHS sections. The focus will be on the structural design of a guyed leg,
which is braced in its center.

In the analysis of both towers, the aim is to check if similar results are obtained from
dierent methods for initial deection. The question of how complex it is to model
the initial deections and how dicult is it to use the methods in everyday design is
discussed. How does the software chosen handle the analysis and what would could
be done next to gain a better understanding of the subject.

2
1.3 Methods

Figure 1.1: a) Statnett tower, b) Landsnet tower

1.3 Methods
Two dierent type of models will be used to analyze the beam-column structures
chosen. One is a linear model using buckling lengths, the other is a geometrically
non-linear model using initial deections. EN1993-1-1 (CEN, 2005) will be used as
a basis for both methods. For the linear model, the structural stability is described
in section 5.2.2(3)c) in EN1993-1-1 and for the geometrically non-linear model in
section 5.2.2(3)a).

Two methods will be used to include the eect of initial deection:

1. Section 5.3.2(3) in EN1993-1-1 (CEN, 2005) describes global initial sway im-
perfections and relative initial local bow imperfections of members for exural
buckling. In section 5.3.2(7), it says that the imperfection eects may be
replaced by as system of equivalent horizontal forces, introduced for each col-
umn.

2. Section 5.3.2(11) in EN1993-1-1 (CEN, 2005) and in EN1999-1-1 (CEN, 2007)

3
1 Introduction

describes how the shape of the elastic critical buckling mode of a structure may
be applied as a unique global and local imperfection. The structure is therefore
initially out of plumb before any load is put on it. This equivalent geometrical
imperfection is called the ηinit,m imperfection or EUGLI imperfection.

The method given in 5.3.2(11) should give almost the same result as the buckling
approximations but the method in 5.3.2(3) in not calibrated to give the same results
in all instances.

4
2 Theoretical background

2.1 Denitions
Elastic analysis
Elastic analysis is based upon the condition that the relationship between stress and
strain is linear. Therefore, the stresses in the cross section need to be lower than
the material yield strength.

Figure 2.1: Stress-Strain relationship (Wikipedia, 2017a)

Elastic deformations are reversible. If the load causing them is removed, then the
object will return to its original shape. These deformations are linear in normal
metals and can be described by Hooke's law

σ =E· (2.1)

where σ is stress, E is the elastic modulus and  is the resulting strain. Hooke's law
is only valid on the elastic range. On that range, the slope of the stress-strain curve
is the material elastic modulus, E. The elastic range ends when the stress reaches
the material yield strength and plastic deformations take over. (Lawrence Martin
& John Purkiss, 2008)

Plastic analysis
Opposite to elastic deformations, plastic deformations are irreversible. However, an
object in the plastic deformation range will rst have undergone elastic deformation,

5
2 Theoretical background

which is reversible, so the object will return part way to its original shape. What
remains is the deformation produced during yield on the plastic range.

Figure 2.2: Stress-strain diagram (Wikipedia, 2017b)

Compression in a column causes shortening. Compressive load in a structural mem-


ber causes stresses in its cross-section. With increased loading, the stresses will
increase until the material yield strength is reached. Steel structures will not col-
lapse when the material yield strength is reached. Instead, plastic hinges start to
form. Generally, the combination of three hinges, plastic or not, will lead to insta-
bility. For ductile materials, the member bends until it fails. For brittle materials,
the member brakes suddenly (Brian Kirke & Iyad Hassan Al Jamel, 2004)

Linear analysis
Linear structural analysis is based on two assumptions, material- and geometric lin-
earity. Material linearity assumes the structure is composed of linear elastic material
and geometric linearity that structural deformations are so small that the equations
of equilibrium can be expressed in the undeformed geometry of the structure.
A simple denition of linear analysis would be that it is the design of a member
(column, beam, etc.) or a whole structure such that even when maximum design
force is applied to the member or structure, the displacements do not exceed the
elastic limit. The structure would therefore, always come back to its initial position
without permanent deformation.

Generally, linear analysis yields accurate results for most common structures under
service loading conditions but it cannot predict the structural response of large
deformations or detect instabilities.

Non-linear analysis
A non-linear analysis consists in the incremental application of loads. During the
calculations, loads are not considered at a specic time, but they are gradually
increased and solutions to successive equilibrium states are performed.

6
2.2 Dierent analysis methods

Stiness denes the fundamental dierence between linear and non-linear analysis.
It is a property of a member or a structure and characterizes its response to an
applied load. The stiness is aected by shape, material properties and supports
conditions. At higher loads, deviations from linear elastic theory are observed and
they can be categorized into two main parts.

• Material non-linearity is when the stress-strain behavior ceases to be linear


and the material reaches its failure limit, the material properties will change.

• Geometric non-linearity occurs if signicant changes in geometry are observed


due to large deformations, the structure stiness changes.

So, if only a small change in stiness is assumed and linear elastic modeling is used,
it may lead to inaccurate results if higher loads are applied (SolidWorks, 2008).

Buckling
Buckling is dened as a sudden sideways failure of a structural member subjected
to high compressive stress. As an applied load on a member is increased, it will
ultimately become large enough to cause the member to become unstable.

Dierent members require dierent amount of loading to become unstable. The


terms stiness and slenderness are used to dene properties of a structural mem-
ber that relate to buckling. Stiness, for a material, either refers to its elastic
modulus, E, or its shear modulus, G = E/(2 + 2ν), where ν is the materials Pois-
son's ratio. Slenderness ratio, for a structural member, is a geometric concept that
has the formula
λ = L/i (2.2)

where L denotes the length of the member and i the radius of gyration of the
members cross-section. The slenderness ratio is a unit-less quantity since both L and
i have the unit of length. The ratio is an useful indicator of the ability of a member
to resist buckling when subjected to compressive load. Increasing slenderness results
in decreasing load resistance (Ed Akin, 2009).

2.2 Dierent analysis methods


Structural analysis is usually performed using elastic theory. Despite that, local plas-
ticity and plastic behavior can be all right in practice. Structural analysis methods
are variable, from using simple approaches to complex nite element analysis. Slow
manual methods have mostly been exchanged for faster and more accurate computer
calculations (Brian Kirke & Iyad Hassan Al Jamel, 2004).

7
2 Theoretical background

Initial imperfections

Non-linear material
Linear material
Perfect shape
Analysis:
(1) Fist-order elastic analysis x x
(2) Second-order elastic analysis x x
(3) First-order plastic analysis x x
(4) Second-order plastic analysis x x

Table 2.1: Dierent structural analysis methods

Figure 2.3: Load-deformation curve for dierent analysis types (Leroy Gardner &
David A. Nethercot, 2011)

First-order elastic analysis


Structure design for linear behavior. The equilibrium equations are written for an
undeformed shape and neither material nor geometric non-linearity is considered.

Linear buckling theory does not give information about the post buckling load-
deection relationship. When a column is loaded, initially it will remain stable.
When the load is larger than the buckling load, the column becomes unstable. Any
deviation from the straight conguration will force the column to buckle. Deforma-
tions remain undetermined after buckling.

8
2.3 Column stability

Second-order elastic analysis


This is also linear analysis but includes geometric non-linearity, i.e. P-Delta eects.
The equilibrium equations are written for a deformed shape and only geometric
non-linearity is considered.

Non-linear elastic buckling theory does not give information about the post buckling
load-deection relationship. A column with minor initial imperfections such as in
form or position of loading, collapses at the same buckling load as the perfect column.
It, however, starts deecting as soon as the load is added and the deection increases
until the buckling load is reached.

First-order plastic analysis


The equilibrium equations are written for an undeformed shape and non-linear ma-
terial properties are considered.

Plastic regions in a structure can develop steadily or suddenly if the plastic hinge
concept is used. If eects on the stability of the structure due to displacements are
insignicant, rst order plastic analysis can give a good account of simple elastic-
plastic behavior and failure.

Second-order plastic analysis


The analysis that shows maximum non-linear behavior. The equilibrium equations
are written for a deformed shape, and both material and geometric non-linearity are
included.

This type of analysis is best suited to simulate the actual behavior of a modeled
structure and calculate the inelastic stability limit, that is when the structures load
resistance limit is reached.

2.3 Column stability


A member under axial pressure, that is signicantly longer than it is wide, could
be called a column. Columns are often dened as long or short, depending on how
likely they are to buckle. Columns can be horizontal, vertical, or oblique.

A short, initially straight, column is applied with an axial load. The load causes
strain in the column cross-section that leads to shortening in the load direction. If
the load is increased incrementally the shortening continues until the column yields
or is crushed.

A long, initially straight, column is applied with an axial load. Column shortening

9
2 Theoretical background

Figure 2.4: Column classication (Ed Akin, 2009)

is only noticeable in the rst steps. When the load increases, the column becomes
unstable and deforms perpendicular to the load direction. The column is no longer
straight and is said to have buckled.

Higher bending stiness reduces buckling probability, while increased member length
increases it. Therefore, long, slender columns are more likely to buckle than short,
stocky ones.

In between the regions of short and long columns, there is a range of intermediate
columns. Columns in this range have a slenderness ratio that is too small for the
elastic stability range, and too large for strength limit range. These columns fail
by inelastic buckling. The maximum stresses in these columns are above the pro-
portional limit (above the proportional limit, the slope of the stress-strain curve is
less than the modulus of elasticity) when buckling occurs. Therefore, in inelastic
buckling, the critical load is always less than the Euler load.

Elastic buckling is the starting point when inspecting structural stability. The most
basic case is a simply supported, axially loaded column. Buckling load of such a
column can be calculated using the Euler equation (Leroy Gardner, 2011).

2.4 Elastic buckling (Euler buckling theory)


In 1757, Euler derived an equation for the maximum axial load on a straight, pris-
matic, simply supported, slender, elastic column. His research is the basis for elastic
buckling theory.

For the simple supported beam on gure 2.5, the following is true:

10
2.4 Elastic buckling (Euler buckling theory)

• The beam material is elastic and homogeneous.

• The beam is perfectly straight, awless and has no residual stresses.

• The beam cross-section, bracing and support conditions are such that only
plane buckling in one direction is relevant.

• The load is applied centrally on the beam cross-section.

Figure 2.5: Simply supported beam (Jørgen Amdahl, 2014)

If gravitational eects are disregarded, the beam will remain straight until the axial
force, N, is equal to the critical buckling load, Ncr , and then the beam will buckle.

To derive an equation for the critical buckling load, gure 2.5 is used as a reference.
In a section at a distance x from the member origin, the external bending moment
Me (x), is
Me (x) = Ncr · v(x) (2.3)

and assuming the member deection is small, the internal bending moment, Mi (x),
is
∂ 2 v(x)
Mi (x) = −EI · (2.4)
∂x2
At equilibrium the following dierential equation governs the deformations

∂ 2 v(x)
Ncr · v(x) = −EI (2.5)
∂x2
and it has the solution

v(x) = A1 · cos(kx) + A2 · sin(kx) (2.6)

Ncr
k2 = (2.7)
EI
In equation 2.6, A1 and A2 are constants. Using the boundary conditions

v(0) = 0 => A1 = 0

11
2 Theoretical background

v(l) = 0 => A2 · sin(kL) = 0


the solutions to 2.5 are
k · l = n · π, n = 1, 2, · · ·
The smallest value of k is at n = 1, and is given as k = π/L. This gives the lowest
value of the critical buckling load

π 2 EI
Ncr = (2.8)
L2
Equation 2.8 is known as the Euler buckling load.

2.5 Boundary conditions


Simply supported columns are not very common but the results for this basic case
can be used to nd the resistance of columns with other end conditions. This is
done through the concept of an eective length.

A column, which is xed at the base and free at the top, see gure 2.6-a) buckles in a
curve that is one-quarter of a complete sine wave. By extending the deection curve,
gure 2.6-b), it becomes one-half of a complete sine wave, which is the deection
curve for a pinned-end column.

The eective length, Le , for any column is the length of the equivalent pinned-
end column with a deection curve that matches all or part of the original column
deection curve.

The eective length could also be expressed as the distance between the inection
points on its elastically deected curve, or points of zero moments, if the curve is
extended (if necessary) until points of inection are reached.

The eective length is calculated by

Le = k · L (2.9)

and the critical buckling load in equation 2.8 (Euler buckling load) becomes

π 2 EI
Ncr = (2.10)
(kl)2

12
2.6 Column imperfections

Figure 2.6: Deection curves showing the eective length Le for a column xed at
the base and free at the top (Vladislav F. Demenko, 2015)

Figure 2.7: Recommended eective lengths for compression members (Leroy Gardner
& David A. Nethercot, 2011)

2.6 Column imperfections


Elastic buckling of a perfect column is a typical textbook example. Real columns are
never perfect and their aws have a big eect on their stability. These deviations
from the perfect shape or material cause the real columns to buckle before the
buckling load, calculated in 2.4, is reached.

13
2 Theoretical background

2.6.1 Initial out-of-straightness


The shape of a column is never perfectly straight. As soon as a column, that has
an initial deection, is loaded with an axial force, bending moments form in every
cross-section along its length, which causes a bending deformation.

Figure 2.8: Simple supported column with initial deection (Rangachari Narayanan
et al., 2012)
.

To simplify the calculations, the initial shape is usually dened by


 πx 
v(x) = A · sin (2.11)
L
where A is the maximum imperfection at the columns center, at x = L/2. The half
sine-wave shape represents the greatest inuence on the actual behavior, but other
initial shapes are of course possible.

Given the material remains elastic, the column deection at the applied force, P, is

1
vmax = v (2.12)
1 − P/Pcr
and the deection will head to innity when P gets close to Pcr . See chapter 2.8 for
a derivation.

As the deection increases, the bending moment increases. On the concave face
of the column, the resulting bending stress is compressive and adds to the stress
from the axial compressive force. The yield starts here and by increasing the axial
force the yield spreads along the column and through the cross-section. The column
stiness decreases and when the whole cross-section is in yield the stiness is gone
and the column buckles. (Baldvin Einarsson, 2005).

14
2.6 Column imperfections

Figure 2.9: Column with an initial deection (Primoº Moºe, 2015)

2.6.2 Load eccentricity


The loading rarely acts exactly on the center of the column cross-section. Initially
there is no deection. When the loading is applied, the eccentricity causes a bending
moment in the column so it starts to deect. The biggest stresses will form at the
columns concave face, mid-span.

The eccentricity is related to the size of the column cross-section, and not to the
column length, so thick columns are also aected.

Figure 2.10: Strength curve for eccentrically loaded columns (Rangachari Narayanan
et al., 2012)

15
2 Theoretical background

2.6.3 Residual stresses


In steel manufacturing processes, high temperatures are used to shape steel mem-
bers. Residual stresses are a consequence of dierential heating and cooling in the
rolling and forming processes.

While the area inside of the section is still hot, the edges cool and start to shrink.
Stresses in each section must be in balance because no external load is applied on
the member. For this reason, while the edges are exposed to compression forces due
to the shrinking, the inner locations are exposed to tensile stresses.

Figure 2.11: Typical distribution of residual stresses (Primoº Moºe, 2015)

The eects of residual stresses on stocky columns is not much, the column stays
straight and reaches yield stress in its whole cross-section. The same applies to
slender columns, they do not reach yield stress before they buckle elastically. In-

16
2.6 Column imperfections

termediate columns are, on the other hand, sensitive to residual stresses. Yield
in the cross-section limits its stiness enough to cause buckling of its elastic part.
(Rangachari Narayanan et al., 2012)

2.6.4 Material eects


Strain hardening eects tend to be ignored by designers, which in fact provides a
margin of safety. The onset of rst yield will not be aected, but the buckling load
may be increased.

Figure 2.12: Stress-strain relationship with strain hardening (Rangachari Narayanan


et al., 2012)

The lack of clearly dened yield point is common in stress-strain curves for high
strength steels. For stresses over the limit of proportionality, σp , material behavior
is non-linear and on loading/unloading the material is linear-elastic.

17
2 Theoretical background

Figure 2.13: a) Lack of a clearly dened yield, b) Lack of a clearly dened yield with
strain hardening (Rangachari Narayanan et al., 2012)

2.7 Structural analysis in EN1993-1-1:2005


Section 5.2.1 "Eects of deformed geometry of the structure" in (CEN, 2005) states
that internal forces and moments of a structure may generally be determined using
either

• a rst-order analysis, with the initial geometry

• a second-order analysis, taking the inuence of the deformations of the struc-


ture into account.

First-order analysis may be used if the increase of the relevant internal forces or
moments or any other change of structural behavior caused by deformations can be
neglected. This requires the following criterion to be satised:

Fcr
αcr = FEd
≥ 10 for elastic analysis

Fcr
αcr = FEd
≥ 15 for plastic analysis

The second-order eects should only be considered if they signicantly increase the
action eects or modify the structural behavior.

Section 5.2.2 Structural stability of frames (CEN, 2005) should be applied to consider
the inuence of the deformations of the structure, if the must be taken into account.
As stated there, stability verication of frames or their parts should be carried out
considering imperfections and second order eects. These imperfections and second-
order eects can be accounted for by one of the following methods:

18
2.7 Structural analysis in EN1993-1-1:2005

5.2.3(3)a) - Both totally by the global analysis.

5.2.3(3)b) - Partially by global analysis and partially by individual stability


checks of members.

5.2.3(3)c) - By individual stability checks of equivalent members using appro-


priate buckling lengths according to the global buckling mode of the structure,
for basic cases.

If individual members second order eects and relevant member imperfections are
totally accounted for in the structures global analysis, no individual stability checks
for the members are necessary.

If individual members second order eects or certain individual member imperfec-


tions are not totally accounted for in the global analysis, the individual stability of
members for the eects not included should be checked according to relevant criteria
in section 6.3. Where the equivalent column method (section 6.3) is used to assess
the stability of a frame, the buckling length values should be based on a global buck-
ling mode of the frame. Internal forces calculated according to rst order theory are
used in this case without considering imperfections.

Section 5.3 Imperfections in (CEN, 2005) states that global imperfections for frames
and bracing systems, and local imperfections for individual members should be con-
sidered. Appropriate allowances should be incorporated in the structural analysis
to include the eects of these imperfections, and equivalent geometric imperfections
should be used with values which reect the possible eects of all types of imperfec-
tions unless these eects are included in the resistance formulae for member design.

Section 5.3.2 Imperfections for global analysis of frames in (CEN, 2005) says the
assumed shape of the global and local imperfections can be derived from elastic
buckling mode of the structure. The most unfavorable conditions, direction and
form of buckling should be used.

Equivalent imperfection in the form of an initial sway imperfection and individual


bow imperfections of members should be used in analysis of frames sensitive to buck-
ling in sway mode. These imperfections may be replaced be systems of equivalent
horizontal forces. The shape of the elastic critical buckling mode of a structure can
be applied as a unique global and local imperfection as an alternative.

19
2 Theoretical background

2.8 Buckling resistance of members


(EN1993-1-1:2005 Section 6.3)
See (Boissonnade & European Convention for Constructional Steelwork Technical
Committee Structural Stability, 2006) for more information.

Continuing with the beam from section 2.4. Now it has the initial deection v0 (x),
which is sinusoidal with a maximum value of e0,d in the beam center

πx
v0 (x) = e0,d · sin (2.13)
L
The sinusoidal shape is often used to represent initial deection. When the beam
is loaded with the axial force, N, an additional deection, v(x), appears which is
related to instability. Given the support conditions, this additional deection can
be written as
πx
v(x) = A · sin (2.14)
L
Here, A is the maximum value of the added deection at the beam center. The
exural equilibrium equation then becomes

NEd
v 00 + (v0 + v) = 0 (2.15)
EI
By inserting 2.13 and 2.14 into 2.15, an expression for A is found to be

NEd
A= e0,d (2.16)
Ncr − NEd
where Ncr is the buckling load of the beam, calculated in section 2.4. The total
deection at the center of the beam can therefore be written as

Ncr 1
vmax = e0,d = e0,d (2.17)
Ncr − NEd 1 − NEd /Ncr

The maximum deection, vmax , is linked to the size of the axial compression force
through the well-known amplication factor K

1
K= (2.18)
1 − NEd /Ncr

An elastic 2nd degree check of the beams most critical cross-section, mid-span, is
made using the following equation

NEd 1 NEd e0,d


+ ≤1 (2.19)
NRd 1 − NEd /Ncr MRd

20
2.8 Buckling resistance of members (EN1993-1-1:2005 Section 6.3)

If the axial force NEd is increased until the beam buckles, it becomes Nb,Rd = χ · NRd
and the left side of equation 2.19 is equal to 1. By using the reduced slenderness
denition
NRk
λ̄2 = (2.20)
Ncr
equation 2.19 can be arranged to the so-called Ayrton-Perry equation

A
(1 − χ)(1 − χλ̄2 ) = e0,d χ = ηχ (2.21)
Wel
In equation 2.21, χ is a reduction factor for column buckling and η represents the
general initial imperfections, that can be used to assess the eects of initial imper-
fections on buckling. The initial imperfection takes residual stresses, shape aws
and load eccentricities into consideration. The beam length and the eects of some
of the imperfections are linked so η is written as

η = α(λ̄ − 0.2) (2.22)

2.8.1 Uniform members in compression


The imperfection factor α is related to the beam cross-section, the buckling axis, the
thickness of the beam plates and the material yield strength. The factor is chosen
from tables 2.2 and 2.3.

When α has been found, then χ can be calculated by

1
χ= p (2.23)
φ + φ2 − λ̄2
where the reduced slenderness, λ̄, is
r
Afy
λ̄ = (2.24)
Ncr
for class 1, 2 and 3 cross-sections and
r
Aef f fy
λ̄ = (2.25)
Ncr
for class 4 cross-sections. The intermediate factor φ is

φ = 0.5(1 + α(λ̄ − 0.2) + λ̄2 ) (2.26)

To take into account the eect of strain hardening in thick columns, the column line
starting point is said to be λ̄ = 0.2. If λ̄ < 0.2, buckling eects can be ignored and
cross-sectional checks are sucient.

21
2 Theoretical background

Table 2.2: Buckling curves (CEN, 2005)

22
2.9 Uniform members in bending and axial compression (EN1993-1-1:2005 Section 6.3.3)

Table 2.3: Imperfection factors (CEN, 2005)

When χ has been calculated, the design buckling resistance can be found by

χAfy
Nb,Rd = (2.27)
γM 1
for class 1, 2 and 3 cross-sections and

χAef f fy
Nb,Rd = (2.28)
γM 1
for class 4 cross-sections. The buckling verication can then be written as

NEd
≤ 1.0 (2.29)
Nb,Rd

When torsional and exural-torsional buckling is relevant, the same verication


method is used, but Ncr in equations 2.24 and 2.25 is replaced by Ncr,T and Ncr,LT ,
respectively. The imperfection factor, α is from tables 2.2 and 2.3 and is based on
buckling about the z-axis

2.9 Uniform members in bending and axial


compression (EN1993-1-1:2005 Section 6.3.3)
In section 6.3.3 in EN1993-1-1:2005 (CEN, 2005), it says that unless second order
analysis is carried out using initial imperfections, the stability of uniform members
which are subjected to combined bending and axial compression should satisfy:

NEd My,Ed + ∆My,Ed Mz,Ed + ∆Mz,Ed


χy NRk
+ kyy M
+ kyz Mz,Rk
≤1 (2.30)
γM 1 χLT γy,Rk
M1 γM 1

NEd My,Ed + ∆My,Ed Mz,Ed + ∆Mz,Ed


χz NRk
+ kzy M
+ kzz Mz,Rk
≤1 (2.31)
γM 1 χLT γy,Rk
M1 γM 1

where:

NEd , My,Ed and Mz,Ed are the design values of the compression force and the
maximum moments

23
2 Theoretical background

∆My,Ed and ∆Mz,Ed are moments due to a shift of the centroidal axis for class
4 sections

χy and χz are reduction factors due to exural buckling

χLT is a reduction factor due to lateral torsional buckling

kyy , kyz , kzy , kzz are interaction factors

The interaction factors have been derived from two alternative approaches and their
values may be found using Annex A or B in EN1993-1-1 (CEN, 2005).

Table 2.4: Values for NRk = fy Ai , Mi,Rk = fy Wi and ∆Mi,Ed (CEN, 2005)

2.10 Sway and bow imperfections


(EN1993-1-1:2005 Section 5.3.2(3))
To include the eects of initial imperfections in structure analysis, they can be mod-
eled directly. This can become dicult for structures of any signicant size and/or
complexity, and require several separate models to capture the most destabilizing
eect on the structure. (Jon A. Schmidt, 1991)

Another option is to use horizontal loads. These loads can be added in a single model
with the original structure geometry and they will produce forces and moments in
the structural members consistent with those which would develop because of the
actual imperfections. These horizontal loads are calculated as a portion of the
applied gravity loads and are added to the structure to account for the eects of
geometric imperfections. (Jason Ericksen, 2011)

In section 5.3.2(3) in (CEN, 2005) it says that if a frame is sensitive to buckling in


a sway mode, the eect of imperfections can be incorporated in the analysis in the
form of initial sway imperfection and individual member bow imperfections.

24
2.10 Sway and bow imperfections (EN1993-1-1:2005 Section 5.3.2(3))

The global initial sway imperfection is calculated using the formula:

φ = φ0 αh αm (2.32)

where

φ0 is the basic value φ0 = 1/200

αh is the√reduction factor for height h applicable to columns.


αh = 2/ h but 2/3 ≤ αh ≤ 1.0

h is the height of the structure

αm is the
p reduction factor for the number of columns in a row.
αm = 0.5(1 + 1/m)

m is the number of columns in a row including only those columns which carry
a vertical load NEd not less than 50% of the average value of the column in
the vertical plane considered.

Figure 2.14: Equivalent sway and bow imperfections (CEN, 2007)


(d) System, load and displacement
(e) Initial local bow and buckling length for exural buckling
(f ) Second order moment including moment from sway imperfection
(g) Initial local bow and buckling length for lateral-torsional buckling

The individual bow imperfections may be determined from relative initial local bow
imperfections of members for exural buckling

e0 /L (2.33)

and recommended values are given in table 2.5

In section 5.3.2(7) in (CEN, 2005) it is stated that the eects of local bow imper-
fections may be replaced by systems of equivalent horizontal forces, introduced for
each column.

25
2 Theoretical background

Table 2.5: Design values of initial local bow imperfection e0 /L (CEN, 2005)
elastic analysis plastic analysis
Buckling curve
e0 /L e0 /L
a0 1/350 1/300
a 1/300 1/250
b 1/250 1/200
c 1/200 1/150
d 1/150 1/100

Figure 2.15: Replacement of initial imperfections by equivalent horizontal forces


(CEN, 2005)

The equations for the loads from gure 2.15 are

8NEd e0,d
q= (2.34)
L2
 
4NEd e0,d L 8NEd e0,d
P = = · (2.35)
L 2 L2

2.11 EUGLI imperfection method


(EN1993-1-1:2005 Section 5.3.2(11))
The most direct approach of including initial imperfections into an analysis of a
structure, is to model them explicitly. In the Equivalent Unique Global and Local

26
2.11 EUGLI imperfection method (EN1993-1-1:2005 Section 5.3.2(11))

Initial Imperfection (abbr. EUGLI imperfection) method, the shape of the elastic
critical buckling mode of a structure, ηcr , is applied as a unique global an local initial
imperfection. Non-linear geometric analysis is easy to perform on such a structure.
The method can be applied to any structural type of geometry since a unique global
and local imperfections can be obtained from the buckling mode.

2.11.1 EUGLI method history


The term "ηinit,m imperfection" stands for "equivalent unique global and local ini-
tial imperfection". All imperfections of a compressed member or a frame can be
expressed by the ηinit,m imperfection in the shape of the elastic critical buckling
mode, ηcr , of the member or the frame.

The imperfection shape ηinit,m (x) shall be of the same shape as the 1st buckling
mode ηcr (x). The maximum amplitude, ηinit,m,max , is based on e0,d (the design value
of the initial bow imperfection) and Mη,cr,m (the bending moments due to the shape
of ηinit,m (x)). (Marian Dallemule, 2015)

The basis of the method using by the ηinit,m imperfection was developed by Professor
Eugen Chladný and showed in his PhD and Habilitation theses. There, the aim
was to derive a formula for the lateral forces acting on U-frames of open truss
bridge structures where he showed the importance of the curvature of the initial
imperfection. The results were used in the Czechoslovak Bridge Standard ƒSN 73
6205 (1984). (Eugen Chladný & Magdalena ’tujberová, 2013)

Prof. Chladný proposed his method in a more generalized form for Eurocode 3 in
2000, and it was accepted in the draft prEN1993-1-1 (June 5, 2002). He derived
the formula for e0,d and is the author of the method given in EN1993-1-1 (2005),
5.3.2(11). Later, he expanded the method for non-uniform cross-sections and non-
uniform compression forces. The expanded version is used in STN EN 1993-1-1/NA
(2007) and in EN 1999-1-1 (2007), 5.3.2(11). (Ivan BalẠ& Yvona Koleková, 2012)

Prof. Chladný used his method in the design of bridges in practice, e.g. the Apollo
bridge in Bratislava and the Pentele bridge in Dunaújváros (gures 2.16 and 2.17),
and in investigations of continuous truss bridges. (Ivan BalẠ& Yvona Koleková,
2012)

The benets of this method include:

• The imperfection shape can be used for various frame structures and is an
eective form compared to other shapes of initial lack of straightness.

27
2 Theoretical background

Figure 2.16: Apollo bridge in Bratislava, Slovakia (Eurocodes, 2015)

Figure 2.17: Pentele bridge in Dunaújváros, Hungary (Magdi51, 2012)

• Frame members can be uniform or tapered, with uniform or non-uniform dis-


tribution of compression forces along its length.

• The buckling shape is very suitable for second-order calculations of deforma-


tions and internal forces in the compressed members

The idea behind the method is that the imperfections aect the structure in the most

28
2.11 EUGLI imperfection method (EN1993-1-1:2005 Section 5.3.2(11))

detrimental way when they are in the shape of the elastic critical buckling mode.
The main problem is determining the amplitude of the equivalent imperfection.

In (Eugen Chladný & Magdalena ’tujberová, 2013) it says that the amplitude should
be based on the experimentally established and statistically evaluated imperfections
for various types of frame structures. If these values are unavailable, the amplitude
may be determined by the fundamental requirement that the buckling resistance
of the frame structure with axially loaded members shall be equal to the exural
buckling resistance of the equivalent member.

The consequence of the fundamental requirement is that if the substitute member


method utilization factor for the member is less than 1.0, then the EUGLI method
utilization factor for the same member is even lower. If the member utilization factor
is equal to 1.0 according to the substitute member method, then it is also equal to
1.0 according to the EUGLI method. This is the result of the behavior of the 2nd
order analysis. (Marian Dallemule, 2015)

The equivalent initial imperfection ηinit,m (x) covers both the global and local im-
perfections for the entire structure. The position of e0,d,m is where the curvature of
ηinit,m (x) is at its maximum. The internal forces should be calculated using the 2nd
order analysis. (Marian Dallemule, 2015)

In general, in iterative process is needed to nd the critical cross-section, xm . It is


located at the point where the combination of NEd,m and MEd,m + Mη,init,m,m has
the overall maximum eect. Only in the case where both cross-section and axial
force are uniform is the calculation not iterative and the critical cross-section can
be found directly. (Marian Dallemule, 2015)

2.11.2 EUGLI method formulation


The deection η II is calculated using second-order analysis. It is the eect of the
imperfection ηcr on a compressed member

ηcr
η II = (2.36)
αcr − 1
where αcr = Ncr (x)/NEd (x).

The equivalent member with equal exural buckling resistance as a frame structure
with axially loaded members, has pinned ends, the same cross-section, and the same
axial force as the frames critical cross-section m. Its length, Lcr,m is such that its
critical force equals the axial force in the critical cross-section m at the critical

29
2 Theoretical background

loading of the structure. Its relative slenderness is then


s s
Am fy NRk,m
λ̄m = = (2.37)
Ncr,m Ncr,m

The determination of the critical cross-sections position is based on the condition


that the utilization at the cross-section, Um , is greater than the utilization at all
other cross-sections

II II
NEd,m Mηinit,m,m
Um = + = UN,m + UM,ηinit,m,m (2.38)
NRd,m MRd,m
The bending moment at the critical cross-section m of the compressed member or
frame due to the ηinit,m imperfection calculated using a second-order analysis is
II
denoted Mηinit,m,m

For the equivalent member with initial imperfection, the bending moment at the
II II
ultimate limit state is Mηinit,m,m = Mηinit,m,max

 00 
II πx 1 1
Mηinit,m,max = −EIm e0d,m sin = NEd,m e0d,m (2.39)
Lcr,m αcr − 1 1 − α1cr

where NEd,m is the design value of the axial force in the critical cross-section m and
e0d,m sin πx/Lcr,m is the equivalent geometrical initial imperfection of the equivalent
member, where

χλ̄ 2
 MRk,m 1 − γMm1
e0d,m = e0k · δe = α λ̄m − λ̄0 ) (2.40)
NRk,m 1 − χλ̄2m
where

NRk = A · fy for class 1, 2 and 3 cross-sections

NRk = Aef f · fy for class 4 cross-sections

MRk = Wpl · fy for class 1 or 2 cross-sections

MRk = Wel · fy for class 3 cross-sections

MRk = Wef f · fy for class 4 cross-sections

The ηinit,m imperfection, for the veried member of frame, in the shape of the elastic
critical buckling mode is given by

ηcr (x)
ηinit,m (x) = ηinit,m,max · (2.41)
|ηcr |max

30
2.11 EUGLI imperfection method (EN1993-1-1:2005 Section 5.3.2(11))

where ηinit,m,max is the maximum amplitude of the imperfection shape and |ηcr |max
is the maximum amplitude of the critical buckling mode ηcr (x), with an arbitrary
value other than zero.

At the critical cross-section m of the veried member or structure initially bent into
the shape of the ηinit,m imperfection, the bending moment with regards to 2.36 will
be
00
II ηinit,m,max |EIm ηcr |m
Mηinit,m,m = 00
(2.42)
αcr − 1 |ηcr |m
II
The veried member or structure should satisfy that Um = 1. The values of NEd ,
NRd and MRd are the same for the equivalent member and the veried member so
from equations 2.39 and 2.42 it can be seen that the design value of the maximum
amplitude for the ηinit,m imperfection is:

Ncr,m |ηcr |max


ηinit,m,max = e0d,m 00 |
(2.43)
EIm |ηcr m

The governing dierential equation for an elastic curve y(x) is

∂ 2y M (x)
2
= (2.44)
∂x EI
and re-writing this second-order linear dierential equation gives the bending mo-
ment due to the shape of y(x)

∂ 2y
M (x) = EI (2.45)
∂x2
For the shape of the critical elastic buckling mode, ηcr (x), the bending moment at
the critical cross-section will be

00
|Mηcr |m = EIm |ηcr (xm )| (2.46)

00
The expression EI |ηcr (x)|m is the value at the critical cross-section m for the bend-
ing moment which would be necessary to bend the structure (in the state without
internal forces) into the form of the buckling mode. Calculating this value can be
done in several ways

• Where the analytical expression for the elastic buckling ηcr is available, the
00
value of EIm |ηcr (x)|m is easily calculated.

• If numerical values of ηcr


are available at particular points of the structure,
00
then the method of dierences can help provide a value for EIm |ηcr (x)|m .

00
• The curvature, ηcr (x), van be calculated by hand with numerical dierentia-
tion. To obtain the curvature of ηcr , either the 2nd derivation of the displace-
ments or the 1st derivation of the rotations must be found.

31
2 Theoretical background

Figure 2.18: An illustrative example. Note the hidden location of e0d,m (Marian
Dallemule, 2015)

• Using the characteristics of imperfection in the form of the elastic buckling


mode, 2.46

Using the characteristics of imperfection, equation 2.43 can then be rewritten as

II

αcr NEd,m α λ̄m − λ̄0 MRk,m |ηcr |max
ηinit,m,max = e0,d,m |ηcr |max = · δ e · (2.47)
|Mη,cr |m λ̄2m |Mη,cr |m

The bending moments on the imperfect structure due to the action of the axial force
II
NEd
00
ηinit,m,max ηcr (x)
Mη,init,m (x) = −EI(x) · · (2.48)
αcr − 1 |ηcr |max
and if the critical point can be located directly, then Mη,init,m (x) can be simplied
into
II
Mη,init,m,m = k · e0,d,m · NEd,m (2.49)

In a uniform member the location of the critical cross-section, m, matches the lo-
II
cation of the maximum bending moment Mη,cr (x). If the position of the critical
cross-section is not surely known, then UM,ηinit,m,m should be determined for all
members that are questionable. The critical cross-section of the structure is where
the value of Um = UN,m + UM,ηinit,m,m is greater than in other cross-sections.

32
2.12 SAP2000

2.12 SAP2000
SAP2000 is a nite element software from Computers and Structures, Inc., ideal
for structural analysis and design. It is object based, meaning that the models are
created using members that represent the physical reality.

SAP2000 can be used for the simplest problems or the most complex projects. It
has a fully integrated system for modeling, analyzing, designing, and optimizing
structures of various types, ranging from 2D to 3D.

SAP2000 has many analytical options. Advanced techniques allow for step-by-step
large deformation analysis, cable analysis, material nonlinear analysis with hinges,
buckling analysis and others make the program a good option for structural analysis
and design needs.

In SAP2000, a Load Case denes how specied loads are applied to a structure
and how the structures response is calculated. Every load case is either linear or
non-linear. The dierence between the two options is signicant.

2.12.1 Linear load case analysis


The structural properties (stiness, damping, etc.) are constant. The analysis starts
with zero stress and does not contain loads from any previous analysis, though it
can use stiness from a previous non-linear analysis. All displacements, stresses,
reactions, etc., are directly proportional to the magnitude of the applied loads and
the results of dierent linear analysis may be superposed. (Computers & Structures
Inc., 2015)

2.12.2 Buckling analysis


During a linear (eigenvalue) buckling analysis in SAP2000, perturbations are applied
to the undeformed structure and buckling factors and corresponding buckling mode
shapes are produced. The multiply of the buckling factors and the applied loading
represents the buckling load. The deections (mode shapes), forces, and reactions of
linear buckling analysis correspond to the normalized buckled shape of a structure.
Buckling must be evaluated explicitly for each set of loads of concern.(Computers
& Structures Inc., 2015)

33
2 Theoretical background

2.12.3 Nonlinear static analysis


Nonlinear static analysis can be used for many purposes, for example to perform an
initial P-delta or large-displacement analysis and to analysis cable structures.

The specied combination of applied loads is applied incrementally. The nonlinear


equations are solved iteratively in each load step. This may require re-forming and
re-solving the stiness matrix. The iterations are carried out until the solution
converges. Several parameters are available for control of the iteration process. The
use of default values for these parameters is recommended.

The structural properties may vary with time, deformation, and loading. The de-
ned properties, magnitude of loading, and parameters specied for the analysis
control the extent of structural non-linearity. Due to varying structural proper-
ties and non-zero initial conditions, the response may not be proportional to the
loading. The results of dierent non-linear analyses should therefore not be super-
posed.(Computers & Structures Inc., 2015)

2.12.4 Initial conditions


For each load case, initial conditions can be specied. Initial conditions describe the
state of the structure at the beginning of a load case and are for example displace-
ments, internal forces, and stresses. With zero initial conditions, the structure has
zero displacement, zero velocity and all elements are unstressed. A load case can
also continue from a previous nonlinear analysis. If it does, it contains all loads,
deformations, stresses, etc., from that preceding case.

For linear buckling load cases, some common reasons for using stiness at the end
of a nonlinear load case are to include p-delta eects from an initial P-delta analysis
or to include tension-stiening eects in a cable structure.(Computers & Structures
Inc., 2015)

2.12.5 Geometric non-linearity


For small loads and subsequent deections, their relationship for the structure is
linear. SAP2000 analyses assume such linear behavior for the most part. The pro-
gram can therefore form the equilibrium equations using the original (undeformed)
geometry of the structure.

34
2.12 SAP2000

The equilibrium equations should refer to the geometry of the structure after de-
formation, strictly speaking. If the load on the structure and/or the resulting de-
ections are large, then the load-deection behavior may become nonlinear. For
nonlinear analysis, you may choose the type of geometric non-linearity to consider
by selecting one of the following:

• None: All equilibrium equations are considered in the undeformed congura-


tion of the structure.

• P-delta only: The equilibrium equations take into partial account the deformed
conguration of the structure. Tensile forces tend to resist the rotation of
elements and stien the structure, and compressive forces tend to enhance
the rotation of elements and destabilize the structure. This may require a
moderate amount of iteration.

• Large displacements: All equilibrium equations are written in the deformed


conguration of the structure. This may require a large amount of iteration;
Newton-Raphson iterations are usually most eective. Although large dis-
placement and large rotation eects are modeled, all strains are assumed to
be small. P-delta eects are included.

When continuing one nonlinear load case from another, it is recommended that they
both have the same geometric non-linearity settings.(Computers & Structures Inc.,
2015)

2.12.6 Guyed towers


For guyed towers, the cables are under a large tension force produced by mechanical
methods that shorten the length of the cables. The catenary cable element does not
require large-displacement analysis except for structures supporting or supported by
cables where signicant deections or rotations are expected. To set the tension force
in the cables, target-force loading is used. It iteratively applies deformation load to
the cables to achieve a desired tension. The relative convergence tolerance of the
cable force is set to be 1% in the modeling of guyed towers.(Computers & Structures
Inc., 2015)

35
2 Theoretical background

2.12.7 Stress Output


Axial stress results can easily be extracted from SAP2000. In the program it is
denoted S11 and is computed at any point in a frame cross section as:

P M3 M2
S11 = − x2 − x3 (2.50)
a i33 i22
where:

P is the axial force and M2 and M3 are the bending moments.

a is the cross-sectional area, and i22 and i33 are the section moments of inertia.

x2 and x3 are the coordinates of the point where the stress is calculated,
measured from the centroid of the section.

Based on this, tensile stresses are always positive and compressive stresses are always
negative, regardless of material.(Computers & Structures Inc., 2015)

2.12.8 Procedures for various methods of analysis


The following abbreviations are used in the following tables

DEAD - Dead load


PRET - Pretension of guys
NLST - Non-linear static
BUCK - Buckling
GNP - Geometric non-linearity parameters are abbreviated
PD - P-delta
LD - Large displacements

SAP2000 procedure for linear analysis

All frame elements are meshed with a 0.5 m maximum length between nodes. The
analysis is run according to table 2.6 and axial forces and bending moments are
extracted from members of interest.

36
2.12 SAP2000

Table 2.6: SAP2000 - Linear analysis


Order Load Case Analysis Prerequisite GNP Load case objective

1 DEAD Nonlinear None None Self weight


2 PRET Nonlinear DEAD None Guy pretension
3 NLST Nonlinear PRET None Loading

SAP2000 procedure for non-linear analysis with sway and bow imperfections

For sway and bow imperfections, transverse loading is added globally to the structure
and locally to individual members to imitate initial sway and bow imperfections
according to section 2.10. This transverse loading is calculated using axial forces
from the analysis procedure in table 2.7.

Table 2.7: SAP2000 - Force calculation


Order Load Case Analysis Prerequisite GNP Load case objective

1 DEAD Nonlinear None PD+LD Self weight


2 PRET Nonlinear DEAD PD+LD Guy pretension
3 NLST Nonlinear PRET PD+LD Loading

After specifying this transverse loading equivalent to the sway and bow imperfection,
a second analysis is run to calculate the eect of the added forces to the structure.
The analysis procedure is shown in table 2.8.

Table 2.8: SAP2000 - Non-linear analysis with sway and bow imperfections
Order Load Case Analysis Prerequisite GNP Load case objective

1 DEAD Nonlinear None PD+LD Self weight


2 PRET Nonlinear DEAD PD+LD Guy pretension
3a SWAY Nonlinear PRET PD+LD Sway forces
3b +BOW Nonlinear SWAY PD+LD +Bow forces
3c -BOW Nonlinear SWAY PD+LD −Bow forces
4a NLST 01 Nonlinear SWAY PD+LD Loading+sway
4b NLST 02 Nonlinear +BOW PD+LD Loading+sway+bow
4c NLST 03 Nonlinear -BOW PD+LD Loading+sway−bow

37
2 Theoretical background

SAP2000 procedure for non-linear analysis with EUGLI imperfection

For EUGLI imperfection analysis, rst the ηinit,m amplication factor must be de-
termined. With the undeformed structure, the procedure in table 2.9 is run. With
the results, the necessary parameters can be found and using equations 2.40 and
2.47, a value for the ηinit,m amplication can be calculated.

The critical buckling factor - αcr


The maximum amplitude of the critical buckling mode - |ηcr |max
The bending moment in the critical cross-section m - |Mη,cr |m
II
The axial force at the critical cross-section - NEd,m

Table 2.9: SAP2000 - ηinit,m amplication calculation


Order Load Case Analysis Prerequisite GNP Load case objective

1 DEAD Nonlinear None PD+LD Self weight


1 PRET Nonlinear DEAD PD+LD Guy pretension
2 BUCK Linear PRET PD+LD Buckling factors
2 NLST Nonlinear PRET PD+LD Loading

Then, the structure is deformed into the shape of the critical buckling mode with the
corresponding amplication factor. The procedure in table 2.10 is used to estimate
the forces in the deformed structure.

Table 2.10: SAP2000 - Non-linear analysis with EUGLI imperfection


Order Load Case Analysis Prerequisite GNP Load case objective

1 DEAD Nonlinear None PD+LD Self weight


2 PRET Nonlinear DEAD PD+LD Guy pretension
3 NLST Nonlinear PRET PD+LD Loading+EUGLI

38
3 Worked examples
This chapter has some worked examples to show how the methods that were ex-
plained in chapter 2 are used.

3.1 Uniform column, pinned supports


A simple supported, axially loaded, IPE300 column can be seen in gure 3.1. Only
buckling about the strong axis (y − y) is inspected and self-weight and shear de-
formations are ignored. According to tables 5.2 and 6.2 in (CEN, 2005), the cross-
section is of class 2 and for buckling about y−y axis, buckling curve a should be used
where α = 0.21. The column properties can be seen in table 3.1. In the computer
analysis, the column is split up into 200 equally sized elements.

Calculating the buckling resistance according to section 2.8 is done in table 3.2

Table 3.1: Example 1 - Steel and column properties


Setup E(M P a) 210000
fy (M P a) 235
γM 0 = γM 1 (−) 1
NEd (kN ) 1000
L(mm) 5000
A(mm2 ) 5380
Iy (mm4 ) 83560000
WRk,y,el (mm3 ) 557000
iy (mm) 125
k(−) 1
α(−) 0.21

Sway and bow imperfections are added to the column according to chapter 2.10.
Since both the top and the bottom of the column are are pinned, sway imperfections
are irrelevant. According to table 2.5, the value selected for a column, following
buckling curve a and using elastic analysis should be e0 /L = 1/300. The column is

39
3 Worked examples

Figure 3.1: Example 1 - Analysis of a simply supported IPE300 column, a) Column


setup, b) ηcr (x), c) Mη,cr (x), d) NEd
II
(x)

Table 3.2: Example 1 - Linear buckling analysis


Setup Lcr(mm) 5000
N cr(kN ) 6927.5
λ̄(−) 0.4272
φ(−) 0.6151
χ(−) 0.9455
Nb,Rd (kN ) 1195.4

Usage NEd (N ) 1000


Usage (−) 0.837

not modeled with the bow imperfection, but equivalent force is added horizontally
to model the bow imperfection. The result are shown in table 3.3.

From gure 3.1, it is clear that the position of the columns critical cross-section, m,
is in its center, where the value of Mη,cr (x) is the greatest. Then by using equations
2.37 and 2.47, a value for ηinit,m amplication can be calculated.

Deforming the structure in the shape of the rst buckling mode with the maximum
displacement equal to ηinit,m,max and running a non-linear analysis gives a combina-
tion of axial force and moment in the column. Using equation 2.38 gives the usage

40
3.1 Uniform column, pinned supports

Table 3.3: Example 1 - Non-linear analysis with bow imperfection equivalent hori-
zontal forces
Setup Analysis elastic
Buckling curve a
e0 /L(−) 1/300
qBOW (kN/m) 5.3
PBOW (kN ) 13.3

Usage NEd (kN ) 1000.0


MEd (kN m) 19.6
σmax (M P a) 221.0
Usage (−) 0.940

values in table 3.4.

Table 3.4: Example 1 - Non-linear analysis with EUGLI imperfection


Setup αcr (−) 6.9275
|ηcr |max (mm) 0.0318
NEd,m (kN ) 1000
Mη,cr,m (kN m) 0.2206
λ̄m (−) 0.4272
ηinit,m,max (mm) 4.9382

Usage NEd,m (kN ) 1000


MEd,m (kN m) 5.8
σmax (M P a) 196.2
Usage (−) 0.835

The linear analysis and the EUGLI imperfection analysis give almost an identical
usage. The local bow imperfection, with an initial bow of e0 /L = 1/300 gives
e0 = 5000/300 = 16.67 mm. This obviously leads to higher usage than the EUGLI
imperfection method, where ηinit,m,max = 4.93 mm.

41
3 Worked examples

3.2 Uniform beam, xed and pinned supports


Example 1 from (Ivan BalẠ& Yvona Koleková, 2012) has the following description:

Uniform member with cross-section HE 260 B (ARBED), steel grade S


355, partial safety factor γM 1 = 1.1, member length L = 4.6m, action:
axial normal force NEd equals to the resistance, which means that for
NEd utilization grade U = 1. Two cases are investigated

a) exural buckling about major axis y-y (buckling curve "`b"', α=


0.34)

b) exural buckling about minor axis z-z (buckling curve "`c"', α =


0.49)

Boundary conditions are the same in both planes: left end is xed, right
end is simple supported.

In the solution to the example in the article in question, an analytical expression


is used for the elastic critical buckling mode ηcr . Comparing these results to the
calculation of ηinit,m imperfection with results from SAP2000 shows that the program
can be used to estimate the amplication. It is assumed that if the results from part
a) match the calculations, so will the results for part b). Part b) is therefore skipped.

Table 3.5: Example 2 - Prole properties


HE260B L(m) 4.6
h(m) 0.26
b(m) 0.26
A(mm2 ) 11840
Iy (mm4 ) 149200000
Wel,y (mm3 ) 1148000
β(−) 0.699
NRk (kN ) 4203
NRd (kN ) 3821
My,Rk (kN ) 407.5

Setting the problem up in SAP2000, creating a buckling load case and running
the analysis gives a normalized set of displacements, with a maximum value of
|ηcr |max , and corresponding moments necessary to bend the structure into the form
of the buckling mode, with a maximum value |Mη,cr |m at the critical cross-section
m. Results from calculating the maximum amplitude of the ηinit,m imperfection,

42
3.2 Uniform beam, xed and pinned supports

ηinit,m,max , and comparison to the results from (Ivan BalẠ& Yvona Koleková, 2012)
are shown in table 3.6.

Figure 3.2: Example 2 - Beam from (Ivan BalẠ& Yvona Koleková, 2012)

Table 3.6: Example 2 - SAP2000 analysis and calculation


Analysis αcr (−) 8.360
|ηcr |max (mm) 0.016
NEd,m (kN ) 3576
Mη,cr,m (kN m) 0.346

Calculation λ̄m (−) 0.3750


φ(−) 0.6000
χ(−) 0.9359
δe () 1.0138
ηinit,m,max (mm) 7.97

The beam is deformed in the shape of the rst buckling mode shown in gure 3.2
with the ηinit,m imperfection amplication according to table 3.6. The results and
comparison with the article is shown in table 3.7. The calculated values are very
similar to those of the article. By increasing the node number in the model, the
results should converge.

Table 3.7: Example 2 - Result comparison


Article Analysis

ηinit,m,max (mm) 7.981 7.97


NEd,m (kN ) -3576.1 -3576.0
MEd,m (kN m) 23.8 23.7
σmax (M P a) -322.7 -322.7

43
3 Worked examples

3.3 Simple frame


In (Eugen Chladný & Magdalena ’tujberová, 2013) the following frame is inspected.The
structural stability of the frame in gure 3.3 is calculated. The beams are IPE A
550 and the columns are HE 300 A. These sections are susceptible to torsional defor-
mations and are not restrained against torsion. Therefore, methods 5.2.2(3)c) and
5.2.2(3)b) in (CEN, 2005) with imperfections according to 5.3.2(3) (sway and bow
imperfections) and 5.3.2(11) (EUGLI imperfections) are appropriate.

Figure 3.3: Example 3 - Analyzed frame, a) scheme and loading, b) sway imperfec-
tion, c) and d) local bow imperfections

The steel in the frame is S355 with E = 210 GPa. The partial factors used are
γM 0 = γM 1 = 1.0. Table 3.8 shows the and cross-sectional characteristics of members
used in the frame. The actions on the frame are shown in table 3.9.

Table 3.8: Example 3 - Section properties


HE 300 A IPE 550 A

A(mm2 ) 11250 11700


Iy (mm4 ) 182600000 599800000
Wpl,y (mm3 ) 1383000 2475000
iy (mm) 127.4 224
NRk (kN ) 3993.8 4153.5
My,Rd,pl (kN m) 491.0 878.6

44
3.3 Simple frame

Table 3.9: Example 3 - Frame loading


Node i 2 3 5 6 7 8

FV,i (kN ) 200 200 160 120 120 160


FH,i (kN ) 30 22.5 20 15 11.25 15

Beam i 2 6 9

qi (kN/m) 70 50 70

The utilization of each column is calculated using equation 3.1, which is equation
2.30 where only the buckling resistance in the plane of the frame is taken into
account. Lateral torsional buckling will be neglected (χLT = 1.0):

NEd My,Ed
Ucolumn = + kyy ≤1 (3.1)
χy NRd My,Rd
   
 NEd NEd
kyy = Cmy 1 + λ̄y − 0.2 ≤ Cmy 1 + 0.8 (3.2)
χy NRk χy NRk
where Cmy = 0.9 according to table B.3 in (CEN, 2005)

3.3.1 Linear analysis


In structural analysis using section 5.2.2(3)c) in (CEN, 2005), imperfections and
second-order eects are accounted for by individual stability checks of equivalent
members, see section 2.7.

Internal forces and moments are calculated using rst-order analysis. The results
are shown in table 3.10. From the results of the analysis, utilization of each column
is calculated using equation 3.1.

3.3.2 Non-linear analysis with sway and bow imperfection


equivalent forces
Structural analysis using section 5.2.2(3)b) with imperfections according to section
5.3.2(3), both from (CEN, 2005), is used. Frame imperfections and second-order
eects are partially accounted for by global analysis and partially by individual

45
3 Worked examples

Table 3.10: Example 3 - Linear analysis


Column no. 1 3 4 5 7 8

NEd (kN ) -1378.5 -1481.5 -860.1 -364.5 -375.5 -899.9


-81.7 -235.0 279.4 241.0 -265.2 -342.6
MEd (kN m)
-24.7 276.8 -199.1 -272.5 327.1 319.7
λ̄y (−) 0.574 0.554 0.727 1.117 1.100 0.711
χ(−) 0.850 0.859 0.768 0.525 0.535 0.778
kyy (−) 1.037 1.038 1.033 1.025 1.027 1.033
U (−) 0.579 1.016 0.868 0.743 0.860 1.011

stability checks. The global initial sway imperfection is calculated using equation
2.26:
1 2
φ = φ0 αh αm = · · 0.866 = 2.88675 · 10−3
200 3
where:

φ0 = 1/200

αh = 2/ 5 + 3 + 3 = 0.603but2/3 ≤ αh ≤ 1.0
p
αm = 0.5(1 + 1/2) = 0.866

For exural buckling, the relative initial local bow imperfection of members accord-
ing to buckling curve b, from table 2.5 is e0 /L = 1/250.

For simplication, the imperfection eects are replaced by a system of equivalent


horizontal forces according to 5.3.2(7) in (CEN, 2005). This is done in three partic-
ular ways, from gure 3.3 with shape b) for the global sway imperfection, and shape
c) and d) for the local bow imperfections.

The local bow imperfections do not need to be allowed according to section 5.3.2(6)
in (CEN, 2005) because

r
L3 Afy
λ̄3 = = 0.5137 < 0.8196 = 0.5
i y λ1 NEd

They are included, nonetheless, in the numerical example in the article and also in
this worked example to give a better comparison.

According to 5.2.2(7)b), in analysis of buckling resistance according to 5.2.2(3)b),


reduction factors χ should be based on the buckling length values equal to the system
length.

46
3.3 Simple frame

λ̄ = 0.5137 and χ = 0.8781 for columns 1 and 3

λ̄ = 0.3082 and χ = 0.9610 for columns 4, 5, 7 and 8

The results of the non-linear analysis with the sway and bow imperfection equivalent
horizontal forces and the corresponding utilization, calculated using equation 3.1 is
shown in table 3.11.

Table 3.11: Example 3 - Non-linear analysis, forces and moments due to loading
and sway and bow imperfections
Column no.
Eect Forces
1 3 4 5 7 8

II
FE d + φ NEd (kN ) -1369.5 -1490.7 -858.3 -364.7 -376.9 -903.3

II -110.8 -265.9 276.2 240.1 -266.2 -347.7


MEd (kN m)
-6.1 299.1 -189.5 -270.4 331.8 330.2
U (−) 0.618 1.046 0.742 0.596 0.713 0.889

II
FE d + φ + e NEd (kN ) -1373.1 -1487.0 -857.8 -365.1 -376.5 -903.9

II -126.8 -283.7 279.3 238.1 -268.0 -344.9


MEd (kN m)
-11.4 280.7 -187.0 272.2 329.9 333.1
U (−) 0.652 1.013 0.748 0.599 0.709 0.884

II
FE d + φ − e NEd (kN ) -1365.9 -1494.4 -858.7 -364.3 -377.3 -902.7

II -94.8 -248.2 273.0 242.1 -264.4 -350.5


MEd (kN m)
23.6 317.4 -192.0 -268.6 333.7 327.4
U (−) 0.583 1.087 0.735 0.592 0.716 0.892

3.3.3 Non-linear analysis with EUGLI imperfections


For structural analysis using 5.2.2(3)b) with imperfections according to 5.3.2(11)
from (CEN, 2005) the same applies as in 3.3.2, the frame imperfections and second-
order eects are partially accounted for by global analysis and partially by individual
stability checks. The imperfections are modeled using EUGLI imperfection method.

From linear buckling analysis, the shape of the rst buckling mode and the corre-
sponding buckling factor are identied. For the frame in gure 3.3, αcr = 8.78329.
The values of the maximum displacement of the frame, |ηcr |max , and the bending
moment, |Mηcr |m , at the critical cross-section m, necessary to bend the frame into
the shape of the buckling mode, are obtained from the buckling analysis. They are
shown in table 3.12 and the upper part of table 3.13.

As αcr < 10, the internal forces and moments are calculated allowing for second-

47
3 Worked examples

Figure 3.4: Example 3 - a) Shape of the rst buckling mode, b) Bending moments
necessary to bend the structure into the shape of the buckling mode

order eects. Running a non-linear analysis with "P-Delta plus Large Displace-
ments" as geometric non-linearity parameters gives the axial forces in the columns,
necessary to calculate the ηinit,m imperfection amplication factor. Figure 3.5 and
table 3.12 display the results.

Table 3.12: Example 3 - Buckling analysis node displacement


Node 1 2 3 4 5 6 7 8

ηcr,x (mm) 0.0 38.2 38.2 0.0 44.9 46.2 46.2 44.9
ηcr,z (mm) 0.0 0.1 -0.1 0.0 0.1 0.1 -0.1 -0.1
|ηcr |max (mm) 0.0 38.2 38.2 0.0 44.9 46.2 46.2 44.9

Table 3.13: Example 3 - Forces and moments from buckling analysis and non-linear
analysis
Column no. 1 3 4 5 7 8

Nη,cr (kN ) 58.2 -58.2 11.3 1.7 -1.7 -11.3


-247.2 -242.8 0.8 4.2 4.1 0.1
Mη,cr (kN m)
236.0 234.9 51.9 8.3 8.3 51.8
II
NEd (kN ) -1373.8 -1486.4 -860.0 -365.1 -376.5 -901.6
-98.0 -253.1 279.1 241.1 -265.2 -344.8
II
MEd (kN m) -4.2 288.8 -194.7 -272.7 329.5 325.0

The columns all have identical cross-sections and each column has a constant axial
force. The position of the critical cross-section, m, can be found by using equation

48
3.3 Simple frame

Figure 3.5: Example 3 - Eects due to external loading using non-linear analysis:
a) Deformation, b) Bending moments MEd II

2.38. The shape of Mη,init,m (x) is the same as that of Mη,cr (x), but of a dierent
scale. From gure 3.4, it can clearly be seen that the biggest value of |Mη,cr (x)| is in
column 1 or 3. The axial force in the columns can be seen in table 3.13. The position
II
of m in column 3 is determined by a larger value of NEd and a small dierence in
II
the value of Mη,cr when comparing members 1 and 3.

With the position of the critical cross-section, and the values of |ηcr |max and |Mη,cr |m ,
the design value of the maximum amplitude of the ηinit,m imperfection can be cal-
culated.

Table 3.14: Example 3 - ηinit,m imperfection calculation


αcr (−) 8.783

cr | max (mm) 46.18
η II (mm) 5.93
max
|M
η,cr |
m (kN m) 242.8
II
Mη,cr
m
(kN m) 31.2
λ̄m (−) 0.5531
ηinit,m,max (mm) 36.64

The frame is deformed in the shape of the rst buckling mode by an amount equal to
the value of ηinit,m,max calculated in table 3.14. The results from non-linear analysis
of the deformed frame due to the loading and imperfections are shown in table 3.15.

According to 5.2.2(3)c), the buckling lengths based on the global buckling mode of
the frame, the associated relative slenderness λ̄ and reduction factor χ are accounted

49
3 Worked examples

for in the analysis

Table 3.15: Example 3 - Non-linear analysis of deformed frame with ηinit,m imper-
fection
Column no. 1 3 4 5 7 8

II
NEd+η,init (kN ) -1367.9 -1492.0 -859.2 -365.0 -376.6 -902.4

II -123.3 -277.8 279.2 241.5 -264.8 -344.9


MEd+η,init (kN m)
19.9 312.6 -189.4 -271.9 330.4 330.4
Um (−) 0.644 1.074 0.748 0.599 0.710 0.883

3.3.4 Method comparison


The utilization results from method are shown in table 3.16. In table 3.17, the results
from the (Eugen Chladný & Magdalena ’tujberová, 2013) are shown. Comparing
the tables, the results are very similar.

Table 3.16: Example 3 - Column utilization according to sections 3.3.1-3.3.3


Method of verication to 5.2.2(3) + assumption of imperfections

Member no. 5.2.2(3)b+ 5.2.2(3)b) + 5.3.2(3)


5.2.2(3)c)
5.3.2(11) FEd + φ FEd + φ + e FEd + φ − e
1 0.579 0.644 0.618 0.652 0.584
3 1.016 1.074 1.046 1.013 1.086
4 0.868 0.748 0.742 0.748 0.736
5 0.743 0.599 0.596 0.599 0.592
7 0.860 0.710 0.713 0.709 0.717
8 1.011 0.883 0.889 0.884 0.894

Table 3.17: Example 3 - Utilization of frame columns analyzed, from (Eugen Chladný
& Magdalena ’tujberová, 2013)

50
4 Statnett tower in Norway

4.1 Introduction
Statnett is the energy system operator in Norway. Statnett is not responsible for
generation of the electricity itself, but rather for ensuring that it reaches the re-
gional distributors and by that, the end user. This means coordinating production,
developing the grid and ensuring good maintenance. The grid is about 11000 km of
high-voltage power lines and 150 stations all over Norway. Statnett is also responsi-
ble for the connections from Norway to Sweden, Finland, Russia, Denmark and the
Netherlands. (Statnett, 2017)

For an analysis of a Statnett tower, a 420kV type NKT was selected. It is a self
supported steel lattice suspension tower with an internal guy system for resisting
transverse loads. It was controlled by EFLA but is a typical Statnett tower design
for a 420 kV transmission tower. It is for a power-line in Norway called Ofoten-
Balsfjord. This line section is approximately 160 km long.

51
4 Statnett tower in Norway

Figure 4.1: Stanett tower - Suspension tower, type BM (courtesy of Statnett)

4.2 Statnett method of analysis


Statnett uses its own method of when analyzing their towers. It is based on research
and experience. Buckling capacity is calculated in accordance with Eurocode 3
(CEN, 2005) and EN 50341, which is a standard for overhead electrical lines. The
relevant buckling curves used are

• Curve c for main members of body and channel proles (UNP) in bridges and
cross-arms with α = 0.49.

• Curve b with α = 0.34 for all other members.

Statnett buckling length coecients are based on dierent guy elds and dierent
direction of load cases. The coecients are displayed in table 4.1. The buckling
length factor is selected from the two options (A or B) based on the support the
compression leg under consideration receives from the other legs through horizontal
redundant planes. Table 4.2 shows the selected option for each load case.

52
4.3 Geometry

Statnett also uses a safety factor γM = 1.1 on steel in design. This factor is omitted
in the comparison in this thesis.

Table 4.1: Statnett tower - Buckling length factors for main body members

Table 4.2: Statnett tower - Suspension towers, buckling length options

4.3 Geometry
The geometry of the tower is relatively simple. It is XY symmetric, so all the tower
legs are of equal length and the size of proles is the same in each leg at the same
height. The bracing in the tower sides is also the same for both side, consisting
of crossing diagonals with prole sizes from L40x4 to L70x6. For each crossing
diagonal, there are 4 L-proles, 2 crosses, bolted to each ange of the UNP tower
leg. The distance between bracing member in the tower sides is 2 meters from the
foundation up to where the leg size changes to UNP180, see gure 4.2. From there,
it shortens with each cross.

The tower has many dierent prole sizes. In the tower chosen, the bottom guy
eld has the same prole size for the main leg members. Table 4.3 shows the main
leg members of the tower and table 4.4 shows the types of guys specied.

53
4 Statnett tower in Norway

Figure 4.2: Statnett tower - Geometry of tower (courtesy of Statnett)

The sections in the bridge is are all L-proles, ranging from L40x4 to L100x10.
Looking at gure 4.3, the bracing is modeled as truss elements (xed with regards
to rotation out of the plane put pinned with regards to in plane rotations) and main
bridge members, the top and bottom sections, as beam elements (moment sti so
rotation is xed in both planes).

54
4.3 Geometry

Table 4.3: Statnett tower - Prole specication


A Iy Iz Wel,y Wel,z
Section type mm2 mm4 mm4 mm3 mm3
x102 x106 x105 x105 x104
UPN160 24.02 9.25 8.54 1.16 1.83
UPN180 27.96 13.53 11.37 1.50 2.24
UPN200 32.19 19.11 14.83 1.91 2.70
UPN240 42.29 35.97 24.79 3.00 3.95

Table 4.4: Statnett tower - Guy specication


Diameter Area Mass Breaking strength Pre-tension
Guy eld 2
mm mm kg/m kN kN
Upper 17.85 190 1.49 249.9 20.0
Lower 21 260 2.04 345.94 16.1

There are two bracing planes in the tower, their main sections are UNP proles
which support the tower legs. They are braced with crossing L-proles. The upper
one is called horizontal redundant plane (no. lettriegel) and is only there to support
the strong axis of the tower legs. The lower one is a horizontal force plane (no.
trykkrigel), designed to withstand compression forces. See gure 4.2.

The insulators for this type of tower are V-strings, one suspended from each cross-
arm and one from the tower bridge.

55
4 Statnett tower in Norway

Figure 4.3: Statnett tower - Geometry of bridge (courtesy of Statnett)

4.4 Loading
For the design of the tower in question, there are 25 dierent load cases. They detail
all possible conditions the tower can experience, from wire breaks or extreme cold,
to extreme wind or ice on the conductors and shield wires.

In the analysis, the focus is on dierent analysis methods used to estimate the
stability of the leg members. Two extreme load cases were selected for inspection.
These load cases can be seen in table 4.5 and the points where they act are shown
on gure 4.4.

The rst load case is called Unbalanced ice. It is a longitudinal force which pulls
the tower in the line direction, causing compression in the legs of the tower in one
face and tension in the opposite face. The second load case is called Wind and ice,
and is a transverse load case which causes compression on one side of the tower, and
tension on the opposite side.

56
4.5 Modeling

Table 4.5: Statnett tower - Load cases


C1, C2, C3 E1, E2
No. Load case
T (kN ) L(kN ) V (kN ) T (kN ) L(kN ) V (kN )
1 Unbalanced ice 0 -12 45 0 -31 16
2 Wind and ice 30 0 31 13 0 9

Figure 4.4: Statnett tower - Load points sketch

4.5 Modeling
The basic geometry is taken from a PLS Tower model used for the tower design. It
is extracted to an Autocad readable le. SAP2000 can import this le and set up a
model with the same geometry as in PLS Tower.

The lattice of the tower is modeled as frame elements in SAP2000. It is made of


S355 steel with E = 210 GPa.

All bracing members in the are dened as moment free with respect to in plane
rotation but moment sti with respect to rotation out of the plane. Both the bracing

57
4 Statnett tower in Norway

planes are moment released in the plane of the tower but moment xed out of their
plane.

The legs are modeled as a single, moment sti, element from the foundation to the
top of the tower. For the tower in this analysis, each leg consists of 6 proles bolted
together. They are only designed for an axial force and not fully moment sti.

At the bottom of the lowest leg proles there is a welded, stiened plate that is
bolted to the foundations of the tower with a few bolts. This could indicate some
degree of xity, but here the leg is modeled as pinned to simplify the comparison
and reduce assumptions.

The guys in the tower are modeled as catenary cable elements, specied as initially
tensioned (see table 4.4). The elastic modulus of steel in the guys is E = 180 GPa.

58
4.6 Analysis

Figure 4.5: Statnett tower - SAP2000 3D model

4.6 Analysis
For the analysis procedure of each method, see chapter 2.12.8.

Since the purpose of this thesis is to compare dierent analysis methods for column
buckling, the focus will be on the tower legs. In the load cases listed in table 4.5,

59
4 Statnett tower in Norway

for both compression and tension, legs 1 and 2 are equally aected for in load case
1-Unbalanced ice, and legs 1 and 4 are equally aected by case 2-Wind and ice. Leg
1 is in compression and is critically loaded in both cases and therefore, it is safe to
limit the inspection to it.

The rst load case, 1-Unbalanced ice, is the critical load case for this tower. It has
a longitudinal eect and causes compression in the front face leg members of the
tower. Its usage should be close to 100%, but not above it.

The second load case, 2-Wind and ice, is less critical. Usually when this load case
is the critical one, the tower has a bracing plane in the bottom guy eld which was
not necessary for the design loading for this tower. The wind pressure on the tower
is omitted due to complex loading and instead the load values from the earth-wires
and conductors are doubled. This is done to get higher usage of the tower legs from
the second load case.

The self-weight of the structure, calculated by SAP2000 and added to the loading,
is increased by a factor of 1.05 to account for coating, bolts and other items not
modeled. The partial coecient factors used in the design are γM 0 = γM 1 = 1.0.
According to table 2.2, UNP proles, such as the leg proles in the tower follow
buckling curve c and have α = 0.49.

When calculating the usage, it is set equal to 1 when the stress at the outer edge
of a cross-section reaches yield strength. Stress above yield strength equals usage
above 1.

4.6.1 Buckling analysis


A buckling analysis is run for both load cases in table 4.5. The rst mode shapes
are shown in gures 4.6 and 4.7. These buckled modes are presumed to be critical in
the analysis. For the sway and bow imperfection, the equivalent forces are modeled
according to these shapes. For the EUGLI method, these shapes are scaled by the
ηinit,m imperfection amplication factor.

Table 4.6: Statnett tower - Load case buckling factors


Load case Mode 1 Mode 2 Mode 3

Unbalanced ice 3.322 3.675 3.813


Wind and ice 2.678 4.468 4.501

Note the shape of the critical buckling mode for load case 1-Unbalanced ice. It seems
to be a combination of transverse and longitudinal buckling. Modeling this with the

60
4.6 Analysis

Figure 4.6: Statnett tower - Unbalanced ice critical buckling load

Figure 4.7: Statnett tower - Wind and ice critical buckling shape

61
4 Statnett tower in Norway

EUGLI method should be relatively simple but more complex with the sway and
bow imperfection equivalent forces. The critical mode shape for load case 2-Wind
and ice is a more simple form of buckling.

Buckling modes 2 and 3 for the for the load cases are shown in gure 4.8.

Figure 4.8: Statnett tower - Buckling shapes, from left, Unbalanced ice mode 2,
Unbalanced ice mode 3, Wind and ice mode 2, Wind and ice mode 3

4.6.2 Linear analysis


Linear analysis of the model returns axial forces and bending moments along leg
1. These forces are then used for calculation of the stability of the tower. The
stability is calculated using equations 2.30 and 2.31. The stability was calculated in
all sections in leg one and the maximum value calculated is displayed in table 4.7.

The interaction factors (see section 2.9) were calculated conservatively. Although
setting their value to 0 did not have much eect on the usage value calculated (ca.
5%). The maximum usage for both load cases is in the lower guy eld according to
this method.

The main reason, perhaps, the usage is higher than 1, is that the eective length
factor, ky . The tower foundation is dened as pinned in the model and the connection
to the lower bracing plane is hinged, so the value of ky used in the calculation of χy
was set to 1 as is done for a pinned-pinned column (see section 2.5).

If the value of ky is lowered to 0.85 (both joints partially restraint), the usage value
drops to 0.77 for 1-Unbalanced ice and 0.81 for 2-Wind and ice.

62
4.6 Analysis

Table 4.7: Statnett tower - Usage from linear analysis


Unbalanced ice NEd My,Ed Mz,Ed Usage
(kN ) (kN ) (kN ) (−)
Lower guy eld (UNP240) -359.8 0.0 0.6 1.048
Upper guy eld (UNP200) -332.5 -0.6 0.1 0.571
Upper guy eld (UNP180) -258.0 -1.5 0.1 0.600

Wind and ice NEd My,Ed Mz,Ed Usage


(kN ) (kN ) (kN ) (−)
Lower guy eld (UNP240) -372.4 -7.2 -0.4 1.167
Upper guy eld (UNP200) -221.5 9.7 -0.1 0.533
Upper guy eld (UNP180) -218.6 -5.4 0.5 0.659

4.6.3 Non-linear analysis with sway and bow imperfection


equivalent forces
Calculation of the necessary parameters to estimate the sway and bow imperfections
is done according to section 2.10. The parameters are shown in table 4.8.

Next, using the buckling modes found in section 4.6.1, the sway and bow imperfec-
tions can be estimated. The shape of the critical buckling mode is used to estimate
the direction of the equivalent forces necessary to model the imperfections. For the
two load cases in question, gure 4.9 shows the equivalent forces.

Table 4.8: Statnett tower - Sway and bow imperfection parameters


Sway imperfection h 32.5
m 2
φ0 0.005
αh 0.6667
αm 0.8660
φ 0.0029

Bow imperfection Buckling curve c


elastic/plastic elastic
e0 /L 0.005

For the transverse and longitudinal forces, the factors in table 4.9 are multiplied by
the axial force, NEd acting in the leg members. The force used is the maximum force
found in each element in each guy eld, which is conservative because the axial force

63
4 Statnett tower in Norway

Figure 4.9: Statnett tower - Sway and bow imperfection equivalent forces.
Upper row: Load case 1-Unbalanced ice 1) Sway forces, 2) +Bow forces
Lower row: Load case 2-Wind and ice 1) Sway forces, 2) +Bow forces
64
4.6 Analysis

is a bit higher in the bottom of a column than in its center, due to dead load. The
force is taken from a non-linear analysis with P-Delta eects and Large deections
included. The axial forces are shown in table 4.10.

Table 4.9: Statnett tower - Factors to calculate sway and bow equivalent forces
Load direction
Guy eld Transverse Longitudinal

PSW AY (N ) ALL 2.9 2.9


PBOW (N ) ALL 20 20
qBOW (N/m) lower 3.2 20
qBOW (N/m) upper 3.33 20
qBOW (N/m) above guy eld 0.01 22.22

Table 4.10: Statnett tower - Axial forces necessary to calculate sway and bow equiv-
alent forces
Load case Unbalanced ice Wind and ice
Guy eld NEd (kN) NEd (kN)

Lower 372.2 385.9


Upper 348.9 230.1
Above guy eld 232.9 tension

The maximum usage of leg one of the tower is shown in table 4.11. As can be seen
in the table, the sway imperfection alone does not yield high usage in the tower leg.
This is expected, because it does not capture the buckling modes very well. For
load case 1-Unbalanced ice, the result from the positive and negative bow eect is
in pretty good agreement. For the load case 2-Wind and ice, the dierence between
the dierent directions of the bow eect is quite large. This could be because, as
can be seen in gure 4.7, the buckling of the leg is into the tower for the specied
load direction. Positive bow force is directed out of the tower so the negative bow
force could be capturing the critical buckling mode better.

Table 4.11: Statnett tower - Usage from non-linear analysis with sway and bow
imperfection equivalent forces
Usage
Load case
FEd + φ FEd + φ + e FEd + φ − e
Unbalanced ice 0.290 0.816 0.791
Wind and ice 0.383 0.543 0.754

65
4 Statnett tower in Norway

4.6.4 Non-linear analysis with EUGLI imperfection


In the EUGLI imperfection method the tower is initially deformed using the critical
mode shapes and buckling factors found in section 4.6.1. The calculation of the
ηinit,m imperfection amplication factor for each load case from table 4.5 is shown
in table 4.12.

Table 4.12: Statnett tower - ηinit,m amplication factor calculation


Load case Unbalanced ice Wind and ice

Column properties Section UPN200 UPN240


A(mm) 3219 4229
fy (M P a) 355 355
α(−) 0.49 0.49
Wel,m (mm3 ) 26998 299750
MRk,m (kN m) 9.6 106.4

Analysis results αcr (−) 3.32 2.68


|ηcr |max (mm) 0.017 0.067
NEd,m (kN ) 328.6 381.7
Mη,cr,m (kN m) 0.002 0.040

Calculated values λm (−) 1.023 1.212


ηinit,m,max (mm) 31.69 60.23

In separate models, the shape of the tower is deformed according to the buckling
shape and scaled by the corresponding ηinit,m,max value.

With nonlinear analysis, the load case matching the deformation is applied to cor-
responding tower, and stresses in leg 1 are inspected. Table 4.13 shows the results
of the analysis.

Table 4.13: Statnett tower - Usage from non-linear analysis with EUGLI imperfec-
tion
Usage
Load case
+ηinit,m
Unbalanced ice 1.046
Wind and ice 0.769

The results in table 4.13 are partially what was expected.

For load case 1-Unbalanced ice, the maximum usage is above 1 which is strange, since
this load case is the critical one and the tower was designed for it. In section 4.6.1,

66
4.6 Analysis

it was noted that the buckling shape for this load case looked like a combination of
transverse and longitudinal buckling. In the calculation of the ηinit,m amplication
factor, the upper guy eld was the critical one. It could be that the buckling in the
lower eld could be reducing the buckling resistance of the weak axis in the upper
eld and therefore causing buckling.

For load case 2-Wind and ice, the usage is very similar to the usage gotten with
sway and bow imperfections. The shape is very straightforward and the result is
believable.

4.6.5 Resistance according to Statnett method


The method Statnett uses in calculation is shown in section 4.2. The buckling
resistances for the tower and load cases in question, according to their conventions,
is shown in table 4.14.

Table 4.14: Statnett tower - Guy eld buckling resistances


Guy eld Upper Upper Lower Lower
Section UNP180 UNP200 UNP240 UNP240

Axis weak strong weak strong


L(mm) 2000 12000 2000 12500
k(−) 1 0.6 1 0.85
Lcr (mm) 2000 7200 2000 10625
Nb,Rd (kN ) 387.7 483.2 742.5 468.8

The axial load for each load case and each size of section was found in the model of
the tower. The analysis was non-linear with P-Delta and Large displacement eects
included. The axial forces and corresponding section usages are shown in table 4.15.

Table 4.15: Statnett tower - Usage according to Statnett method of analysis


UNP180 UNP200 UNP240

Buckling resistance Nb,Rd (kN ) 387.7 483.2 468.8


Unbalanced ice NEd (kN ) -282.1 -328.8 -370.8
Wind and ice NEd (kN ) -222.0 -227.5 -384.7

Unbalanced ice U (−) 0.728 0.681 0.791


Wind and ice U (−) 0.573 0.471 0.821

The maximum axial forces calculated in the tower are similar for the two load cases

67
4 Statnett tower in Norway

and they are found in the same member (bottom of leg 1). The usage between the
load cases is therefore similar and is well comparable to the other methods.

4.7 Method comparison


Comparing the methods described in sections 4.6.2 to 4.6.5, the critical usage calcu-
lated with each method is taken and compared to the next. The results are shown
in table 4.16 and on gure 4.10.

Table 4.16: Statnett tower - Comparison of dierent analysis methods


Load case Unbalanced ice Wind and ice

Linear 1.048 1.167


Sway + bow 0.816 0.754
EUGLI 1.046 0.769
Statnett 0.791 0.821

Figure 4.10: Statnett tower - Comparison of methods

First for load case 1-Unbalanced ice, using the Statnett usage as a benchmark, it is
interesting that the Sway+Bow results show so much similarity due to the complex
modeling of the equivalent horizontal forces, see gure 4.9.

The linear analysis and EUGLI method yield similar result, maximum section usage
just above 1. It was discussed in section 4.6.2 that the usage according the the linear
analysis is probably to high due to a high eective length factor. It was said that

68
4.7 Method comparison

Figure 4.11: Statnett tower - Location of the critical cross section by method

with a lower factor the usage would drop to about 0.77 which is on par with sway
and bow analysis and the statnett method.

From gure 4.11, it can be seen that the EUGLI method estimates the critical
cross section to be in the upper guy eld, contrary to the other methods. We are
therefore not comparing the same member between the methods. It is most likely, as
was discussed in section 4.6.4, that the buckling in the lower guy eld is amplifying
the buckling in the upper guy eld somehow, causing lower buckling resistance. If
this is the case, this would be a very interesting result. In gure 4.8, buckling mode
2 for load case 1-Unbalanced ice can be seen. That is the buckling form that designs
are based on today. Further research is needed to prove that the critical buckling
shape found is the right one, but if that is the case then the design may need to be
adjusted.

For the second load case, 2-Wind and ice, the shape of the critical buckling mode
is relatively simple and therefore, the EUGLI method and the Sway+Bow method
should yield a good result. As shown on gure 4.10, the result of the Statnett method
gives a slightly higher usage value. Perhaps it is a bit conservative and the methods
modeling the imperfections are more correct. The linear analysis result is also well
above 1.0 for this load case. As was stated in section 4.6.2, the eective length
factor used for the linear analysis is probably to large. With a lower factor, the
usage dropped to 0.81. That is slightly higher that the two imperfection methods,
but slightly lower that the Statnett method. Overall, the methods are in agreement
for load case 2-Wind and ice.

69
5 Guyed tubular tower

5.1 Introduction
Interest for the use of a new tower type in the building of new 220 kV overhead
transmission lines has been growing in Iceland recently due to visual aspects. The
type is called guyed tubular M-tower. During the writing of this thesis, Landsnet
(the grid owner and operator in Iceland) is building 220 kV lines, using this new
tower type.

Regularly, full-scale tower test are performed. This is done to verify the resistance
of the tower to loading. Landsnet decided to have a full-scale tower test for the
guyed tubular M-tower carried out to gain a better understanding of the stability
of the tower and to see how the test results compare with calculations with the goal
to conrm and/or improve the calculation methods.

This type of tower is inspected here with the methods previously described in this
thesis. Special focus is on tower performance using tower the tower test loading.

71
5 Guyed tubular tower

Figure 5.1: Landsnet tower - Guyed tubular M-tower (courtesy of Landsnet)

5.2 Geometry
The suspension towers, for the line being built, are built out of hollow steel sections
and are of varying height, ranging from 17 m to 28 m, measured from the ground
level to the underside of the cross-arm. In the tower selected for the full-scale test,
L1=L2=26 m.

In table 5.1, all sections used in the tower are listed and their main properties are
shown. All members are cold formed so the tower legs follow buckling curve c with
α = 0.49 according to tables 2.2 and 2.3.

72
5.2 Geometry

A technical drawing of the tower can be seen in gure 5.2 and a drawing of the
location of foundation in gure 5.3.

Figure 5.2: Landsnet tower - Outline drawing of the suspension tower (courtesy of
Landsnet)

Table 5.1: Landsnet tower - Prole specication


A Iy Iz Wel,y Wel,z
Section type mm2 mm4 mm4 mm3 mm3
x103 x107 x107 x105 x105
Cross-arm SHS 350x8 10.94 21.40 21.40 12.20 12.20
Davit arms RHS 350x250x8 9.34 16.70 9.91 9.52 7.93
Bracing SHS 160x4 2.50 1.01 1.01 1.27 1.27
Steel poles CHS 355.6x8 8.74 13.20 13.20 7.42 7.42

73
5 Guyed tubular tower

Figure 5.3: Landsnet tower - Location of foundations (courtesy of Landsnet)

The tower can be split up into two sections, a primary and a secondary structure.
The primary structure consist of the main tower members and carries most of the
loading. The secondary structure consist of the secondary guys and the bracing
between the legs. The purpose of the secondary structure is to shorten the buckling
length of the tower legs and therefore, increase the primary structures buckling
resistance. The structure division can be seen in gure 5.4.

The main guys are double (2x21 mm) and connected to the cross-arm. The sec-
ondary guys are single (1x11.5 mm) and are connected to the middle of the tower
legs. The pre-tension in the guys is set equal to 6.67% of the ultimate tension ca-
pacity for the main stability guys and 3.0% for the secondary guys. The pre-tension
values can be seen in table 5.2. Note that the guys connect to the surface of the
tower proles and therefore cause a bending moment in the leg members.

74
5.3 Loading

Figure 5.4: Landsnet tower - Primary/secondary structure

Table 5.2: Landsnet tower - Guy specication


Diameter Area Mass Breaking strength Pre-tension
2
mm mm kg/m kN kN
Main guys 21 261.5 2.16 357 2x24
Secondary guys 11.5 80.65 0.66 113 1x3.4

5.3 Loading
In the full-scale tower test, the tower was be tested for 8 load cases. Table 5.3 shows
the load acting at each attachment point. After testing of load cases 1-8, the loading
shall be continued and the tower shall be tested to failure. Table 5.4 describes how
the loading shall be increased from load case 8 to failure. In the actual test the
holding period of 100% specied load shall be 3 minutes for load cases 1-8 and 2
minutes for load cases 8b-8k. See gure 5.5 for the location of the loading points.

75
5 Guyed tubular tower

Table 5.3: Landsnet tower - Load cases


C1 C2 C3

No. Load case T L V T L V T, L, V


kN kN kN kN kN kN kN
1 Transverse wind 59 5 20 Same as C1 Same as C1
2 Maximum ice 0 5 85 Same as C1 Same as C1
3 Maximum ice no EW 0 17 93 Same as C1 Same as C1
4 Wind and ice no EW 70 10 60 Same as C1 Same as C1
5 Cond. Break - left phase 0 74 11 0 0 16 Same as C2
6 Cond. Break - center phase 0 0 16 0 74 11 Same as C1
7 EW. Break - left 0 0 16 Same as C1 Same as C1
8 Wind and ice 65 0 52 Same as C1 Same as C1

E1 E2 W1

No. Load case T L V T L V T


kN kN kN kN kN kN kN
1 Transverse wind 18 0 7 Same as E1 18
2 Maximum ice 0 5 55 Same as E1 0
3 Maximum ice no EW 0 0 0 Same as E1 0
4 Wind and ice no EW 0 0 0 Same as E1 9
5 Cond. Break - left phase 0 0 5 Same as E1 0
6 Cond. Break - center phase 0 0 5 Same as E1 0
7 EW. Break - left 0 55 4 0 0 5 0
8 Wind and ice 38 0 26 Same as E1 9

76
5.3 Loading

Table 5.4: Landsnet tower - Load case 8 stepped increase


C1, C2, C3 E1, E2 W1

No. Load case T L V T L V T


kN kN kN kN kN kN kN
8a Wind and ice 65 0 52 38 0 26 9
8b Wind and ice 65 0 52 42 0 30 9
8c Wind and ice 65 0 52 48 0 32 9
8d Wind and ice 65 0 52 53 0 35 9
8e Wind and ice 65 0 52 61 0 40 9
8f Wind and ice 70 0 52 61 0 40 9
8g Wind and ice 75 0 52 61 0 40 9
8h Wind and ice 80 0 52 61 0 40 9
8i Wind and ice 85 0 52 61 0 40 9
8j Wind and ice 90 0 52 61 0 40 9
8k Wind and ice 95 0 52 61 0 40 9

Figure 5.5: Landsnet tower - Location of load point attachments

77
5 Guyed tubular tower

5.4 Modeling
The tower structure is modeled in SAP2000. The tower legs, leg bracing, cross-arm
and earth-wire peak are dened as dened as beam elements in the model, made of
S355 steel with E = 210GP a. The guys are catenary cable elements, specied with
initial tension (see table 5.2). Maximum segment length in the model is 0.5 m.

Looking at a 2D view of the tower, for a leg of the tower structure the following
applies

• Its top connection (leg to cross-arm) is rigid with respect to rotation out of
the plane shown but free with respect to rotation in the plane.

• Its central connection (to the bracing between the legs) is rigid with respect to
rotation out of the plane shown but free with respect to rotation in the plane.

• Its bottom connection (to the foundation) is pinned, i.e. free with respect to
rotation in and out of the plane.

• All the guys are also modeled as pinned to the ground.

78
5.5 Buckling of a intermediately restrained leg

Figure 5.6: Landsnet tower - SAP2000 3D model

5.5 Buckling of a intermediately restrained leg


The two legs in the tower have a intermediate elastic restraint at mid height. There-
fore, it can be assumed that the eective buckling length of each leg is half the leg
height, if the elastic restraint (i.e. elastic spring) has sucient stiness, see gure
5.7.

The elastic buckling capacity of a column with a elastic central spring restraint can
be expressed with an approximate relationship between the critical buckling force,

79
5 Guyed tubular tower

Ncr,spring , and the stiness of the spring restraint, KT , as:

3
Ncr,spring = Ncr + · L · min (KT ; 16 · Ncr /L) (5.1)
16
where

Ncr is the Euler load of the unrestrained column


KT is the stiness of an elastic spring

See (Leroy Gardner, 2011) for more information.

Figure 5.7: Column with intermediate elastic restraint

In gure 5.7, a) shows the deformation in braced rst mode and b) shows the defor-
mation in second mode, where the elastic restraint has sucient stiness.

With no stiness of the elastic spring, the column will buckle in rst mode. The
buckling capacity increases linearly with the elastic spring stiness and reaches the
buckling capacity of the second mode when KT = 16 · Ncr /L, the buckling capacity

80
5.5 Buckling of a intermediately restrained leg

is then four times the capacity of the rst mode. Further increase in stiness of KT
does not give additional buckling capacity, see gure 5.8.

Figure 5.8: Inuence of elastic spring stiness on buckling capacity

The critical load case for the leg in the tower test had high transverse loading acting
at the level of bridge and the earth-wire peak, it was combined with vertical loading
on the attachment points. The section loading on the legs can be described by
combination of three factors:

(i) Axial load in leg

(ii) Deformation of leg due to coercive forces from the secondary structure, leads
to moment in leg

(iii) Deformation of leg due to initial deformation, leads to moment in leg

Transverse wind on the tower legs is not included in the tower test loading. Accord-
ing to the designers, it was decided to skip it since it could have favorable inuence
on the buckling strength, i.e. it could increase the leg buckling capacity.

81
5 Guyed tubular tower

Figure 5.9: Three factors leading to section loading in the right leg

5.6 Analysis
For the analysis procedures, see chapter 2.12.8.

The calculated self-weight of the structure is increased by a factor of 1.05 to account


for coating, bolts and other items not modeled. The partial coecient factors used
in the design are γM 0 = γM 1 = 1.0

The right leg of the tower is the critical member of load case 8. Reported usage for
dierent methods and loading stages is calculated the critical section in the right
leg.

When calculating the usage, it is set equal to 1 when the stress at the outer edge
of a cross-section reaches yield strength. Stress above yield strength equals usage
above 1.

5.6.1 Buckling analysis


Buckling analysis was performed for each load case and corresponding buckling
shapes and buckling factors were found. Load case 8 had the lowest buckling factor.
In the full-scale test of the tower, load case 8 has multiple stages that should in
the end, cause the tower to buckle. These same stages are used in the analysis
comparison.

Figures 5.10 and 5.11 show the critical buckling shapes for all load cases in table
5.3.

82
5.6 Analysis

Table 5.5: Landsnet tower - Load case buckling factors


Buckling factors
No. Load case
Mode 1 Mode 2

1 Transverse wind 4.80 4.88


2 Maximum ice 6.70 6.81
3 Maximum ice no EW 8.50 8.67
4 Wind and ice no EW 4.27 4.36
5 *Cond. Break - left phase 24.36 24.38
6 *Cond. Break - center phase 24.82 25.29
7 *EW. Break - left 24.16 24.56
8 Wind and ice 3.32 3.40

*Abnormally high buckling factors, not used for further analysis.

Figure 5.10: Landsnet tower - Buckling modes, in top row modes 1 and 2, in bottom
row 3 and 4

5.6.2 Bracing stiness


For the leg to buckle in the second buckling mode, the stiness in the intermediate
elastic restraint needs to be at least

KT = 16 · Ncr /L = 243.3N/mm

Using a unit force, the bracing stiness was calculated in the transverse and longi-
tudinal directions for the structures initial conditions (where dead load and cable

83
5 Guyed tubular tower

Figure 5.11: Landsnet tower - Buckling modes, in top row modes 5 and 6, in bottom
row 7 and 8

pre-tension are in eect) and for the structure under loading from load case 8.

Initially the stiness is well above the ca. 240 N/mm limit, ca. 650 N/mm and 450
N/mm for the transverse and longitudinal directions respectively. After running
load case 8a and calculating the stiness again, it has dropped to about 400 N/mm
and 100 N/mm respectively.

The stiness provided by the secondary structure is therefore reduced by increased


load acting on the structure. The reason for this is not really clear. In the longitu-
dinal direction, this could be an eect caused by some guys slackening while others
tighten. For the transverse direction, the cause of this is not yet known.

In gure 5.12, the stiness in the secondary structure is decreasing from left to right.
The buckling mode in the rst (far left) tower is the second mode for a column, the
two towers in the middle the mode turns from the second mode to a braced rst
mode and nally to the rst buckling mode in the fourth (far right) tower.

5.6.3 Linear analysis


Structural analysis according to 5.2.2(3)c) in EN1993-1-1 (CEN, 2005) is performed
and the stability of the tower is calculated with equation 2.30.

84
5.6 Analysis

Figure 5.12: Landsnet tower - Secondary guy elastic modulus eect on the critical
buckling mode

Using the calculated values of axial force and bending moments from the linear
SAP2000 analysis results in an estimation of stability of the structure. The maxi-
mum usage for each load case is shown in table 5.6.

Figure 5.13: Landsnet tower - Linear analysis of load case 8a, a) Axial force NEd ,
b) Bending moment My,Ed

In table 5.6, the axial loads and bending moments in the right leg are shown and
corresponding usage values calculated with equation 2.30. For each load case, the
usage was calculated in every section of the leg prole and the maximum values
collected and displayed in the table. The moment shown is most likely coercive
moment mentioned in section 5.5 and gure 5.9.

The interaction factors were calculated conservatively. The moment calculated in


the tower leg is not very high and therefore their eect is not substantial. The usage

85
5 Guyed tubular tower

Table 5.6: Landsnet tower - Linear analysis, critical section usage


Load case NEd (kN ) My (kN m) Usage

8a -740.5 41.1 0.812


8b -762.5 42.6 0.838
8c -791.4 44.8 0.874
8d -816.8 46.7 0.905
8e -857.5 49.6 0.954
8f -888.2 52.7 0.995
8g -918.9 55.8 1.036
8h -949.4 58.9 1.076
8i -979.9 61.9 1.117
8j -1010.3 65.0 1.159
8k -1040.6 68.0 1.200

in the section from the linear analysis is perhaps rather low.

5.6.4 Non-linear analysis with sway and bow imperfections


To model the sway and bow imperfection, equivalent forces are used. To estimate
the size of these forces , sway and bow parameters and axial forces are needed.
The necessary parameters are calculated according to section 2.10. Their values are
shown in table 5.7. The axial forces used for the imperfection equivalent forces are
taken from the linear buckling analysis in section 5.6.3.

An example of an imperfect column loaded with equivalent forces is shown in gure


5.14. In part a) imperfect column, b) Sway forces, c) Bow forces.

Table 5.7: Landsnet tower - Sway and bow imperfection parameters


Sway imperfection h 26
m 2
φ0 0.0050
αh 0.6667
αm 0.8660
φ 0.0029

Bow imperfection Buckling curve c


elastic/plastic elastic
e0 /L 0.005

The equations for the imperfection equivalent forces for this tower are

86
5.6 Analysis

Figure 5.14: Landsnet tower - Modeling of sway and bow imperfection on the tower
leg

PSW AY = 2.9 · NEd (N )

PBOW = 20.0 · NEd (N )

qBOW = 3.05 · NEd (N/m)

The forces are added to both tower legs. Figure 5.15 shows the direction and shape
of the assumed deformations the equivalent forces are modeled after. The maximum
usage of the right leg is shown in table 5.8.

Figure 5.15: Landsnet tower - Sway and bow imperfection form

The sway imperfection alone does not capture the right imperfection of the structure.

87
5 Guyed tubular tower

Figure 5.16: Landsnet tower - Maximum stress for load case 8, a) Sway, b)
Sway+Bow, c) Sway-Bow

Table 5.8: Landsnet tower - Non-linear analysis with sway and bow imperfections,
critical section usage
Usage
Load case
φ φ+e φ−e
8a 0.453 0.801 0.779
8b 0.472 0.842 0.819
8c 0.499 0.900 0.876
8d 0.499 0.954 0.929
8e 0.563 1.049 1.021
8f 0.604 1.151 1.105
8g 0.604 1.172 1.124
8h 0.827 1.401 1.369
8i 0.841 1.428 1.396
8j 0.795 1.597 1.543
8k 0.856 1.759 1.759

For the inspected load case, the right leg is expected to fail in the second buckling
mode of a straight column. Using local bow imperfections captures this better.

The dierence between the usage of the two, opposite directed, bow imperfection
cases seems to be minimal. The usage is however much higher with this method
than in the linear analysis.

88
5.6 Analysis

5.6.5 Non-linear analysis with EUGLI imperfection


As stated in section 5.6.1, load case 8 has the lowest buckling factor.

Using results from the buckling analysis of load case 8 and a static nonlinear analysis
of the same load case, the amplitude of the ηinit,m imperfection can be calculated.
This is done in table 5.9. Buckling mode shape number 1 is shown in gure 5.17
and corresponding axial forces in the structure and bending moments necessary to
bend the structure into the shape of the critical buckling mode can be seen in gure
5.18.

Figure 5.17: Landsnet tower - Elastic critical buckling mode shape number 1 for load
case 8

Table 5.9: Landsnet tower - ηinit,m amplication factor calculation


Column properties A 8736 mm2
fy 355 MP a
α 0.49 -
Wel,m 742485 mm3
MRk,m 263.6 kN m
Analysis results αcr 3.32 -
|ηcr |max 57.57 mm
NEd,m 763.6 kN
Mη,cr,m 90.53 kN m
Calculated values λm 1.106 -
ηinit,m,max 60.82 mm

The model is deformed in the shape in gure 5.17 with the maximum displacement

89
5 Guyed tubular tower

Figure 5.18: Landsnet tower - Load case 8, a) Axial force NEd


II
, b) Bending moment
II
Mη,init

equal to the calculated value of ηinit,m,max in table 5.9. For each loading step shown
in table 5.4, the maximum usage reached in the right leg is calculated and displayed
in table 5.10.

Table 5.10: Landsnet tower - Non-linear analysis with EUGLI imperfection, critical
section usage
Usage Um
Load case
+ηinit,m −ηinit,m
8a 0.72 0.69
8b 0.77 0.73
8c 0.83 0.78
8d 0.87 0.83
8e 0.96 0.90
8f 1.03 0.97
8g 1.14 1.05
8h 1.37 1.17
8i 1.84 1.45
8j 1.87 1.65
8k 1.86 1.65

Analyzing the tower in SAP2000 was not straightforward. For load cases 8a to 8g,
the usage increase is steady and believable. At load cases 8h and above, the analysis

90
5.7 Results

had a hard time reaching the maximum loading. In the end the program could not
converge on a result for the highest loading values and stranded a few steps from the
maximum loading for each load case, increasing from 8h to 8k. The reason for this
is not completely clear. Perhaps the reducing stiness in the secondary structure
discussed in section 5.6.2 had an impact on this.

5.7 Results
Sections 5.6.3 - 5.6.5 show values of the estimated maximum section usage in the
tower per loading stage from table 5.4. The results are summed up in table 5.11.

Table 5.11: Landsnet tower - Critical section usage from dierent analysis methods
Load case Linear Sway & Bow EUGLI

8a 0.81 0.80 0.72


8b 0.84 0.84 0.77
8c 0.87 0.90 0.83
8d 0.90 0.95 0.87
8e 0.95 1.05 0.96
8f 0.99 1.15 1.03
8g 1.04 1.17 1.14
8h 1.08 1.40 1.37
8i 1.12 1.43 1.84
8j 1.16 1.60 1.87
8k 1.20 1.76 1.86

Figure 5.19: Landsnet tower - Usage comparison of dierent methods

91
5 Guyed tubular tower

As can be seen on gure 5.19, the results of the linear analysis are linear. The load
increase in table 5.3 is very close to linear, so is the response. It does probably not
capture the instability that the other methods suer from when analyzing the tower.

The EUGLI method gives lower usage than the linear analysis for load cases 8a-8d.
For case 8e, the usage is very close and just below 1. From there on, the EUGLI
method gives higher usage than the linear analysis. At load case 8h and 8i, the
usage from the EUGLI method jumps up and converges at approximately 1.85.

The Sway & Bow method gives slightly higher values than the EUGLI method, and
it has a similar climb up to case 8f. After that, the Sway & Bow usage continues to
climb, in an erratic way though.

For both the EUGLI and Sway & Bow method, the analysis in SAP was halted by
some numerical instability at the higher load cases. Denite reasons for this are not
known. It could very well have to do with the decreasing stiness at higher loading.
With decreasing stiness, the column stops heading for the second buckling mode
and starts heading for a braced version of the rst buckling mode, see gure 5.12.
When this occurs, the buckling capacity of the tower leg drops signicantly. The
movement of the tower leg when heading from buckling mode 2 to mode 1 would
most likely cause a snap-through eect in the leg. This could also be aecting the
analysis adversely.

5.8 Tower test


During May 9th and 10th, 2017, a full-scale tower test was performed in a Eucomsa
testing facility in Seville, Spain. The purpose was to inspect the overall performance
of the tower with special emphasis on the buckling resistance of the tower leg.
Therefore, the loading was tailored so that one of the tower legs would buckle and
the tower would collapse.

When this is written, a nal test report is not yet available. Therefore, verication
test of the true steel quality is not known. What is known is the test load at failure
and the assessment of inspectors that the production of the cold formed tubular legs
seemed very good. Initial deections were minimal, no welding was visible and the
cross section looked perfectly circular.

In the tower test, the tower buckled at load step 8k (the last step, see table 5.4). In
gure 5.20 the buckling of the tower is shown.

1. The top right picture is taken just before the tower starts buckling. Looking

92
5.8 Tower test

closely at the bracing between the legs, it can be seen that it is a bit bent.

2. The top left picture is taken just after the right leg of the tower (left leg in
the picture) starts buckling. The buckling occurs just above a connection in
the upper part of the leg. After the compression side buckles, the bolts on the
tension side of the leg break. The bend in the bracing between the legs is also
quite clear.

3. The lower left picture shows the leg continuing to buckle and that the bracing
between the legs has fully buckled.

4. The lower right picture the tower has completely failed. The cross-arm is
broken and the connection between the buckled leg and the cross-arm also
breaks.

93
5 Guyed tubular tower

Figure 5.20: Landsnet tower - Buckling of the full-scale tower test (courtesy of
Landsnet)

94
6 Concluding remarks
Drawing some conclusions from this thesis, the starting point would be the three
dierent methods used to estimate the stability of columns under combined axial
load and bending moment. From the results of the analyses, all the methods give a
good indication of the stability and usage in the structures analyzed. To compare
the results of the dierent methods, understanding of the basis of each method is
necessary.

Using sway and bow imperfection equivalent forces can be complex. According to
the method, the structure should be inspected with the loading acting in all relevant
horizontal direction. For transmission towers, many load cases acting in all direc-
tions, estimating these forces and modeling them could become a time consuming
task and with many possibilities for errors.

The EUGLI imperfection is simpler and a more straightforward method. It could


be made easier by creating a programming routine in the SAP2000 OAPI that could
calculate the amplication factor of the imperfection. From the results in the thesis,
the method is giving good results for simple buckling shapes and usage close to 1.
For unstable structures, such as the Landsnet tower with high loading, it is not clear
what the method produces. It would be interesting to run a series of increasing loads
on a more stable structure and see the results.

The analysis of the Statnett tower gave relatively good results with the two load
cases selected. Using the EUGLI method for the Statnett tower is much simpler
than the sway and bow imperfections, because modeling the bow forces can require
a lot of dierent loads for a tower with so many elements.

The Landsnet tower is a much simpler structure. In the analysis, it was discovered
that something strange was happening to the stiness in the secondary structure
of the tower and further analysis is needed to get some denitive answers. Perhaps
this could be a project for someone in the future.

SAP2000 was the chosen analysis program and it was adequate for the modeling and
analysis in the thesis. Its user interface could use some work but the data extraction
works very well. Modeling EUGLI imperfections is relatively easy in the program
by modeling the analyzed structure after a specic buckling mode with a scaled

95
6 Concluding remarks

factor. Linear analysis is also very easily set up in SAP2000. For the sway and bow
imperfection forces, some handiwork is required to set up the correct forces.

96
Bibliography
Baldvin Einarsson. 2005. Súlur. Lecture notes for graduate course Steel Structures
1 in the University of Iceland.

Boissonnade, N., & European Convention for Constructional Steelwork Technical


Rules for Member Stability in EN 1993-1-
Committee Structural Stability. 2006.
1: Background Documentation and Design Guidelines. ECCS design guides and
european recommendations. ECCS.

Brian Kirke, & Iyad Hassan Al Jamel. 2004. Steel Structures Design Manual to AS
4100. First Edition edn.

CEN. 2005. Euocode 3: Design of steel structures - Part 1-1: General rules and rules
for buildings. Tech. rept. European Committee for Standardization, Brussels.

CEN. 2007. Eurocode 9: Design of aluminnium structures - Part-1-1: General


structural rules. Tech. rept. European Committee for Standardization, Brussels.

Computers & Structures Inc. 2015 (July). SAP2000 v18 CSI Analysis Reference
Manual. Berkley, California, USA.

Ed Akin, J. 2009. Finite Element Analysis Concepts via SolidWorks. [Online;


accessed 20-April-2017].

Eugen Chladný, & Magdalena ’tujberová. 2013. Frames with unique global and local
imperfection in the shape of the elastic buckling mode (Part 1). Der Stahlbau,
82(8), 609617.
Eurocodes, Building the future. 2015. Apollo bridge picture. [Online; accessed 30-
May-2017].

Ivan Baláº, & Yvona Koleková. 2012. Structures with UGLI imperfections. Pages
6186 of: 18th International Conference Engineering Mechanics 2012.

Jason Ericksen. 2011. A How-To Approach to Notional Loads. Modern Steel Con-
struction, January.

Jon A. Schmidt. 1991. Design of Steel Columns in Unbraced Frames Using Notional
Loads. Practice Periodical on Structural Design and Construction, February.

97
BIBLIOGRAPHY

Jørgen Amdahl. 2014. TMR4205 Buckling and Ultimate Strength of Maring Struc-
tures, Chapter 2: Buckling of Bars and Frames. University lecture notes.

Lawrence Martin, & John Purkiss. 2008. Structural Design of Steelwork to EN 1993
and EN 1994. Third edn. Linacre House, Jordan Hill, Oxford OX2 8DP, UK:
Butterworth-Heinemann.

Leroy Gardner. 2011. Stability of Steel Beams and Columns. Silwood Park, Ascot,
Berkshire, SL5 7QN UK: SCI.

Leroy Gardner, & David A. Nethercot. 2011. Designers' guide to Eurocode 3: Design
of steel buildings. Second edn. 40 Marsh Wall, London E14 9TP: ICE Publishing.

Magdi51. 2012. Pentele bridge picture. [Online; accessed 30-May-2017].

Marian Dallemule. 2015. Equivalent imperfections in arched structures. Slovak


Journal of Civil Engineering, 23(3), 915.

Primoº Moºe. 2015. ESDEP Course. Teaching material [Online; accessed 31-May-
2017].

Rangachari Narayanan, V.Kalyanaraman, Santhakumar, A.R., S.Seetharaman,


S.R.Satish Kumar, S.Arul Jayachandran, & R.Senthil. 2012. Introduction to Col-
umn Buckling. Teaching material [Online, accessed 15-April-2017].

SolidWorks. 2008. Understanding Nonlinear Analysis. [Online; accessed 20-April-


2017].

Statnett. 2017. About Statnett. [Online; accessed 20-05-2017].

Vladislav F. Demenko. 2015. Lecture 25 Buckling of Columns (Part 1) Elastic Colmn


Behavior. [Online; accessed 20-February-2017].

Wikipedia. 2017a. Deformation (engineering)  Wikipedia, The Free Encyclopedia.


[Online; accessed 20-April-2017].

Wikipedia. 2017b. Plasticity (physics)  Wikipedia, The Free Encyclopedia. [Online;


accessed 23-May-2017].

98

Potrebbero piacerti anche