Professional Documents
Culture Documents
CABLE-STAYED BRIDGES
BY
FUZZY - FINITE ELEMENT MODELLING
By
A Thesis Submitted in
Fulfilment of the requirements
for the award of the Degree of
DOCTOR OF PHILOSOPHY
This thesis or part there of, has not been submitted to any other University or
I do hereby declare that the research work reported in this thesis has been
University, Hyderabad for Ph.D. degree in Civil Engineering and that this thesis,
neither in full nor in part, has been submitted for the award of any Degree or Diploma
Date: 16-03-2004
Place: Hyderabad.
ACKNOWLEDGEMNTS
indulgence, academic and emotional support. I am grateful to him for his guidance
and suggestions from time to time throughout the progress of the present work.
I thank Prof. L.V.A. Sesha Sayi, Head, Department of Civil Engineering, University
Technology, Kanpur for his valuable advice and help throughout the course of my
work and particularly for the help and support extended to me in my literature survey.
Savannah, USA and Dr. Andrzej Pownuk, Chair of Theoretical Mechanics, Faculty of
iv
I thank the Management and the Principal of Vasavi College of Engineering,
Hyderabad for the help and support extended to me through the staff development
I would like to express gratitude to Prof. E.S.Rao profusely for his valuable advice
and support throughout the course of my work. I also thank him for his careful
colleagues Sri. M.Bhasker, Sri. B.Veeranna and Sri. K.V.Ramana Reddy for their
encouragement and support. I thank Mr. M.V.Suresh for his help in typing the
manuscript.
I am thankful to my mother Mrs. Vijaya Lakshmi, brother Dr. M.V.Krishna Rao and
my wife Prameela Rani for their help and support. I particularly appreciate the
v
ABSTRACT
accurately in order to predict their behaviour during their design life period.
Normally, various parameters of a structure that influence its behaviour, are identified
and are incorporated in a mathematical model of the structure. Classical finite element
analysis is performed using the mathematical model under the influence of the given
loads and forces are evaluated. However, uncertainties associated with the material
and geometric properties of a real life structure as well as the range of service loads
acting on it could not be accounted. Classical finite element analysis, despite its
advantages, is not well suited to handle such uncertainties in the structural parameters
and response quantities i.e. material and geometric properties, loads, displacements,
moments and forces. Thus there is a need to incorporate the effect of these
and independent variation of uncertainties of material property (E), live load and mass
Fuzzy finite element analysis has not been hitherto applied to the study of
complex structures including cable-stayed bridges. In the light of the above, it is felt
that there is a need to apply fuzzy finite element methodology to evaluate the effect of
vi
of its behaviour. Also, there is a need to study the effect of these uncertainties on the
post-process the solution in order to obtain axial forces, bending moments and shear
mass and material uncertainties subjected to fuzzy interval loading is evaluated. The
preliminary case studies and the results are validated with the available literature. The
structures response of beams, trusses and frames under fuzzy interval loads is
evaluated and the structures are observed to behave uniquely (Appendix C).
finite element analysis in order to evaluate the effect of multiple uncertainties on the
structural response. The nature of the fuzzy membership functions of the structural
a given node as well as over a set of given nodes. The sensitivity (relative variation
vii
of uncertainty) of structural response quantities to unit change in variation of load and
The present work demonstrates the effectiveness of fuzzy finite element model
in evaluating the structural response in the presence of multiple uncertainties over the
viii
CONTENTS
Certificate ii
Declaration iii
Acknowledgements iv
Abstract vi
Contents ix
Notation xiii
List of Tables xvi
List of Figures xix
List of Tables in Appendices xxiv
List of Figures in Appendices xxvi
ix
2.2 Assembled Finite Element Model 34
2.2.1 Uncertainty of live load 35
2.2.2 Post-processing of solution 37
2.2.3 Uncertainty of mass density 38
2.3 Material Uncertainty 38
2.4 Sources of Overestimation 39
2.4.1 Evaluation of overall displacement vector 42
2.5 Concomitant Presence of Load and Material Uncertainties 43
2.5.1 Cumulative effect of material and load uncertainties (α and β) 45
2.5.2 Evaluation of {λ}cc from assembled FEA model 46
2.5.3 Solution of linear interval matrix equations 47
2.6 Post-processing of Solution 48
2.7 Concomitant Presence of Mass Density and Material Uncertainties 49
2.8 Concomitant Presence of Mass density and Material and Live Load 50
Uncertainties
2.9 Sensitivity Analysis 50
2.10 Summary 51
Chapter 3 Case studies
3.1 Introduction 54
3.2 Case Study 1 – Cable-Stayed Bridge with Uncertainties of Material 57
Property and Mass Density
3.3 Case Study 2 – Cable-Stayed Bridge with Uncertainties of Material 59
Property and Live Load
3.4 Case Study 3–Cable-Stayed Bridge with Uncertainties of Material 61
Property, Live Load and Mass Density
3.5 Cumulative Effect of Multiple Uncertainties α, β and γ on Displacements 62
and Forces
3.6 Sensitivity Analysis 63
3.7 Summary 64
Chapter 4 Results and discussion
4. Results and discussion 71
4.1 Case study 1 – Cable-Stayed Bridge with Uncertainties of Material 71
Property and Mass Density
x
4.1.1 Effect of concomitant variation of α and γ on displacements and rotations 72
4.1.1.1 Uncertain horizontal displacement at node 2 73
4.1.1.2 Uncertain vertical displacement at node 3 73
4.1.1.3 Uncertain rotation at node 4 73
4.1.2 Effect of concomitant variation of α and γ on shear forces and bending 74
moments
4.1.2.1 Axial force in deck in element 9 75
4.1.2.2 Axial force (kN) in deck in element 12 75
4.1.2.3 Shear force (kN) in deck just to the left of node 3 76
4.1.2.4 Bending moment (kNm) in deck at node 4 76
4.1.2.5 Axial force (kN) in pylon in element 17 (at node 11) 77
4.1.2.6 Axial force (kN) in cable 2 (at node 2) 77
4.1.2.7 Axial force (kN) in cable 3 (at node 3) 77
4.2 Case study 2 – Cable-Stayed Bridge with uncertainties of material 93
property and live load.
4.2.1 Effect of concomitant variation of α and β on displacements and rotations 93
4.2.1.1 Uncertain horizontal displacement at node 2 93
4.2.1.2 Uncertain vertical displacement at node 3 94
4.2.1.3 Uncertain rotation at node 4 94
4.2.2 Effect of concomitant variation of α and β on shear forces and bending 95
moments
4.2.2.1 Axial force (kN) in deck in element 9 95
4.2.2.2 Axial force (kN) in deck in element 12 96
4.2.2.3 Shear force (kN) in deck just to the left of node 3 96
4.2.2.4 Bending moment (kNm) in deck at node 4 97
4.2.2.5 Axial force (kN) in pylon in element 17 (at node 11) 97
4.2.2.6 Axial force (kN) in cable 2 (at node 2) 98
4.2.2.7 Axial force (kN) in cable 3 (at node 3) 98
4.3 Case Study 3 – Cable-Stayed Bridge with Uncertainties of Material 113
Property, Live Load and Mass Density
4.3.1 Effect of concomitant variation of α and β on displacements and rotations 113
4.3.1.1 Uncertain horizontal displacement at node 2 114
4.3.1.2 Uncertain vertical displacement at node 3 115
xi
4.3.1.3 Uncertain rotation at node 4 115
4.3.2 Effect of concomitant variation of α and β on shear forces and bending 116
moments
4.3.2.1 Axial force (kN) in deck in element 9 117
4.3.2.2 Axial force (kN) in deck in element 12 118
4.3.2.3 Shear force (kN) in deck just to the left of node 3 118
4.3.2.4 Bending moment (kNm) in deck at node 4 119
4.3.2.5 Axial force (kN) in pylon in element 17 (at node 11) 120
4.3.2.6 Axial force (kN) in cable 2 (at node 2) 121
4.3.2.7 Axial force (kN) in cable 3 (at node 3) 122
4.4 Sensitivity Analysis 151
4.4.1 Sensitivity analysis of displacements and rotations at a given node 151
4.4.2 Sensitivity analysis of forces and moments at a given node 152
4.4.3 Sensitivity analysis of a given response quantity at different nodes 152
4.5 Discussion 159
Chapter 5 Conclusions and Recommendations for Future Work
5.1 Conclusions 161
5.2 Recommendations for Future Work 163
References 164
Appendices Appendix- ‘A’ Fuzzy Sets and Membership Functions 169
Appendix- ‘B’ Fuzzy Finite Element Model- Muhanna’s Approach 174
Appendix- ‘C’ Preliminary Case Studies 176
Bio-Data 197
Publications 198
xii
NOTATION
βl, βu : Lower and upper bounds of β-cut on membership function for live load
{λαβ} : Fuzzy interval vector of internal forces under the influence of multiple
uncertainties α and β
{λcβ},{λβ} : Fuzzy interval vector of internal forces under the influence of live load
uncertainty β alone
{λ} α c : Fuzzy interval vector of internal forces under the influence of material
uncertainty β alone
{λαβ}(e) : Fuzzy interval vector of internal forces under the influence of multiple
uncertainties α and β at element level
{λ}(e) : Overall internal force vector for the element
model
xiii
A1, A2 : Axial Forces at the ends of a plane frame element
E : Young’s modulus
xiv
{P} : Point interval load vector
xv
LIST OF TABLES
xvi
Table. 4.2.3 Cable stayed Bridge - Concomitant Variation of Rotation at node 4 99
(×10-3 radians) w.r.t α and β
Table. 4.2.4 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 100
deck in element 9 w.r.t α and β
Table. 4.2.5 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 100
deck in element 12 w.r.t α and β
Table. 4.2.6 Cable stayed Bridge - Concomitant Variation of Shear Force (kN) in 100
deck just to the left of node 3 w.r.t α and β
Table. 4.2.7 Cable stayed Bridge - Concomitant Variation of Bending Moment 101
(kNm) in deck at node 4 w.r.t α and β
Table. 4.2.8 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 101
pylon in element 17 w.r.t α and β
Table. 4.2.9 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 102
cable 2 w.r.t α and β
Table. 4.2.10 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 102
cable 3 w.r.t α and β
Table. 4.3.1 Cable stayed Bridge - Concomitant Variation of Horizontal 124
displacement of node 2 (×10-4 metres) w.r.t α and β at γ =1.0
Table. 4.3.2 Cable stayed Bridge - Concomitant Variation of Horizontal 124
displacement of node 2 (×10-4 metres) w.r.t α and β at γ =0.8
Table. 4.3.3 Cable stayed Bridge - Concomitant Variation of Vertical 124
displacement of node 3 (×10-2 metres) w.r.t α and β at γ =1.0
Table. 4.3.4 Cable stayed Bridge - Concomitant Variation of Vertical 125
displacement of node 3 (×10-2 metres) w.r.t α and β at γ =0.8
Table. 4.3.5 Cable stayed Bridge - Concomitant Variation of Rotation at node 4 125
(×10-3 radians) w.r.t α and β at γ =1.0
Table. 4.3.6 Cable stayed Bridge - Concomitant Variation of Rotation at node 4 125
(×10-3 radians) w.r.t α and β at γ =0.8
Table. 4.3.7 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 126
deck in element 9 w.r.t α and β at γ =1.0
Table. 4.3.8 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 126
deck in element 9 w.r.t α and β at γ =0.8
xvii
Table. 4.3.9 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 126
deck in element 12 w.r.t α and β at γ =1.0
Table. 4.3.10 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 127
deck in element 12 w.r.t α and β at γ =0.8
Table. 4.3.11 Cable stayed Bridge - Concomitant Variation of Shear Force (kN) in 127
deck just to the left of node 3 w.r.t α and β at γ =1.0
Table. 4.3.12 Cable stayed Bridge - Concomitant Variation of Shear Force (kN) in 127
deck just to the left of node 3 w.r.t α and β at γ =0.8
Table. 4.3.13 Cable stayed Bridge - Concomitant Variation of Bending Moment 128
(kNm) in deck at node 4 w.r.t α and β at γ =1.0
Table. 4.3.14 Cable stayed Bridge - Concomitant Variation of Bending Moment 128
(kNm) in deck at node 4 w.r.t α and β at γ =0.8
Table. 4.3.15 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 128
pylon in element 17 w.r.t α and β at γ =1.0
Table. 4.3.16 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 129
pylon in element 17 w.r.t α and β at γ =0.8
Table. 4.3.17 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 129
cable 2 w.r.t α and β at γ =1.0
Table. 4.3.18 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 129
cable 2 w.r.t α and β at γ =0.8
Table. 4.3.19 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 130
cable 3 w.r.t α and β at γ =1.0
Table. 4.3.20 Cable stayed Bridge - Concomitant Variation of Axial Force (kN) in 130
cable 3 w.r.t α and β at γ =0.8
xviii
LIST OF FIGURES
xix
Fig. 4.1.6 Membership Functions of Shear Force in deck to left of node 3 at (a) 88
γ=1.0, (b) γ=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.1.7 Membership Functions of Bending Moment in deck at node 4 at (a) 89
γ=1.0, (b) γ=0.8, (c) α=1.0, (d) α=0.6
Fig. 4.1.8 Membership Functions of Axial Force in pylon in element 17 at (a) 90
γ=1.0, (b) γ=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.1.9 Membership Functions of Axial Force in cable 2 at (a) γ=1.0, (b) 91
γ=0.8, (c) α=1.0, (d) α=0.6
Fig. 4.1.10 Membership Functions of Axial Force in cable 2 at (a) γ=1.0, at (a) 92
γ=1.0, (b) γ=0.8, (c) α=1.0, (d) α=0.6
Fig. 4.2.1 Membership Functions for horizontal displacement at node 2 at (a) 103
β=1.0, (b) β=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.2.2 Membership Functions for vertical displacement at node 3 at (a) β=1.0, 104
(b) β=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.2.3 Membership Functions for rotation at node 4 at (a) β=1.0, (b) β=0.8, 105
(c) α=1.0, (d) α=0.8
Fig. 4.2.4 Membership Functions for Axial Force in deck in element 9 at 106
(a) β=1.0, (b) β=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.2.5 Membership Functions for Axial Force in deck in element 12 at 107
(a) β=1.0, (b) β=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.2.6 Membership Functions for Shear Force in deck to left of node 3 at 108
(a) β=1.0, (b) β=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.2.7 Membership Functions for Bending Moment in deck at node 4 at 109
(a) β=1.0, (b) β=0.8, (c) α=1.0, (d) α=0.6
Fig. 4.2.8 Membership Functions for Axial Force in pylon in element 17 at 110
(a) β=1.0, (b) β=0.8, (c) α=1.0, (d) α=0.8
Fig. 4.2.9 Membership Functions for Axial Force in cable 2 at (a) β=1.0, 111
(b) β=0.8, (c) α=1.0, (d) α=0.6
Fig. 4.2.10 Membership Functions for Axial Force in cable 3 at (a) β=1.0, 112
(b) β=0.8, (c) α=1.0, (d) α=0.6
xx
Fig .4.3.1 Membership Functions for horizontal displacement at node 2 at 131
(a) β=1.0, γ =1.0, (b) β=0.8, γ =1.0, (c) α=1.0, γ =1.0 (d) α=0.6, γ =1.0
Fig. 4.3.3 Membership Functions for vertical displacement at node 3 at (a) β=1.0, 133
γ =1.0, (b) β=0.8, γ =1.0, (c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.4 Membership Functions for vertical displacement at node 3 at (a) β=1.0, 134
γ =0.8, (b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8, (d) α=0.6, γ =0.8
Fig. 4.3.5 Membership Functions for rotation at node 4 at (a) β=1.0, γ =1.0, 135
(b) β=0.8, γ =1.0, (c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.6 Membership Functions for rotation at node 4 at (a) β=1.0, γ =0.8, 136
(b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8 (d) α=0.6, γ =0.8
Fig. 4.3.7 Membership Functions for Axial Force in deck in element 9 at 137
(a) β=1.0, γ =1.0, (b) β=0.8, γ =1.0,(c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.8 Membership Functions for Axial Force in deck in element 9 at 138
(a) β=1.0, γ =0.8, (b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8 (d) α=0.6, γ =0.8
Fig. 4.3.9 Membership Functions for Axial Force in deck in element 12 at 139
(a) β=1.0, γ =1.0, (b) β=0.8, γ =1.0,(c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.10 Membership Functions for Axial Force in deck in element 12 at 140
(a) β=1.0, γ =0.8, (b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8 (d) α=0.6, γ =0.8
Fig. 4.3.11 Membership Functions for Shear Force in deck to left of node 3 at 141
(a) β=1.0, γ =1.0, (b) β=0.8, γ =1.0,(c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
xxi
Fig. 4.3.12 Membership Functions for Shear Force in deck to left of node 3 at 142
(a) β=1.0, γ =0.8, (b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8 (d) α=0.6, γ =0.8
Fig. 4.3.13 Membership Functions for Bending moment at node 4 at (a) β=1.0, 143
γ =1.0, (b) β=0.8, γ =1.0,(c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.14 Membership Functions for Bending moment at node 4 at (a) β=1.0, 144
γ =0.8, (b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8 (d) α=0.6, γ =0.8
Fig. 4.3.15 Membership Functions for Axial Force in pylon in element 17 at 145
(a) β=1.0, γ =1.0, (b) β=0.8, γ =1.0, (c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.16 Membership Functions for Axial Force in pylon in element 17 at 146
(a) β=1.0, γ =0.8, (b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8 (d) α=0.6, γ =0.8
Fig. 4.3.17 Membership Functions for Axial Force in cable 2 at (a) β=1.0, γ =1.0, 147
(b) β=0.8, γ =1.0,(c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.18 Membership Functions for Axial Force in cable 2 at (a) β=1.0, γ =0.8, 148
(b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8, (d) α=0.6, γ =0.8
Fig .4.3.19 Membership Functions for Axial Force in cable 3 at (a) β=1.0, γ =1.0, 149
(b) β=0.8, γ =1.0, (c) α=1.0, γ =1.0, (d) α=0.6, γ =1.0
Fig. 4.3.20 Membership Functions for Axial Force in cable 3 at (a) β=1.0, γ =0.8, 150
(b) β=0.8, γ =0.8, (c) α=1.0, γ =0.8, (d) α=0.6, γ =0.8
Fig. 4.4.1 Sensitivity Analysis for displacements (for dead load) at node 3 at 154
γ=1.0
Fig. 4.4.2 Sensitivity Analysis for displacements (for dead load) at node 3 at 154
γ = 0.8
xxii
Fig. 4.4.3 Sensitivity of displacements at node 3 in deck at α=1.0 155
Fig. 4.4.4 Sensitivity Analysis of Forces and Moments in Element 12 (at node 5) 155
at γ = 1.0
Fig. 4.4.5 Sensitivity Analysis of Forces and Moments in Element 12 (at node 5) 156
at γ = 0.8
Fig. 4.4.6 Sensitivity Analysis for Bending Moment (due to dead load) in deck 156
slab at γ =1.0
Fig. 4.4.7 Sensitivity Analysis for Axial Force (due to live load) in deck slab at 157
β =1.0
Fig. 4.4.8 Sensitivity Analysis for Axial Force (due to live load) in deck slab at 157
β =0.8
Fig. 4.4.9 Sensitivity Analysis for Shear Force (due to live load) in deck slab at 158
β =1.0
xxiii
LIST OF TABLES IN APPENDICES
xxiv
Table. C-18 Plane Frame - Concomitant Variation of Rotation at node 3 (×10-3 194
radians) w.r.t α and β
Table. C-19 Plane Frame-Concomitant Variation of Bending Moment (kNm) at 194
node 2 w.r.t α and β
Table. C-20 Plane Frame - Concomitant Variation of Axial Force (N) in member 194
1 w.r.t α and β
xxv
LIST OF FIGURES IN APPENDICES
xxvi
Fig. C-21 Plane Frame 195
Fig. C-22 Plane Frame – Membership Function for Horizontal Displacement at 195
node 2 at β=1.0
Fig. C-23 Plane Frame – Membership Function for Vertical Displacement at node 195
3 at β=0.8
Fig. C-24 Plane Frame – Membership Function for Rotation at node 3 at α=0.6 195
Fig. C-25 Plane Frame – Membership Function for Bending Moment at node 2 at 196
β=1.0
Fig. C-26 Plane Frame – Membership Function for Shear Force just to the left of 196
at node 3 at α=1.0
Fig. C-27 Plane Frame – Membership Function for Axial Force in Member 1 at 196
β=0.8
Fig. C-28 Plane Frame – Membership Function for Axial Force in Member 1 at 196
α=0.6
xxvii
Chapter 1
Introduction
and
Literature Review
Chapter 1
1.1 Introduction
Civil Engineering. Modern day structures are usually complex in geometry and are
missiles, nuclear power reactors and dams are of strategic importance. The failure of
structures do not fail during their intended design life period, proper analysis and
design are mandatory. Design of a structural system requires that the performance of
the system be guaranteed over its lifetime. Accordingly, the objective of structural
analysis is to predict the behaviour of the structure during its estimated life period
accurately.
moment distribution, Kani’s method etc. were widely used. However, with the
also underwent a sea change. Matrix methods of structural analysis came to be used
widely with a great degree of computational precision coupled with high speed.
2
Classical finite element analysis is presently the most popular mathematical
tool for the analysis of structures. Finite element analysis attempts to construct a
identified and used in constructing a mathematical model for the structure. The
conditions and applied loads. The mathematical model constructed could be analysed
for the given structural configuration under the influence of the given loads. The
parameters used in generating the mathematical model are normally crisp and certain
closely corresponds to the behaviour of the actual structure. Any variation in the
response of the structure predicted using the mathematical model and the response of
the physical structure could be due to the following factors (Muhanna and Mullen,
1999):
The errors due to above uncertainties can neither be handled nor eliminated by
the use of classical finite element analysis. Therefore, there seems to be a lot of
parameters in the analysis. Thus, there is a need to improve the existing mathematical
model of the structure in order to bring its behaviour as close as possible to the actual
3
The uncertainty introduced in the mathematical model of the structure needs to
be reflected in the method of analysis and its output as well. This requires the
redefinition and extension of the classical finite element model to a fuzzy finite
element model, which allows the use of fuzzy interval variables in order to account
and features which are vague, imprecise, qualitative, linguistic or incomplete (Rao
One of the major difficulties a designer faces is that both the external demands
of the system and its manufacturing variations are not known exactly. In order to
overcome this uncertainty, the designer requires to provide wider allowance and
devise a conservative design for the system. As new analysis tools continue to be
4
developed, the predictive skills of the designer become sharper. The demands of the
market require more efficient designs and innovations (Muhanna and Mullen, 1999).
uncertainties.
Classical finite element analysis, despite its advantages, is not suited to handle
uncertainties in the design variables of the structural system i.e. displacements, loads,
bending moments and shear forces. All structures in reality possess physical and
the analysis and the output generated is crisp and non-fuzzy in nature. Thus, idealised
classical finite element model for analysis and design does not truly represent the
to evolve a refined and reliable model. Thus, there is a need for an uncertainty-based
cancellation, and range of tolerance are among the main sources of overestimation in
interval arithmetic as they can lead to catastrophic results. (Muhanna and Mullen,
2001)
5
It may thus be surmised that any uncertainty introduced in the mathematical
model requires to be reflected in the analytical approach as well as in the final result.
This classical finite element model needs to be reformulated to include the concept of
uncertainty.
system, three different approaches are used for its analysis. (Rao and Sawyer, 1995)
the use of possibility theory based on the theory of fuzzy sets. If the
1.2.2.1.Anti-optimisation
In this approach (McWilliam, 2001), the least favourable response under the
this purpose, which relies upon knowledge of the bounds of uncertainty. In practical
applications of this approach, it is usual to use (elliptical) convex sets to model the
6
uncertain phenomena; while more recent work has used interval sets. This method
takes some account of the interactions between uncertain parameters in the stiffness
matrix and force vector. As the uncertain parameters are varied between predefined
upper and lower bounds, the displacement of the structure in each degree of freedom
varies, according to the values of the uncertain parameters, between certain upper and
Probability theory has a long history that can be traced back to the work of
much more recent endeavour. The use of finite element methods for random variables
has been developed over the last two decades. Stochastic finite elements have been
second-order methods (FORM and SORM) and are applied to both static and transient
stochastic finite elements have been developed (Muhanna and Mullen, 1999).
Although there is no suitable technique available for the analysis of all types
of imprecision, the stochastic finite element method can be used to handle uncertain
probabilistic finite element method was developed in the 1980s to account for
applied loads. The uncertain variables are spatially distributed over the region of the
7
Underlying the work of probabilistic analysis of uncertain data is the existence
functions have small but non-zero values for all ranges of dependent values (e.g.
that any outcome is possible, albeit with a very small chance. The design and
determination of the reliability of system using probability methods often use the
region of very low probabilities. In most cases, it is the region of the probability
density function that is the least accurately assessed (Muhanna and Mullen, 1999).
structural parameters and loads is scanty, there has been increased interest in the
application of models of uncertainty that need not depend upon such detailed
knowledge.
calculations is the use of possibility theory based on the theory of fuzzy sets
(Muhanna and Mullen, 1999). The fuzzy or imprecise information may be present in
structural system. In the traditional (deterministic) finite element approach, all the
parameters of the system are taken to be precisely known (Akpan et al, 2001). In
practice, however, there is always some degree of uncertainty associated with the
8
actual values for structural parameters and applied loads. In addition, the pattern of
from above and below, and can be considered to be defined with in envelope bounds.
As a consequence of this, the response of the structure will always exhibit some
values are known to lie between two limits, but the exact values are unknown. The
might be to consider the values of unknown variables, which are defined within
intervals that possess known bounds or limits (Muhanna and Mullen, 2001).
quantities, displacements, internal forces, and stresses under uncertain loads in the
imprecisely defined systems using fuzzy logic. Uncertainty in the input data as well
terms in the definition of design variables of the problem. Use of fuzzy logic in order
Concerted efforts were made since then to handle uncertainty in engineering problems
systems and also service loads to which the structures are exposed to during their
design life period. In order to understand the applicability of fuzzy logic to solve
9
fuzzy logic and fuzzy set theory. Some fundamentals of fuzzy set theory are
presented in Appendix A.
numbers and their relation with fuzzy logic are described in Appendix-A
provides a link between design and analysis where uncertainty may be represented by
software packages developed in the past decade. Comparing with traditional floating-
1. Computational results obtained with interval arithmetic are reliable since there
is no loss of information.
interval analysis.
Interval algebra has been applied to develop several methods for the solution
of linear interval systems with a given interval vector on right hand side. Among
them is the inclusion theory developed by Gay (1982), Neumaier (1987, 1989, 1990),
Jansson (1990) and Rump (1990). In their theories, the exact hull for the interval
solution set of the linear interval system is found to be sharply bounded by outer and
inner estimates.
10
1.2.4 Fuzzy Finite Element Model of a Structure
The uncertainty model for analysis and design is constructed by combining the
concepts of fuzzy interval numbers and classical finite element analysis. In traditional
finite element methods, the development of the element stiffness matrices and load
vectors requires the evaluation of integrals over the domain or region of the element.
The resulting assembled finite element equations are to be solved using a suitable
technique such as Guassian Elimination or one of its variants (Rao and Sawyer,
1995).
equations are to be redefined to handle fuzzy information. All of these theoretical and
computational aspects can be developed starting from the basic definitions of a fuzzy
quantity and fuzzy arithmetic operations. In particular, the system of linear interval
matrix equations that arise out of the fuzzy integrals is solved using several
Fuzzy finite element analysis is of recent origin. The studies made in the field
of fuzzy finite element analysis by various researchers are set out in the succeeding
modelling using fuzzy finite element analysis has been made. The limitations of the
existing methodologies are identified and the need for further improvement is
underlined. The causes and the nature of various uncertainties are also identified and
discussed.
11
1.2.5 State of the Art
Mid-nineties can be considered as the period when the main activities for
Computational Mechanics and when the efforts for development of interval finite
utilizing finite element method to deal with pattern loading and structural
elastic beam. This yielded the elements of the stiffness matrix with uncertainties, to
be defined in bounded intervals. Then, the response quantities become uncertain, yet
bounded. The solutions for the system in the form of linear interval equations utilized
the triangle inequalities and linear programming. The results were conservative
within the bounds for the response quantities. Also, the problem of pattern-loading
was solved using this approach. Sample frames were worked out to illustrate the
theory and to examine the sharpness of the bounds in the studied examples, for which
finite elements as applied to the problem of shear frames with uncertain material
and interval methods to quantify the uncertainties in the response quantities. A shear
frame of multi-degree of freedom was considered, and the bending rigidities of the
storey columns were taken as uncertain fields. The uncertain bending rigidity of each
12
beam-column element is modelled using two methods. One is a probabilistic method,
known first order and second order statistical information. The other method is a set-
bounded from above and below. Galerkin finite elements with cubic deterministic
Rao, S.S. and Sawyer, J.P. (1995), Rao, S.S. and Berke L (1997), and Rao,
S.S. and Li Chen (1998) have developed different versions of interval based finite
Rao, S.S. and Sawyer, J.P. (1995) developed a fuzzy finite element approach
from the basic concepts of fuzzy numbers in fuzzy arithmetic and implemented
suitably defined concepts of fuzzy calculus and a solution to finite element equations.
properties, external loads, and boundary conditions were solved to establish and
illustrate the new procedure. The approach developed is applicable to systems that
are defined in linguistic terms as well as those that are described by incomplete
Rao, S.S. and Berke L. (1997) made a study of analysis of uncertain structural
systems using interval analysis. This work considered the modelling of uncertain
the form of a system of linear interval equations. In addition to the direct and
13
Guassian-elimination based methods, a combinatorial approach (based on an
base method are presented for finding the solution of interval equations. The range or
interval of the solution vector was found to increase with the increasing size of the
examples were presented to illustrate the computational aspects of the methods and
also to indicate the importance of the truncation approach in practical problems. The
structures is underlined.
A sincere effort was made by Rao, S.S. and Li Chen (1998) to develop a new
algorithm for solution of linear interval equations. The algorithm used search-based
operations were needed to obtain the interval solution, where 128 operations were
required by the combinatorial method (exact). This shows the inefficiency of the
algorithm especially in large size problems, besides the fact that the sharpness of the
Tonon and Bernardini (1996 and 1999) have developed a mono and multi-
objective optimisation approach for structural uncertainty using fuzzy set and random
set theories. In their work, they utilised a fuzzy set to model the engineer’s judgment
numbers and the meaning was clarified by random set and possibility theories.
14
Nakagiri and Yoshikawa (1996) have developed finite element interval
the change of structural responses due to the change of structural parameters that are
assumed to exist in a convex hull. The Lagrange multiplier method was employed to
interval mathematics. In their work, they investigated the load identification as one of
parameters was limited and a combinatorial solution was used for the interval system
approach was applied to trusses and frames. The purpose of their study was to
some truss structures and frames. Their study gave an introduction to interval
methods for finding interval estimates for the set of solutions of such systems. The
problems using interval methods for analysis of such linear systems were identified,
with promising avenues for further research indicated as a result. The problems
15
discussed include estimation of inaccuracy of the algorithms (especially the
(FFEM) for treating uncertain loads in static structural problems was developed.
Using this approach, the problem domain is first discretised, which resulted in a
guaranteed inclusions for the solution of such fuzzy systems was implemented. In the
case of uncertain loading, the resulting system of equations is linear and only the right
hand side vector contains fuzzy values. Solutions due to all load combinations, as a
computationally intractable for typical structures. Examples show that the possible
interval values calculated using FFEM provide a sharp bound on possible nodal
possible loading combinations. Thus the extreme values of element forces for all
Comparison of extreme values of element forces with typical design loading schemes
show that the current design procedures can lead to nonconservative predictions of
element forces. The method is thus applicable to static analysis of any structural
16
Muhanna and Mullen (1999) have developed interval finite element analysis
procedure utilizing the concept of fuzzy sets through interval calculations. They have
also computed the response of different structural systems due to geometric and
loading uncertainties. Results were exact in the case of load uncertainty where all
stated in this work, the interval algorithm guarantees that all stationary global
solutions have been found. According to him, existing optimisation methods usually
robust, universally applicable and fully reliable. The interval algorithm guarantees
that all stationary global solutions have been found. The convergence of this method
constructing the inclusion function was presented. This method was based on special
systems. The preceding results indicated that the presented algorithm was an
quite slow if many local minima have values of function differing very little from the
global value. The pure interval algorithm guaranteed that all stationary global
solutions (in the initial interval) had been found. The bounds on the solution(s) were
guaranteed to be correct. Error from all sources was accounted for. The algorithm
could solve the global optimisation problem also when the objective function is non-
17
A practical approach for analysing the structures with fuzzy parameters was
structural properties were represented by convex normal fuzzy sets. Vertex solution
methodology that was based on α-cut representation was used for the fuzzy analysis.
determine the binary combinations of the fuzzy variables that resulted in fuzzy
responses at an α- cut level. These binary combinations of the fuzzy variables were
More recently, Muhanna and Mullen (2001) have developed an interval finite
uncertainty in solid and structural mechanics. The method allows the treatment of
avoids most sources of overestimation and computes a very sharp hull for the solution
set of interval linear equations in the field of solid and structural mechanics, with
1.3.1 General
The concept and practical application of the cable-stayed bridge date back to
the 1600’s, when a Venetian engineer named Verantius built a bridge with several
points by cables extending from one or more towers. The concept attracted to
engineers and builders for many centuries and experimentation and development
continued until its modern-day version evolved in 1950 in Germany. The renewal of
18
the cable-stayed system in modern bridge engineering was due to the tendency of
performance from material which was in short supply during the post-war years
(Troitsky, 1972).
During the past three decades, cable-stayed bridges have found wide
applications all over the world, especially in Western Europe and United States. In
particular, the cable-stayed girder type of design is fast gaining popularity among the
in bridge construction as demonstrated by several bridges existing all over the world,
built of different materials and techniques. The Stromsund Bridge, which was
constructed in Sweden in 1955 with a central span of 183 m is the world’s first cable-
constructed all over the world in many countries. The Second Hoogly Bridge over the
river Ganga at Howrah is one of the longest bridges in the world with a span of 457.2
m, the Tatara Bridge in Japan being the longest with a span of 890 m. Efforts are on
to increase the span further beyond 1000 m. For medium spans of 100 - 300 m, cable-
and other notable advantages, cable-stayed bridges have gained popularity in recent
decades. This fact is due, on one hand, to the relatively small size of the substructures
required and on the other hand, to the advent of efficient construction techniques apart
from the rapid progress in the analysis and design of this type of bridges.
19
development of welding techniques and progress in structural analysis. The
unlimited possibilities for the exact solution of these highly statically indeterminate
longer but also lighter and slender bridges. Thus, very long span slender cable-stayed
bridges are being built, and the aim is to further increase the span length and use
shallower and more slender girders for future bridges. To achieve this, accurate
realistic prediction of the structural response to not only wind and earthquake loads
comprises an orthotropic deck and continuous girders, which are supported by stays
i.e. inclined cables passing over or attached to towers located at the main piers.
are dictated to a great extent by the relative stiffnesses of the load-bearing elements –
bridges can be divided into four basic systems as shown in Fig. 1.1 (Walther, 1981).
20
i. Harp system
ii. Fan system
iii. Semi-Harp system
iv. Asymmetric system
With respect to the various positions in space, which may be adopted for
planes in which cable-stays are disposed, there are two basic arrangements:
21
i. Two plane systems
a) Two vertical plane systems
b) Two inclined plane systems
ii. Single plane systems
22
The various possible types of tower construction are:
Most cable-stayed bridges have orthotropic decks, which differ from one
another only as far as the cross sections of the longitudinal ribs and the spacing of the
i. Parallel-bar cables
ii. Parallel-wire cables
iii. Stranded cables
iv. Locked-coil cables
23
(b) Parallel-wire cables
(a) Parallel-bar cables
1.3.3 Loads
There are wide disparities throughout the world concerning the loads on
bridges. There are two considerations governing the loads on the bridges i.e. the load
24
For serviceability, it is only necessary to consider cases of loads likely to be
encountered. For major structures, it may be advisable to base the design loads on a
probabilistic traffic analysis. There is a wide range of statistical data available for the
estimation of probable loads. It is still necessary to take note of the fact that the
permanent loads themselves are often decisive, setting aside the phenomena of
however ambitious it may be, given the fact that the majority of serious accidents
causing structures to collapse are due to non-stochastic causes (for example, human
error). These unforeseeable risks must be covered, at least partly, by the margin of
safety. It matters little whether this safety margin is ensured indirectly by overall or
partial safety factors, provided that it reduces the risks to an acceptable level
(Walther, 1988).
For the serviceability limit state, the permanent loads require to be taken as the
actual loads, allowing normal tolerances for the materials used and for the method of
be paid to these loads. Agreement between calculations and actual practice, notably
in the deflection of the deck and the strengths of the stays, narrowly depends on the
precision of the self-weight and the permanent loads (surfacing and kerbs)
(Walther, 1988).
The loads to be considered when checking the ultimate limit state depend, in
the first instance on the safety philosophy. According to FIP recommendations, the
25
γc = 1.50 for concrete and
coefficients given as
Following FIP recommendations, the factor of safety for variable loads (live
Depending on the complexity of the structure and the stage of the design,
different models may be used. These may be plane or spatial systems, covering the
whole structure or only a part, and can comprise of a wide range of members. The
pylons generally can be represented by beam elements. The same can be said of the
deck if this actually behaves like beam. The cables can be represented by bars by
assigning them a very small bending inertia and an idealised modulus of elasticity,
which makes it possible to take into account the effects of cable sag. In this model the
cables are sufficiently tensioned under permanent loads. Thus any compression,
which is likely to arise under live loads, results only in a reduction of initial tension
(Walther, 1988).
26
The behaviour of cable-stayed bridges under the action of live loads is difficult
initial design stages to model the cable-stayed bridges as plane frames. Final
dimensioning can also be done on the basis of a plane frame model in case of
structures where pylons experience no transverse bending under dead weight plus live
computer is necessary for the solution of this type of structure, its use being primarily
Computer programs are necessary to generate the influence diagrams for cable
forces, stiffening girder, bending moments and shear, and tower and pier reactions.
The computer is also required for the rapid solution of various parametric efforts and
loadings that have to be taken into account in achieving a reasonably efficient design.
Probably, the most important problems are the determination of the optimum section
elastic system, which may be analysed using the standard stiffness or flexibility
method. Several general computer programs are available which use this approach,
of several stages. The first involves calculations to give preliminary sizes to the deck,
27
pylons and stays. The general aim is to check the feasibility of the work, as well as to
estimate the quantities required for its completion. At this stage of the scheme, it is
consideration secondary influences or long term effects. In the second stage, final
calculations are prepared, determining the strength and deformations based on the
A review of the work done by the previous researchers reveals the following
points:
2. In particular three major approaches are used to model uncertainty i.e. anti-
3. Of the above three methods, the review of literature reveals that anti-
smaller magnitude are solved using these approaches, as is evident from the
literature review.
28
6. Making use of this methodology, the above researchers were able to study the
mechanics. Simple cases such as propped cantilever beam, plane truss, and
plate problems were handled and fuzzy structural response was evaluated in
7. However, the above studies were confined to the studies of the individual
(simultaneous presence of more than one uncertainty in the system) was not
displacements, bending moments, axial forces and shear forces. Further, the
uncertainties.
29
A review of literature of the analysis of the cable-stayed bridges reveals the
following points:
2. The service loads acting on the bridge are highly variable in nature. The
Elishakoff, 1998).
cable stays and concrete deck may also be uncertain in nature. Owing to the
Young's modulus has been reported in the literature (Muhanna and Mullen,
2001).
(Walther, 1988). Further, during the lifetime of the bridge, the value of the
mass density may undergo variation owing to the effect of corrosion, etc.
in the analysis and design of cable-stayed bridges. This apart, fuzzy finite element
analysis has not been applied to the study of cable-stayed bridges in order to
understand its structural response. In the light of the above, it is felt that there is a
30
need to apply fuzzy finite element methodology to evaluate the effect of multiple
the presence of multiple uncertainties using fuzzy finite element analysis. The nature
1.6. Summary
31
process the solution in order to obtain axial forces, bending moments and
2. The proposed method is tested on beams, frames and trusses and then
validated with the available literature (Appendix C). The validated new
Chapter 3.
3. The fuzzy-finite element model of the structure is analysed for three possible
32
Chapter 2
Methodology
Chapter 2
2. METHODOLOGY
2.1 Introduction
material property (E), live load and mass density respectively. These uncertainties are
by Muhanna (Muhanna and Mullen, 2001) (Appendix B). This model is suitably
uncertainties (0≤α,β,γ≤1).
The proposed fuzzy finite element model of the structure in the presence of the
34
where [K̃] is a crisp (non-interval) assembled stiffness matrix, {U} is the
and subjected to a fuzzy interval live load at a level β, Eq. 2.1 is reformulated as
respectively and {Pβ} is the fuzzy-interval live load vector at the level of
uncertainty β.
interval equations in the presence of live load (Muhanna and Mullen, 1999).
Uncertainty of mass density, if present in the system, gets reflected as the uncertainty
In this case, the structural system has crisp material property (Young’s
modulus) and is subjected to a fuzzy interval live load. Accordingly, Eq. 2.2 is
recast as
35
Eq. 2.4 represents a system of linear interval equations. Solution of this system
linear interval matrix equations (Hansen, 1965, Gay, 1982, Alefeld and Herzberger,
1983, Neumaier, 1987,1989 and 1990, Jansson, 1991, Muhanna and Mullen, 1999).
dependency by multiplying all the non-interval values first and performing operations
on fuzzy values as a final step. The procedure suggested by Muhanna (Muhanna and
matrix [M̃] and an interval element live load vector {q} as shown below.
where n is the kinematic indeterminacy for the structure and m is the number
of elements in the structure. The jth column of [M̃] matrix contains the contribution to
element forces due to a unit live load on the jth element of the structure.
Thus all non-interval values are multiplied first, and the last multiplication
involves the fuzzy-interval quantities. If this order is not maintained, the resulting
In case the structure is acted upon by set of fuzzy point loads, the point load
36
{Uβ} n×1= [K̃] -1n×n {P´β} n×1 - (2.8)
where P´i = [P´il , P´iu] is the fuzzy point load applied at the ith degree of freedom.
After solving for the fuzzy displacement vector {Uβ} n×1 (primary unknown),
the values of fuzzy axial force, fuzzy shear force and fuzzy bending moment
level.
[K̃] (e) md×md [T] (e) md×md {Uβ}(e) md×1 = {Pβ}(e) md×1 + {g}(e) md×1 - (2.9)
where md is the number of degrees of freedom for the element, [T](e) is the
transformation matrix for the element and {g}(e) is the vector of element-end forces.
where A1, A2, V1, V2, M1, M2 are the axial forces, shear forces and bending moments
at the first node and the second node of the element respectively.
Substituting Eq. 2.10 and Eq. 2.7 in Eq. 2.9 for uniformly distributed loads,
[K̃](e) md×md [T](e) md×md [L] md×n [K̃]-1n×n [M}n×m{qβ}m×1 ={Pβ}(e) md×1+{g}(e) md×1 - (2.11)
Substituting Eq. 2.10 and Eq. 2.8 in Eq. 2.9 for point loads,
[K̃] (e) md×md [T] (e) md×md [L] md×n [K̃]-1n×n {P′}n×1 = { P′β}(e) md×1+{g} (e) md×1 - (2.12)
37
From Eq. (2.11) and Eq. (2.12), {g}(e) is computed and the fuzzy values of A1,
uncertainty. In this case, the governing equation, given by Eq. 2.3, is modified as
The displacement vector {Uγ} and the vector of internal forces {λγ} are obtained by an
Considering the case of a structural system with a crisp load (β=1) and with
where [Kα] is the structure matrix with material uncertainty α for an assembled finite
element model.
The solution of the set of equations given by Eq. 2.14 for the assembled finite
element model using the methodology described in section 2.2.1 and section 2.2.2
(Appendix B). In this model, the governing equations for assembled finite element
38
model given by Eq. 2.2 are modified in order to impose constraints on the
a) The set of elements of the model is dissembled. Each element has its own set
b) A set of additional constraints (in addition to the constraints that are imposed
50kNm
50kN
100kN 100kN
25kN/m
4m
25kN/m
3m 3m
Several problems involving uncertain Young’s modulus and crisp live load
attempt was made in the present study to utilise the above methodology to problems
involving multiple uncertainties (α and β). However, it was found that the application
39
overestimated solution. A detailed numerical investigation and error tracking led to
the case of a structure with a crisp Young’s modulus and uncertain interval
load (α=1, 0≤β≤1), the assembled finite element model (section 2.2) and
investigations made in the present study could not validate this assumption by
In the case of classical finite element analysis using crisp loads, the
sum total of the fixed end reactions at a particular degree of freedom at the
given element (for a given load-case) is computed. The element load vector is
obtained by reversing the sign of the vector of fixed end forces for a given
element. But, it is observed in the present study that the application of this
procedure in the case of several interval loads acting on the elements causes
40
However, this problem was not encountered in the problems solved by
b) Even in the case of single uncertain load acting on the structure, the result of
the assembled model did not exactly match the result of the element-by-
the approximation of the interval vector of internal forces {λ} by its mid-point
structures internal forces depend on both the structural stiffness and external
load applied. Thus in the presence of material uncertainty (at a level α) and
load uncertainty (at a level β), the internal forces are expected to exhibit the
uncertainty (0≤α≤1 and β=1), it is observed that the use of the mid-point
vector of the interval vector {λ} i.e. {λc} in Eq. 7 of Appendix-B yields the
sharp solution.
ii) A new approximation to the vector of internal forces {λαβ} (under the
41
By incorporating the modifications mentioned in points (i) and (ii) above, a
+ +
The structure is analysed by keeping the uncertain loads acting on each of the
elements separate throughout the course of the solution in order to prevent load
coupling at the element level. The overall fuzzy displacement vector {U} which
represents combined effect of all joint loads and element loads is obtained by
vectors {Uαβ} obtained for individual load cases. As brought out later in section 2.5,
42
In the case of structures with uncertainty of load (Young’s modulus being
crisp), it was found that the results obtained using assembled finite element model and
where {Uαβ} is the fuzzy-interval displacement vector, {Pβ} is the fuzzy-interval load
vector, [C̃] is a constraint matrix , [Kα] is the fuzzy-interval stiffness matrix and {λαβ}
owing to the dissembled state of elements in the EBE model (Muhanna and Mullen,
2001). If the unknowns of each element are numbered consecutively, then the
stiffness coefficients of [K]α will be clustered in the form of a square diagonal blocks
each of size equal to the number of unknowns per element. The size of [K]α is n×n.
The interval vector of Lagrange multipliers {λ}αβ represents the vector of internal
forces that are exposed due to the dissembled nature of elements in the EBE model.
The value of {λ}αβ depends on only applied loads and boundary conditions in the case
43
of statically determinate structures. But in the case of statically indeterminate
structures, the value of {λ}αβ depends on structural stiffness, applied loads and
where [D]α is a diagonal matrix of size n×n containing interval terms corresponding to
material uncertainty at level α and [S] is a deterministic singular matrix of size n×n..
If Eq. 2.17 is multiplied on either side by [D]α[C]T and the result is added to
If the interval vector {λ}αβ is determined exactly, then the solution of Eq. 2.21
represents the exact hull of interval system of equations given by Eq. 2.2
depends only on applied loads and boundary conditions and does not depend on
exactly. Thus, the procedure described by Muhanna (Muhanna and Mullen, 2001) is
44
2.5.1 Cumulative Effect of Material and Load Uncertainties (α and β)
At a given level of uncertainty β, the interval load vector {P}β and vector of
where βl and βu are the lower and upper bounds of the normalised variable for interval
vectors {P} β or {λ}αβ. The exact value of {λ}αc is not known and is approximated
with {λ}cc.
Thus,
Substituting the approximate value of {λ}αβ in Eq. 2.21, the following approximate
{
[D]α[R̃]{U}αβ = {P}β -[βl, βu] [C]T{λ}cc} - (2.26)
where {δ}α is an interval vector of size n×1 containing interval material properties of
n elements taken from the diagonal entries of [D]α. The solution of this equation
yields an exact solution in the presence of load uncertainty and a sharp solution if
In order to solve Eq. 2.27, it is required to compute the value of the mid-point
internal force vector {λ}cc. This is done by considering the assembled finite element
45
model of the structure with deterministic value of structural stiffness subjected to
unassembled discrete elements in the EBE model by assembling elements that are
kept unassembled hitherto. In this process, each set of hitherto coincident nodes of
the BE model is collated and treated as a single node common to all the coincident
elements. Owing to this process of assembling the elements, the internal force vector
taken as crisp (or equivalently, point interval) and only load uncertainty is allowed.
Accordingly, substituting α=1 and {λαβ} = {0} in Eq. 2.16, and assembling all
Replacing the point interval stiffness matrix [K] by its equivalent floating-
46
problem). Hence, the equations are solved by using the procedure mentioned in
section 2.2.1.
as per the procedure outlined in section 2.2.2 and the vector of element-end forces
{λ}cβ(e) is calculated element-wise. The vector {λ}cβ(e) is obtained from the following
equations for body forces and surface tractions respectively (by modifying
{λ}cβ(e) md×1=- {Pβ}(e) md×1+ [K̃] (e) md×md [T] (e) md×md [L] md×n1[K] -1n×n1 [M̃] n1×m {qβ}m×1 - (2.30)
and
{λ}cβ(e) md×1= -{Pβ}(e) md×1+ [K̃] (e) md×md [T] (e) md×md [L] md×n1 [K] -1n1×n1 {P′}m×1 - (2.31)
The vector of internal forces {λcβ} is obtained from the calculation of vector
approximate the value of the displacement vector {U}αβ. This is owing to the
Muhanna’s approach (Muhanna and Mullen, 2001) and using Eq. 2.21 and Eq. 2.26,
{
{δU} =[R̃]-1 [D]α-1 {{P}β -[C]T{λ}αβ}- [R̃]-1 [D]α-1 {P}β -[βl, βu] [C]T{λ}cc } - (2.32)
A sharp solution to Eq. 2.27 is obtained by minimising the value of {δU} using
1982, Neumaier, 1987,1989 and Rump, 1990). The theory of inclusion aims at
obtaining an optimum interval enclosure for the solution of the given interval system.
47
Solution of these equations using the inclusion theory results in an optimal enclosure
AHb known as the hull of the solution. The lower bound vector {x} and the upper
xx
AHb
yy
Making use of Jansson’s algorithm (Jansson 1991, Muhanna 2001), the lower
bound vector {x} and the upper bound vector {y} enclosing hull of the solution are
In the present study, {Uαβ} is taken as the mean value of the inner bound
The [M] matrix in Eq. 2.27 has n rows and m columns. The matrix contains
48
Eq. 2.9 is reformulated as
[Kα] (e) md×md [T] (e) md×md {Uαβ}(e) md×1 = {Pβ}(e) md×1 + {λαβ}(e) md×1 - (2.36)
Eq. 2.14 as
Here {Uαβ}(e) contains displacements appropriate to the given element with reference
{λαβ}(e) md×1 = [βl, βu]([Kα] (e) md×md [T](e)md×md[L] md×n[R̃]-1[M]{δ}α–{P}(e) md×1) - (2.40)
As explained in the section 2.5.2, the vectors {λαβ}(e) obtained for individual
load cases are superimposed to obtain the overall internal force vector for the element
{λ}(e), which represents the combined effect of all loads acting on the structure.
A methodology similar to the one that was adopted in the previous case
(simultaneous presence of load and material uncertainties) is adopted in this case also.
Accordingly, the set of equations givens by Eq. 2.15 through Eq. 2.40 are modified by
49
2.8 Concomitant Presence of Mass Density, Material Uncertainty and Live
Load Uncertainty
In this case the results are obtained by superimposing the following cases:
the presence of dead load with material and mass density uncertainty.
ii) Structural response in the presence of live load with material and live load
uncertainty.
quantity (displacement or force) about its mean value with reference to a unit
property, live load and mass density) about their respective mean values.
In order to perform the sensitivity analysis, the material property, live load and
quantities (Forces and moments) are also normalised. Normalisation refers to the
process of dividing the lower and upper values of a fuzzy interval number by its mid
value.
mid = 0.5 (a + b). The corresponding normalised interval is [φ,η] where φ and η
represent a/mid and b/mid respectively. Obviously, the mid-value of this normalised
interval µ is equal to 1 where as φ and η are less than 1 and greater than 1
50
indicates that the normalised triplet (φ,µ, η) equals (0.98,1.00,1.02) and also that the
variation of the given response quantity about its mean value is ±2 percent.
Sensitivity graphs can be plotted for each response quantity by taking the
percentage variation of characteristic values of the structure on the x-axis and the
respective mean values) on the y-axis. A larger value of sensitivity for a given
response quantity indicates that the level of uncertainty associated with that particular
response quantity is quite significant. This information may help the designer to
2.10 Summary
explained.
conducting certain preliminary case studies (Appendix C). After due validation, the
51
proposed methodology is applied to obtain the response of a cable-stayed bridge as
explained in Chapter 3.
52
Chapter 3
Case Studies
Chapter 3
3. CASE STUDIES
3.1 Introduction
and validity of the proposed methodology, several simple problems are solved and the
results are compared with the published work of previous researchers, wherever
and the independent variation of uncertainties of material property (E), live load and
mass density. These uncertainties are considered as fuzzy interval values. The levels
The fuzzy interval quantities are represented at a specified level of presumption (α,β
approach. The concept of fuzzy interval numbers, fuzzy membership functions and
membership function for a fuzzy quantity, a fuzzy interval variable in the form a
closed interval [a, b] (a≤x≤b) is obtained. In order to avoid confusion, the cuts made
on the membership functions of material property (E), live load and mass density are
denoted as α-cut, β-cut and γ-cut respectively. The uncertainties and the
54
Fig. 3.1 shows the membership function of uncertainty of material property
(Young’s modulus E). The membership function for material uncertainty shows the
normalized values of uncertain Young’s modulus, taking the crisp value as 1.0. The
values of these uncertainties are shown on the y-axis at various levels of α ranging
from α=0.0 to α=1.0 in steps of 0.2. A maximum of ±5% variation about the mean
corresponding to the mean value and the width of the interval corresponding to
various α-cut levels is equally distributed about the mean (crisp) value. This is
because of the uncertainty in the value of Young’s modulus arising out of minor
sample has an equal chance of being above or below the mean value. Thus, an equal
At the level α=0.0, the α-cut on the membership function yields a fuzzy
interval number [0.95,1.05]. Similarly, for other values of α also, the corresponding
of ±1% about the mean value), the normalized α-cut value is [0.99,1.01]. At α=1.0
(corresponding to 0 percent variation about the mean value), the normalized α-cut
value is a fuzzy point interval number [1.0,1.0]. This corresponds to total certainty.
Fig. 3.2 depicts membership function for load uncertainty. The membership
function for load uncertainty shows the normalized values of uncertain load, taking
the mean (crisp) value as 1.0. The load, which is being referred to, is a service load
and is subject to wide fluctuation about its mean value. In fact, load variation as
much as 100% about the mean value has been reported in the literature (Mullen and
55
Muhanna 1999, Muhanna and Mullen, 1999). Thus, ±100% variation about the mean
triangle with a central peak (corresponding to the crisp load). This is because the
service load at any instant has an equal chance of falling below or going above the
nominal (mean) value of the load. Fig.2.2 depicts the β-cut levels corresponding to
steps of 0.2. For example at β=0.4, the normalized β-cut value is [0.4, 1.6] and at a
Fig. 3.3 presents membership function for mass uncertainty. The membership
function for mass uncertainty shows the normalized values of uncertain mass, taking
the mean (crisp) value as 1.0. The mass, which is being referred to, is the nominal
mass and is subject to fluctuation about its mean value owing to the uncertainty
associated with mass density of the materials during the manufacturing process. In
addition, corrosion may affect the mass of the structure during its design life. This
fact can also be taken into consideration during analysis by introducing uncertainty in
mass. Therefore, ±5% variation about the mean value has been chosen as maximum
whereas material uncertainty is denoted by α). In Fig. 3.3, the membership function
shown to represent mass uncertainty is triangle with a central peak. This is because
the mass density has an equal chance of falling below or going above the nominal
In Fig. 3.3 the γ-cut levels corresponding to various levels of γ ranging from
γ=0.0 (total uncertainty) to γ=1.0 (total certainty) in steps of 0.2 are presented. For
example at γ=0.4, the normalized γ-cut value is [0.970,1.030] and at a level of γ = 0.8
56
the corresponding γ cut value is [0.990,1.010]. Just as in the case of membership
function for material uncertainty, the membership function for mass uncertainty also
is a central peak triangle with the peak corresponding to the crisp mass density.
self-weight). This is because dead load intensity is the product of mass density and
cross-sectional area for a prismatic element. Thus, the uncertainty associated with
dead load is the same as the uncertainty due to mass density. Thus membership
function adopted for mass density together with the corresponding γ-cut can be
3. Structural response to a combination of dead load and fuzzy live load in the
(Young's modulus).
The case study considered here is a cable-stayed bridge with fan configuration
of cable-stays, shown in Fig. 3.4. This problem is adopted from the configuration of
pedestrian bridge with two parallel planes of cables arranged in fan configuration.
The bridge is symmetric about the longitudinal axis. The properties of the bridge are
57
mentioned in Table 3.1. Owing to the symmetry of the bridge deck about the
longitudinal axis of the bridge, only one half of the bridge along with a single plane of
structural elements belonging to bridge deck and the pylon are idealized as plane
frame elements while cables are modelled as bar elements. The structure is hinged at
nodes 1 and 9 on the bridge deck and is fixed at node 14 at the bottom of the pylon.
A roller support is provided at node 5, at the point where the deck slab passes over the
In this case study, the cable-stayed bridge described above is subjected to the
action of a uniformly distributed dead load (owing to the self-weight).The dead load is
computed on the basis of mass density of steel and concrete as provided in Table 3.1.
58
Also, the bridge is subjected to simultaneous uncertainties in material property E at a
combinations of α and γ ranging from total certainty (α=1, γ=1) to total uncertainty
(α=0, γ=0) are considered. Fuzzy finite element model of the structure is used to
obtain fuzzy interval values of displacements and rotations. These results are further
post-processed to obtain fuzzy interval values of shear force and bending moment at
various nodes. The values of structural response (displacement and forces) obtained
structural response.
cumulative effect of uncertainties associated with material property and mass density
investigated and the results are shown in the form of tables and figures (for
membership functions). The tables and the membership functions represent the
cumulative effect of material property and mass density uncertainties on the values of
the displacements (horizontal, vertical and rotation), and forces (axial and shear forces
In the second case study, a cable-stayed bridge (shown in Fig. 3.4) with the
configuration given in Table 3.1 is taken up for analysis. The bridge is subjected to
59
uncertain live load due to traffic at a level of uncertainty β. Uncertainty is also
introduced in the Young’s modulus of the materials (Steel and Concrete) at a level α.
2
As mentioned in Table 3.1 a live load intensity of 4.0 kN/m (for a pedestrian
bridge) is adopted to act all over the deck slab. For half the width of the slab in the
uncertainty for load βmax is taken as ±100% about mean value of load as is reported in
(α=1, β=1) to total uncertainty (α=0, β=0) are considered. Fuzzy finite element
model of the structure is used to obtain fuzzy interval values of displacements and
rotations. These results are further post-processed to obtain fuzzy interval values of
shear force and bending moment at various nodes. The values of structural response
cumulative effect of uncertainties associated with material property and live load (α
60
material and load uncertainties (α and β) on displacements and forces is investigated
and the results are shown in the form of tables and figures (for membership
functions). The tables and the membership functions represent the cumulative effect
(horizontal, vertical and rotation), and forces (axial and shear forces and bending
moments).
Here the Cable stayed Bridge with fan configuration of cable-stays (taken in
case study 3.1 and case study 3.2) subjected to a combination of dead load and live
live load (owing to traffic loading). The dead load uncertainty arises out of
in material property (α), load (β) and mass density (γ) respectively.
uncertainty (α=β=0) are considered at specified values of γ. The fuzzy finite element
These results post-processed to obtain fuzzy interval values of axial force, shear force
cumulative effect of uncertainties associated with material, load and mass density (α,
61
at a specific level of mass uncertainty γ and the results are shown in the form of tables
and figures (for membership functions). The tables and the membership functions
represent the combined effect of material, live load and mass uncertainties on the
values of the displacements (horizontal, vertical and rotation), and forces (axial and
Table 3.2 and Table 3.3 present the sample output of the case studies
explained in section 3.1, section 3.2 and section 3.3 for various combinations of α,β
and γ respectively. Table 3.2 represents the variation of uncertain rotation (×10-3
radians) at node 4 for various combinations of material and mass density uncertainties
(α and γ). The value of the rotation shown at the top left corner of the table
corresponds to the combination α=γ=1. The value of the rotation shown at the bottom
right corner of the table corresponds to the combination α=γ=0. The uncertain values
of rotation in the first row of the table correspond to the presence of mass density
uncertainty alone (0≤γ≤1 and α=1). The uncertain values of rotation in the first
column of the table correspond only to the presence of material uncertainty (0≤α≤1
and γ=1).
Table 3.3 represents the variation of uncertain axial force in deck (kN) in
element 12 for various combinations of material and live load uncertainties (α and β).
The value of the axial force shown at the top left corner of the table corresponds to the
combination α=β=1. The value of the axial force shown at the bottom right corner of
the table corresponds to the combination α=β=0. The uncertain values of axial force
in the first row of the table correspond to the presence of live load uncertainty
62
(0≤β≤1 and α=1), while the uncertain values of axial force in the first column of the
table correspond to the presence of material uncertainty alone (0≤α≤1 and β=1).
Fig. 3.6 represents the membership function for axial force (kN) in cable 3 (at node 2)
at α=0.6 at various levels of β (0≤β≤1) in steps of 0.2. These results show that the
The detailed discussion of the results obtained for case studies 1,2 and 3 are
vertical displacements and rotations together with axial force, shear force and
Fig. 3.7 depicts the sensitivity analysis of displacements of the deck at node 3
under the action of dead load at γ=1.0 with respect to percentage variation of material
property (E) about its mean value. Fig. 3.8 depicts the sensitivity analysis of forces
and moments of the deck at node 5 under the action of dead load at γ=1.0 with respect
The results and the detailed discussion of this investigation are presented and
63
3.7 Summary
bridge.
response quantities.
64
1 1.000
0 0.950 1.050
0.950 0.960 0.970 0.980 0.990 1.000 1.010 1.020 1.030 1.040 1.050
Normalised Young's Modulus
1 1
0 0 2
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Normalised load
65
1 1.000
0 0.950 1.050
0.950 0.960 0.970 0.980 0.990 1.000 1.010 1.020 1.030 1.040 1.050
Normalised mass density
1
2 3 4 5 6 7 8
66
1 -2.775E-02 -1.730E-03
0 -3.263E-02 2.670E-03
-3.50E-02 -3.00E-02 -2.50E-02 -2.00E-02 -1.50E-02 -1.00E-02 -5.00E-03 0.00E+00 5.00E-03
Vertical Displacement (m)
Fig. 3.5 Cable stayed bridge- Membership Function for vertical displacement at node
3 at beta=0.8
1 120.06 139.54
0 -30.58 303.31
-50.00 0.00 50.00 100.00 150.00 200.00 250.00 300.00
Axial Force (kN)
Fig. 3.6 Cable stayed Bridge - Membership Function for Axial Force in cale 3 at
alpha=0.6
67
35.000
30.355
30.000
Percentage Variation of displacement about mean
27.953
24.296
25.000
22.290
20.000 18.237
16.667
15.000 15.330
12.148
12.214
11.084
10.000
9.122
6.089
Horizontal Displacement
6.054
5.000 5.502 Vertical displacement
Rotation
3.021
0.000
0.000 0.000
0.0 1.0 2.0 3.0 4.0 5.0
Percentage Variation of E about mean
Fig. 3.7 Sensitivity Analysis for displacements (for dead load) at node 3 at gamma=1.0
18.0
16.40
16.0
Percentage Variation of Force/moment about mean
14.51
14.0 13.17
12.0 11.75
9.92 10.64
10.0
8.93
8.45
8.0
6.65
6.29
6.0 6.03
0.00
0.0 0.00
0.0 1.0 2.0 3.0 4.0 5.0
Percentage Variation of E about mean
Fig 3.8 Sensitivity Analysis of Forces and Moments in Element 12 (at node 5) at gamma=1.0
68
Table. 3.2 Cable-Stayed Bridge –
Concomitant Variation of Rotation at node 4 (×10-3 radians) w.r.t α and γ
γ→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [0.846,0.846] [0.766,0.925] [0.686,1.005] [0.606,1.085] [0.526,1.165] [0.447,1.244]
69
Chapter 4
The present study involves the study of the cumulative effect of uncertainties
in material property (E), live load and mass density (α,β and γ) on the structural
presented in chapter 2 is applied to EBE model of the structure in order to obtain the
displacements and rotations thus obtained are further post-processed using the
methodology described in chapter 2 and fuzzy values of axial forces and shear forces
and bending moments are evaluated. The results obtained are tabulated and
membership functions for the structural response quantities are constructed for
the relative variation of structural response to the corresponding change in the values
Young’s modulus of the materials (steel and concrete) and also in mass density.
combinations of α and γ ranging from total certainty (α=1, γ=1) to total uncertainty
(α=0, γ=0). The results are presented in Table 4.1.1 through Table 4.1.10. The
71
a) The values of structural response in the first row correspond respectively
to the presence of mass density uncertainty alone (0≤γ≤1 and α=1) and
(0≤α≤1 and γ=1). It is observed that the width of the interval increases
[1-ε1, 1+ε2] indicates that lower bound and upper bound variations of the
respectively.
c) Fuzzy membership functions (a), (b),(c) and (d) are constructed from
described below.
Table 4.1.1, Table 4.1.2 and Table 4.1.3 show the concomitant variation of
horizontal and vertical displacements and rotations at nodes 2,3 and 4 respectively, for
72
4.1.1.1 Uncertain horizontal displacement at node 2
Fig. 4.1.1 (a) and (b) represent the membership functions for horizontal
[1.0,1.0] in Fig. 4.1.1(a) and [0.868,1.137], [0.984,1.016] in Fig. 4.1.1 (b). Fig. 4.1.1
(c) and (d) represent the membership functions for horizontal displacement at node 2
(m) at α=1.0 and α=0.6. The uncertain horizontal displacements at γ=0.0 and γ=1.0
Fig. 4.1.2 (a) and (b) represent the membership functions for vertical
displacement at node 3 (m) at γ=1.0 and γ=0.8. In these figures, the uncertain vertical
[-1.0,-1.0] in Fig. 4.1.2 (a) and [-1.353,-0.653] , [-1.044,-0.957] in Fig. 4.1.2(b). Fig.
4.1.2 (c) and (d) represent the membership functions for vertical displacement at node
3 (m) at α=1.0 and α=0.6. The uncertain vertical displacements at γ=0.0 and γ=1.0
Fig. 4.1.3 (a) and (b) represent the membership functions for rotation (radians)
at node 4 at γ=1.0 and γ=0.8. The uncertain rotations at α=0.0 and α=1.0 correspond
[0.905,1.093] in Fig. 4.1.3 (b). Fig. 4.1.3 (c) and (d) represent the membership
73
functions for rotation at node 4 at α=1.0 and α=0.6. The uncertain rotations at γ=0.0
[0.935,1.065] and [0.919,1.081]. Similarly, the normalised values in the first column
observed that the variation of horizontal displacement about its nominal value is more
(8.1%). In the case of vertical displacement (Table 4.1.2) the corresponding variations
about its nominal value are 30.9% for material uncertainty and 21.8% for mass
uncertainty. In the case of rotation (Table 4.1.3), the corresponding variations about
its nominal value are 68.0% for material uncertainty and 47.0% for mass uncertainty.
moments
Table 4.1.4 through Table 4.1.10 show the concomitant variation of axial
forces, shear forces and bending moments for various combinations of α and γ. Table
4.1.4 and Table 4.1.5 represent the variation of axial force (kN) in element 9 and
element 12 respectively. Table 4.1.6 represents the variation of shear force in the
deck just to the left of node 3. Table 4.1.7 represents the variation of bending moment
in the deck at node 4. Table 4.1.8 represents the variation of axial force (kN) in pylon
in element 17 (at node 2). Table 4.1.9 and Table 4.1.10 represent the variation of
74
axial force (kN) in cable 2 (at node 2) and cable 3 (at node 3) respectively. In the
above tables, tensile force and sagging moment are considered positive.
observed that variation of axial force in deck is more in the case of material
bound 8.1%). Similar behaviour is observed from all the other uncertain
Fig. 4.1.4 (a) and (b) represent the membership functions for axial force in
deck in element 9 (kN) at γ=1.0 and γ=0.8. The uncertain axial forces at α=0.0 and
and [0.795,1.221], [0.984,1.016] in Fig. 4.1.4 (b). Fig. 4.1.4 (c) and (d) represent the
membership functions for axial force in deck in element 9 (kN) at α=1.0 and α=0.6.
The uncertain axial forces at γ=0.0 and γ=1.0 correspond to normalised values
Fig. 4.1.4(d).
Fig. 4.1.5 (a) and (b) represent the membership functions for axial force (kN)
in deck in element 12 at γ=1.0 and γ=0.8. The uncertain axial forces at α=0.0 and
and [-1.177,-0.846], [-1.01,-0.99] in Fig. 4.1.5 (b). Fig. 4.1.5 (c) and (d) represent the
membership functions for axial force in deck in element 9 at α=1.0 and α=0.6. The
75
[-1.05,-0.95], [-1.0,-1.0] in Fig. 4.1.5(c) and [-1.117,-0.89], [-1.06,-0.94] in
Fig. 4.1.6 (a) and (b) represent the membership functions for shear force just
to the left of node 3 (kN) at γ=1.0 and γ=0.8. The uncertain shear forces at α=0.0 and
and [-1.116,-0.891] , [-1.012,-0.988] in Fig. 4.1.6 (b). Fig. 4.1.6 (c) and (d) represent
the membership functions for shear force in deck in element 9 (kN) at α=1.0 and
α=0.6. The uncertain shear forces at γ=0.0 and γ=1.0 correspond to normalised values
Fig. 4.1.7 (a) and (b) represent the membership functions for bending moment
(kNm) in deck at node 4 at γ=1.0 and γ=0.8. The uncertain bending moments at
[-1.000,-1.000] in Fig. 4.1.7 (a) and [-1.206,-0.802] , [-1.016,-0.984] in Fig. 4.1.7 (b).
Fig. 4.1.7 (c) and (d) represent the membership functions for bending moment at node
4 at α=1.0 and α=0.6. The uncertain bending moments at γ=0.0 and γ=1.0 correspond
76
4.1.2.5 Axial force (kN) in pylon in element 17 (at node 11)
Fig. 4.1.8 (a) and (b) represent the membership functions axial force (kN) in
pylon in element 17 (at node 11) at γ=1.0 and γ=0.8. The uncertain axial forces at
Fig. 4.1.8 (a) and [-1.126,-0.891] , [-1.010,-0.990] in Fig. 4.1.8 (b) . Fig. 4.1.7 (c) and
(d) represent the membership functions for axial force in deck in element 9 (kN) at
α=1.0 and α=0.8. The uncertain axial forces at γ=0.0 and γ=1.0 correspond to
Fig. 4.1.9 (a) and (b) represent the membership functions axial force (kN) in
cable 2 (at node 2) at γ=1.0 and γ=0.8. The uncertain axial forces at α=0.0 and α=1.0
correspond to normalised values [0.823,1.195] and [1.000,1.000] Fig. 4.1.9 (a) and
[0.812,1.208], [0.988,1.012] in Fig. 4.1.9 (b). Fig. 4.1.9 (c) and (d) represent the
membership functions for axial force in cable 2 (at node 2) at α=1.0 and α=0.6. The
uncertain axial forces at γ=0.0 and γ=1.0 correspond to normalised values [0.94,1.06] ,
[1.0,1.0] in Fig. 4.1.9 (c) and [0.87,1.138] , [0.928,1.075] in Fig. 4.1.9 (d) .
Fig. 4.1.10 (a) and (b) represent the membership functions axial force (kN) in
cable 3 (at node 3) at γ=1.0 and γ=0.8. The uncertain axial forces at α=0.0 and α=1.0
[0.812,1.209], [0.989,1.011] in Fig. 4.1.10 (b). Fig. 4.1.10 (c) and (d) represent the
membership functions for axial force in cable 3 at α=1.0 and α=0.6. The uncertain
77
axial forces at γ=0.0 and γ=1.0 correspond to normalised values [0.944,1.056],
[1.0,1.0] in Fig. 4.1.10 (c) and [0.874,1.135], [0.927,1.076] in Fig. 4.1.10 (d).
represented by figures (a) and (c) of Fig. 4.1.1 through Fig. 4.1.10 are triangular
(d) of Fig. 4.1.1 through Fig. 4.1.10 are trapezoidal in the presence of multiple
more in comparison with the variation with reference to mass uncertainty. This is
because the material uncertainty causes uncertainty in the stiffness matrix [Kα] in Eq.
2.3 where as uncertainty of mass density causes uncertainty in the right hand side
78
Table. 4.1.1 Cable stayed Bridge -
Concomitant Variation of Horizontal displacement of node 2 (×10-4 metres) w.r.t α and γ
γ→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [7.169,7.169] [7.052,7.285] [6.936,7.401] [6.820,7.517] [6.704,7.633] [6.588,7.749]
0.8 [-3.468, -3.069] [-3.611, -2.927] [-3.754, -2.785] [-3.897, -2.643] [-4.040, -2.501] [-4.183, -2.359]
0.6 [-3.669, -2.871] [-3.812, -2.729] [-3.956, -2.587] [-4.099, -2.445] [-4.243, -2.304] [-4.386, -2.162]
0.4 [-3.871, -2.672] [-4.015, -2.531] [-4.159, -2.389] [-4.303, -2.248] [-4.447, -2.107] [-4.590, -1.965]
0.2 [-4.073, -2.474] [-4.218, -2.333] [-4.362, -2.192] [-4.507, -2.051] [-4.652, -1.910] [-4.796, -1.768]
0.0 [-4.277, -2.276] [-4.422, -2.135] [-4.567, -1.995] [-4.713, -1.854] [-4.858, -1.713] [-5.002, -1.572]
79
Table. 4.1.4 Cable stayed Bridge – Concomitant Variation
of Axial Force in deck in element 9 (kN) w.r.t α and γ
γ→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [385.17,385.17] [378.93,391.41] [372.69,397.65] [366.45,403.90] [360.21,410.14] [353.96,416.37]
80
Table. 4.1.7 Cable stayed Bridge – Concomitant Variation
of Bending Moment in deck at node 4 (kNm) w.r.t α and γ
γ→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
81
Table. 4.1.9 Cable stayed Bridge – Concomitant Variation
of Axial Force (kN) in cable 2 (at node 2) w.r.t α and γ
γ→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [286.1,286.1] [282.7,289.6] [279.2,293.0] [275.8,296.5] [272.4,299.9] [268.9,303.3]
82
1 7.169E-04 1 7.052E-04 7.285E-04
alpha
alpha
gamma
gamma
alpha
alpha
1 -3.268E-02
1 -3.669E-02 -2.871E-02
gamma
gamma
0 -4.386E-02 -2.162E-02
0 -3.980E-02 -2.556E-02 -4.50E-02 -4.25E-02 -4.00E-02 -3.75E-02 -3.50E-02 -3.25E-02 -3.00E-02 -2.75E-02 -2.50E-02 -2.25E-02 -2.00E-02
-4.000E-02 -3.800E-02 -3.600E-02 -3.400E-02 -3.200E-02 -3.000E-02 -2.800E-02 -2.600E-02 Vertical Displacement (m)
Vertical Displacement (m) (d) alpha=0.6
(c) alpha=1.0
alpha
alpha
Rotation(radians) Rotation(radians)
gamma
0.4 6.061E-04 1.085E-03 0.4 3.784E-04 1.315E-03
alpha
0.4 340.9 431.4
0.4 334.9 437.9
0 312.3 463.5
310 330 350 370 390 410 430 450 470 0 306.3 470.2
Axial Force (kN) 305 325 345 365 385 405 425 445 465
(a) gamma=1.0 Axial Force (kN)
(b) gamma=0.8
86
gamma
gamma
alpha
0.4 -305.7 -253.8 0.4 -311.8 -248.8
gamma
0.4 -287.1 -270.4 0.4 -305.3 -254.1
alpha
0.4 -70.300 -62.300 0.4 -71.1 -61.5
0 -73.100 -59.800
0 -73.9 -59.0
-74.000 -72.000 -70.000 -68.000 -66.000 -64.000 -62.000 -60.000
-74.000 -72.000 -70.000 -68.000 -66.000 -64.000 -62.000 -60.000
Shear Force (kN)
Shear Force(kN)
(a) gamma=1.0
(b) gamma=0.8
88
gamma
0.4 -68.6 -63.8 0.4 -71.4 -61.2
alpha
0.4 -191.6 -153.3 0.4 -194.4 -150.6
gamma
0.4 -180.5 -163.9 0.4 -193.5 -151.4
alpha
0.4 -962.3 -846.4 0.4 -971.9 -837.9
1 -941.5 -864.3
1 -901.8
gamma
gamma
alpha
0.4 306.6 383.4 0.4 303.0 387.6
gamma
0.4 307.8 382.1 0.4 332.3 355.4
alpha
alpha
gamma
0.4 275.8 296.5 0.4 255.4 318.5
Young’s modulus of the materials (steel and concrete) and also in live load. The
α and β ranging from total certainty (α=1, β=1) to total uncertainty (α=0, β=0). The
1 The uncertain values of structural response in the first row of correspond to the
presence of live load uncertainty alone (0≤β≤1 and α=1). Further, the uncertain
2 Fuzzy membership functions (a), (b), (c) and (d) are plotted as described in the
Table 4.2.1, Table 4.2.2 and Table 4.2.3 show the concomitant variation of
horizontal and vertical displacements and rotations at nodes 2,3 and 4 respectively, for
Fig. 4.2.1 (a) and (b) represent the membership functions for horizontal
93
[1.0,1.0] in Fig. 4.2.1(a) and [0.566,1.456], [0.675,1.325] in Fig. 4.2.1 (b). Fig. 4.2.1
(c) and (d) represent the membership functions for horizontal displacement at node 2
(m) at α=1.0 and α=0.8. The uncertain horizontal displacements at β=0.0 and β=1.0
Fig. 4.2.2 (a) and (b) represent the membership functions for vertical
displacement at node 3 (m) at β=1.0 and β=0.8. In these figures, the uncertain
Fig. 4.2.2(b). Fig. 4.2.2 (c) and (d) represent the membership functions for vertical
displacement at node 3 (m) at α=1.0 and α=0.8. The uncertain vertical displacements
Fig. 4.2.3 (a) and (b) represent the membership functions for rotation at node 4
at β=1.0 and β=0.8. The uncertain rotations at α=0.0 and α=1.0 correspond to
[-0.963,2.963] in Fig. 4.2.3 (b). Fig. 4.2.3 (c) and (d) represent the membership
functions for rotation at node 4 (radians) at α=1.0 and α=0.8. The uncertain rotations
94
The normalised values of uncertain horizontal displacement in the first row of
above values, it is observed that the variation of horizontal displacement is less in the
case of material uncertainty (upper bound 12.1%) compared to live load uncertainty
displacement and rotation as well. This is because the material uncertainty adopted
moments
Table 4.2.4 through Table 4.2.10 show the concomitant variation of axial
forces, shear forces and bending moments for various combinations of α and β. Table
4.2.4 and Table 4.2.5 represent the variation of axial force (kN) in element 9 and
element 12 respectively. Table 4.2.6 represents the variation of shear force in the
deck just to the left of node 3. Table 4.2.7 represents the variation of bending moment
in the deck at node 4. Table 4.2.8 represents the variation of axial force (kN) in pylon
in element 17 (at node 2). Table 4.2.9 and Table 4.2.10 represent the variation of
axial force (kN) in cable 2 (at node 2) and cable 3 (at node 3) respectively. In the
above tables, tensile force and sagging moment are considered positive.
observed that the variation of axial force in deck is less in the case of material
95
uncertainty (upper bound 20.4%) compared to live load uncertainty (upper bound
162.5%). Similar behaviour is observed from all the other uncertain forces/moments
Fig. 4.2.4 (a) and (b) represent the membership functions for axial force in
deck in element 9 (kN) at β=1.0 and β=0.8. The uncertain axial forces at α=0.0 and
and [0.501,1.550], [0.675,1.325] in Fig. 4.2.4 (b). Fig. 4.2.4 (c) and (d) represent the
membership functions for axial force in deck in element 9 (kN) at α=1.0 and α=0.8.
The uncertain axial forces at β=0.0 and β=1.0 correspond to normalised values
4.2.4 (d).
Fig. 4.2.5 (a) and (b) represent the membership functions for axial force (kN) in deck
in element 12 at β=1.0 and β=0.8. The uncertain axial forces at α=0.0 and α=1.0
and [-1.399,-0.684] , [-1.200,-0.800] Fig. 4.2.5 (b) . Fig. 4.2.5 (c) and (d) represent
the membership functions for axial force in deck in element 9 (kN) at α=1.0 and
α=0.8. The uncertain axial forces at β=0.0 and β=1.0 correspond to normalised
values [-2.001,0.000], [-1.0, -1.0] in Fig. 4.2.5 (c) and [-2.063,0.0], [-1.032,-0.970] in
Fig. 4.2.6 (a) and (b) represent the membership functions for shear force just
to the left of node 3 (kN) at β=1.0 and β=0.8. The uncertain shear forces at α=0.0 and
96
α=1.0 correspond to normalised values [-1.102,-0.905] , [-1.0,-1.0] in Fig. 4.2.6 (a)
and [-1.360,-0.673] , [-1.243,-0.756] in Fig. 4.2.6 (b). Fig. 4.2.6 (c) and (d) represent
the membership functions for shear force in deck in element 9 (kN) at α=1.0 and
α=0.8. The uncertain shear forces at γ=0.0 and γ=1.0 correspond to normalised values
Fig. 4.2.7 (a) and (b) represent the membership functions for bending moment
(kNm) in deck at node 4 at γ=1.0 and γ=0.8. The uncertain bending moments at
Fig. 4.2.7 (a) and [-1.512,-0.522] , [-1.313,-0.686] in Fig. 4.2.7 (b). Fig. 4.2.7 (c) and
(d) represent the membership functions for bending moment at node 4 at α=1.0 and
α=0.6. The uncertain bending moments at β=0.0 and β=1.0 correspond to normalised
Fig. 4.2.8 (a) and (b) represent the membership functions axial force (kN) in
pylon in element 17 (at node 11) at β=1.0 and β=0.8. The uncertain axial forces at
Fig. 4.2.8 (a) and [-1.342,-0.718] , [-1.200,-0.800] in Fig. 4.2.8 (b) . Fig. 4.1.7 (c) and
(d) represent the membership functions for axial force in deck in element 9 (kN) at
α=1.0 and α=0.8. The uncertain axial forces at β=0.0 and β=1.0 correspond to
97
normalised values [-2.00,0.00], [-1.000,-1.00] Fig. 4.2.8 (c) and [-2.044,0.00] ,
Fig. 4.2.9 (a) and (b) represent the membership functions axial force (kN) in
cable 2 (at node 2) at β=1.0 and β=0.8. The uncertain axial forces at α=0.0 and α=1.0
[0.620,1.454] , [0.775,1.225] in Fig. 4.2.9 (b). Fig. 4.2.9 (c) and (d) represent the
membership functions for axial force in cable 2 at α=1.0 and α=0.6. The uncertain
Fig. 4.2.10 (a) and (b) represent the membership functions axial force (kN) in
cable 3 (at node 3) at β=1.0 and β=0.8. The uncertain axial forces at α=0.0 and α=1.0
[0.601,1.468] , [0.758,1.242] in Fig. 4.2.10 (b). Fig. 4.2.10 (c) and (d) represent the
membership functions for axial force in cable 3 at α=1.0 and α=0.6. The uncertain
[1.0,1.0] in Fig. 4.2.10 (c) and [-0.236,2.34], [0.926,1.077] in Fig. 4.2.10 (d).
represented by figures (a) and (c) of Fig. 4.2.1 through Fig. 4.2.10 are triangular
membership functions represented by figures (c) and (d) of Fig. 4.2.1 through Fig.
98
Table. 4.2.1 Cable stayed Bridge – Concomitant Variation
of Horizontal displacement of node 2 (×10-4 metres) w.r.t α and β
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
99
Table. 4.2.4 Cable stayed Bridge – Concomitant Variation
of Axial Force in deck in element 9 (kN) w.r.t α and β
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [175.99,175.99] [118.80,233.19] [61.61,290.38] [4.42,347.57] [-52.77,404.77] [-109.96,461.94]
100
Table. 4.2.7 Cable stayed Bridge – Concomitant Variation
of Bending Moment in deck at node 4 (kNm) w.r.t α and γ
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
101
Table. 4.2.9 Cable stayed Bridge – Concomitant Variation
of Axial Force in cable 2 (kN) w.r.t α and β
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
102
1 3.275E-04 1 2.211E-04 4.340E-04
alpha
0.4 3.046E-04 3.511E-04
0.4 1.995E-04 4.593E-04
0 2.896E-04 3.672E-04
2.8E-04 2.9E-04 3.0E-04 3.1E-04 3.2E-04 3.3E-04 3.4E-04 3.5E-04 3.6E-04 3.7E-04 0 1.854E-04 4.767E-04
1.75E-04 2.25E-04 2.75E-04 3.25E-04 3.75E-04 4.25E-04 4.75E-04
Horizontal Displacement (m)
Horizontal Displacement (m)
(a) beta=1.0
(b) beta=0.8
103
beta
beta
alpha
0.4 -1.750E-02 -1.200E-02 0.4 -3.066E-02 9.100E-04
beta
beta
alpha
alpha
alpha
0.4 155.72 197.17 0.4 100.27 256.46
beta
0.4 4.42 347.57 0.4 -0.57 356.49
alpha
0.4 -140.51 -116.71 0.4 -168.61 -93.37
beta
beta
alpha
0.4 -33.54 -29.81 0.4 -41.51 -22.34
beat
beta
Fig. 4.2.6 Membership Functions for Shear Force in deck to left of node 3
1 -83.2 1 -109.2 -57.1
alpha
0.4 -92.2 -74.3 0.4 -119.0 -48.9
beta
beta
alpha
0.4 -387.33 -339.53 0.4 -464.79 -271.63
beta
0.4 -579.82 -144.96 0.4 -592.73 -141.82
0 -724.78 0 0 -740.91 0
-800 -710 -620 -530 -440 -350 -260 -170 -80 10 -825 -725 -625 -525 -425 -325 -225 -125 -25
Axial Force (kN) Axial Force (kN)
(c) alpha=1.0 (d) alpha=0.8
alpha
0.4 138.26 173.35 0.4 105.67 211.03
beta
beta
alpha
0.4 115.41 144.65 0.4 85.87 178.13
beta
beta
dead load and uniformly distributed live load (owing to service loads). Uncertainty is
introduced in the Young’s modulus of the materials (Steel and Concrete), in live load
and also in mass density (α,β and γ). The structural response of the cable-stayed
combinations of α and β ranging from total certainty (α=1, β=1) to total uncertainty
(α=0, β=0) are considered at γ=1.0 (crisp) and γ=0.8 (±1% variation) respectively.
Table 4.3.1, Table 4.3.3 and Table 4.3.5 show the concomitant variation of
for various combinations of α and β at γ=1.0. Also, Table 4.3.2, Table 4.3.4 and
Table 4.3.6 show the concomitant variation of horizontal and vertical displacements
γ=0.8.
In Table 4.3.1, the uncertain values of horizontal displacement in the first row
above values, it is observed that the variation of horizontal displacement is less in the
113
case of material uncertainty (upper bound 12.2%) compared to live load uncertainty
displacement and rotation as well. This is because the material uncertainty adopted
Table 4.3.1 and Table 4.3.2 represent the variation of uncertain horizontal
displacement (×10-4m) at node 2 for various combinations of material and live load
Fig. 4.3.1 (a) and (b) represent the membership functions for horizontal
[1.0,1.0] in Fig. 4.3.1 (a) and [0.784,1.226], [0.898,1.102] in Fig. 4.3.1 (b).
Fig. 4.3.1 (c) and (d) represent the membership functions for horizontal displacement
Fig. 4.3.2 (a) and (b) represent the membership functions for horizontal
[0.989,1.011] in Fig. 4.3.2 (a) and [0.774,1.238], [0.887,1.114] in Fig. 4.3.2 (b).
Fig. 4.3.2 (c) and (d) represent the membership functions for horizontal displacement
114
4.3.1.2 Uncertain vertical displacement at node 3
Fig. 4.3.3 (a) and (b) represent the membership functions for vertical
[-1.0,-1.0] in Fig. 4.3.3 (a) and [-1.591,-0.424] , [-1.274,-0.726] in Fig. 4.3.3 (b).
Fig. 4.3.3 (c) and (d) represent the membership functions for vertical displacement at
Fig. 4.3.4 (a) and (b) represent the membership functions for vertical
[-1.030,-0.970] in Fig. 4.3.4 (a) and [-1.620,-0.395] ,[-1.304,-0.694] in Fig. 4.3.4 (b).
Fig. 4.3.4 (c) and (d) represent the membership functions for vertical displacement at
Table 4.3.5 and Table 4.3.6 represent the variation of uncertain rotation
(×10-3 radians) at node 4 for various combinations of material and live load
Fig. 4.3.5 (a) and (b) represent the membership functions for uncertain rotation
α=0.0 and α=1.0 correspond to normalised [0.317,1.686] , [1.0,1.0] in Fig. 4.3.5 (a)
115
and [-0.289,2.305] , [0.390,1.610] in Fig. 4.3.5 (b). Fig. 4.3.5 (c) and (d) represent the
α=0.6,γ=1.0. In these figures, the uncertain vertical displacements at β=0.0 and β=1.0
Fig. 4.3.6 (a) and (b) represent the membership functions for uncertain rotation
Fig. 4.3.6 (a) and [-0.353,2.371] , [0.325,1.674] in Fig. 4.3.6 (b). Fig. 4.3.6 (c) and (d)
and α=0.6,γ=0.8. The uncertain vertical displacements at β=0.0 and β=1.0 correspond
Table 4.3.7 through Table 4.3.10 represent the variation of axial force (kN) in
element 9 and element 12 respectively. Table 4.3.11 and Table 4.3.12 represent the
variation of shear force in the deck just to the left of node 3 at γ=1.0 and γ=0.8
respectively. Table 4.3.13 and Table 4.3.14 represent the variation bending moment
in the deck at node 4 at γ=1.0 and γ=0.8 respectively. Table 4.3.15 and Table 4.3.16
represent the variation of axial force (kN) in pylon in element 17 (at node 2) at γ=1.0
and γ=0.8 respectively. Table 4.3.17 and Table 4.3.18 represent the variation of axial
force (kN) in cable 2(at node 2) at γ=1.0 and γ=0.8 respectively. Table 4.3.19 and
116
Table 4.3.20 represent the variation of axial force (kN) in cable 2(at node 2) at γ=1.0
In Table 4.3.7, the uncertain values of axial force in element 9 in the first row
values of axial force in element 9 in the first column correspond to normalised values
[0.810,1.204] respectively. From the above values, it is observed that the variation of
axial force in element 9 is less in the case of material uncertainty (upper bound
31.0%) compared to live load uncertainty (upper bound 137.1%). Similar behaviour
Fig. 4.3.7 (a) and (b) represent the membership functions for uncertain axial
force (kN) in element 9 at β=1.0,γ=1.0 and β=0.8,γ=1.0. The axial forces at α=0.0
and α=1.0 correspond to normalised values [0.810,1.204], [1.0,1.0] in Fig. 4.3.7 (a)
and [0.713,1.312], [0.898,1.102] in Fig. 4.3.7 (b). Fig. 4.3.7 (c) and (d) represent the
α=0.6,γ=1.0. The uncertain axial forces at β=0.0 and β=1.0 correspond to normalised
Fig. 4.3.8 (a) and (b) represent the membership functions for uncertain axial
force (kN) in element 9 at β=1.0,γ=0.8 and β=0.8,γ=0.8. The axial forces at α=0.0
and α=1.0 correspond to normalised values [0.8,1.215], [0.989,1.011] in Fig. 4.3.8 (a)
and [0.703,1.324], [0.887,1.113] in Fig. 4.3.8 (b). Fig. 4.3.9 (c) and (d) represent the
117
α=0.6,γ=0.8. The uncertain axial forces at β=0.0 and β=1.0 correspond to normalised
Fig. 4.3.9 (a) and (b) represent the membership functions for uncertain axial
force (kN) in element 12 at β=1.0,γ=1.0 and β=0.8,γ=1.0. The axial forces at α=0.0
and [-1.239,-0.801] , [-1.063,-0.937] in Fig. 4.3.9 (b). Fig. 4.3.9 (c) and (d) represent
the membership functions for axial force (kN) in element 12 at α=1.0,γ=1.0 and
α=0.6,γ=1.0. The uncertain axial forces at β=0.0 and β=1.0 correspond to normalised
Fig. 4.3.10 (a) and (b) represent the membership functions for uncertain axial
force (kN) in element 12 at β=1.0,γ=0.8 and β=0.8,γ=0.8. The axial forces at α=0.0
Fig. 4.3.10 (a) and [-1.247,-0.795] , [-1.070,-0.930] in Fig. 4.3.10 (b). Fig. 4.3.10 (c)
and (d) represent the membership functions for axial force (kN) in element 12 at
α=1.0,γ=0.8 and α=0.6,γ=0.8. The uncertain axial forces at β=0.0 and β=1.0
Fig. 4.3.11 (a) and (b) represent the membership functions for the uncertain
shear force (kN) in deck just to the left of node 3 at β=1.0,γ=1.0 and β=0.8,γ=1.0. The
118
uncertain shear forces at α=0.0 and α=1.0 correspond to normalised values
in Fig. 4.3.11 (b). Fig. 4.3.11 (c) and (d) represent the membership functions for
shear force (kN) in deck just to the left of node 3 at α=1.0,γ=1.0 and α=0.6,γ=1.0. The
Fig. 4.3.11 (d ).
Fig. 4.3.12 (a) and (b) represent the membership functions for the uncertain
shear force (kN) in deck just to the left of node 3 at β=1.0,γ=0.8 and β=0.8,γ=0.8.
The uncertain shear forces at α=0.0 and α=1.0 correspond to normalised values
[-1.087,-0.913] in Fig. 4.3.12 (b). Fig. 4.3.12 (c) and (d) represent the membership
functions for shear force (kN) in deck just to the left of node 3 at α=1.0,γ=0.8 and
α=0.6,γ=0.8. The uncertain shear forces at β=0.0 and β=1.0 correspond to normalised
Fig. 4.3.13 (a) and (b) represent the membership functions for the uncertain
Fig. 4.3.13 (b). Fig. 4.3.13 (c) and (d) represent the membership functions for bending
119
moments at β=0.0 and β=1.0 correspond to normalised values [-1.511,-0.489],
[-1.0,-1.0] in Fig. 4.3.13 (c) and [-1.594,-0.425] ,[-1.074,-0.928] in Fig. 4.3.13 (d).
Fig. 4.3.14 (a) and (b) represent the membership functions for the uncertain
[-1.113,-0.887] in Fig. 4.3.14 (b). Fig. 4.3.14 (c) and (d) represent the membership
functions for bending moment in deck at node 4 at α=1.0,γ=0.8 and α=0.6,γ=0.8. The
Fig. 4.3.15 (a) and (b) represent the membership functions for the uncertain
Fig. 4.3.15 (b). Fig. 4.3.15 (c) and (d) represent the membership functions for axial
[-1.0,-1.0] in Fig. 4.3.15 (c) and [-1.344,-0.684] , [-1.045,-0.958] in Fig. 4.3.15 (d).
Fig. 4.3.16 (a) and (b) represent the membership functions for the uncertain
axial force (kN) in pylon in element 17 at β=1.0,γ=0.8 and β=0.8,γ=0.8. The uncertain
120
Fig. 4.3.16 (a). Fig. 4.3.16 (c) and (d) represent the membership functions for axial
axial forces at β=0.0 and β=1.0 correspond to normalised values [-1.294, -0.706],
[-1.007, -0.993] in Fig. 4.3.16 (c) and [-1.352, -0.677], [-1.052, -0.951] in
Fig. 4.3.17 (a) and (b) represent the membership functions for the uncertain
axial force (kN) in cable 2 (at node 2) at β=1.0,γ=1.0 and β=0.8,γ=1.0. The uncertain
[1.000,1.000] in Fig. 4.3.17 (a) and [0.752,1.286] , [0.921,1.079] in Fig. 4.3.17 (a).
Fig. 4.3.17 (c) and (d) represent the membership functions for axial force (kN) in
cable 2 (at node 2) at α=1.0,γ=1.0 and α=0.6,γ=1.0. The uncertain axial forces at
Fig. 4.3.18 (a) and (b) represent the membership functions for the uncertain
axial force (kN) in cable 2 (at node 2) at β=1.0,γ=0.8 and β=0.8,γ=0.8. The uncertain
[0.993,1.009] in Fig. 4.3.18 (a) and [0.745,1.295] , [0.914,1.088] in Fig. 4.3.18 (b).
Fig. 4.3.18 (c) and (d) represent the membership functions for axial force (kN) in
cable 2 (at node 2) at α=1.0,γ=0.8 and α=0.6,γ=0.8. The uncertain axial forces at
121
4.3.2.7 Axial force (kN) in cable 3 (at node 3)
Table 4.3.19 and Table 4.3.20 represent the variation of uncertain axial force
(kN) in cable 3 (at node 3) for various combinations of material and live load
Fig. 4.3.19 (a) and (b) represent the membership functions for the uncertain
axial force (kN) in cable 3 (at node 3) at β=1.0,γ=1.0 and β=0.8,γ=1.0. The uncertain
[1.0,1.0] in Fig. 4.3.19 (a) and [0.762,1.271] , [0.934,1.066] in Fig. 4.3.19 (b).
Fig. 4.3.19 (c) and (d) represent the membership functions for axial force (kN) in
cable 3 (at node 3) at α=1.0,γ=1.0 and α=0.6,γ=1.0. The uncertain axial forces at
Fig. 4.3.20 (a) and (b) represent the membership functions for the uncertain
axial force (kN) in cable 3 (at node 3) at β=1.0,γ=0.8 and β=0.8,γ=0.8. The uncertain
[0.992,1.008] in Fig. 4.3.20 (a) and [0.755,1.280], [0.926,1.074] in Fig. 4.3.20 (b).
Fig. 4.3.20 (c) and (d) represent the membership functions for axial force (kN) in
cable 3 (at node 3) at α=1.0,γ=0.8 and α=0.6,γ=0.8. The uncertain axial forces at
It is observed that membership functions (a) and (c) depicted in Fig. 4.3.1,
Fig. 4.3.3, Fig. 4.3.5, Fig. 4.3.7, Fig. 4.3.9, Fig. 4.3.11, Fig. 4.3.13, Fig. 4.3.15,
Fig. 4.3.17, Fig. 4.3.19 are triangular owing to the presence of a single uncertainty
(α or β).
122
It is further observed that membership functions (b) and (d) depicted in
Fig. 4.3.1, Fig. 4.3.3, Fig. 4.3.5, Fig. 4.3.7, Fig. 4.3.9, Fig. 4.3.11, Fig. 4.3.13,
Fig. 4.3.15, Fig. 4.3.17, Fig. 4.3.19 are trapezoidal in the presence of multiple
4.3.2, Fig. 4.3.4, Fig. 4.3.6, Fig. 4.3.8, Fig. 4.3.10, Fig. 4.3.12, Fig. 4.3.14, Fig. 4.3.16,
Fig. 4.3.18, Fig. 4.3.20 are trapezoidal in the presence of multiple uncertainties (α and
β and γ).
123
Table. 4.3.1 Cable stayed Bridge - Concomitant Variation
of Horizontal displacement of node 2 (×10-4 metres) w.r.t α and β at γ=1.0
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
124
Table. 4.3.4 Cable stayed Bridge - Concomitant Variation
of Vertical displacement of node 3 (×10-2 metres) w.r.t α and β at γ=0.8
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
125
Table. 4.3.7 Cable stayed Bridge - Concomitant Variation
of Axial Force in deck in element 9 (kN) w.r.t α and β at γ=1.0
β→ 1.0 0.8 0.6 0.4 0.2 0.0
α↓
126
Table. 4.3.10 Cable stayed Bridge - Concomitant Variation
of Axial Force in deck in element 12 (kN) w.r.t α and β at γ=0.8
β→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [-409.7,-404.1] [-435.3,-378.4] [-460.9,-352.8] [-486.5,-327.2] [-512.2,-301.6] [-537.8,-275.9]
0.8 [-99.8,-95.9] [-107.6, -88.3] [-115.4, -80.7] [-123.1, -73.0] [-130.9, -65.4] [-138.7, -57.8]
0.6 [-101.8, -94.0] [-109.7, -86.5] [-117.6, -78.9] [-125.4, -71.4] [-133.3, -63.8] [-141.2, -56.3]
0.4 [-103.8, -92.1] [-111.8, -84.6] [-119.8, -77.2] [-127.8, -69.7] [-135.7, -62.2] [-143.7, -54.8]
0.2 [-105.9, -90.2] [-114.0, -82.8] [-122.0, -75.4] [-130.1, -68.0] [-138.2, -60.6] [-146.2, -53.2]
0.0 [-108.0, -88.4] [-116.1, -81.1] [-124.3, -73.8] [-132.5, -66.5] [-140.6, -59.1] [-148.8, -51.8]
0.8 [-100.7, -95.1] [-108.5, -87.5] [-116.3, -79.9] [-124.0, -72.2] [-131.8, -64.6] [-139.6, -57.0]
0.6 [-102.6, -93.2] [-110.5, -85.7] [-118.4, -78.1] [-126.2, -70.6] [-134.1, -63.0] [-142.0, -55.5]
0.4 [-104.6, -91.3] [-112.6, -83.8] [-120.6, -76.4] [-128.6, -68.9] [-136.5, -61.4] [-144.5, -54.0]
0.2 [-106.7, -89.5] [-114.8, -82.1] [-122.8, -74.7] [-130.9, -67.3] [-139.0, -59.9] [-147.0, -52.5]
0.0 [-108.8, -87.6] [-116.9, -80.3] [-125.1, -73.0] [-133.3, -65.7] [-141.4, -58.3] [-149.6, -51.0]
127
Table. 4.3.13 Cable stayed Bridge - Concomitant Variation
of Bending Moment in deck at node 4 (kNm) w.r.t α andβ at γ=1.0
β→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [-255.4, -255.4] [-281.4, -229.3] [-307.5, -203.2] [-333.6, -177.1] [-359.7, -151.0] [-385.8, -124.9]
0.8 [-264.7, -246.1] [-291.1, -220.2] [-317.4, -194.4] [-343.7, -168.5] [-370.0, -142.7] [-396.4, -116.8]
0.6 [-274.3, -236.9] [-300.8, -211.2] [-327.4, -185.6] [-354.0, -159.9] [-380.6, -134.3] [-407.1, -108.6]
0.4 [-283.8, -227.6] [-310.6, -202.2] [-337.5, -176.7] [-364.3, -151.3] [-391.2, -125.8] [-418.0, -100.4]
0.2 [-293.5, -218.4] [-320.6, -193.2] [-347.7, -167.9] [-374.8, -142.6] [-402.0, -117.4] [-429.1, -92.1]
0.0 [-303.2 -209.3] [-330.6, -184.2] [-358.0, -159.1] [-385.4, -134.1] [-412.8, -109.0] [-440.3, -83.9]
0.8 [-267.5, -243.3] [-293.9, -217.4] [-320.2, -191.6] [-346.5, -165.7] [-372.8, -139.9] [-399.2, -114.0]
0.6 [-277.1, -234.1] [-303.6, -208.4] [-330.2, -182.8] [-356.8, -157.1] [-383.4, -131.5] [-409.9, -105.8]
0.4 [-286.6, -224.9] [-313.4, -199.5] [-340.3, -174.0] [-367.1, -148.6] [-394.0, -123.1] [-420.8, -97.70]
0.2 [-296.3, -215.7] [-323.4, -190.5] [-350.5, -165.2] [-377.6, -139.9] [-404.8, -114.7] [-431.9, -89.40]
0.0 [-306.1, -206.6] [-333.5, -181.5] [-360.9, -156.4] [-388.3, -131.4] [-415.7, -106.3] [-443.2, -81.20]
1.0 [-1264.2, -1264.2] [-1336.7, -1191.7] [-1409.1, -1119.2] [-1481.6, -1046.8] [-1554.1, -974.3] [-1626.6, -901.8]
0.8 [-1291.9, -1237.4] [-1366.0, -1166.4] [-1440.0, -1095.5] [-1514.1, -1024.6] [-1588.2, -953.7] [-1662.3, -882.8]
0.6 [-1320.3, -1211.2] [-1396.0, -1141.9] [-1471.8, -1072.5] [-1547.5, -1003.1] [-1623.3, -933.7] [-1699.0, -864.3]
0.4 [-1349.6, -1185.9] [-1427.1, -1118.0] [-1504.6, -1050.1] [-1582.0, -982.2] [-1659.5, -914.3] [-1737.0, -846.4]
0.2 [-1379.9, -1161.2] [-1459.1, -1094.8] [-1538.3, -1028.3] [-1617.6, -961.8] [-1696.8, -895.4] [-1776.0, -828.9]
0.0 [-1411.1, -1137.2] [-1492.1, -1072.2] [-1573.2, -1007.1] [-1654.2, -942.0] [-1735.3, -877.0] [-1816.3, -811.9]
128
Table. 4.3.16 Cable stayed Bridge - Concomitant Variation
of Axial Force in pylon in element 17 (at node 11) (kN) w.r.t α and β at γ=0.8
β→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [-1273.3,-1255.19] [-1345.77,-1182.71] [-1418.24,-1110.23] [-1490.7,-1037.76] [-1563.2, -965.28] [-1635.68, -892.8]
129
Table. 4.3.19 Cable stayed Bridge - Concomitant Variation
of Axial Force (kN) in cable 3 (at node 3) w.r.t α and β at γ=1.0
β→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [473.5,473.5] [442.1,504.9] [410.7,536.3] [379.4,567.7] [348.0,599.1] [316.6,630.5]
130
1 1.044E-03 1.044E-03 1 9.380E-04 1.151E-03
alpha
alpha
beta
beta
alpha
0.4 9.599E-04 1.131E-03 0.4 8.548E-04 1.239E-03
beta
0.4 7.134E-04 1.375E-03 0.4 6.675E-04 1.429E-03
alpha
alpha
beta
0.4 -8.64E-02 -8.40E-03 0.4 -9.26E-02 -2.80E-03
alpha
0.4 -5.76E-02 -3.73E-02 0.4 -7.08E-02 -2.44E-02
beta
0.4 -8.79E-02 -6.90E-03 0.4 -9.40E-02 -1.40E-03
alpha
alpha
beta
beta
alpha
0.4 6.440E-04 1.811E-03 0.4 -1.000E-04 2.565E-03
beta
beta
alpha
alpha
Fig. 4.3.7 Membership Functions for Axial Force in deck in element 9 at γ=1.0
1 554.92 567.4 1 497.73 624.6
alpha
alpha
beta
beta
Fig. 4.3.8 Membership Functions for Axial Force in deck in element 9 at γ=0.8
1 -406.9 1 -432.5 -381.2
alpha
0.4 -446.2 -370.6 0.4 -474.3 -347.2
beta
beta
Fig. 4.3.9 Membership Functions for Axial Force in deck in element 12 at γ=1.0
1 -409.70 -404.07 1 -435.27 -378.44
alpha
0.4 -449.22 -368.02 0.4 -477.32 -344.68
beta
beta
0.5
0.25
0.2 -512.16 -301.56 -544.69 -283.39
Fig. 4.3.10 Membership Functions for Axial Force in deck in element 12 at γ=0.8
1 -97.8
1 -105.5 -90.1
alpha
0.4 -103.8 -92.1 0.4 -111.8 -84.6
beta
0.4 -121.0 -74.7 0.4 -125.4 -71.4
Fig. 4.3.11 Membership Functions for Shear Force in deck to left of node 3 at γ=1.0
1 -98.6 -97.0 1 -106.3 -89.3
alpha
alpha
beta
0.4 -121.8 -73.9 0.4 -126.2 -70.6
Fig. 4.3.12 Membership Functions for Shear Force in deck to left of node 3 at γ=0.8
1 -255.4 1 -281.4 -229.3
alpha
beta
beta
0.4 -333.6 -177.1 0.4 -354.0 -159.9
alpha
0.4 -286.6 -224.9 0.4 -313.4 -199.5
beta
0.4 -336.4 -174.3 0.4 -356.8 -157.1
alpha
alpha
beta
0.4 -1481.6 -1046.8 0.4 -1547.5 -1003.1
Fig. 4.3.15 Membership Functions for Axial Force in pylon in element 17 at γ=1.0
1 -1273.3-1255.19 1 -1345.77 -1182.71
alpha
alpha
1 -1273.3 -1255.19
1 -1329.77 -1202.64
beta
0.4 -1490.72 -1037.76 0.4 -1557.03 -994.47
alpha
0.4 393.6 492.2 0.4 361.0 529.8
beta
beta
alpha
alpha
beta
beta
alpha
0.4 422.0 528.1 0.4 392.5 561.5
beta
0.4 379.4 567.7 0.4 348.5 607.8
alpha
alpha
beta
beta
i.e. displacements, rotations, forces and moments is performed and the relative
sensitivity is ascertained. The results are presented and discussed in the following
sections.
Fig. 4.4.1 depicts the sensitivity analysis of displacements of the deck under
the action of dead load at γ=1.0 with respect to percentage variation of material
property (E) about its mean value. In this figure, the relative sensitivity of horizontal
displacement, vertical displacement and rotation are studied. It is observed from this
plot that that the slopes of the plots of horizontal displacement, vertical displacement
and rotation are 3.054, 6.071,5.591 respectively. Fig. 4.4.2 depicts the sensitivity
analysis of displacements of the deck under the action of dead load at γ=0.8 with
respect to percentage variation of material property (E) about its mean value. It is
observed from this plot that that the slopes of the plots of horizontal displacement,
Fig. 4.4.3 depicts the sensitivity analysis of displacements of the deck under the action
of live load at α=1.0 with respect to percentage variation of live load about its mean
151
value. It is observed from this plot that that the slopes of the plots of horizontal
displacement, vertical displacement and rotation are 2.23, 4.413 and 4.372
respectively. Thus it is observed that in all these cases, vertical displacement has the
Fig. 4.4.4 depicts the sensitivity analysis of forces and moments of the deck
under the action of dead load at γ=1.0 with respect to percentage variation of material
property (E) about its mean value. In this figure, the relative sensitivity of axial force,
shear force and bending moment are studied. It is observed from this plot that that the
slopes of the plots of axial force, shear force and bending moment are 2.902, 2.218
Fig. 4.4.5 depicts the sensitivity analysis of forces and moments of the deck
under the action of dead load at γ=0.8 with respect to percentage variation of material
property (E) about its mean value. It is observed from this plot that that the slopes of
the plots of axial force, shear force and bending moment are 3.074, 2.41 and 3.566
respectively. Thus it is observed that in all these cases, bending moment has the
greatest sensitivity and shear force has the lowest sensitivity. Further, the sensitivities
of all the above quantities are more at γ=0.8 compared to γ=1.0 because the
corresponding slopes of axial force, shear force and bending moment are more in case
Fig. 4.4.6 depicts the sensitivity analysis of bending moment at nodes 2,3,4
and 5 on the deck slab under the action of dead load at γ=1.0 with respect to
152
percentage variation of material property (E) about its mean value. It is observed
from this plot that that the slopes of the plots of bending moment at these nodes are
16.208, 3.046, 3.280 and 3.648 respectively. Thus it is observed that the sensitivity of
bending moment varies along the span of the deck slab. The sensitivity at the node 2
is the highest because the bending moment increases from a value of zero at node 1 to
Fig. 4.4.7 and Fig. 4.4.8 depict the sensitivity analysis of axial force at nodes
2, 4 and 5 on the deck slab under the action of live load, at β=1.0 and β=0.8
respectively, with respect to percentage variation of material property (E) about its
mean value. It is observed from Fig. 4.4.7 that that the slopes of the plots of axial
force at these nodes are 4.068, 4.904 and 2.902 respectively. Thus it is observed that
the sensitivity of axial force varies along the span of the deck slab in this case. The
sensitivity at the node 4 is the highest and the sensitivity at node 5 is the lowest.
It is observed from Fig. 4.4.8 that that the slopes of the plots of axial force at
the nodes 2, 4 and 5 on the deck slab are 4.498, 2.246 and 2.592 respectively. Thus it
is observed that the sensitivity of axial force varies along the span of the deck slab in
this case. The sensitivity at the node 2 is the highest and the sensitivity at node 4 is
the lowest.
Fig. 4.4.9 depicts the sensitivity analysis of shear force at nodes 1,3, 4 and 5
on the deck slab under the action of live load, at β=1.0, with respect to percentage
variation of material property (E) about its mean value. It is observed from Fig. 4.4.9
that that the slopes of the plots of shear force at these nodes are 1.698, 2.012, 2.396
and 2.128 respectively. Thus it is observed that the sensitivity of shear force varies
along the span of the deck slab in this case. The sensitivity at the node 4 is the highest
153
35.000
30.355
30.000
Percentage Variation of displacement about mean
27.953
24.296
25.000
22.290
20.000 18.237
16.667
15.000 15.330
12.148
12.214
11.084
10.000
9.122
6.089 Horizontal Displacement
6.054 Vertical displacement
5.000 5.502
Rotation
3.021
0.000
0.000 0.000
0.0 1.0 2.0 3.0 4.0 5.0
Percentage Variation of E about mean
Fig. 4.4.1 Sensitivity Analysis for displacements (for dead load) at node 3 at gamma=1.0
35.0 34.670
32.282
30.0 28.611
Percentage Variation of displacement about mean
26.618
25.0
22.552
20.955
20.0
16.493 17.593
15.0 15.372
14.465
10.435
11.373
10.0 9.790
8.293
Horizontal displacement
4.345 Vertical displacement
5.0 5.248
4.248 Rotation
2.227
0.0
0.0 1.0 2.0 3.0 4.0 5.0
Percentage Variation of E about mean
Fig. 4.4.2 Sensitivity Analysis for displacements (for dead load) at node 3 at gamma=0.8
154
450.00 441.32
437.15
400.00
Percentage variation of displacement about mean
353.05
350.00
349.69
300.00
264.79
250.00 262.31
223.07
200.00 176.53
178.58
150.00 174.84
133.86
88.27
100.00 Horizontal displacement
89.24
87.38 Vertical displacement
50.00 Rotation
44.62
0.00
0.00 0.00
0.00
0.0 20.0 40.0 60.0 80.0 100.0
Percentage variation of live load about mean
18.0
16.40
16.0
Percentage Variation of Force/moment about mean
14.51
14.0 13.17
12.0 11.75
9.92 10.64
10.0
8.93
8.45
8.0
6.65 Axial Force
6.29 Shear Force
6.0 6.03
Bending Moment
3.05
2.0 2.07
0.00
0.0 0.00
0.0 1.0 2.0 3.0 4.0 5.0
Percentage Variation of E about mean
Fig 4.4.4 Sensitivity Analysis of Forces and Moments in Element 12 (at node 5) at gamma=1.0
155
18.0 17.83
16.0
15.37
Percentage Variation of Force/moment about mean
14.62
14.0
12.64
12.0 12.05
11.38
8.0 8.12
7.67
6.97
6.0
5.54 Axial Force
4.83 Shear Force
4.0 4.02 Bending Moment
3.43
2.0
1.51
1.34
1.00
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Percentage Variation of E about mean
Fig 4.4.5 Sensitivity Analysis of Forces and Moments in Element 12 (at node 5)at gamma=0.8
90.0
80.0 81.04
Perecntage variation of Bending moment about mean
70.0
64.50
60.0
50.0 Node 2
48.15
Node 3
Node 4
40.0
Node 5
31.97
30.0
20.0 18.24
14.62 15.23
15.92
11.00 16.40
12.25
10.0 7.35 13.17
9.24
3.69 6.20 9.92
0.00 3.12 6.65
0.0 0.00 3.34
0.00
0.0 1.0 2.0 3.0 4.0 5.0
Percentage variation of E about mean
Fig 4.4.6 Sensitivity Analysis for Bending Moment (due to dead Load) in deck slab at gamma=1
156
24.52
25.0
19.74
20.0 20.34
Percentage variation of axial force about mean
14.91 16.13
15.0
14.51
12.00 Node 2
10.01 11.75 Node 4
10.0 Node 5
7.93 8.93
5.04
5.0 6.03
3.94
3.05
0.00
0.0 0.00
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Percentage Variation of E about mean
Fig 4.4.7 Sensitivity Analysis for Axial Force (due to Live Load) in deck slab at beta=1.0
60.00
54.99
50.30
50.00
Percentage variation of Shear Force about mean
45.72
41.22
40.00 36.82
34.78 37.33
32.50 32.21 35.53
29.62
30.00 33.19
27.01
30.91
24.37 28.66
26.46
24.30 node 2
20.00
node 4
node 5
10.00
0.00
0.0 1.0 2.0 3.0 4.0 5.0
Percentage variation of E about mean
Fig. 4.4.8 Sensitivity Analysis of Axial Force due to live load in deck slab at beta=0.8
157
14.0
12.0 11.98
Perecntage variation of Shear Force about mean
10.64
10.0 9.53 10.06
8.45 8.49
8.0 7.98
7.11
6.74
6.29
6.0 5.93
4.72 Node 1
5.01
Node 3
4.0 4.17 Node 4
3.93
3.32 Node 5
2.35
2.0 2.07
1.95
1.65
0.00
0.0 0.00
0.0 1.0 2.0 3.0 4.0 5.0
Percentage variation of E about mean
Fig 4.4.9 Sensitivity Analysis for Shear Force (due to Live Load) in deck slab at beta=1.0
4.5 Discussion
uncertainties on the structural response of cable-stayed bridges and the results are
tabulated. The following common characteristics are be observed in all the tables:
a) It is observed that the width of the interval increases along and across each
table.
158
b) The variation with reference to material uncertainty is found to be more in
3. Membership functions are plotted for various combinations of α,β and γ in the
moments. Triangular and trapezoidal membership functions are obtained under the
and moments at a given node as well as over a set of given nodes. The following
c) Bending moment is found to have the greatest sensitivity while the shear force
d) Sensitivity of structural response is found to vary from node to node along the
The conclusions and recommendations for future work are addressed in the
succeeding chapter.
159
Chapter 5
Conclusions
and
Recommendations for future work
Chapter 5
5.1 Conclusions
On the basis of the work carried out on cable-stayed bridges using fuzzy-finite
far in the earlier studies. In the present study, overestimation of the structural
by
3. Sensitivity analysis is a new and useful concept proposed in the present study.
161
a) Structural response is more sensitive to variation in material property
that even a smaller variation in Young’s modulus may cause the actual
behaviour predicted by the designer. This brings out the need for a
and live load may change from their nominal values, thus leading to
behaviour.
162
5.2 Recommendations for Future Work
163
REFERENCES
Abdel-Tawab, K. Noor, A. K., (1999). "A Fuzzy-set Analysis for a Dynamic Thermo-
elasto-viscoplastic Damage Response," Computers and Structures Vol. 70, pp 91-107.
Akpan, U.O., Koko, T.S., Orisamolu, I.R., Gallant, B.K., (2001), "Practical fuzzy
Finite element analysis of structures", Finite Elements in Analysis and Design 38
(2001) 93}111. (Elsevier Publishing).
Bernardini, A. and Tonon, F. (1996). "A combined fuzzy and random-set approach to
the multi-objective optimization of uncertain systems," Proc, 7th ASCE EMD/STD
Joint Specialty Conference on Probabilistic Mechanics and Structural Reliability,
WPI, MA, August 7-9, 314-317.
Köyluoglu, U., Cakmak, A.S., and Nielsen, S. R. K. (1995). "Interval Algebra to Deal
with Pattern Loading and Structural Uncertainties, " Journal of Engineering
Mechanics, November, 1149-1157.
164
Köyluoglu, U. and Elishakoff, I. (1998). "A Comparison of Stochastic and Interval
Finite Elements Applied to Shear Frames with Uncertain Stiffness Properties",
Computers and Structures, Vol. 67, No. 1-3, pp.91-98.
Kulpa Z., Pownuk A., Skalna I., (1998) Analysis of linear mechanical structures with
uncertainties by means of interval methods. Computer Assisted Mechanics and
Engineering Sciences, vol. 5, pp.443-477.
Muhanna, R. L. and Mullen, R. L., (2000). “Sharp Enclosure for Material Uncertainty
in Solid and Structural Mechanics-Interval Based Approach”, Proc., 8th ASCE
Specialty Conference on Probabilistic Mechanics and Structural Reliability,
University of Notre Dame, Notre Dame, Indiana, July, 24-26.
165
Nakagiri, S. and Yoshikawa, N. (1996). "Finite Element Interval Estimation by
Convex Model, " Proc., 7th ASCE EMD/STD Joint Specialty Conference on
Probabilistic Mechanics and Structural Reliability, WPI, MA, August 7-9, 278-281.
Nakagiri, S. and Suzuki, K. (1999). "Finite Element Interval analysis of external loads
identified by displacement input uncertainty," Comput. Methods Appl. Mech. Engrg.
168, pp. 63-72.
Rao, S. S., Sawyer, P. (1995). "Fuzzy Finite Element Approach for Analysis of
Imprecisely Defined Systems, " AIAA Journal, Vol. 33, No. 12, pp 2364-2370.
166
Rao, S. S., Li Chen, (1998). "Numerical Solution of Fuzzy Linear Equations in
Engineering Analysis, " Int. J. Numer. Meth. Engrg. Vol. 43, pp 391-408.
Rump, S. M. (1990). " Rigorous Sensitivity Analysis for Systems of Linear and
Nonlinear Equations, "Mathematics of Computations, Vol. 54, 190, pp. 721-736.
Troitsky. M.S. DSC; “Cable-Stayed Bridges: Theory and Design”, Crosby Lockwood
Staples, London, 1972.
167
Appendices
APPENDIX-A
1. Introduction
Fuzzy set theory was first developed by Zadeh (Zadeh, 1965) to model
Thus the fuzzy set theory encompasses and extends the scope of classical set theory.
The difference between an ordinary subset A and a fuzzy subset A can be explained
as follows:
µA(x) ∈ [0,1]
which shows that an element of E belongs to or does not belong to A, according to the
For the same referential set E, a fuzzy subset A will be defined by its
characteristic function, the membership function µA(x) , which takes its values in the
interval [0,1].
There are several methods for the representation of fuzzy numbers. In one
such representation, the possible range of a value is used to represent the set
where A represents the set of x each with membership function µA(x). Alternately,
where 0,0.1,0.2,0.3 etc. represent the membership function (or in other words, the
degree of belongingness to the given fuzzy s et ) of the values 0,1,2 etc. respectively.
169
1 1.000
0 0.950 1.050
0.950 0.970 0.990 1.010 1.030 1.050
Variation about the mean value
In general, for discrete fuzzy sets, the number of elements in the fuzzy sets
express fuzzy numbers as sets of lower and upper bounds of a finite number of α-cut
subsets.
α-Cut Representation
At a level of α from the x- axis a cut is made (α-cut) to extract an ordered pair
in a closed interval form [xl, xu]. The α-cut can be taken anywhere ranging from α = 0
ordered pair [xl, xu] where xl ≤ xu. In case xl = xu, the interval is called a fuzzy- point
170
considered to be 1 and the minimum level of presumption is considered to be 0. The
Aα = {x∈R, µA(x)≥α}.
(α1<α2) ⇒ (Aα2⊂Aα1)
or
The fuzzy numbers thus defined are known as intervals. Thus in the
Interval Algebra
Herzberger, 1983 and Jansson, 1991) . However, some of the properties of interval
171
Validity of laws of algebra
Commutativity
Associativity
A+ (B+C) = (A+B) +C --- Addition
A× (B×C) = (A×B)×C --- Multiplication
Sub-Distributivity
A× (B+C) ⊆ A×B + A×C
causes overestimation of result. Further, when two interval numbers, which represent
the same physical quantity are operated upon, obtaining a sharp enclosure for the
This is because of failure of some algebraic laws that are valid in real
arithmetic. An overestimated solution encloses the solution hull but the width of the
solution obtained is very large, there by rendering the result very conservative,
overestimation arise owing to the following reasons (Muhanna and Mullen, 2001):
in interval algebra).
172
2. When one or several variables occur more than once in an interval expression.
quantity, interval algebra treats them as separate variables having their own
the uncertain Young’s modulus needs to be factored out from the element
stiffness matrix and be presented as a single multiplier for the entire stiffness
matrix
Young’s modulus), some elements have common nodes and the elements of
(Muhanna and Mullen, 2001) in which elements are kept disassembled throughout the
course of solution. By doing so, the coupling that usually occurs in the conventional
finite element formulation is delayed. The Lagrange multiplier method imposes the
173
APPENDIX- B
Here, [C] is a crisp constraint matrix. Using the Lagrange multiplier method,
δ∏* = 0 - (4)
Here, [K̃], {U} and {λ} are deterministic in nature. Also, {λ} is the vector of
Lagrange multipliers.
model to force the displacements of the slave nodes to match those of the master
Eq. 1.
[C]{U}={0} - (6)
174
Using Eq. 6, Eq. 5 can be modified to in the case of element-by element
model as
where [K] is the interval stiffness matrix, [D] is an diagonal matrix containing the
value of uncertain Young’s modulus and [S] is a crisp singular matrix, each of size
n×n. Also, {λ} = interval internal force vector of size n×1. Here, n is the product of
number of degrees of freedom per each element and the total number of elements in
the structure.
where [M] is an interval matrix of size n×m where m is the number of elements. Also,
{δ} is an interval vector of size m×1 and contains the inverse values of the Young’s
modulus for each of m elements. The solution of Eq. 10 represents the solution of
Eq. 9 by approximating the interval force vector of internal forces {λ} by its mid-point
175
APPENDIX ‘C’
1. Introduction
distinct structures as case studies in the following section. All structures considered
1. A fixed beam with a uniformly distributed load over the entire span
3. A plane frame with point loads and moments acting at the joints and uniformly
In all the above case studies, the simultaneous effect of material and live load
uncertainties is considered. All the case studies are modelled by using modified
bending moments obtained using the above approach are plotted with reference to
material and live load uncertainties. Each of the case studies is explained in detail
below:
problem was adapted from the work of Akpan (Akpan et al, 2001). The fixed beam
has the following properties (after conversion from U.S. Customary units) (Table.
C-1). The fixed beam is subjected to a uniformly distributed load over the entire
span. The uncertainty associated with Young’s modulus is denoted by α where as the
176
Table. C-1 Material and Geometric Properties of Fixed beam
Span 10.16 metres (400 inches)
Area of Cross section 0.07165 m2 (111.06 inch2)
Moment of Inertia 1.2486×10-3 m4 (3000 inch4)
Uniformly distributed load 70.076 kN/m (400lb/inch)
Location of load Entire span
Young’s Modulus (E) 206.9 GPa (30×10-6 psi)
Material Uncertainty of E (αmax) 4 percent (±2% about mean value of E)
Load Uncertainty (βmax) 20 percent (±10% about mean value of load)
Fig. C-2 and Fig. C-3 depict the membership functions of material and load
the normalized values of uncertain Young’s modulus, taking the crisp values as 1.0.
The membership function for material uncertainty shows the normalized values of
live load, taking the crisp value as 1.0. Table. C-2 represents the simultaneous
observed that the width of the interval increases along and across the table. The
variation with reference to material uncertainty is less compared to the variation with
reference to load uncertainty because more uncertainty (±10%) is associated with load
Table. C-3 shows a comparison of the results obtained by the present approach
with the results obtained by Akpan (Akpan et al, 2001) using the response surface
methodology. Fig. C-4 represents the membership function for mid-span displacement
(mm) for three cases of uncertainty (E and Load, Load only and E only). It is observed
that the present results compare very well with the results of Akpan et al for all the
Table. C-4 and Table. C-5 represent the simultaneous variation of fixed end
moment and fixed end shear force for various combinations of α and β. The
177
displacement for various levels of material uncertainty α while the load is crisp
(β=1.0). Fig. C-6 depicts the variation of mid-span displacement at various levels of
load uncertainty β at α=0.5. Fig. C-7 and Fig.C-8 represent the membership functions
for fixed end moment at β=0.75 and α=1.0 respectively. In general it is observed that
178
Table. C-2 Fixed Beam – Concomitant Variation of mid-span displacement (×10-3 metres)
w.r.t. α and β
β→
1.0 0.75 0.50 0.25 0.0
α↓
1.0 [-7.527,-7.527] [-7.715,-7.339] [-7.904,-7.151] [-8.092,-6.963] [-8.280,-6.774]
0.75 [-7.565,-7.490] [-7.754,-7.302] [-7.943,-7.115] [-8.132,-6.928] [-8.322,-6.741]
0.50 [-7.603,-7.453] [-7.793,-7.266] [-7.984,-7.080] [-8.174,-6.893] [-8.364,-6.707]
0.25 [-7.642,-7.416] [-7.833,-7.230] [-8.024,-7.045] [-8.215,-6.859] [-8.406,-6.674]
0.0 [-7.681,-7.379] [-7.873,-7.195] [-8.065,-7.010] [-8.257,-6.826] [-8.450,-6.641]
Table. C-3 Comparison of present results with Response Surface Approach at Degree of
Belief (α or β) = 0.0 (Mid-span displacements)
Akpan (mm) Present (mm)
Uncertainty Akpan (inch) Normalised Normalised
Interval Interval
interval interval
Load and E [-0.332, -0.2615] [-8.433,-6.642] [-1.119, -0.881] [-8.450, -6.641] [-1.120,0.880]
Load only [-0.326,0.267] [-8.280,-6.782] [-1.099,-0.901] [-8.280, -6.774] [-1.100,-0.900]
E only [-0.303, -0.290] [-7.671,-7.366] [-1.02,0.980] [-7.681, -7.379] [-1.02,0.9799]
Table. C-4 Fixed Beam – Concomitant Variation of Fixed end moment (kNm)
w.r.t. α and β
β→
1.0 0.75 0.50 0.25 0.0
α↓
1.0 [-602.80, -602.80] [-617.87, -587.73] [-632.94, -572.66] [-648.01, -557.59] [-663.08,-542.52]
0.75 [-608.68,-597.00] [-623.90, -582.08] [-639.12, -567.15] [-654.34, -552.23] [-669.55,-537.30]
0.50 [-614.66, -591.29] [-630.02, -576.51] [-645.39, -561.73] [-660.75, -546.94] [-676.12,-532.16]
0.25 [-620.72, -585.66] [-636.24, -571.02] [-651.75, -556.38] [-667.27, -541.74] [-682.79,-527.10]
0.0 [-626.87, -580.12] [-642.54, -565.61] [-658.21, -551.11] [-673.89, -536.61] [-689.56,522.10]
Table. C-5 Fixed Beam - Concomitant Variation of Fixed end Shear Force (kN)
w.r.t. α and β
β→
1.0 0.75 0.50 0.25 0.0
α↓
1.0 [355.98,355.98] [347.08,364.88] [338.18,373.78] [329.28,382.68] [320.38,391.58]
0.75 [352.82,359.19] [344.00,368.17] [335.18,377.15] [326.36,386.13] [317.54,395.11]
0.50 [349.71,362.44] [340.96,371.50] [332.22,380.56] [323.48,389.62] [314.74,398.68]
0.25 [346.63,365.74] [337.97,374.88] [329.30,384.02] [320.64,393.17] [311.97,402.31]
0.0 [343.60,369.08] [335.01,378.31] [326.42,387.54] [317.83,396.77] [309.24,405.99]
179
1 1.000
70.076 kN/m
10.16 m
0 0.980 1.020
0.980 0.985 0.990 0.995 1.000 1.005 1.010 1.015 1.020
Design Quanity
Fig. C-2 Fixed Beam - Membership Function for Material Uncertainty
1 1
-7.527
180
1.00
-7.715 -7.339
-7.984 -7.080
0.50 -7.603 -7.453
0.5 0.95 1.05
-7.904 -7.151
-8.215 -6.859
0.25 -7.642 -7.416
0.25 0.925 1.075 -8.092 -6.963
Fig. C-3 Fixed Beam - Membership Function for Load Uncertainty Fig. C-4 Fixed Beam - Displacement (mm) at the mid-span under fuzzy-static load
-7.603E-03 -7.453E-03
1.00 -7.527E-03 1.00
beta
0.50 -7.984E-03 -7.080E-03
0.50 -7.603E-03 -7.453E-03
beta
alpha
Fig. C-7 Fixed Beam - Membership Function for Fixed end moment at beta=0.75 Fig. C-8 Fixed Beam - Membership Function for Fixed End Moment at alpha=1.0
3. Preliminary Case Study-2 – Propped Cantilever
The second preliminary case study considered here is a propped cantilever
beam shown in Fig. C-9. This problem was adapted from the work of Muhanna
(Muhanna and Mullen, 2001). The beam has the following properties (Table C-6)
The propped cantilever shown above is subjected to a point load at the mid-
span. Fuzziness is introduced in the Young’s modulus and the applied load. Fig. C-10
and Fig. C-11 depict the membership functions of material and load uncertainties
respectively.
Table C-7 and Table C-8 represent the simultaneous variation of mid-span
displacement and propped end rotation respectively for various combinations of α and
β. Fig. C-12 depicts the variation of mid-span displacement at various levels of load
uncertainty β at α=0.4.
Table C-9 and Table C-10 indicate the results of the simultaneous variation of
bending moment and shear force respectively at fixed end for various combinations of
α and β. Fig. C-13 depicts the variation of fixed end bending moment at various
levels of material uncertainty α at β=0.8. Fig. C-14 depicts the variation of fixed end
Fig. C-15 and Fig. C-16 show the variation of shear force (N) and bending
moment (kNm) along the span for various levels of α ranging from α=1.0 to α=0.0 at
β=1.0. The diagrams (shear force or bending moment) for the crisp load
(corresponding to α=1.0) are shown as thick black lines. The corresponding diagrams
182
for other values of α envelop the crisp diagrams from above and below. Therefore the
entire set of diagrams can be named as shear force and bending moment envelops.
The first column of Table C-7 represents the variation of mid-span displacement
corresponding to crisp load and uncertain material property (E) (β=1, 0≤α≤1). These
results are found to be in close agreement with the results given by Muhanna.
183
Table. C-7 Propped Cantilever Beam
Concomitant Variation of Mid-span displacement (×10-3 metres) w.r.t α and β
β→
α↓
1.0 0.8 0.6 0.4 0.2 0.0
184
1 1.000
1 1
1 -4.698E-03 -4.424E-03
beta
0.4 0.4 1.6 0.4 -7.520E-03 -1.767E-03
0 0 2 0 -9.400E-03
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 -1.0E-02 -9.0E-03 -8.0E-03 -7.0E-03 -6.0E-03 -5.0E-03 -4.0E-03 -3.0E-03 -2.0E-03 -1.0E-03 0.0E+00
Fig. C-11 Membership Function for Load Uncertainty Fig. C-12 Propped Cantilever- Membership Function for Mid-span displacement at
alpha = 0.4
1 -22499 -15000
0 -24930 -13605
-25000 -23000 -21000 -19000 -17000 -15000 -13000
Bending Moment (Nm) at Fixed End
1 5500 8250
0 4988 9141
4900 5400 5900 6400 6900 7400 7900 8400 8900 9400
Shear Force (N) at Fixed End
Fig. C-14 Propped Cantilever- Membership Function of Shear Force
at Fixed End at beta =0.8
186
7617
7300
6875
6300 6236
alpha=1.0
alpha=0.8
5300
alpha=0.8
4300 alpha=0.6
alpha=0.6
3300 alpha=0.4
Shear Force (N)
alpha=0.4
2300 alpha=0.2
alpha=0.2
1300 alpha=0.0
alpha=0.0
300
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
-700
-1700
-2700 -2834
-3125
-3462
-3700
Span (m)
16500
14000
11500
9000
6500
4000
Bending Moment Nm
1500
-1000
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
-3500 alpha=1.0
alpha=0.8
-6000
alpha=0.8
-8500 alpha=0.6
alpha=0.6
-11000
alpha=0.4
-13500 alpha=0.4
alpha=0.2
-16000
alpha=0.2
-18500 alpha=0.0
alpha=0.0
-21000
Span (m)
187
4. Preliminary CaseStudy-3 – Plane Truss
The third case study considered here is a two bay plane truss shown in
Fig. C-17. This problem is adopted from the work of Muhanna (Muhanna and
Mullen, 2001). The plane truss has the following properties (Table C-11)
Uncertainty is introduced in the Young’s modulus and the applied loads. The material
and load uncertainties are given by membership functions shown in Fig. C-10 and
Fig. C-11.
Table C-12 and Table C-13 represent the simultaneous variation of horizontal
Table C-14 and Table C-15 represent the simultaneous variation of axial force
Fig. C-19 indicates the variation of axial force in member 4 at various levels of
material uncertainty α at β=0.8. Fig. C-20 depicts the variation of axial force in
188
In general it is observed that membership functions are triangular in the
The results presented in first column of Table C-12 and Table C-13 (horizontal
and vertical displacement) represents the presence of a single uncertainty (0≤α≤1 and
β=1). These results are found to be in close agreement with those given by Muhanna
189
Table. C-12 Plane Truss-Concomitant Variation of Horizontal displacement of node 4
(×10-5 metres) w.r.t α and β
β→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [3.951,3.951] [3.161,4.741] [2.371,5.532] [1.580,6.322] [0.790,7.112] [0.000,7.902]
0.8 [3.845,4.058] [3.076,4.870] [2.306,5.682] [1.537,6.494] [0.768,7.306] [0.00,8.118]
0.6 [3.740,4.166] [2.991,4.999] [2.243,5.833] [1.494,6.667] [0.745,7.501] [-0.003,8.334]
0.4 [3.635,4.274] [2.907,5.130] [2.179,5.986] [1.451,6.841] [0.724,7.697] [-0.004,8.552]
0.2 [3.531,4.384] [2.824,5.262] [2.117,6.139] [1.409,7.017] [0.702,7.895] [-0.005,8.772]
0.0 [3.427,4.495] [2.741,5.395] [2.054,6.294] [1.368,7.194] [0.681,8.094] [-0.005,8.993]
Table. C-14 Plane Truss- Concomitant Variation of Axial Force (kN) in Member 4
w.r.t α and β
β→
1.0 0.8 0.6 0.4 0.2 0.0
α↓
1.0 [-7.902,-7.902] [-9.483,-6.322] [-11.063,-4.741] [-12.64,-3.161] [-14.22,-1.580] [-15.805,0.0]
0.8 [-8.124,-7.687] [-9.748,-6.150] [-11.373,-4.612] [-12.99,-3.075] [-14.62,-1.537] [-16.247,0.0]
0.6 [-8.351,-7.477] [-10.02,-5.982] [-11.69,-4.486] [-13.36,-2.991] [-15.03,-1.495] [-16.702,0.0]
0.4 [-8.585,-7.273] [-10.30,-5.818] [-12.019,-4.364] [-13.74,-2.909] [-15.45,-1.455] [-17.170,0.0]
0.2 [-8.826,-7.074] [-10.59,-5.659] [-12.36,-4.244] [-14.12,-2.830] [-15.89,-1.415] [-17.652,0.0]
0.0 [-9.074,-6.880] [-10.89,-5.504] [-12.70,-4.128] [-14.52,-2.752] [-16.33,-1.376] [-18.148,0.0]
Table. C-15 Plane Truss-Concomitant Variation of Axial Force (kN) for Member 10
w.r.t α and β
β→
α↓
1.0 0.8 0.6 0.4 0.2 0.0
190
1 -2.03E-04 -1.95E-04
4 6
0.8 -2.44E-04 -1.56E-04
3 4
10
8
0.6 -2.85E-04 -1.17E-04
5 6 11 7 5m
beta
0.4 -3.26E-04 -7.81E-05
9 1 2
1 3
10 m 10 m 0.2 -3.66E-04 -3.89E-05
20 kN 0 -4.07E-04
-4.1E-04 -3.6E-04 -3.1E-04 -2.6E-04 -2.1E-04 -1.6E-04 -1.1E-04 -6.0E-05 -1.0E-05
Fig. C-17 Plane Truss Vertical displacement (m)
Fig. C-18 Plane Truss- Membership Function for vertical displacement of node 2 at
alpha=0.6
191
1 -9.483 -6.322
1 8.389 9.305
beta
0.4 -10.3 -5.818 0.4 3.356 14.887
Fig. C-19 Plane Truss- Membership Function for axial force in Member 4 at beta=0.8 Fig. C-20 Plane Truss- Membership Function for Axial Force in Member 10 at
alpha=0.6
5. Preliminary Case Study -4 –Plane Frame
The fourth case study considered here is a single bay-single storey portal
frame shown in Fig. C-21. The plane frame has the following properties (Table. C-16)
levels of material uncertainty α at β=1.0. Fig. C-23 represents the variation of vertical
α=0.6. Table. C-19 represents the simultaneous variation of bending moment at node
variation of axial force in member 1 for various combinations of α and β. Fig. C-25
uncertainty α at β=1.0. Fig. C-26 represents the variation of shear force just to the left
Fig. C-27 depicts the variation of axial force in member 1 at various levels of
load uncertainty α at β=0.8. Fig. C-28 depicts the variation of axial force in member 1
192
6. Conclusions
In all the above case studies, uncertain structural response is evaluated in the
presence of multiple uncertainties. In all the cases, it is found that the uncertainty of
structural response (such as displacement, rotation, moment, shear force and axial
force) is found to vary simultaneously with the independent variation of material and
be triangular in the presence of a single uncertainty (Fig. C-5, Fig. C-8, Fig. C-22,
Fig. C-25 and Fig. C-26) and trapezoidal in the presence of multiple uncertainties
(Fig. C-6, Fig. C-7, Fig. C-12, Fig. C-13, Fig. C-14, Fig. C-18, Fig. C-19, Fig. C-20,
193
Table. C-17 Plane Frame-Concomitant Simultaneous Variation of Vertical displacement of
node 3 (×10-5 metres) w.r.t α and β
β→
α↓
1.0 0.8 0.6 0.4 0.2 0.0
194
50 kN m 1 -4.462
100 kN
2 2 3 0.8 -4.587 -4.338
alpha
25 kN/m 4m
1 3 0.4 -4.84 -4.092
1 4
0 -5.097 -3.849
-5.2 -5.1 -5 -4.9 -4.8 -4.7 -4.6 -4.5 -4.4 -4.3 -4.2 -4.1 -4 -3.9 -3.8
3m Horizontal Displacement(mm)
Fig. C-22 Plane Frame- Membership Function for Horizontal Displacement at
Fig. C-21 Plane Frame node 2 at beta=1.0
195
beta
0.4 1.153 2.7 0.4 -0.244 1.580
0 -0.804 2.160
0 1.088 2.788
-1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8
rotation (X0.001 radians)
Vertical Displacement(X0.01mm)
Fig. C-23 Plane Frame- Membership Function for Vertical Displacement at node Fig. C-24 Plane Frame- Membership Function for rotation at node 3 at alpha=0.6
3 at beta=0.8
1 81.33 1 -57.35
beta
0.4 71.41 91.73 0.4 -116.78 2.08
Fig. C-25 Plane Frame-Membership Function for Bending Moment at node 2 at Fig. C-26 Plane Frame- Membership Function for Shear Force just to left of
beta=1.0 node 3 at alpha=1.0
196
beta
0.4 32.06 84.42 0.4 -4.83 122.59
Fig. C-27 Plane Frame- Membership Function for Axial Force in member 1 at beta=0.8 Fig. C-28 Plane Frame- Membership Function for Axial Force in member 1 at
alpha=0.6
BIO-DATA
Mr. M.V.Rama Rao, the author of the present work, completed B.Tech in Civil
1988. He worked as a Junior Research Fellow for a brief period in a CSIR sponsored
As a student Mr. Rama Rao was an active participant in extra-curricular activities like
elocution, debate, essay writing and poetry. He won a number of merit certificates,
cups, medals and shields throughout the school and college studies. Mr. Rama Rao
worked as a software engineer in M/s IMI Engineering Ltd., Hyderabad during 1992.
His work included development and maintenance of finite element software related to
Mr. Rama Rao joined Vasavi College of Engineering, Hyderabad towards the end of
Assistant Professor in the same college. His areas of specialisation include Structural
197
PUBLICATIONS
Rama Rao, M.V. and Ramesh Reddy, R. (2003). “Fuzzy Finite Element Analysis
of Structures with Uncertainty in Load and Material Properties”, Journal of
Structural Engineering, Structural Engineering Research Centre (SERC), Chennai
(accepted for publication).
Rama Rao, M.V. and Ramesh Reddy, R. (2003). “ Fuzzy Finite Element Analysis
of a Cable-Stayed Bridge”, Proc, 8th International Conference on Innovations in
Planning, Design and Construction Techniques in Bridge Engineering, IIBE,
Hyderabad, pp 321-331.
Rama Rao, M.V. and Ramesh Reddy, R. (2003). “Fuzzy Finite Element Analysis
of a Plane Frame with Multiple Uncertainties.”, Proc, National Conference on
Emerging Trends in (ETSMC-2003), NIT, Rourkela, December, 01-02.
Rama Rao, M.V. and Ramesh Reddy, R. (2004). “Fuzzy Finite Element Analysis
of a Cable Stayed Bridge”, Proc, National Conference on Materials and
Structures (MAST-2004), NIT, Warangal, January 23-24, 214-218.
Rama Rao, M.V. and Ramesh Reddy, R. (2004). “Fuzzy Finite Element of a Plane
Truss with Multiple Uncertainties”, Proc, National Conference on Materials and
Structures (MAST-2004), NIT, Warangal, January 23-24, 250-254.
198