Вы находитесь на странице: 1из 286

Behavior, Design and Construction of Horizontally Curved Composite Steel Box Girder Bridges

by

M uayad W hyib A ldoori

A thesis submitted in conformity with the requirements for the degree of

Doctor of Philosophy
Graduate Department o f Civil Engineering

University of Toronto

Copyright by M uayad Aldoori 2004

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

1*1

National Library of Canada Acquisitions and Bibliographic Services


395 W ellington Street Ottawa ON K1A 0N4 Canada

Bibliotheque nationale du Canada Acquisisitons et services bibliographiques


395, rue W ellington Ottawa ON K1A 0N4 Canada Your file Votre reference ISBN: 0-612-94268-6 Our file Notre reference ISBN: 0-612-94268-6

The author has granted a non exclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis in microform, paper or electronic formats.

L'auteur a accorde une licence non exclusive permettant a la Bibliotheque nationale du Canada de reproduire, preter, distribuer ou vendre des copies de cette these sous la forme de microfiche/film, de reproduction sur papier ou sur format electronique. L'auteur conserve la propriete du droit d'auteur qui protege cette these. Ni la these ni des extraits substantiels de celle-ci ne doivent etre imprimes ou aturement reproduits sans son autorisation.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.

In compliance with the Canadian Privacy Act some supporting forms may have been removed from this dissertation. While these forms may be included in the document page count, their removal does not represent any loss of content from the dissertation.

Conformement a la loi canadienne sur la protection de la vie privee, quelques formulaires secondaires ont ete enleves de ce manuscrit. Bien que ces formulaires aient inclus dans la pagination, il n'y aura aucun contenu manquant.

Canada
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Behavior. D esign and Construction of Horizontally C urved Composite Steel Box Girder Bridges Doctor o f Philosophy, 2004, Muayad Aldoori Department of Civil Engineering, University o f Toronto

Abstract
Horizontally curved girder bridges have been used considerably in recent years in highly congested urban areas. However, although significant research on physical testing and advanced analysis has been underway for the past decade, the practical employment o f many recommendations has not been achieved by the engineering community nor have standards reflecting this work been brought into practice. The design process o f curved composite bridges involves tracking the stresses and the potential failure change in the girders during erection, construction and service loading stages. For structural safety and ser viceability, the designer estimates the stresses induced within the bridge and assure that they do not exceed the applicable specified limit state as required in bridge design standards. However, the designer m ay be concerned about the level o f approximation that is used in his estimate or even the applicability o f the underlying theory. To answer this question and provide the designer with more insight into the behavior of the curved bridges, the field testing during construction and service loading o f a curved bridge located near Baltimore, M aryland is re-examined here using linear elastic three-dimensional finite element modeling. Comparisons are made between the finite element results and the measured results. The level o f safety or reliability that would be available during the erection and the construction pro cesses o f horizontally curved girder bridges represents another major concern for the designer. A three span continuous curved box girder bridge in Houston, Texas is used in this study as an example reflecting current detailing and fabricating practice and it is chosen for a detailed evaluation of the structural safety/ reliability during the erection and construction process. This task involves simulating the girder erection and concrete slab placement sequence o f the bridge using comprehensive nonlinear three dimensional finite element modeling.

ii

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Finally, to facilitate the finite element modeling effort for use by a designer, ANSYS Parametric Design Language (APDL) capabilities are used here to develop an analysis/design tool for "Bath-Tub" style curved steel girder bridges. This tool is then used to evaluate the effects o f several important design variables on the response and behavior o f the girders during the construction phase. This study demonstrates the ability o f finite element modeling to assess the stiffness, serviceability performance, buckling behavior and ultimate strength o f curved bridges during construction and it is a major step towards a performance based approach to design for stability.

iii

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Acknowledgements
I am indebted to my supervisor, Professor Peter Birkemoe, for his direction and support throughout my graduate studies at the University o f Toronto. His interest in my work and appreciation o f my efforts pro vided me with the constant motivation needed to achieve my goal. I would like to extend my sincere thanks to Professor Todd Helwig and Dr. Zhanfei Fan from Univer sity o f Houston for providing me with all the data and information about Houston Bridge. I would like also to thank Professor Chai Yoo for his permission to use the engineering drawing o f Baltimore Bridge. Thanks to Professor Joseph Yura o f University o f Texas at Austin for making the software and the thesis of his students available for me. I would like to acknowledge the financial support o f the John. Kellerman Fellowship Committee and the University o f Toronto. The support o f the Department o f Civil Engineering at the University o f Toronto and the partial support from the operating grants program of the Natural Sciences and Engineering Research Council o f Canada (NSERC) are also gratefully acknowledged. I would like to thank the Cana dian Institute o f Steel Construction (CISC) for supporting my attendance in the two-day course Steel Bridges, Design, Fabrication and Construction Based on CHBDC-S6-2000. Thanks to my wife Sana, my brothers, Waael, Raeed and Yasir, and my sister Eman for their support and caring and to my beautiful daughter M ina who fills m y heart with happiness. Thanks to my colleague Dr. Deena Dinno for her help, support and proof reading my thesis. Finally, I must express my gratitude to my parents. Their encouragement, love and support have always been a source o f strength and inspiration.

iv

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Table o f Contents
A bstract.............................. ........................................................................................... ................ ii Acknow ledgem ents ...... iv Table of C o n ten ts.............................................. ...................... ......... ...................... .....................v List of Figures. ..... viii ....... xi List of T a b le s Notation ..... xii

C hapter 1 Introduction.................................................................................................................................... 1 1.1 General...................................................................................................................................1 1.2 Construction of Box Girder Bridges.................................................................................... 4 1.3 Objectives and Scope.............................................................................................................5

C hapter 2 L iterature Review ..........................................................................................................................7 2.1 General Behavior of Curved Box Girders...........................................................................7 2.2 Behavior of Curved Girder Bridges during Construction...................................................14 2.3 Static Analysis of Curved Box Girder Bridges................................................................... 29 2.3.1 The Plane Grid Method................................................................................................ 29 2.3.2 Space Frame M ethod................................................................................................... 30 2.3.3 Folded Plate Method.................................................................................................... 31 2.3.4 Finite Strip Method...................................................................................................... 32 2.3.5 Finite Element M ethod................................................................................................ 33 2.3.6 Thin Walled Curved Beam Theory............................................................................. 36 2.4 Experimental Studies on Elastic Response of Box Girder Bridges...................................38 2.4.1 Field Studies................................................................................................................. 38 2.4.2 Model Studies...............................................................................................................40 2.5 Ultimate Response of Box Girder Bridges..........................................................................43 2.6 Design Aids for Curved Box Girder Bridges....................................................................... 51 2.7 Design codes and specifications........................................................................................... 63 2.7.1 AASHTO, Guide Specification for Horizontally Curved Bridges (2003) [40]...... 63 2.7. 1.1. Strength of Curved Top Flange Plates............................................................ 63 2.7. 1.2. Strength of Curved Bottom Flange Plates......................................................67 2.7.2 Canadian Highway Bridge Design Code (CHBDC) (2000) [18]............................ 71 2.7.3 Hanshin Guidelines.......................................................................................................71

C hapter 3 Curved Box G irder Bridge Field Test...................................... .............................................. . 76 3.1 Introduction......................................................................................................................... 76

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

3.2 Description of the Field T est........................................... ................................................... 77 3.2.1 Instrumentation.................................................. .........................................................77 3.2.2 Analytical Procedure....................................................................................................86 3.3 Finite Element analysis.........................................................................................................90 3.3.1 General.......................................................................................................................... 90 3.3.2 Finite Element M ethod................................................................................................90 3.4 The Finite Element Program: ANSYS................................................................................. 93 3.4.1 Shell Elements............................................................................................................. 93 3.5 Model Study By Fam and Turkstra [34]..............................................................................94 3.6 Finite Element Modeling of the Baltimore Bridge............................................................. 101 3.6.1 Non-Composite Bridge M odeling.............................................................................. 101 3.6.2 Composite Bridge modeling........................................................................................105 3.7 Comparison of Results for Construction Loading.............................................................. 108 3.7.1 Deflection...................................................................................................................... 108 3.7.2 Longitudinal Stress Distribution........................................................... Ill 3.8 Comparisons of Results for the Live Load T est..................................................................120 3.8.1 Deflection...................................................................................................................... 120 3.8.2 Longitudinal Stress Distribution..................................................................................123

Chapter 4 Design Safety o f H orizontally Curved Bath-tub G irder B ridges during Erection and C onstruction.................................................................................................................................................. 133

4.1 Introduction............................................................................... :.............................. 133 4.2 Objectives............................................................................................................................... 135 4.3 Detailed Description of the Houston Bridge........................................................................ 135 4.4 Nonlinear Finite Element Capabilities of ANSYS..............................................................141 4.4.1 Shell 181: Characteristics............................................................................................ 143 4.4.2 Nonlinear Analysis.......................................................................................................145 4.4.3 Geometric Nonlinearity:.............................................................................................. 146 4.4.4 Material Nonlinearity:...................................................................................................147 4.4.5 Iterative Solutions.........................................................................................................149 4.5 Finite Element Modeling of Houston B ridge...................................................................... 152 4.5.1 Erection of Segment 905.............................................................................................. 157 4.5.2 Erection of Segment 907.............................................................................................. 159 4.5.3 Slab Construction, Stage 1:.......................................................................................... 165 4.5.4 Slab Construction, Stage 2:.......................................................................................... 170 4.5.5 Slab Construction, Stage 3:.......................................................................................... 170

Chapter 5 Behavior o f H orizontally Curved Bath-T ub G irder B ridges during C onstruction............ 176

5.1 Introduction............................................................................................................................. 176 5.2 Objectives................................................................................................................................177 5.3 Finite Element Model for Construction Staging.................................................................. 177 5.4 Parametric Study...................................................................... 181

vi

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

5.4.1 Span Length ....... ....183 5.4.2 Cross-Frame Spacing................................... 184 5.4.3 Girder D epth................................................... 184 5.4.4 Web Inclination:.......................................................................................................... 187 5.4.5 Top Flange Width:....................................................................................................... 187 5.5 Parametric Effects on Ultimate Behavior............................................................................ 191 5.5.1 Span Length:.................................................................................................................192 5.5.2 Top Flange Width:....................................................................................................... 194 194 5.5.3 Cross-Frame Spacing:.................................. 5.6 Modes of Failure....................................................................................................................196 200 5.7 Summary and Conclusions......................................................

C hapter 6 Sum m ary an d C o n clu sio n s.................................

201 6.1 Summary of Results, Observations and Conclusions........................................................ 203 6.1.1 Literature Review ........................................................................................................ 204 6.1.2 Linear Elastic Modeling of Curved Bath-Tub Girder Bridge Structures.................205 6.1.2.1. Fam and Turkstra Experiments.........................................................................205 6.1.2.2. Baltimore Bridge Field T e st.............................................................................206 6.1.3 Design Safety of Horizontally Curved Bath-Tub Girder Bridges during Erection and Construction............................................................................................208 6.1.4 Behavior o f Horizontally Curved Bath-Tub Girder Bridges during Construction 210 6.2 Principal Achievements of the Current Study.................................................................... 211 6.3 Future Research............................................................................... 212 214

R eferen ces......................................

A PPEN D IX I A NSYS Input File for Two Spans C ontinuous C urved Bath-tub G irder Bridge during C onstruction ........ 223

vii

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

List of Figures
Fig. 1.1. Cross-Section of Box Girder Bridge................................................................................4 Fig. 2.1. General Behavior of an Open Box Section under Gravity Load Showing Separate Effect................................................................................................................... 8 Fig. 2.2. Normal and Shear Components of Longitudinal Bending Stress................................ 10 Fig. 2.3. Shear Flow in Box Girder from Saint-Venant Torsion.................................................10 Fig. 2.4. Bracing System terminology: Common Configurations.............................................. 13 Fig. 2.5. Top Bracing Arrangements.............................................................................................16 Fig. 2.6. Diagonal Brace Forces from Torsion According to (EPM) [50].................................26 Fig. 2.7. Horizontal Component of Applied Loads on the Top Flanges....................................27 Fig. 2.8. Angular Deformation of a Box Girder Under Distortion............................................. 53 Fig. 2.9. Beam-on-Elastic-Foundation Analogy of Box Girder Distortion [50]....................... 54 Fig. 2.10. Box Girder Cross Section Dimensional Parameters...................................................56 Fig. 2.11. Torsional Parameter k Versus Central Angle q for Actual Curved Bridges [60]....59 Fig. 2.12. Maximum Strength of Curved Bottom Flange Plate in Compression. (AASHTO 1993) [39].................................................................................................... 71 Fig. 3.1. Plan View of the Baltimore Bridge................................................................................ 78 Fig. 3.2. Longitudinal View of the Baltimore Bridge..................................................................79 Fig. 3.3. Cross Sectional Dimension at Splice Locations (Baltimore Bridge).......................... 80 Fig. 3.4. Typical Cross Section of the Baltimore Bridge............................................................ 81 Fig. 3.5. Concrete Placement Sequence........................................................................................ 82 Fig. 3.6. Longitudenal and Transverse strain Gage Locations....................................................84 Fig. 3.7. Lane Locations for Live Load Test................................................................................ 85 Fig. 3.8. Typical Nodal Point Variation for Field Splice............................................................ 88 Fig. 3.9. Analytical Load Point Location for Static Condition of Live Load T est...................89 Fig. 3.10. The Current FE Model of Fam and Turkstra [34] Plexiglas Curved Box Girder.... 96 Fig. 3.11. Dimensions and Transverse Cross-Section Discretization of Fam and Turkstra [34] Model..................... 97 Fig. 3.12. FE (Shell63, Shell93) and Experimental Deflection at Mid-span for Fam and Turkstra [34] Series A Plexiglas Box Girder Bridge............................................... 98 Fig. 3.13. FE (Shell63, Shell93) and Experimental Deflection at Mid-span for Fam and Turkstra [34] Series B Plexiglas Box Girder Bridge............................................... 98 Fig. 3.14. FE (Shell93, Shell63) and Experimental Mid-Span Stresses in Top Flange of Fam and Turksrta [34] Series A Plexiglas Box Girder Bridge............................... 99 Fig. 3.15. FE (Shell93, Shell63) and Experimental Mid-Span Stresses in Top Flange of Fam and Turksrta [34] Series B Plexiglas Box Girder Bridge................................99 Fig. 3.16. FE (Shell93, Shell63) and Experimental Mid-Span Stresses in Bottom Flange of Fam and Turksrta [34] Series A Plexiglas Box Girder Bridge..............................100 Fig. 3.17. FE (Shell93, Shell63) and Experimental Mid-Span Stresses in Bottom Flange of Fam and Turksrta [34] Series B Plexiglas Box Girder Bridge.............................. 100 Fig. 3.18. Typical FE Mesh for Baltimore Curved Box Girders..............................................103 Fig. 3.19. Distributed Loads Used to Simulate the Weight of Concrete S lab........................ 106 Fig. 3.20. FE Model Used in Simulating Live Load Tests........................................................ 107

viii

with perm ission of the copyright owner. Further reproduction prohibited without permission.

Fig. 3.21. Comparison of Vertical Deflections for Exterior Girder Under Concrete (Pour I) Load..................................................... 109 Fig. 3.22. Comparison of Vertical Deflections for Exterior Girder Under Concrete (Pour II) load.................................................................................................................110 Fig. 3.23. Comparison o f Vertical Deflections for Exterior Girder Under Concrete (Pour III) Load.............................................................................................................. 110 Fig. 3.24. Longitudinal Stress Dead Load Test, Section AA-A, Pour I (ksi).......................... 114 Fig. 3.25. Longitudinal Stress Dead Load Test, Section BB-B, Pour I (ksi)...........................114 Fig. 3.26. Longitudinal Stress Dead Load Test, Section CC-C, Pour I (ksi)...........................115 Fig. 3.27. Longitudinal Stress Dead Load Test, Section DD-D, Pour I (ksi)..........................115 Fig. 3.28. Longitudinal Stress Dead Load Test, Section AA-A, Pour II (ksi).........................116 Fig. 3.29. Longitudinal Stress Dead Load Test, Section BB-B, Pour II (ksi)......................... 116 Fig. 3.30. Longitudinal Stress Dead Load Test, Section CC-C, Pour II (ksi)......................... 117 Fig. 3.31. Longitudinal Stress Dead Load Test, Section DD-D, Pour II (ksi).........................117 Fig. 3.32. Longitudinal Stress Dead Load Test, Section AA-A, Pour III (ksi)....................... 118 Fig. 3.33. Longitudinal Stress Dead Load Test, Section BB-B, Pour III (ksi)........................118 Fig. 3.34. Longitudinal Stress Dead Load Test, Section CC-C, Pour III (ksi)........................119 Fig. 3.35. Longitudinal Stress Dead Load Test, Section DD-D, Pour III (ksi)....................... 119 Fig. 3.36. Live Load Deflection, Gage (1-4), Load Point.1..................................................... 121 Fig. 3.37. Live Load Deflection, Gage (1-4), Load Point.2 ..................................................... 121 Fig. 3.38. Live Load Deflection, Gage (1-4), Load Point.3 ..................................................... 122 Fig. 3.39. Live Load Deflection, Gage (1-4), Load Point.4 ..................................................... 122 Fig. 3.40. Longitudinal Stress Live Load Test, Section AA-A, Load Point 1.........................124 Fig. 3.41. Longitudinal Stress Live Load Test, Section AA-A, Load Point 2.........................124 Fig. 3.42. Longitudinal Stress Live Load Test, Section AA-A, Load Point 3.........................125 Fig. 3.43. Longitudinal Stress Live Load Test, Section AA-A, Load Point 4.........................125 Fig. 3.44. Longitudinal Stress Live Load Test, Section BB-B, Load Point 5......................... 126 Fig. 3.45. Longitudinal Stress Live Load Test, Section BB-B, Load Point 6......................... 126 Fig. 3.46. Longitudinal Stress Live Load Test, Section BB-B, Load Point 7......................... 127 Fig. 3.47. Longitudinal Stress Live Load Test, Section BB-B, Load Point 8......................... 127 Fig. 3.48. Longitudinal Stress Live Load Test, Section CC-C, Load Point 9......................... 128 Fig. 3.49. Longitudinal Stress Live Load Test, Section CC-C, Load Point 10....................... 128 Fig. 3.50. Longitudinal Stress Live Load Test, Section CC-C, Load Point 11....................... 129 Fig. 3.51. Longitudinal Stress Live Load Test, Section CC-C, Load Point 12....................... 129 Fig. 3.52. Longitudinal Stress Live Load Test, Section DD-D, Load Point 13.......................130 Fig. 3.53. Longitudinal Stress Live Load Test, Section DD-D, Load Point 14.......................130 Fig. 3.54. Longitudinal Stress Live Load Test, Section DD-D, Load Point 15.......................131 Fig. 3.55. Longitudinal Stress Live Load Test, Section DD-D, Load Point 16........... 131 Fig. 4.1. General Layout of Houston Bridge. [50]..................................................................... 136 Fig. 4.2. Cross-Section of the Box Girder Bridge. [50]..............................................................136 Fig. 4.3. Erection Sequence of the Interior Girder. [50].............................................................139 Fig. 4.4. Construction Sequence of the Concrete Slab. [50]...................................................... 140 Fig. 4.5. Isotropic and Kinematic Hardening.............................................................................. 149 Fig. 4.6. Traditional Newton-Raphson Method vs. Arc-Length Method Diagrammatic Description. [4]................................................................................................................152 Fig. 4.7. Typical Finite Element Discretization for the Houston Bridge................................. 154

ix

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

4.8. Models Used to Simulate the Erection of Segments 905 and 907. [50] ....... ..156 4.9. Distributed Loads Used to Simulate the Weight of Concrete Slab. [50]...... 156 4.10. Typical Modeling of Stress-Strain Curve Including Strain Hardening.................. 157 4.11. Erection of Segment 905.............................................................................................160 4.12. Applied Load-Vertical Deflection Curve for Segment 905 Erection..................... 161 4.13. Deformed Shape of the Houston Bridge (Segment 905 Erection Phase) at Ultimate..................................................................................................................... 162 Fig. 4.14. Analytical Application of Restraints at the Supports of the Single Interior Girder Model ................................................................................................... 163 Fig. 4.15. Applied Load-Vertical Deflection Curve for Segment 907 Erection..................... 164 Fig. 4.16. Deformed Shape of Houston Bridge (Segment 907 Erection Phase) at Ultimate. 166 Fig. 4.17. Applied Load-Vertical Deflection Curve for Slab Construction, Stage 1..............168 Fig. 4.18. Deformed Shape of the Houston Bridge (Slab Construction, Stage 1) at Ultimate.................................................................................................................... 169 Fig. 4.19. Applied Load-Vertical Deflection Curve for Slab Construction, Stage 2.............. 171 Fig. 4.20. Deformed Shape of the Houston Bridge (Slab Construction, Stage2) at Ultimate.................................................................................................................... 173 Fig. 4.21. Applied Load-Vertical Deflection for Slab Construction, Stage 3......................... 174 Fig. 4.22. Deformed Shape of the Houston Bridge (Slab Construction, Stage 3) at Ultimate.....................................................................................................................175 Fig. 5.1. Cross-Section of the Box Girder Bridge Model.......................................................... 178 Fig. 5.2. Construction Staging in FE Continuous Model...........................................................180 Fig. 5.3. Cross-Section of the Bridge Model.............................................................................. 181 Fig. 5.4. Finite Element Discretization, Non-composite Model............................................... 182 Fig. 5.5. Effect of Span Length on Deflection Ratio................................................................. 185 Fig. 5.6. Effect of Span Length on Bending Stress Ratio..........................................................185 Fig. 5.7. Effect of Cross-Frame Intervals on Deflection Ratio................................................. 186 Fig. 5.8. Effect of Cross-Frame Intervals on Warping-Bending Stress Ratio......................... 186 Fig. 5.9. Effect of Girder Depth on Deflection Ratio.................................................................188 Fig. 5.10. Effect of Girder Depth on Bending Stress Ratio............................................... 188 Fig. 5.11. Effect of Girder Depth on Warping-Bending Stress Ratio...................................... 188 Fig. 5.12. Effect of Web Inclination on Deflection Ratio..........................................................189 Fig. 5.13. Effect of Web Inclination on Bending Stress Ratio ....................................... 189 Fig. 5.14. Effect of Web Inclination on Warping-Bending Stress Ratio................................. 189 Fig. 5.15. Effect of Top Flange Width on Deflection Ratio......................................................190 Fig. 5.16. Effect of Top Flange Width on Bending Stress Ratio.............................................. 190 Fig. 5.17. Effect of Top Flange Width on Warping-Bending Stress Ratio..............................190 Fig. 5.18. Span Length Effect on Load-Deflection Curves....................................................... 193 Fig. 5.19. Span Length Effect on Ultimate Strength Ratio........................................................193 Fig. 5.20. Top Flange Width Effect on Load -Vertical Deflection Curve............................... 195 Fig. 5.21. Cross-Frame Spacing Effect on Load-Vertical Deflection Curves......................... 197 Fig. 5.22. Load-Vertical Deflection Curve for Model with (L/R=0.2).................................... 198 Fig. 5.23. Local Lateral Buckling of the External Web of Exterior Girder for Panels Near Mid Span.................................................................................................................................199

Fig. Fig. Fig. Fig. Fig. Fig.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

List of Tables
Table 2.1: Curved Box-Girder Methods of Analysis...................................... 37 Table 3.1: Comparison of Experimental, Analytical and Design Stresses Due to Dead Load... 120 Table 3.2: Comparison of Experimental, Analytical and Design Stresses Due to Live Load+Impact Factor (28.6%)...................................................................................132

xi

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Notation
Ab = cross sectional area of lateral bracing member (in ) as defined in Eq. 2.9 A j = cross sectional area o f diagonal bracing member A f = area o f actual top flange plus one quarter o f web as defined in Eq. 2.6 A0 = enclosed area o f box section As = cross sectional area of the strut a = spacing between transverse bracing members as defined in Eq. 2.6 and shown in Fig. 2.5 a = longitudenal stiffener span B = half width o f the bridge deck b = bottom flange width between webs (in.), (Eq. 2.9) b = smaller o f the two plate dimensions (Eq. 2.18) b = width o f cross-section at bracing level (Eq. 2.6). b = distance between centers o f the top flanges (Eq. 2.15b) bf = top flange width

bs = distance between longitudinal flange stiffeners Dtot = total forces in the diagonals of the top flange truss
D epm = diagonal forces from the torsional moment determined using EPM (Equivalent Plate Method) as shown in Fig. 2.6 Dbend = frces in a diagonal caused by vertical bending of the box girder Djat = forces in a diagonal caused by lateral components of applied loadings DFC= curved girder load distribution factor DFg = load distribution factor for straight girders d = length of diagonal bracing member

xii

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

d = depth o f the box (Eq. 2.38) ds = width o f the given plate in the box E = m odulus o f elasticity o f steel Fbc= m axim um average stress in the curved flange Fbs = equivalent straight girder stress Fws = ratio o f maximum warping stress in a curved bridge to the maximum bending stress in a straight bridge (Eq. 2.7) Fy = yield stress (psi) (Eq. 2.44) f = bending normal stress fL tot = total top flange stress fx t = longitudinal stress at the middle o f the top flange

i) lat = lateral bending stress caused by lateral component o f applied loadings f| bend = lateral bending stress caused by the axial forces in the struts that result from the vertical bending o f the box girder. fv = shear stress associated with moment gradient. G = shear modulus of the steel I = moment o f inertia o f the section. ID = distortional warping constant o f the box girder If = moment o f inertia o f actual top flange. Is = actual moment o f inertia o f one longitudinal flange stiffener about an axis parallel to the flange at base o f stiffener (Eq. 2.21) Is = moment o f inertia o f transverse bracing member (strut) which is assumed to perform like a beam member (Eq. 2.6) Iw = warping constant.

xiii

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

K = plate buckling coefficient K d = distortional stiffness of the box girder cross section Ks = shear buckling coefficient Kt = torsional constant o f the section L = length o f the outside girder (Eq. 2.7) L = span length L = length in inches o f the unsupported compression flange between cross frames (Eqs. 2.45 and 2.46) M = bending moment M p = full plastic m oment xnD = distortional load on the box girder. N g = num ber o f box girders n = num ber o f equally spaced longitudinal flange stiffeners p = entire lateral component of applied loading as shown in Fig. 2.7 Q = first moment o f area under consideration q = uniform shear flow (force/length) R = radius o f the outside girder (Eq. 2.7) R = radius o f curvature S = section modulus S = diaphragm spacing, expressed as decimal fraction o f span length (Eqs. 2.23 and 2.24) S = cross frame spacing (in.) (Eq. 2.37) S = distance between the center line o f the deck and that of the girder under consideration (Eqs. 2.40 and 2.41) Smax = maximum cross frame spacing Stot = total forces in the struts

xiv

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Sbend = forces in the struts caused by the vertical bending o f the box girder

Siat = forces in the struts caused by lateral component o f applied loading s = spacing o f struts (panel length) (Eq. 2.15b) T = torque on the cross section o f the member Tp = full plastic torque t = thickness of the given plate in the box tf= top flange thickness tf = thickness of bottom flange plate ( Eq. 2.21) tw = web thickness teq = thickness o f equivalent top plate U = internal strain energy V = shear force W = external energy Wc = roadway width between curbs w = bottom flange width or width between longitudinal stiffeners z = longitudenal axis o f the member

a = acute angle between the top flange and the diagonal (Eq. 2.15b) a = top flange slenderness
A p = magniude o f the maximum out o f plane deformation o f stiffened panels o f the bottom flange A s = maximum out-of straightness of the longitudinal stiffener
0 = angular distortion o f the box girders shown in Fig. 2.9 a) = twist angle o f the cross section

A = top flange slenderness as defined in (Eq. 2.43).

v = poissons ratio

XV

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

II = total potential energy <7b = bending stress corresponding to the distortional stress (Kips/in2) (Eqs. 2.23 and 2.24) <7bc = bending compressive stress (crba) c = allowable compressive stress for lateral buckling

(jbao = upper limit o f the allowable compressive stress


(o'c)cr= critical compressive stress
'j

(Tf= maximum distortional stress at the bottom flange o f the member (Kips/in ) (Eqs. 2.23 and 2.24) crwc = additional warping compressive stress
T

= shear stress

( r c)cr = critical shear stress = extended central angle [B] = strain- displacement matrix. [D] = stress-strain constitutive matrix. {F} = nodal load vector [K] = global stiffness matrix assembled from the elem ent stiffness matrices [N] = interpolation function or shape function {u} = nodal displacement vector {u(e)} = node displacement vector o f the element {} = strain vector Note: The parameters o f the analysis/design tool that was developed in the current study are listed in chapter 5.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 1 Introduction

1.1 General

Horizontally curved girder bridges have been used considerably in recent years in highly congested urban areas. M odem highway bridges are often subjected to severe geometric restrictions, especially in dense cities where elevated highways and multi-level structures are necessary; therefore, they must be built in curved alignment. In the past, curved girders were generally composed o f a series o f straight segments. These segments were used as chords in forming a curved alignment. Currently curved girders have replaced straight seg ments. There are several reasons for using a continuous curved girder over several spans. Even though the cost of the super-structure for the curved girder is higher, the total cost o f the curved girder system is reduced considerably since the number o f intermediate supports, expansion joints and bearing details is reduced. Using continuous curved girders also permits the use o f shallower sections as well as permits a reduction in the slab overhang at the outside girder [58], The continuous curved girder also provides more esthetically pleasing structures.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 1

Introduction

Horizontally curved girders, despite all the advantages mentioned above, are generally more complex than straight girders. Curved girders are subjected to vertical bending plus torsion caused by the girder cur vature. In the sixties, several approximate analysis methods were developed to assist designers to deal with such complexities. De Saint Venant published his memoir during the first half o f the 19th century [81], which has been considered as the first work on the static analysis o f horizontally curved beams. Since then, considerable advancement has been done, however, a serious research on the subject of horizontally curved beams started only at the beginning o f the seventies when the Federal Highway Administration (FHWA) formed The Consortium o f University Research Team (CURT). Researchers from Carnegie M ellon University, The University o f Pennsylvania, The University o f Rhode Island and Syra cuse University participated in (CURT). The study was a funded investigation sponsored by twenty five state highway departments. The objective o f (CURT) was to conduct a comprehensive analytical and experimental research program on the behavior o f horizontally curved highway bridges and to perform a survey on all published works pertaining to horizontally curved bridges. In addition, investigations were conducted at the University o f Maryland. These research efforts resulted in the publication o f the first edition o f the Guide Specification (a working stress design guide) for Horizontally Curved Bridges by the American Association o f State Highway and Transportation Officials (AASHTO) in 1976 [41]. The load factor design criteria were adopted by (AASHTO) in 1980 was based on the working stress results. The latest edition o f the guide specification was published in 2003 [40]. Until recently, there has been little research performed on horizontally curved bridges since the early seventies. The current design methods, which are essentially based on the research conducted more than 20 years ago, have a number o f deficiencies. As indicated by the National Cooperative Highway Research Program (NCHRP 1998)[41], many provisions in the Guide Specification are overly conservative, and many may be difficult to implement. Other provisions lend themselves to misinterpretation, which may lead to uneconomical designs or designs with a lower factor o f safety than intended.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 1

Introduction

Recently, research projects have been sponsored by a number o f highway administrations and organi zations to improve design methods for curved highway bridges. For instance, a (NCHRP) project 12-38 was done to develop a complete load and resistance factor design (LRFD) specifications for horizontally curved bridges. The Federal Highway Administration (FHWA) sponsored other projects to increase knowl edge of the behavior o f curved steel bridges. There are basically two types o f steel cross sections currently being in use for curved alignment: an open section consisting o f a number o f I-shaped cross sections braced with a heavy transverse bracing sys tem. The other type o f section currently in use is a closed section consisting o f few box girders. Steel box girders that serve to transfer loads directly from the concrete deck to the abutments and piers have emerged as the most common application in North America today [15]. Box girder cross sections may take the form o f single cell (one box), multi-spline (separate boxes) or multi-cell with a common bottom flange. Due to the high torsional stiffness o f the closed cross section o f the box girders, which often ranges from 100 to 1000 times larger than the torsional stiffness o f comparable I-shaped sections, the torsional moment induced by the curvature o f the girder can be resisted by the box girder with much less transverse bracing than the I-shaped girders. The fabrication o f the box girder is more expensive compared to the I-shaped girder, but this additional cost is usually balanced by the reduction in sub-structuring for the box girder. In addition to the large torsional stiffness, box girders provide higher corrosion resistance because a high percentage o f the steel surface including the top o f the bottom flange is not subjected to the environ mental attack. The box girder also has a smooth shape that leads to better bridge aesthetics. The trapezoidal shape in particular, offers several advantages over rectangular shaped cross section. The trapezoidal box girder (bath-tub girder) provides a norrow bottom flange. N ear the abutments where the bending moment is low, narrow flanges allow for steel savings. The narrow flange as well, provides plate stockiness in the compression areas o f continuous bridge. In addition, bath-tub girders are more aesthetically pleasing.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 1
1.2 C onstruction o f B ox G irder Bridges

Introduction

A multi-spline composite box girder bridge cross section as shown in Fig. 1.1, usually consists o f two or more (bath-tub) trapezoidal shape steel cross sections. The steel section commonly consists o f two top flanges and one bottom flange, which is connected by the inclined web. The concrete slab that is usually cast in place is connected to the steel section using shear connectors to ensure full interaction between them. Therefore during construction, the steel girders are subjected to the w et concrete load in addition to other construction loads without the composite action that results from the hardened concrete deck. The open section o f the bath-tub girder is a major concern because o f its relatively low torsional stiff ness. Therefore, the top flanges have to be braced to increase the torsional stiffness of the section. In this case the section is called quasi-closed section because the it is not fully closed and the top flange is still susceptible to lateral buckling. Bracing systems commonly consist o f a horizontal truss attached to the girder near its top flange to increase its torsional stiffness. The distortion o f the cross section is reduced by using internal cross frames and diaphragms. The box girder with the horizontal truss bracing is usually referred as a quasi-closed box girder. The box girder cross section possesses a high torsional stiffness after the concrete deck gain its full strength since the cross section is considered as a fully closed section.

Top Flange

Concrete Slab

Shear Connector

Web Bottom Flange

Fig. 1.1. Cross-Section o f B ox G irder Bridge.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 1 1 3 O bjectives and Scope

Introduction

Construction o f horizontally curved steel bridges is generally more complex than the construction of straight bridges o f similar span. Up to date, there is essentially no research or documented construction plan to help the fabricators with the construction o f curved steel girders. The recommended AASHTO specifications [41] indicate that a construction plan must be provided including one possible scheme to construct the bridge, outlining the sequence o f girders and deck placement. Although significant research on physical testing and advanced analysis o f horizontally curved girders has been underway for the past few decades, the mobilization o f m any recommendations has not achieved practical employment by the engineering community nor have standards incorporated the findings. Significant additional research and development are required to accomplish the transition to use the advanced analytical techniques in the design and construction processes. Principle features o f this study include studying the behavior o f the trapezoidal composite bath-tub girder bridges during construction and demonstrating the ability o f finite element modeling to assess the stiffness, serviceability performance and ultimate strength o f curved bridges both during construction and service loading phases. The main objectives o f this thesis are as outlined below: 1. Prepare and summarize a literature review on the experimental and theoretical research work pertain ing to curved girder bridges. 2. 3. Review the available specifications and standards pertaining to the curved girder strength. Provide the designer with more insight into the behavior o f the curved bath-tub girder bridges. This will be fulfilled by linear elastic finite element modeling o f a curved bath-tub girder bridge field test. The finite element results represented by the strain and the deformations will be compared with m ea sured field values, as well as a comparison will be made with the two dimensional analytical results. 4. Assess the level o f safety and reliability for the erection and construction sequence o f curved box girder bridges. This task will be accomplished using nonlinear finite element modeling. The various modes of failure o f every erection step and every construction stage will be examined.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 1

Introduction

5. Employ th e advanced analytical techniques such as the finite element method in the analysis and design process and thus make this method more popular and easy to use by a designer. ANSYS Para metric D esign Language (APDL) capabilities are utilized to develop an analysis/design tool for bath tub style curved composite steel girder bridges. This tool will be used to evaluate the effects of sev eral im portant design variables on the response and behavior o f the girders during the construction phase. The general description o f the curved box girder behavior is presented in chapter 2 with emphases on the behavior during construction. This is followed by a comprehensive review to the research related to girder erection and construction and the various available static analysis methods. Chapter 3 follows the simulation o f the curved box girder field test using linear elastic three-dimensional finite element model ing. Comparisons are made between the finite element results, the two-dimensional analytical results and the field test measurements. The level o f safety and reliability o f the curved box girder bridges that would be available at various stages o f the erection and construction processes is evaluated in chapter 4 using nonlinear finite element modeling. Chapter 5 presents the development o f an analysis/design tool for the bath-tub style curved steel girder bridges during construction utilizing ANSYS Parametric Design Language (APDL). The work is summarized in Chapter 6 where all important conclusions are presented and recommendations are made for future work. The analysis/design tool (input file) for two-span continuous curved bath-tub girder bridges during construction is listed in Appendix I.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2 Literature Review

2.1 G eneral B ehavior o f Curved Box Girders

An arbitrary line loading on a simple span box girder (Fig. 2.1a) has bending and torsional components. Under the bending load (Fig. 2.1b), the section deflects rigidly (longitudinal bending), and deforms (bending distortion), while the torsional loading Fig. 2.1c rotates the section rigidly (mixed torsion) and deforms the section (torsional distortion). Curved bridges will develop bending and associated shear stresses as well as torsional stresses because o f the horizontal curvature even if they are only subjected to their own gravitational load. Longitudinal Bending According to a survey conducted by ASCE task committee on horizontally curved steel box girder bridges [47], box girders typically have an average span to depth ratio o f 23 for single spans and 25 for continuous girder spans. Because o f the large span/depth ratio, transverse load causes significant bending stresses in the girder.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

P/2

P/2 P/2

P/2

I
X -

(a) Loading Components. P/2 P/2

Long. Bending

Bending Distortion

+
(b) Bending Load Actions. 72 P/2

f
Mixed Torsion | j Torsional Distortion

J/

(c) T orsional Load Actions.

Fig. 2.1. General Behavior of an Open Box Section under Gravity Load Showing Separate Effect.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

Assuming elastic behavior and that plane sections remain plane under bending, bending normal stresses, f, arising from the equilibrium o f the cross section (Fig. 2.2a) are calculated by:

Where, M is the bending moment, S is the section modulus. Shear stresses associated with the moment gradient also occur (Fig. 2.2b) and are calculated by:

fy - X ..............................................................................................(2-2)

Where, V is the shear force, I is the moment o f inertia o f the section, Q is the first moment o f area under consideration, t is the thickness of the segment.

Mixed Torsion
Because of the bridge curvature, the transverse loads acting on the girder cause twisting about its longitudinal axis. Uniform torsion occurs if the rate of change o f the angle o f twist is constant along the girder and warping is constant and unrestrained. St. Venant analyzed this problem and found that the St. Venant shear stresses occur in the cross section (Fig. 2.3). If there is a variation o f torque or if warping is prevented or altered along the girder, longitudinal torsional warping stresses develop.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

-I !
a iI

(a) Normal Stress

(b) Shear Stress

Fig. 2.2. Normal and Shear Components of Longitudinal Bending Stress.

Fig. 2.3. Shear Flow in Box Girder from Saint-Venant Torsion.

10

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

In general, both St.Venant torsion and the warping torsion are developed when thin-walled members are twisted. B ox girders are usually dominated by St.Venant torsion because the closed cross section has a high torsional stiffness. Kollbrunner and Basler [54] indicated that the longitudinal normal stresses resulting from the non-uniform warping torsion are usually negligible. Box girders have large St.Venant stiffness, which may be 100-1000 times larger than that o f a comparable I-section. St.Venant stiffness of the box section is a function o f the shear modulus o f the steel, G, and the torsional constant Kt, which related to the cross section geometry. Box girders resist the applied torque (in part) by St.Venant torsion that is given by:

T - G K t^

......................................................................................................(2.3)

Where, T is the torque on the cross section o f the member, t9 is the twist angle o f the cross section, z is the longitudinal axis o f the member. The torsional constant for a single cell box girder is given by:

4 Ao

A A2

Where, A0 is the enclosed area o f the box section, ds is the width o f the given plate in the box, t is the thickness o f the given plate in the box.

11

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

Using P ran d tls membrane analogy, the shear stress resulting from pure torsion can be solved [54]. Using Bredts equation, the uniform shear flow (force/length), q, for a single cell box girder can be determined as follows:

( 2

' 5 )

Where,
T

is the shear stress, and it is essentially uniform through the thickness o f the plates.

Bending Distortion
Because o f the out o f plane bending o f the plates forming the girder, the cross section shape changes. Bending distortion (spreading) occurs when transverse loads are applied to the open box (without top bracing). In open box girders, this distortion causes outward bending o f the webs, upward bending of the bottom flange and in-plane bending o f the top flange (Fig. 2.1b). Ties (Struts) as shown in Fig. 2.4 are usually placed between top flanges to prevent bending distortion.

Torsional Distortion
Torsional load tends to deform the cross section through bending of the walls (Fig. 2.1c). If the box girder has no cross frames or diaphragms, the distortion is restrained only by the transverse stiffness o f the plate elements. Due to the lack o f distortional stiffness of the open box girder cross section, cross frames connecting top and bottom flanges are used to prevent the torsional distortion o f the cross section. Horizontal bracing can be placed at a small distance below the top flanges to increase the torsional stiffness o f the open box cross section and reduce the twist.

12

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

L ateral Ties (Struts)

,x \

Solid Diaphragm

Cross-Frame

Top Bracing System (X-Type)

Top Bracing System (Single Diagonal Type)

Fig. 2.4. Bracing System terminology: Common Configurations.

13

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

This chapter summarizes the research conducted to investigate the static behavior o f curved steel girder bridges. The research on curved girders covers a wide range o f topics. These can be itemized as follows: 1. Behavior o f curved girder bridges during construction. 2. 3. 4. 5. 6. Static analysis o f curved box girder bridges. Experimental studies on curved box girder bridges. Ultimate response o f curved box girder bridges Design aids for curved box girder bridges Design codes and specifications.

2.2 Behavior o f Curved Girder Bridges during C onstruction

Despite the fact that the torsional stiffness o f a constructed composite box girder is very large, the quasi-closed section o f a box girder during fabrication, erection and casting concrete stages has a relatively low torsional stiffness. The bottom flange and webs during the early construction stages are very flexible elements and high distortion and twist o f the cross section can occur. Therefore, top bracing systems are commonly used to increase the torsional stiffness o f the steel girder. The distortion can be also controlled by using internal cross frames and solid diaphragms. Top bracing, internal cross frames and diaphragms act as secondary and passive bracing members in straight bridges, but in a curved girder, the interaction of bending and torsion requires them to be considered as primary load carrying elements. The flexural rigidity o f the steel section o f a composite steel-concrete box girder prior to deck hardening is as little as one third that o f the composite section. Similarly, without top lateral bracing, the torsional rigidity o f the resulting open section before deck hardening may be less than 0.01 to 0.001 that o f the closed section after the deck has hardened [5]. Curved girders have been erected based generally on experience with straight girders and this has sometimes led to unexpected problems. Guide specification (AASHTO 93) [39] has some critical deficiencies and these include lack o f fabrication and erection provisions and provisions related to the

14

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

construction o f curved bridges [41]. The specification (AASHTO 93) commonly does not provide designers or construction engineers enough information about the behavior o f the girder during construction to design for bracing. Few experimental and analytical research studies have been carried out to understand the behavior of curved bridge at various construction stages or enhance the design methods pertaining to top bracing systems, internal cross frames and diaphragms. In 1976, McDonald, Chen, Yilmaz and Yen [57] studied two single cell model box girders. The purpose o f their study was to investigate the behavior o f such girders and to examine the stresses in a single cell braced open (quasi-closed) box section under vertical loads eccentric to its longitudinal center line; the other goal o f their study was to evaluate the stresses in the bracing members. The open cross section o f the models was rectangular in shape with various arrangements o f longitudinal and transverse stiffeners in the bottom flange and the webs respectively and with various patterns o f top bracing that was placed at the top flange level o f both models. The specimens had simple span with a cantilever portion. Both specimens had plate diaphragms at the loading points and the support points. Since the braced open section of a box girder is loaded primarily during the construction phases and the stresses in the section are normally within the elastic range, the loads on the model specimens were kept within the elastic range. Horizontal and vertical deflections o f the test box girders were measured with dial gages at the supports and loading points. Afterward, rotations were calculated from the measured deflections. Stresses at various points o f the specimens were estimated using electrical resistance strain gage measurements. The open cross section with top bracing was analyzed using the equivalent thickness concept by Kollbrunner and Basler [54]. Kollbrunner and Basler [54] considered the bracing system as an equivalent plate and converted the braced open cross section into an equivalent closed box section. The thickness, teq, o f the equivalent plate for different patterns of bracing can be calculated through the consideration o f strain energy results from shear in the equivalent plate and the strain energy o f the bracing members. The equations for the thickness, teq, o f the bracing patterns shown in Fig. 2.5 are as follows:

15

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

H a

- Ad
_

.v
Af (a) Ad Af Af
(b>

b!

(c)

Af

Is
/X / 1 \ I f ^ r !

Y
;

Ad

(d)

(e)

As Ad

Af

(f>

Fig. 2.5. Top B racing Arrangements.

16

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

ab
eq

.(2 .6 a )

G d3 2a + A j 3A f

ab ^eq G d3 + + A a 4A c 6A f

.( 2 .6 b )

ab
*eq G

.(2 .6 c )

2Aa

6Af

E
eq

ab b3 a

,( 2 .6 d )

G d3

Aj

A,

6At

eq

,(2 .6 e ) 2 b + a 24L 121


,

,
eq

E
Q
j

ab
3 ,3 3 ^ ^

2A d Where,

A s 6A f

teq is the thickness o f theequivalent top plate,

17

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
E is the modulus o f elasticity o f steel, G is the shear modulus o f steel, a is the spacing between transverse bracing members, b is the width of the cross section at the bracing level, d is the length o f diagonal bracing member, A j is the area o f diagonal bracing member, A f is the area o f the actual top flange plus one quarter o f web, A s is the area o f the strut,

Litrature Review

Is is the moment of inertia o f the transverse bracing member (strut) that is assumed to perform like a beam member, If is the moment of inertia of the actual top flange. The theory o f thin-walled elastic beams has been used by M cDonald et. al. to analyze the equivalent closed box section. Normal and shearing stresses at various points in the webs and flanges o f the specimens were analytically determined using the equivalent closed section. The study showed that the computed normal and shear stresses were in good agreement with the measured stresses. The study indicated that the measured rotations and deflections o f the braced girders were smaller than those computed for the unbraced members. This indicates the influence o f the bracing. Measured and estimated axial forces in some top bracing members were compared; the results showed that these forces were in good agreement. The study concluded that using the concept o f an equivalent closed box section was a good method to approach the calculation o f the forces in the top bracing o f the quasi-closed box girder. In 1985, Branco and Green [10] investigated the influence o f the bracing systems on the behavior of open box girders. A test program was developed with a one-quarter scale model o f an open box girder. The girder was extensively instrumented at the mid span, three eights, quarter, and one-eight points in such a way that a complete strain profile o f the section could be obtained. Several tests were conducted on the girder with concentric and eccentric loading for different types and distributions o f bracing systems. The

18

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

test results w ere compared with analytical results using both the finite strip method and thin-walled curved beam theory. It was concluded that ties and cross frames were effective in preventing distortionof the

section. Top truss bracing that is usually placed near the top flange level was found to be effective in reducing the tw isting of the section. Yoo and Littrell [78] in 1986 used three-dim ensional finite element models composed entirely of eightnode brick elem ents and truss members to analyze curved I-girder steel bridges with varying curvature, length and cross frame intervals. The analyzed bridges were single span structures horizontally curved in a circular arc. Boundary conditions at the ends consisted of pinned-roller bearings radially oriented under the bottom flanges. The material o f the bridge superstructures was assumed to be linearly elastic and the deformations w ere assumed to remain within the limits o f small displacement theory. Results from this parametric study were utilized to develop empirical equations using linear and non-linear regression. These equations predict the ratio o f the maximum bending stress, the maximum warping stress and the maximum deck deflection for a curved bridge to corresponding parameters o f a straight bridge o f equal length. In addition, an equation suitable for a preliminary estimate o f maximum bracing spacing was developed as follows:

Smax

(2 .7 )

Where, Smax is the maximum bracing spacing, L is the length of the outside girder, R is the radius of the outside girder, Fws is the ratio o f the maximum warping stress to the maximum bending stress in a curved bridge.

19

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

Davidson, Keller and Yoo [30] in 1996 created three dimensional finite element models of a horizontally curved steel I-girder bridge connected by cross frames. ABAQUS (a Finite Element Program) was used in the bridge idealization; six shell elements were used to model the webs, while beam elements were used to m odel the flanges. All geometry, boundary conditions and loading conditions were modeled in the cylindrical coordinate system. As in the previous paper o f Yoo et. al. (1986), the material o f the models was assum ed to remain linearly elastic and deformations were assumed to remain within the limits of small displacem ent theory. The authors investigated the effects o f various parameters (number o f girders, girder spacing, span length, radius o f curvature, flange width and cross frame spacing) on the warping-bending stress ratio, and an equation for the preliminary cross frame spacing was developed from regression as follows:

Smax = L

Z FwsRbfX -In
V 509L2;

-1.52

. ( 2 .8)

Where, bf is the flange width. In contrast to the previous study by Yoo et. al., the authors o f this study believed that the flange width parameter was a major parameter in relation to reduction o f the warping stress. Therefore, flange width appears in the equation for maximum cross frame spacing (Eq. 2.8). The results o f the developed equation were compared and verified by the finite element m odel results. Davidson et.al. indicated in their study that the Yoo-Littrell equation (Eq. 2.7) consistently gave values for the cross frame spacing that were unconservative compared with the values used in the actual design, while for those spans with a high degree o f curvature, the values generated by Davidson et. al. equation compared more favorably with those in the AASHTO design [39],

20

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

In 1996, Z hang [79] constructed a three dimensional finite element model for a simple span system of horizontally curved non-composite twin box girder bridge using NASTRAN V.68 (a finite element analysis program). Quadrilateral plate elements were used to model the webs, the top and bottom flanges, and the internal and external solid diaphragms while beam elements were used to model the cross frame and lateral bracing members. The author studied the effects o f the arrangement of the cross frames, the lateral bracing and the external solid diaphragms on the box girder system behavior. The equivalent thickness form ula was verified. The study showed that as the number o f cross frames increased, the behavior o f the box girder with lateral bracing approached that o f the quasi-closed box girder (equivalent closed box girder). In addition, the smaller the ratio o f span length to radius of curvature was, the closer the behavior o f the box girder with lateral bracing would agree with that o f the quasi closed box girder. AASHTO gives the lower bound value o f the member size o f the lateral bracing as:

A b > 0 .0 3 b ................................................................................................................(2 .9 )

Where, 'y Ab is the cross sectional area o f lateral bracing m em ber (in ), b is the bottom flange width between webs (in.). This formula is based on the ratio of the induced warping stress to the bending stress that is less than five percent. The results o f this study indicated that this lower bound is too low, and thus, inadequate. The bending moment and the torque induced in the box girders were calculated by both finite element and M /R method, and the results were compared. It was shown that the M /R method gave reasonable results for the bending moments and the torques in comparison with the results given by NASTRAN. In 2000, Helwig and Fan [50] conducted a field study on a curved composite trapezoidal twin box girder bridge. Girder stresses were measured during erection, construction o f the concrete slab and

21

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

subsequent live loading. Three-dimensional finite element models were developed to simulate the construction o f the bridge. It was shown that the finite element solution and the measured results were in reasonable agreement. The finite element model was also used to conduct a number o f parametric studies on the top flange lateral truss and the internal bracing systems for box girders. The results indicated that current design methods often yield unconservative brace forces and girder stresses in box girders. Based on equilibrium and compatibility criteria, design equations were developed to evaluate the bending induced brace forces in the horizontal trusses and the lateral bending stresses o f the top flange. The forces that are developed in the top flange horizontal truss were shown to be influenced by three parameters: 1) The torsional m om ent on the girder. 2) The vertical bending o f the box girder, and 3) The lateral component of the applied load Therefore, the total forces in the truss system members and the total stresses in the top flange o f the box girder can be expressed as follows:

S-T ot

S b e n d +

> l a t ................................................................................................................................................................ ( 2 . 1 0 )

M ot

M p M + M e n d + M a t ............................................................................... ( 2 - 1 1 )

fL LTot = fL + fL + fx xTop ............................................................................ (2 .1 2 ) ^b en d M at

Where, Stot is the total forces in the struts,

22

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
Sbend

Litrature Review
the forces in the struts caused by the vertical bending o f the box girder,

Sjat is the forces in the struts caused by the lateral component o f applied loading, Dtot is the total forces in the diagonals o f the top flange truss, Dgpjy[ is the diagonal forces from the torsional moment determined using EPM (Equivalent Plate M ethod) as shown in Fig. 2.6, Dbend is foe frce in a diagonal caused by the vertical bending o f the box girder, Dlat is the force in a diagonal caused by the lateral components o f applied loadings, fL tot is the total top flange stress, fx top is the longitudinal stress at the middle o f the top flange, fp iat is the lateral bending stress caused by the lateral component o f applied loadings, f( bend is the lateral bending stress caused by the axial forces in the struts that result from the vertical bending o f the box girder. Two types of top flange truss systems are commonly used: 1) Single Diagonal (SD) and 2) Xdiagonals, see Fig. 2.4. For SD-Type Trusses: fv sco sa (2 .1 3 a )

x Top_________

Sbend

~ b e n d S*n a

(2 .1 3 b )

23

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

1.5s c
9 bend'

. (2 .1 3 c )

(bf)tf

For X-Type Trusses:

fT
D bend

-'bend

SC O Sa
.(2 .1 4 a ) '

k;

S bend = - 2 D b en ds i n a ......................................................................................................( 2 ' 1 4 b )

-'bend

.(2 .1 4 c )

In which K] and K2 are parameters defined by:

_d_ + (sin a ) + A , A

- (s in a ) ^ 2 ( b f) t f

.(2 .1 5 a )

2 b (sin a )

.(2 .1 5 b )

24

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
Where, s is the spacing of struts (panel length),

Litrature Review

a is the acute angle between the top flange and the diagonal,
bf is the top flange width, t f is top flange thickness, d is length o f the diagonal, b is the distance between centers o f the top flanges, A d and A s are the respective cross sectional areas o f the diagonals and the struts. In a SD-type truss, the struts carry the majority o f the lateral component o f the applied loading, however, the struts and the diagonals in the X-type share the lateral component. For simplicity, the authors recommended that each strut should be designed to carry the entire lateral load component since the diagonal force due to the lateral load component is relatively small compared to the bending and torsional components. Therefore, the same expressions were recommended for both SD-type and X-type truss system:

(2 .1 6 a )

S lat = P s

(2 .1 6 b )

25

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

ps

fL,.t =
f

(2 J 6 c )

Where, p is the entire lateral component o f applied loading as shown in Fig. 2.7.

Fictitious

Dsd= qb/sina

DX 1= DX 25sqb/(2sina)

(b)

Fig. 2.6. Diagonal Brace Forces from Torsion According to (EPM) [50].

26

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

s jg 5 a -S * s o V 5 TJ S 3 O iJ 'S a a cu < a o a S u 3 s

S tj)

ie< J

fS .Sf

27

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

In 2002, Chavel and Earls [17] presented examples of difficulties that commonly happen during construction o f horizontally curved steel I-girder bridges. Construction difficulties and instabilities can result during bridge erection from unexpected and excessive out o f plane displacements. As part of their study, the authors implemented a nonlinear finite element model of the Ford city veterans bridge and analyzed each stage o f the curved span erection sequence. It was shown that the out o f plane displacements could be m inimized by the use o f temporary supports. In 2002, Jetann, Helwig and Lowery [52] conducted a laboratory test on Permanent Metal Deck Forms (PMDF) systems that had been recently used in the bridge construction as a concrete formwork. These forms are generally not allowed to be considered for bracing in bridge construction. The study proved that a relatively simple modification to the connection details between the forms and top flange o f the girder enhances the shear strength, stiffness and overall performance o f such systems. The shear strength and stiffness improvements o f PMDF systems have encouraged the bridge industry to consider this kind of formwork as lateral bracing if it is properly detailed. Chen and Yura in 2003 [19], presented the results o f experimental tests conducted on a straight trapezoidal steel box girder under pure torsion with a permanent metal deck serving as the top flange lateral bracing. The corrugated metal decking was modeled by an equivalent flat plate that is based on shear stiffness predictions from the Steel Deck Institute (SDI) Diaphragm Design Manual. The specimen used in the experimental test was a straight trapezoidal steel box girder measuring 54 in. in depth. Three bracing configurations that incorporated the metal deck panels as the top lateral bracing system were tested. The three test configurations were 16-gauge (0.0598 in.) deck, 20-gauge (0.0359 in.) deck and 20gauge deck with stiffening angles. A three dimensional finite element model was developed using the commercially available program ABAQUS to analyze the behavior o f the test specimen. The experimental test results demonstrated that permanent metal deck forms could substantially increase the torsional stiffness o f a trapezoidal steel box girder. In addition, modeling o f the metal decking using the equivalent flat plate approximation and the modified SDI shear stiffness formulations appeared reasonable. The test

28

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

showed that th e use o f the stiffening angle detail in the 20-gauge test configuration produced an overall torsional stiffness increase up to approximately 12%.

2.3 Static A nalysis o f Curved Box G irder Bridges


Torsional and bending moments are considered the to be major structural m em ber forces in curved bridges. The horizontally curved beam theory that combines the effects o f bending and torsion was developed by Vlasov [73] and Dabrowski [28]. Although the closed form solutions are complex and difficult to apply in practice, these solutions are considered as the basis for m any analysis methods. The analysis methods of curved bridges are usually simplified by some assumptions; the accuracy of such methods relies generally on the validity o f these assumptions. Several numerical methods o f analysis were developed based on the general stiffness method o f structural analysis. M ost o f these methods are considered as the basis for a number o f computer programs that are being used in the analysis and design o f curved bridges today. The different methods that are commonly used in the analysis o f curved box girder bridges will be discussed in the following sub-sections.

2.3.1 The Plane Grid M ethod


The palne grid method was first used in the analysis o f curved I-girder bridges. The method treats the curved girder bridge as a planar frame and the loads are applied normally to the frame. A further refinement o f this method was made so that box girders could be analyzed. The grid method assumes that bracing is provided at the top flange level to make the section like a closed section and assumes that the girder cross section does not change shape (i.e. no distortion). Therefore, an adequate number o f cross frames is required to minimize distortional stresses. The method is widely used in design offices because it requires minimal modeling efforts. In 1974, Brennan [11] tested small scale horizontally curved models elastically with several configurations. The models were o f multi-I girders or multi-spline boxes, in two or three spans, with or without skew ends, composite or non-composite and with variable intermediate diaphragm spacing. They

29

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

were subjected to a total of 76 loading conditions. Strain, displacements and support reactions were recorded. The models were analyzed using a planar grid computer program and covered the effects o f 180 loading conditions were covered. The correlation between vertical displacements and moments obtained by analytical methods and those obtained experimentally was good, but there were indications that for structures with small radii o f curvature, the values o f torsional constant used in the theoretical analysis were not sufficiently accurate.

2.3.2 Space Fram e M ethod


The space frame method idealizes the bridge as a three-dimensional model. This idealization can be accomplished by using a three-dimensional space frame matrix program. The model commonly permits the consideration o f three moments and three normal forces at the end o f each member. Thereafter, bending, torsion and warping effects can be incorporated into the curved girder analysis. In 1984, Heins and Jin [49] examined the live load distribution o f single and continuous curved composite I-girder bridges using the three-dimensional space frame method. The top and bottom flanges of the girder were modeled by longitudinal top and bottom space frame members, these flanges were then connected with rigid vertical and diagonal members representing the webs. The properties o f the top flange consisted o f the base plates plus the effective composite deck. The web element properties were then assumed and the space frame representing the girder was examined under the effects o f known loadings. The deformations and stresses produced were compared with the stresses and deformations based on curved beam theory. The web plates were then modified based on this comparison, until complete agreement between curved beam theory and the space frame model occurred. The authors conducted a parametric study on the response o f the curved bridges subjected to AASHTO live loading. In studying the response o f each model, bottom lateral bracing was: 1) not included 2) placed in every other bay and 3) placed in all bays. Since the grid analysis method could not account for the effect o f these bracings, examination of the response o f these models permitted the authors to develop empirical equations that can

30

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

be applied to the results obtained by the grid method to account for the load redistribution happening because o f the existence of bottom lateral braing.

2.3.3 Folded Plate M ethod


The plane stress elasticity theory and the classical bending theory are considered as the bases for the folded plate method. In the folded plate method, the curved bridge is made up o f an assembly o f curved strip elements (plates) interconnected at joints along their longitudinal edges and simply supported at their ends by diaphragms. These diaphragms are infinitely stiff in their own plane but perfectly flexible normal to their own plane. According to the Canadian Highway Bridge Design Code (CHBDC) (2000) [18], this method is applicable to bridges with support conditions that are closely equivalent to line supports at both ends and also line intermediate supports in the case o f multi-span bridges. In 1960, Scordelis [65] developed an analytical procedure for determining longitudinal stresses, transverse moments and vertical deflections in folded plate structures by utilizing matrix algebra. The procedure can be easily programmed for digital computers. Meyer and Scordelis [59] in 1971 described a method o f analysis called the finite strip analysis of curved folded plate structures, or simply the curved strip method that is closely related to the finite strip analysis o f plates. The authors developed a general stiffness matrix and load vector for a general conical shell element. Each element had two nodal joints; each joint had four degrees o f freedom (three translations and one rotation). All loads, displacements and forces were developed into Fourier series. A computer program called CURSTAR based on the curved strip theory was written. Two-cell simply supported box girder bridge was analyzed with CURSTAR for a standard AASHTO truck in two different transverse positions. The differences between the results o f CURSTAR and the results obtained from the curved beam theory were relatively small. The authors attributed this small difference to the slightly different truck simulation between the theories.

31

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
2.3.4 Finite Strip M ethod

Litrature Review

Currently, the finite element and the finite strip methods are the most commonly used methods for the analysis o f curved box girders. However, these methods usually require significant computer time and resources as w ell as extensive data preparation. This makes them suitable for the final analysis o f a curved girder. The finite strip method has an advantage over the finite element method that the amount of data input is reduced drastically because o f strip idealization and according to that, it requires shorter computer time and sm aller computer storage. Computation time, a direct measure o f the cost, can be as low as about one tenth o f that for the finite element method. The finite strip method of analysis can be considered as a special form o f finite element using the matrix displacement approach. Using a strain-displacement relationship, the strain energy o f the structure and the potential energy of external loads can be expressed by displacement parameters. At equilibrium, the values o f the displacement parameters should make the total potential energy o f the structure become minimal. The box girder section can be discretized into circular finite strips running between the two ends and connected along their edges by longitudinal nodal lines. The displacement functions are assumed to consist o f harmonics (Fourier series) and polynomials. The harmonics usually satisfy the boundary conditions. The external loads are also resolved into the same Fourier series for the corresponding displacement components. Application o f the finite strip method to the analysis o f curved box girders was pioneered by Cheung and Cheung (1971) [20]. The authors programmed the method and used the program to solve numerical examples o f curved bridges a well as straight bridges by making the radius o f curvature very large and the subtended angle very small. In 1992, Bradford and Wong [9] used the finite strip method with one harmonic to study the local buckling o f the bottom flange and web o f a composite box girder in negative bending. Design graphs o f the elastic buckling coefficients were produced. These graphs can be used to obtain accurate values o f the web depth-thickness ratio that separates the boundary between slender and semi-compact sections. In 1995, using the finite strip method, Cheung and Foo [21] presented the results o f a parametric study on the relative behavior o f curved and straight box girder bridges using the finite strip method. The

32

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

parameters considered in the study included types o f cross section, type/location and magnitude o f loads, span length and radius o f curvature. Empirical equations were developed for longitudinal moment ratios between curved and straight box girder bridges.

2.3.5 Finite E lem ent M ethod


The finite elem ent method has now widely become a powerful and versatile analytical tool. It is very effective in the analysis of structures with complex geometry, material properties and support conditions and subjected to a variety o f loading conditions. The method can be considered an extension o f analysis techniques mentioned earlier, in which the structure is discretized into a num ber o f elements interconnected at a finite number o f nodal points. In 1971, Chu and Pinjarkar [22] developed a finite element approach for analyzing curved box girder bridges. The top and bottom flanges were modeled as horizontal sector plates while the web was idealized as vertical cylindrical shell elements. The sloped web elements (Bath-Tub girders) were not considered as they require conical shell elements. Membrane and bending actions were both considered for the plate and shell elements, but no interaction between them was assumed. The developed method is applicable to simply supported bridges with infinitely stiff end diaphragms in their own plane. A simply supported box girder bridge provided with diaphragms only at the ends was analyzed by this approach with different radii o f curvature. Sisodiya, Cheung and Ghali in 1970 [69] presented finite element analyses of single box girder skew bridges that were curved in plan. The bridge that could be analyzed by this method may be o f varying width, curved in any shape, not just a circular shape and with any support conditions. They used rectangular elements for the webs and parallelogram or triangular elements for top and bottom flanges. The results o f the analyses were verified by tests on an aluminum bridge model. Two different finite element mesh divisions were used in the analyses, a coarse mesh and a fine m esh obtained by subdividing all the coarse elements into two in the span direction. They concluded that the deflections obtained from the two different meshes were practically identical, and, that the experimental values for the deflection were

33

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

somewhat higher. This was attributed to the fact that the stiffness o f the model beam has been reduced by the imperfect epoxy bond between the web and flanges. In 1971, Chapman, Dowling, Lim and Billington [16] analyzed steel and concrete box girder bridges with different cross section shapes using the finite element method to investigate the effect of intermediate diaphragms on the warping and distortional stresses. They showed that curved steel boxes even with symmetrical load components gave rise to distortional stresses, and showed that the use o f sloping webs resulted in an increase in distortional stresses. In the analysis o f curved box girder bridges, curvilinear boundaries were usually approximated by a series o f straight lines. The structure commonly used was idealized by quadrilateral or rectangular flat elements for the webs and the flanges. It is obvious that the geometrical errors produced by modeling curvilinear boundaries using straight lines tend to decrease as the size o f the elements decreases. Fam and Turkstra in 1975 [33] developed a finite element scheme for static and free vibration analysis o f box girders having arbitrary combinations o f straight and curved alignments. They idealized the top and bottom flanges by rectangular elements for the straight portion and annular elements for the curved one. The webs were idealized with rectangular elements for the straight portion o f the girder and conical elements for the curved portion; the conical elements reduce to a cylindrical elements when the web is vertical. A threedimensional finite element program entitled CBRIDG that follows the above mentioned scheme was written in FORTRAN programming language. The program was used to analyze few well-known experiments on curved box girders. An excellent agreement between the analytical and experimental results was obtained. In 1976, Fam and Turkstra [34] conducted an experimental investigation on a simply supported plexiglass model o f a curved box girder. They developed a three dimensional finite element m odel o f the experiment utilizing the program CBRIDG . The finite element results were found to be in good agreement with the experimental results for deflections, radial stresses and tangential stresses. They

34

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

concluded that the distribution o f diaphragms has a pronounced effect on the magnitude and distribution of both tangential and radial stresses in the flanges for conditions of severe curvature. In 1983, Zhang and Lyons [80] developed a thin walled box beam element, which has nine degrees of freedom at each beam node. The thin walled beam theory was combined with the finite element technique in developing this element. In addition to the usual six degrees o f freedom, three extra degrees o f freedom were incorporated in the finite element formulation to account for the longitudinal warping and transverse distortions. These additional degrees o f freedom are: the rate of change o f twisting angle, the distortional angle o f the cross section and the rate o f change of distortional angle. The entire theoretical scheme was incorporated in a finite element stress analysis system that was written in FORTRAN. The performance of the element w as compared with the results o f a three-dimensional shell element model and with the results o f model experiments and it was in close agreement. In 1995, Galuta and Cheung [37] developed an analytical solution that combines the boundary element method (BEM) with the finite element method (FEM) to analyze box girder bridges. Numerical techniques for solving engineering problems can be categorized as domain or boundary techniques; the domain method is the method that involves a discretization of the dom ain like the finite element method or the finite strip method, while the boundary method is the one that involves a discretization of the boundary. For bridges subjected to moving loads, the finite strip method requires a large number of longitudinal elements and many terms to simulate the local effects o f the concentrated loads, while in the finite element method the slab mesh requires more refinement in the vicinity o f point loads to capture the local effects. The authors overcame these difficulties with the domain type m ethod by developing a hybrid analytical solution. The finite element method was used to model the webs and bottom flange while the deck was modeled by the boundary element method. The performance o f the BEM -FEM was tested through two numerical examples. The results obtained from this m odel were found to be in good agreement when compared with the results o f the finite strip method.

35

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
2.3.6 Thin W alled Curved Beam Theory

Litrature Review

Saint-Venant established the curved beam theory in the 19th century for the case o f a solid curved bar loaded in a direction normal to the plane o f curvature. In non-circular solid sections, the warping stresses caused by torsion are often assumed negligible [28], while such stresses are primary in thin walled sections and particularly in open sections. For the above-mentioned reason, the solutions for a curved girder o f solid cross section can not be applied to the analysis of thin walled sections. The thin-walled beam theory was developed by Vlasov [73] for axi-symmetric sections. The main results o f V lasovs investigations were contained in his book Thin-Walled Elastic Beams that was published in 1940. Dabrowski [28] derived the fundamental equations that account for warping deformations caused by the gradient o f normal stresses in individual box element. The theory assumes also that the thin-walled members preserve their cross section shape under all loads (non-deformable cross section). In 1975, Oleinik and Heins [63], and 1976 Heins and Oleinik [42] divided the analysis o f curved box girders into two parts. In the first part o f the analysis, the box sections were assumed to retain their shape under the load. The load-deformation response o f such a curved box that considers bending, torsion and warping deformations was developed by Vlasov. V lasovs differential equations were solved using a finite difference approach to calculate the normal bending and normal warping stresses. In the second part o f the analysis, the effect o f cross sectional deformations was considered. These cross sectional deformations were calculated using a differential equation developed by Dabrowski. This equation was also solved using the finite difference approach and the normal stresses that resulting from cross-sectional deformations were calculated. The effects o f both parts were summed to give the total normal stress distribution. The above-mentioned formulations and the final solutions o f these basic differential equations were programmed by Heins and Sheu in 1982 [43], A single straight or curved box girder with prismatic or nonprismatic section can be analysed using this program. The box girder may have internal transverse diaphragms spaced along the box and top lateral bracing. A parametric study was conducted using this

36

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

program to investigate the effect of internal diaphragms on the induced normal stresses in curved box girder bridges due to dead and live loads. In 1995, F u and Hsu [36] generated a new finite element based on Vlasov's theory o f curved thinwalled beams. The horizontally curved thin walled beam element stiffness was developed directly in the cylindrical coordinate system. The element stiffnesses o f a curved thin-walled beam with four degrees of freedom (torsion, bending, vertical shear and bimoment) per node were presented in this study. The results produced using this element were in good agreement with Dabrowski's closed form solution. A summary o f the methods o f analysis that were discussed earlier along with corresponding references are listed in Table 2.1

Table 2.1: Curved Box-G irder M ethods o f Analysis

Method Stiffness

Solution Plane Grid Space Frame Folded Plate Brennan (1974) [11]

Reference

Heins & Jin (1984) [49] Scordelis (1960) [65] Meyer & Scordelis (1971) [59] Cheung & Cheung (1971) [20] Bradford & Wong (1992) [9] Cheung & Foo (1995) [21] Sisodiya, Chueng & Ghali (1970) [69] Chu & Pinjarkar (1971) [22] Chapman et.el. (1971) [16] Fam & Turkstra (1975) & (1976) [33]&[34] Zhang & Lyons (1983) [80] Galuta & Cheung (1995) [37] Vlasov (1940) [73] Dabrowski (1968) [28] Oleinik & Heins (1975) [63] Heins & Oleinik (1976) [42] Heins and Sheu (1982) [43]

Finite Strip

Finite Element

Differential Equation

Closed Form

Finite Difference

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
2.4 E xperim ental Studies on Elastic R esponse o f B ox G irder Bridges

Litrature Review

Many experimental studies have been conducted to investigate the behavior o f curved steel plate girder bridges while only few studies related to similar bridges using box girders. The material presented in this section covers experimental studies on box girderr bridges and is separated into two categories: field studies and m odel studies.

2.4.1 Field Studies


The primary objectives o f field tests that usually involve comprehensive instrumentation o f the steel superstructure are to measure loads, strains and deformations in the curved steel bridges. These measurements usually commence during the erection stage and track the strain history throughout the erection, concrete deck casting and application o f live loading. In 1973, Buchanan, Yoo and Heins [14] conducted a field test on a twin-box girder bridge with a composite 10 inches thick deck located near Baltimore, Maryland. The bridge has three continuous spans o f 100-135-122 ft. with a centerline radius o f approximately 1318 ft. Strains and displacements were recorded using strain gages and deflection meters. The study reported two sets o f tests. The first one measured the response o f the steel section to the placement o f the concrete deck. This was accomplished with three separate pours. The second test measured strains, deflections and rotations induced during live loading o f the completed structure. The finite difference approach was used to analyze the bridge for both test measurements. The experimental results obtained from the test programs were then compared with the analytical solutions, and where possible, with the design estimates. Analytical results showed good agreement with experimental behavior, while a comparison with design estimates showed that the design calculations were generally conservative. In 1975, Kissane and Beal [53] presented an experimentally determined behavior o f an asymmetric, two-span continuous horizontally curved steel box girder bridge located on the Avoco-Bath section o f the Southern Tier Expressway in the state o f New York. The evaluated bridge is a three-box girder bridge with

38

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

a centerline span length and radius o f 112.7 and 290 ft. respectively. Strains, deflections, rotations and cross-sectional deformations in one span o f the structure were recorded. The first phase of the test program was to measure the structure's response to the placement o f the 8-inch concrete deck that was placed in one continuous pour. The second phase of the test program was to evaluate the behavior of the completed structure under live load conditions. The structure was analyzed using a curved girder analysis program that idealized the structure as a planar grid and used the stiffness method to solve for the unknown joint displacements and subsequently the moment, shear and torque at each joint. Test results showed that the experimental in-plane bending moments for dead load and static live loads were about 86 percent o f their respective theoretical values. However, comparisons between experimental and theoretical results concerning the proportion o f the total load carried by the individual girder were within 6 percent. The experimental deflection compared well with the theoretical deflection. The authors believed that the high torsional stiffness resulting from heavy and closely spaced internal diaphragms was the reason behind getting the low live load distribution factor compared to the factor used in the design o f straight box girder structures. Heins and Lee in 1981 [45] presented the experimental results obtained from the field testing o f a twospan continuous curved composite steel box girder located in Seoul, Korea. These results were then compared with results from several analytical schemes that were being used for the design o f such structures (CURSGL and STDUDL planar grid program; SAP finite element program). In general, correlation between the three different analytical schemes and the test data was excellent. Because o f the high torsional stiffness o f the box, warping was negligible and the beam idealization was acceptable. In 2000, Helwig and Fan [50] conducted field studies on a three-span continuous curved composite trapezoidal twin box girder bridge located in Houston, Texas. Girder strains were measured using strain gages during girder erection, construction o f the concrete slab and subsequent live loading application. The construction of the box girder bridge was idealized comprehensively with a three-dimensional finite element analysis model o f the quasi-closed steel section using ANSYS (finite element program). A line

39

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

element m odel as well as a detailed three-dimensional finite element model o f the composite section were used to sim ulate live loading o f the bridge. The finite element model was able to reasonably approximate the magnitude and distribution of the girder stresses and the brace forces during girder erection. The line element m odel was effective in predicting the largest bending stresses; however, live load distribution factors would have to be applied.

2.4.2 M odel Studies


In 1972, Heins, Bonakdarpour and Bell [44] presented an investigation o f the behavior o f a small plexiglass curved box beam model bridge. They tested a three-box girder bridge model as both a single span and two-span structure. Strain and dial gages were used to measure strains, deflections and rotations under the effect o f a series o f concentrated loads. The models were analyzed using the slope deflection theory. This theory assumes no cross-sectional distortions and thus adequate internal cross frames should be provided to minimize cross sectional distortion. The cross-sectional properties (moment o f inertia Ix, St.Venant torsional constant Kt and warping constant Iw) required for the analytical solution were computed for the bridge section. It was expected that the warping stresses would be negligible. Thus, the warping constant Iw was computed for both the individual girder section and multi-girder section. Correlation with the experimental using both Iw values provided good results. For this reason, it was concluded that the warping phenomenon was negligible for these closed box girder sections and the warping constant, Iw need not be considered in the analysis. In 1975, Evans and Al-Rifaie [32] presented experimental and theoretical studies on the behavior of curved box girder models. They tested eighteen models o f different curvatures and different cross-sectional dimensions subjected to different loading conditions. They used a sand/araldite material to build some o f the models and the rest were built from steel plates. All the models were sim ply supported with rigid end diaphragms. The models were instrumented with strain gages. Deflections, longitudinal stresses and transverse moments were measured. The models were analyzed using the finite element method. It was

40

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

concluded that the finite element method was capable o f accurately representing the behavior o f curved box girders. A fter that, the finite element method was used successfully as a tool to conduct a parametric study to show the curvature effect on the deflections, longitudinal stresses and transverse moments of the models. In 1975, Aslam and Godden [6] presented the results o f an experimental study on the static response of a series o f four-cell aluminum curved box girder bridges. The main purpose o f this study was to develop an accurate experimental data for checking the validity o f computer analyses o f such structures. The models were tested elastically, with and without a mid-span radial diaphragm for different locations o f a single point load. The models were analyzed by the computer program CURSTAR developed by M eyer and Scordelis [59]. Experimental and analytical results related to tangential plate forces and their distribution across the section, radial plate bending moments and deflections were in very close agreement. In 1976, Fam and Turkstra [34] performed two series of tests on plexiglass simply supported box girder models. The first series was with end diaphragms only while the second series was with both end and intermediate diaphragms. The models were tested under the effect o f concentrated and line loadings. Strain gages and vertical dial gages were used to measure strains and vertical deflections. A threedimensional finite element program was developed to analyze these models. The results obtained from the finite element program were found to be in good agreement with experimental results for deflections, radial stresses and tangential stresses. It was concluded that the introduction o f diaphragms had an effect on the magnitude and distribution o f both longitudinal and radial stresses in the flanges o f the girders with severe curvature. In 1988, Siddiqui and N g [67] conducted tests on two plexiglass box girder models to examine the effect o f rigid diaphragms in reducing the warping and distortional stresses that usually developed in box girders from deformation o f the cross section. One o f the models had a rectangular cross section while the other one was trapezoidal. Experimental results were compared with analytical results obtained from the beam on elastic foundation (BEF) analogy [74]. Both models were tested as a simply supported beam with

41

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 2

Litrature Review

a 1168.4 mm. span. The testing program was divided into three major parts: case I, the models were tested using end diaphragms only; Case II, two interior diaphragms were inserted in the models and case III; six interior diaphragm s were inserted in the models. The models were subjected to a concentrated load at different positions along the span and with different eccentricities. Both models were instrumented with dial gages and strain gages to measure vertical and rotational displacements as well as longitudinal strains. The experimental results indicated 1) that the deformation of a box girder cross section eccentrically loaded m ay cause substantial warping and distortional stresses and 2 ) that these stresses can be effectively controlled by the installation of rigid diaphragms along the span o f the girder. Warping and distortional stresses that w ere calculated by the BEF analogy compared reasonably well with the experimental stresses. In 1988, Yasunori, Hamada and Oshiro [76] tested three curved composite box girders with end diaphragms to investigate their distortional response and slip behavior. The test specimens had different radii of curvature, cross sections and placements o f shear connectors. The test results were compared with analytical results based on the curved beam theory and the distortional theory proposed by Dabrowski, as well as with analytical results from the finite strip method. The authors concluded that the cross-sectional deformations o f curved composite box girders with only end diaphragms were considerably significant and produced large additional longitudinal stresses. The major conclusion from the investigation was that if a sufficient number o f diaphragms was provided, the effects o f cross-sectional deformations might be disregarded in the design o f curved composite box girders. In 1993, Ng, Heung and Hachem [62] conducted a model study to evaluate the elastic response o f a curved composite bridge. The specimen was a 1/24 linear scale model o f the Cyrville road bridge over passing the Queensway, east o f Ottawa, Ontario. The prototype bridge was a two-lane, two-span concrete curved box girder structure, while the model bridge was a composite construction o f concrete and aluminum. The model bridge was instrumented by strain gages and dial gages to measure strain and vertical deflections. ADINA (finite element program) was used to examine the bridge analytically. Analytical results o f both vertical displacements and normal stresses at critical sections compared fairly

42

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 2

Litrature Review

well with those observed experimentally; the authors concluded that ADINA could be successfully applied in modeling an accurate elastic response o f composite curved box girder bridges.

2.5 U ltim ate R esponse o f Box Girder Bridges

Much experimental and analytical information concerning nonlinear behavior up to collapse is found for straight box beams, but far less is available for horizontally curved box beams. In 1970, Culver and Mozer [25] presented the results o f a series o f static tests on horizontally curved steel box girders as part o f research efforts that were conducted by (CURT). The main objective o f these tests was to evaluate the local buckling characteristics o f the compression flange. In addition to this main objective, the tests were to serve as preliminary data on the behavior o f transversely stiffened webs of curved box girders in the elastic and inelastic ranges. In these series o f tests, two box girders with unstiffened compression flanges and four with longitudinally stiffened compression flanges were tested. The stiffened boxes had rectangular cross sections, while the unstiffened boxes had sloping or inclined webs. The stiffened boxes were used with different centerline span lengths and different radii o f curvature. All the specimens had transverse web stiffeners spaced from 0.3 to 1.0 tim es the web depth. A steel plate was used as a tension flange that represented an uncracked concrete deck. The boxes were tested as simply supported spans. Two concentrated loads were applied at the centerline of the cross section symmetric with respect to the middle o f the girder. Because o f the high torsional rigidity o f the specimens, the torque produced by the curvature o f the boxes was extremely small. Therefore, an external torque was applied through a torque beam at one end o f the specimen to artificially increase the torsional shear stresses in the specimens. Twisting about the longitudinal axis at the other end o f the specimen was prevented by using a fully restrained torque beam. The applied loads and end torque were applied separately and were increased monotonically up to failure. Based on this process, different ratios o f vertical and torsional loads could be applied. The tests results were compared with analytical solutions using the curved beam theory by

43

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

Dabrowski a n d the calculated critical loads. The experimental results were also compared with existing information on straight girders to assess the differences in behavior resulting from curvature. Based on th e tests results, the report concluded that the web strength in a curved box girder was lower than that in straight plate girders. This decrease in strength was related to the curvature o f the web. It was also believed that the inclination of the box webs might have contributed to this decrease in strength. Longitudinal stiffeners were observed to twist, permitting the flange to twist out o f plane. The report indicated that the width to thickness ratio o f stiffened and unstiffened compression flanges based on AASHTO specifications for straight box girders was adequate for A36 steel curved box girders. Note that the AASHTO lim it would insure that the compression flange in the box girder would reach the fully yielded condition prior to local buckling failure. If ultimate strength design criteria were adopted, where the factored design loads are based on the full plastic moment resistance (Limit State Design), a smaller width to thickness ratio for the stiffened boxes should be used because o f the influence of the twisting of the longitudinal stiffeners on the flange strength. It was indicated also that without adequate diaphragms, the cross section o f curved box girders deformed under even concentric loads and normal stresses caused by such cross-sectional deformations existed. In 1971, Culver and M ozer [26] continued their efforts to study the behavior o f curved box girders under the CURT research program. The report presented a total o f 13 tests in the elastic range on two curved box girders. Subsequently, tests up to failure were performed. One o f the boxes had a rectangular cross section while the other had a trapezoidal (inclined web) cross section. The main objective o f this series of tests was to obtain additional information on the web strength particularly for more slender webs. The influence of intermediate diaphragms or cross frames on the behavior o f the curved box girders was also studied. Three different loading conditions were used: 1) a simply supported span with concentrated loads at quarter points, 2) a simply supported span with concentrated loads at m id span, and 3) two equal spans with equal concentrated loads at the middle o f each span. The report concluded that the internal cross frames were effective in reducing box deformations. M easured vertical deflections and rotations o f the box

44

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

L itratu re Review

girder were greater than the computed ones that assumed no deformation o f the box. The report also identified the effectiveness o f slender curved webs in contributing to the overall flexural rigidity o f the girder. In 1973, Abdel-Sayed [1] examined the problem of pre-buckling behavior as well as the critical limit o f loading for webs of curved girders under the combined action of shear and normal stresses. He indicated that except for the case o f pure shear, the problem o f a curved web subjected to in-plane membrane loading was not a buckling problem, but rather one o f deflection amplification. However, the membrane loading could not exceed a certain critical limit. Approximate conservative formulae were developed to estimate the curved web buckling coefficient as a function o f the straight web buckling coefficient for pure normal loading, pure shear and for combined normal and shear loading. In 1973, Corrodo and Yen [24] conducted failure tests on two slender web rectangular steel box girder models having a vertical axis of symmetry. The main goal o f the study was to evaluate the post-buckling behavior, modes o f failure and the load carrying capacity o f single span rectangular steel box girders. The specimens were instrumented with strain gages and dial gages; dial gages were utilized to m onitor the vertical and rotational displacement o f each specimen. The specimens were tested as simply supported members. The models were tested under both symmetrical loading and unsymmetrical loading to create a torsion effect. The authors concluded that when one o f the two webs o f a box girder panel failed, redistribution o f shear within the box panel took place provided that it was properly braced to prevent distortion o f the girder cross section, also the torsional stiffness o f a box girder panel was greatly reduced when the box panel experienced web failure as a result o f yielding along the tension diagonal. In 1979, Heins and Humphreys [46] developed an interaction equation to be used in the load factor design o f curved box girders. They conducted a series o f model box beam tests. Three specimens were used in the study: the first specimen was tested as a simple beam subjected to positive moment without a concrete deck and flange stiffener plate. The second specimen had the same structural details as the first specimen except that a wide reinforced concrete deck was added. The third specimen was tested as a

45

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

simple span beam subjected to negative moment without a concrete deck but the compression flange had a longitudinal stiffener. Cross diaphragms and top lateral bracing were provided in all the three specimens. All models w ere instrumented with strain gages along the top and bottom flanges, web plates, top lateral bracing and diaphragms. Deflection gages were placed at different locations to measure support movements, maximum vertical deformations and rotations. The models were tested to failure under a combination o f increasing torsional and bending forces. The results o f these tests provided information on the verification o f the elastic classical torsional theory, also information about the failure response o f the girders under combined bending and torsion. The following non-dimensional empirical interaction equation was developed to agree reasonably well with the test data:

M.V'5 +fXj 15 = ]
V
......................................................................... ( 2 ' 1?)

Where, M is the applied bending moment, T is the applied torque, Mp is the full plastic moment, Tp is the full plastic torque. In 1978 and 1979, Yonezawa et al and Dogaki et al [77, 31] in Kansai University (Japan) tested two curved steel box girder models. The first model had a longitudinally stiffened plate deck, while the second one had a longitudinally and transversely stiffened plate deck. The models were simply supported. The study investigated the behavior and ultimate strength o f the curved orthotropic steel plate deck under static loading. The experimental stresses and deflections were compared with the theoretical values obtained from the nonlinear orthotropic sector plate theory. This theory is considered for the curved decks instead o f the flat plate theoiy to account for the curvature. In addition, the ultimate strength o f curved orthotropic

46

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

plate decks subjected to compression was examined experimentally, and the results were compared with collapse loads calculated from straight orthotropic plate deck theory. In 1985, Daniels [29] presented ultimate strength test results for a horizontally curved single cell closed composite steel-concrete box girder. The girder was tested as a simply supported beam with a 36 ft. span along the centerline of the girder and with a 120 ft. radius o f curvature. An 8 -inch concrete slab was attached to the webs using 3/4 inch diameter shear studs. The slab was constructed using a steel deck form between the webs. The girder was instrumented with strain gages located on the two webs and the bottom flange at three cross sections along the girder. Dial gages were used to measure vertical and horizontal displacements o f the box girder and out-of-plane displacements o f the outside web. Before casting the concrete deck, the non-composite steel box girder was also tested at low elastic response levels with and without the m etal deck forms. The test results consisted o f the load-deformation behavior and stresses at selected locations along the girder. The test results o f the non-composite steel box girder indicated that the metal deck form significantly reduced the vertical deflections o f the outside web. In the ultimate strength test for the composite section, the premature failure o f the concrete slab and shear connection near the two ends o f the girder did not permit the attainment o f a higher ultimate strength o f the girder that would have involved yielding o f the steel girder near the mid-span. The manner in which the ends o f the inside web were supported to prevent uplift caused by the large applied torque was believed to have resulted in the premature splitting and fai lure o f the concrete slab. In 1988, Thimmhardy and Korol [70] conducted an investigation to measure the magnitude o f geometric imperfection induced during the fabrication o f welded steel box girders. Based on over 9600 measurements o f geometric imperfections on plate panels o f webs and bottom flanges and 2200 measurements o f out-of-straightness o f stiffeners undertaken on nine steel box girder bridges; the authors concluded that the following tolerance limits o f initial imperfection were realistic for the Canadian steel industry to achieve.

47

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
For out-of-plane imperfection o f plate panel.

Litrature Review

= , ! , ................................................................................

Where,

A p is the magnitude o f the maximum out o f plane deformation o f stiffened panels and
b is the smaller o f the two plate dimensions. For out-of-straightness o f the stiffener. As : 1 7 - 5 5 0 ...................................................................................... ( 2 J 9 )

Where, As is the maximum out-of straightness of the stiffener and a is the stiffener span. In 1988, Korol, Thimmhardy and Cheung [55] conducted experimental tests up to failure on single cell, one-fourth scale box girders having identical cantilevers to model the geometry o f a pier girder o f the Hunt Club-Rideau bridge structure in Ottawa. The purpose o f the tests was to investigate the effect of geometric imperfections and residual stresses on the strength o f the compression flange (bottom flange) o f the single box girder. To develop disparities o f imperfection in the two cantilever ends o f the girder, different fillet weld sizes were used in attaching longitudinal stiffeners to the bottom flange plate. A tolerance imperfection level o f (b/205) was measured in one end compared with (b/153) in the other end. The two ends were loaded up to failure. The resulting difference in failure load for the two ends was only 3.5%. It was concluded that the effect o f imperfection on the failure loads o f Canadian built girders was not excessive. Thimmhardy in 1991 [71] presented the results o f the numerical study on the buckling behavior and ultimate strength o f stiffened compression flanges o f steel box girder bridges. An elasto-plastic large

48

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 2

Litrature Review

deflection finite element model was used in which both geometric and material nonlinearities were included. As w ell residual stresses, initial out-of-plane deformations o f the plate panels, out-of-straightness of longitudinal stiffeners, plate panel and stiffener slenderness ratios were included in the modeling. ADINA (Finite Element Program) was used to perform the numerical study. A three-node shell element was used to m odel the plate panels, while a two-node beam element was used to model the longitudinal stiffeners. The authors concluded that good agreement was achieved between the calculated and experimental membrane compression stresses that developed in the stiffened compression flange at maximum load. A parametric study was conducted to assess the effects o f out-of-plane deflection o f plate panel, out-of-straightness o f stiffeners, residual stress, plate panel slenderness and stiffener slenderness on the ultimate strength of the stiffened compression flange. The study recommended that the tolerance limits o f initial imperfections o f Eqs. 2.18 and 2.19 should be specified in the fabrication codes and should be considered in the design o f box girder bridges. In 1995, Yabuki, Arizumi and Vinnakota [75] proposed an analytical procedure to calculate the ultimate strength and non-linear behavior o f thin-walled, welded, steel box girders curved in plan and stiffened by intermediate interior diaphragms. A nonlinear finite element analysis was used to study the nonlinear flexural behavior o f curved girders. The Newton-Raphson iterative approach and the tangent stiffness method were used in the nonlinear finite element analysis. A modified element stress-strain curves allowing for plate local buckling were proposed based on regression analysis o f the stress-strain data; constitutive models based on this modified stress-strain curves were incorporated into the nonlinear analysis. The effect o f distortional warping was simulated in an iterative m anner at the initial strain state in the nonlinear analysis, where the strain increments for distortional warping that were calculated from the BEF analogy were applied in a step-by-step procedure until an external distortional moment was reached. Two tests on large scale curved steel box girders were also conducted. A comparison o f the analytical and the experimental results for the ultimate strength was judged satisfactory by the authors.

49

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

In 2002, Lee, Yoo and Lee [56] investigated the elastic buckling behavior of unstiffened and longitudinally stiffened compression flanges o f horizontally curved box girders. The compression flange of a horizontally curved steel box girder was subjected to combined action of compression and Saint Venant torsional shear. AASHTO Guide Specification (1993) [39] adopted the following interaction equation for evaluating the buckling strength of curved girder flanges:

( 2 .20 )

Where,
(crc)cr and (Tc)cr are the critical stress, respectively under compression or shear alone.

The authors modeled unstiffened rectangular panels with all four edges simply supported using 4-node plate/shell elements (ADINA). The panels were subjected to compression and shear. The finite element analysis results were compared with Eq. (2.20), and the authors concluded that it produced results that were in good agreement with finite element buckling analyses results. According to the AASHTO Guide Specifications (1993) [39], the shear buckling coefficient, Ks, should be taken as:

5.34 + 2.84
vbtjo

K.s

(n + 1)

( 2 .21 )

Where. n is the number o f equally spaced longitudinal flange stiffeners, Is is the actual moment o f inertia o f one longitudinal flange stiffener about an axis parallel to the

50

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
flange at base o f stiffener (in4), bs is th e distance between longitudinal flange stiffeners (in) and tf is the thickness o f flange plate (in).

Litrature Review

Lee, Yoo and Lee conducted an elastic linear buckling analysis for several flange panels with one and two longitudinal stiffeners, having varying aspect ratios and stiffener rigidities using the finite element method. The authors, based

on the finite element analyses results, concluded that Eq. (2.21) was valid for

the determination o f the shear buckling coefficient o f longitudinally stiffened flange and that it could safely be used for practical designs.

2.6 D esign A ids for Curved B ox Girder B ridges

With their large torsional stiffness, bridges o f concrete decks over steel boxes are more efficient and more economical than those o f composite steel-concrete I girders since they exhibit better transverse load distribution. Approximate procedures have been developed in the past three decades to help the designers in the analysis and design o f curved box girders based on extrapolating the design calculations for straight box girders, usually by 5%-20% for curved box design. The most popular approximate analysis method for curved box girder bridges is the M/R method. This method depends on basic statics to analyze the girder where the bending and torsion actions are uncoupled and solved independently. In 1970, Tung and Fountain [72] presented the M /R method for the torsional analysis o f single span and continuous curved box girders. They specified in their study the limits in which this m ethod would give results with sufficient accuracy for practical purposes. The basic assumptions considered in this method are: 1) the cross section is symmetrical with respect to the vertical axis, 2 ) the line o f bearings at each support is radial, 3) the thickness o f each plate element is small compared with its width and the width is small compared with the span length, 4) adequate diaphragms are provided to prevent distortion o f the cross section and 5) warping stresses are considered to be negligible. The bending and torsional analysis of a curved box girder m ay be uncoupled and investigated independently in this method. The bending

51

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

moment and vertical shear forces are determined by straightening the curved girder to its full-developed length. A conjugate beam procedure is used in the torsional analysis. In the torsional moment analysis, the straight conjugate beam is subjected to a distributed load equal to the bending moment to radius ratio (M/
R ) and the applied distributed torque (t) in the span direction. The shear diagram o f the conjugate beam

represents the torsional moment diagram o f the developed girder; and the vertical displacement o f the conjugate beam loaded with [M/R+(EI/GJ)(M/R-t)] represents the angles o f twist o f the developed girder. For a simple span, the accuracy of this method is dependent upon the central angle, the flexural-torsional stiffness ratio and end restraints. The method can also be used to analyze continuous curved girders. The accuracy o f curved continuous girder analysis depends upon the flexural-torsional stiffness ratio, the total central angle o f the entire girder, the central angle o f each individual span and the torsional restraint provided at the supports. The study recommended that the central angle o f each span should not exceed 30, the average flexural-torsional stiffness ratio should not exceed 2.5 and the central angle o f the entire girder should not exceed 90. In 1968, Wright, Abdel-Samad and Robinson [74] presented a simplified analytical procedure for the determination o f stresses caused by the deformation o f box girder cross sections. The differential equation that governs box girder distortions was developed by Dabrowski [28], Nakai and Yoo [61] using energy methods as:

d 4 E I d 2_ ^ + K d 0 = dx

fflr,

( 2 .2 2 ) z

Where, ID is the distortional warping constant o f the box girder, K d is the distortional stiffness o f the box girder cross section, 0 is the angular distortion of the box girders shown in Fig. 2.8 and

52

Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
mD is the distortional load on the box girder.

Litrature Review

The form o f the above equation is similar to the governing differential equation for the Beam on Elastic Foundation (BEF). Therefore, the distortion of a box girder can be described by BEF analogy. The distortional stiffness o f the box cross section is analogous to the effect o f the foundation modulus o f the BEF, while the stiffness o f diaphragms or cross frames o f the box girder corresponds to intermediate supports for the BEF. The analogy between the box girder subjected to distortional loading and the BEF is demonstrated in Fig. 2.9. In 1975, Oleinik and Fleins [63] developed a computer program that solves both the differential equation for horizontally curved girders developed by Vlasov and the differential equation for crosssectional deformation by Dabrowski using the finite difference technique. A parametric study was then conducted that utilized the computer program to evaluate the distortional response o f series o f curved single cell box sections when subjected to dead and live loads. Empirical design equations were developed based on this parametric study for use in evaluating the distortional effects in curved box beams. These equations estimated the ratio o f the maximum distortional stress to the corresponding bending stress for dead and live loads as follows:

Wi

r~ W i I

i.

' Hi

: (V2-V i)/h+(W2 -Wi)/b

Fig. 2.8. Angular Deformation o f a Box Girder Under D istortion.

53

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

Q
---- ..- | *~ i EIbef~ Kf ' < ^
Ks

T
t
//

^j
Ks

i
Ks

Kf"

E lastic 7 F oundation

Beam-on-Elastic-Foundation
B ending Stiffness (E I b e f ) Stiffness o f the E lastic F ou n d atio n (K F). Stiffness o f th e D iscrete S upports (K s).

Box Girder
W arping S tiffness o f B o x G irder (E ID). D istortional S tiffness o f C ro ss-S ectio n (K D). Stiffness o f the Internal D iaphragm s and C ross-F ram es. D isto rtio n a l L o ad s on B ox G irder. => D isto rtio n al S tresses (O ut-of-P lane B e n d ing S tresses) F orces in th e In tern al D iaphragm s and C ross-F ram es. W arping S tresses in B o x G irder. A n g u lar D isto rtio n o f B o x G irder ( 0 )

T ransverse L o ad s

(Q, q).

R eactions on the E lastic Foundation.

R eactions at the D iscrete Supports.

B ending Stresses in the B eam . D eflection o f the B eam .

Fig. 2.9. B eam -on-Elastic-Foundation A nalogy o f Box G ird er D istortion [50].

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
For Dead -Load

Litrature Review

af _
d

(A ) ( s ) ' (W A )
"1/4

ub

.(2.23)

(R )

For Live-Load:

o_f = (28-1)(L)3( s)2 , (105)(L )1/2(S)


r j,

(W A )

" 1/4

(R )

(W A )

" 1/4

Where, fff is the maximum distortional stress at the bottom flange o f the m em ber (Kips/in ),

crb is the bending stress corresponding to the distortional stress defined above (Kips/in ),
A" is the functional value o f the span length = 0.22 L 3+30.8 L2-697.0L, S is the diaphragm spacing, expressed as decimal fraction o f span length, L is the span length (ft.), R is the radius o f curvature (ft.), WA is the distortional stiffness (Kip-in4). Where

'y

E a V i r A v( 2 p - l ) + A u(31
WA =

48 JL

.(2.25)

1+P

55

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

{A o [l + ( 2 e / a ) ] + 3 A V}
P =

(A + 3 A v)

'

.(2 .2 6 )

A 0 = (a + 2 e ) 5 0 A v = b5v A = a .(2 .2 7 )

Where variables are as shown in Fig. 2.10.

C.L.

Fig. 2.10. Box Girder Cross Section Dimensional Parameters.

Another set o f design equations that relate the bending m om ent in a straight girder to the corresponding bending moment in a curved girder through modification factors for both dead and live loadings were developed Oleinik and Heins as follows:

56

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

-curved

ve da(w y) _ W v c xu 'crur ve ^
t
J

/o oo\

(2 .2 8 )

For Dead Load:

c . rved = <K D ) ( i r ) .............................................................................. ( 2 -2 9 >

Where,

K D = 1 + ........ O0 ) 2

(2 .3 0 )

For Live Load:

,d

= ( K L ) ( M x)stra|ght..................................................................(2 .3 1 )

.Where,

K L = 1 + .......

(2 .3 2 )

0 )
Tables were developed also that give the required num ber o f diaphragms for various radii o f curvature and bending stresses to keep the distortional stresses within certain limits. The use of the above-mentioned equations was restricted to girders with an R/L ratio not greater than or equal to 1.0 where R is the radius o f curvature and L is the span length.

57

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

L itratu re Review

In 1977, N akai and Heins [60] presented design criteria that help the designers in establishing the need for designing the structure as a complex curved girder system (both pure torsion and warping torsion are considered), a m odified curved system (only pure torsion is considered) or a straight girder. The authors presented stiffness parameter equations and limiting central angle equations for various girder types, which provide information to the designer in determining the need for a curved girder analysis. The nondimensionalized stiffness parameter was presented as:

(2 .3 3 )

Where, L is the span length, G is the shear modulus, K t is the pure torsional constant, E is the modulus o f elasticity and Iw is the warping constant. A parametric study o f actual curved bridges with different cross section configurations was undertaken by Nakai and Heins and shown in Fig. 2.11. The critical stiffness parameter values can be approximately given by the following formulae:

K,cr
k, cr

10 + 4 0 30

0 < < 0.5 > 0.5 (2 .3 4 )

Where, is the central angle.

58

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

200

100

Range where warping stress can be ignored

rrrr
3.0

Twin bon

Multiple I .

0.50

. 1 J ____________ ____ 1 __ 1 .
0
0.1 0.2 0.3 0.4 0.5 0.6 Central nrtgie, 9 0.7 0.8 .0.9

Fig. 2.11. Torsional Parameter k Versus Central Angle

for Actual Curved Bridges [60].

The evaluation o f

tc follows the following criteria: when tc is less than 0.4, stresses from pure torsion tc is greater than or equal to /ccr the stresses from warping may be omitted. An

m ay be omitted. When

implification factor for deflections was computed for a curved girder bridge when compared with a straight girder bridge having identical span length and cross section shape. This factor was also related to the flexural and torsional rigidities o f the girder cross section in addition to the girder radius o f curvature and central angle.

59

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

In the analysis o f composite box girder bridges under live load, the concept o f using a distribution factor was adopted. The distribution factor represents the lateral distribution o f live load between girders in a multi-girder bridge system. The type o f the structure, the number of girders and the spacing of the girders are considered as the factors that the distribution factor depends upon. The distribution factor for the straight composite box girder specified in AASHTO 1996 [2] is based upon work conducted by Fountain and Mattock [35] on simply supported straight bridges. Tests were conducted on a one-quarter scale model to simulate a tw o lane, three-girder prototype bridge with an 80ft span and HS20 truck loading. The tests were also simulated numerically using a computer program based on folded plate theory. The behavior o f a series o f thirty one composite box girder bridges with a wide range o f spans was studied using this program. The number o f traffic lanes and the number o f girders were also considered. The results were used to develop an equation for the live load bending moment distribution factor for each box girder as a function o f the roadway width and number o f boxes as follows: N, o 85
nl

D F = 0.1 +

l J - + p ....................................................................... (2 .3 5 )
ng

Where, DF is the wheel load distribution factor, N l = Wc/12, reduced to the nearest whole number, Wc is the roadway width between curbs (ft.) and N q is the number of box girders, 0.5 <N l/N q <1 .5. In 1978 and 1983, Heins [47, 48] proposed an adjustment to the straight multiple box girder moment distribution equation provided by Fountain and M attock to make it applicable to horizontally curved composite multiple box girder bridges. A modification factor was proposed to the straight girder moment distribution. This factor is a function o f the radius o f curvature provided that cross bracing is used inside the boxes:

60

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

D F C = (1 4 4 0 X 2 + 4 .8 X + 1 ) D F S

(2 .3 6 )

Where, DFCis the curved girder load distribution, DF is the load distribution factor for straight girders and X = l/R , where, R is the radius o f the center line o f the bridge (ft.). Empirical relationships were also developed to select the total area o f the top and bottom flanges for preliminary design o f single, two and three spans bridges. Heins also introduced the required cross frame spacing and the area o f one cross frame diagonal for the curved box girder that perm it a maximum distortional-bending stress ratio o f 10%:

(2 .3 7 )

A b = 75

d 2 (d + b)

(2 .3 8 )

Where, S is the cross frame spacing (in.), Ab is the area o f one diagonal brace (in.), R is the radius o f curvature (ft.), L is the span length (ft.),

61

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
d is the depth o f the box (in.), b is the width o f the box (in.) and is the web thickness (in.).

Litrature Review

In 1993 Cheung and Foo [21], using the finite strip method, proposed expressions that relate the behavior o f curved and straight multiple box girder bridges, these expressions are functions o f the span length, half w idth o f the bridge deck, box spacing and the radius o f curvature. For centre girder or whole section:

F , = F o ( f 1......................................................................................(2 -3 9 ) If(L > 50) T hen^ = 1 Ju

For interior girder:

F2 = l + 0 . 7 ( D ( g X f ; ) * F , .................................................. (2.40)
I f ( L < 50 ) T hen^ = 1
J /

For exterior girder:

r , = Fo - O . 3 ( | ) ( f ) ( 0 < F , .............................................................. (2.41)


If( L < 50 ) T hen^ = 1

Where,

F 0 = , + 0 -0 8

2-

.(2 .4 2 )

62

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
and,

L itratu re Review

F l5 F 2 and F 3 are the moment ratios between simply supported curved and straight box girder bridges, L is the span length, R is the radius o f curvature and B is the half width o f the bridge deck, S is the distance between the centerline o f the deck and that o f the girder under consideration. The drawback o f this study is that it did not consider the beneficial effect o f interior diaphragms and the cross bracing between boxes.

2.7 Design codes and specifications

2.7.1 A ASH TO , G uide Specification for H orizontally C urved Bridges (2003) [40].

2.7.1.1. Strength of Curved Top Flange Plates


In 1971, M cManus and Culver [27] developed a mathematical model for determining the lateral buckling behavior o f horizontally curved girders o f I-shaped cross section loaded normal to the plane of curvature. Strength equations resulting from their study are being used as the flexural strength design criteria in the current version o f the AASHTO Guide Specifiactions. They used the second-order departure equations in their derivation; they also considered in their derivation the rapid build-up o f deflections and stresses upon loading in the vicinity o f the critical buckling load. Strength equations that were developed in this study were used as the flexural strength design criterion in the model that was chosen to be a single span, simply supported curved girder with equal applied end moments and bimoments. The equal forces on each end gave conservative results and were considered as a worst possible case o f load application. For the elastic range, the lim it state or failure for a non-compact flange was considered to occur when the yield stress was reached at the flange tip while for compact sections failure was considered to occur by full

63

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

plastification o f the critical cross section. McManus and Culver introduced the flange stress o f the curved girder as a reduction in the stress o f an equivalent straight flange using the following design formula:

^bc

^bsPbPw

Fbs = F y( l - 3 p

(2 .4 3 )

Where, Fbc is the maximum average stress in the curved flange, Fbs is the equivalent straight girder stress, p b and p w are reduction factors and A is the flange slenderness. When the warping stress is equal to zero, p w=1.0 and when the warping stress exists, the additional parameter p w is included. Both factors are functions o f (A ^/A f, d/tf, L/b, L/R and crw/(7b). McManus and Culver [27] reported that the compression flange plates width to thickness ratios should not exceed the value determined by the formula: Compact section b ^ 3200

(2 .4 4 )

Where, Fy is the yield stress (psi) This ratio has been liberalized to 18 for compact flanges with Fy= 50 Ksi in the current Guide Specifications [40],

64

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2
The values o f p b and p w are:

Litrature Review

p b = --------------------^ ..........................

(2 .4 5 )

Pw = 0.95 + 18(0.1 - g j

L V , fb ^

mf 0 .3 -0 .1 ' Rb. ------ .........


7bs
Pbv p y

(2.46)

Where,

*i = fw + fn.......................................................................................................(2.47)

fm is the factored lateral flange bending stress at critical brace point due to the effects other than curvature, fw is the factored warping stress due to curvature at critical brace point, fb is the factored average axial flange stress at either brace poit L is the length in inches o f the unsupported compression flange between cross frames, R is the radius o f curvature in inches to the web o f the girder and b is the total flange width in inches. p b p w must not exceed 1.0 and fw/fb is the ratio o f warping to bending stress at cross frame locations. This ratio should be taken as negative if fw is compression on the flange tip further away from the center of curvature and positive if fw is compression on the flange tip closer to the center o f curvature.

65

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

For a non-compact flange, McManus and Culver [27] indicated that the flange slenderness should satisfy the following formula:

3 2 0 0 ^ 1 , ^ 4 4 0 0 ......................................................................

f t

Jfy

In the current Guide specifiactions (2003)[40], the flange slenderness should satisfy the following:

rf - 1 0 2 V It t+ t ft i t (fb w)

- 23 .............................................................( 2 -4 9 )

Values o f Pb and p w are:

Pb = T T ................................................................................ (2-50) Rb

Pw,

f
1- ^ i l

' 75

.............................................................. (2-51)

fb v

0.95 + --------------

--------------

Pw2

30 + 8000(0.1 - L / R ) 1 + 0.6(fm/ f b)

If fw/fb is negative then p w = p wl and if it is positive then p w is the smaller o f p wl or p w2.

66

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2 2.7.1.2. Strength of Curved Bottom Flange Plates

Litrature Review

The strength of curved bottom flange plates is calculated as for straight plates except that the shear stress that results from torsion is considered through the following interaction equation:

fi b

v .(F v ) cr.

= 1 .0 ............................................................................(2 .5 3 )

( F b) cr

The elastic critical stress under uniform compressive stress on a plate is:

( F b ) ^ life
Where,

2 ^ 2 6 ' 21

1o6k

2 ............................................... ( 2

'5 4 )

E is the elastic modulus=29000 000 psi,

v is the Poissons ratio=0.3,


t is the plate thickness, w is the plate width or width between stiffeners, K is the plate buckling coefficient= 4 (no stiffeners); 2 < K < 4 when stiffeners present. The elastic critical shear stress is;

* EKS m 2 . 6 n v = 7 ^ 7 i d w J = 2 6 '21 x 10 H w J 12(1 - v * )

........................................ ( 2 -55)

Where, K s is the shear buckling coefficient =5.34 (no stiffeners); W hen the stiffeners are present:

67

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

5.34 + 2.84
Vbt37
=

.(2.56)

(n+ iy

Where, n is the number o f stiffeners, Is is the moment o f inertia o f a stiffener about an axis parallel to the flange and at the base of the stiffener. Substituting Eqs. (2.54), (2.55) into Eq. (2.53) gives the elastic critical stress for the curved flange plate:

(f,)2K
F =r 26.21 x 1 0 6K ( i ) -

.(2.57)

26.21

t 10 (K ) I w-

The Von Mises yield criterion is used to determine the normal stress at full yielding o f the flange plate with the existence o f shear stress:

(fb)z + 3(fv)z = ( F V

f> = %

e ) 2 = FyA

.(2.58)

A=

1 -3

f ,v 2

The maximum normal stress is again calculated using the Von Mises yield criterion and considering the assumed residual stress=0.4 Fv

68

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

( f b + 0 .4 F y) 2 + 3 ( f v) 2 = ( F y) 2 .(2 .5 9 )

fb = Fv(A-0.4)

To determine the limiting slenderness ratio of the flange plate, the ratio (w/t) shown in Fig. 2.12 is defined as 0.6 tim es (w/t)3, where (w/t )3 is the plate slenderness when Fcr=AFy is substituted in the elastic formula that gives:

3070(7k)/(T f;)

= j^l_
Fy

.(2 .6 0 )

The ratio (w/t )2 (Fig. 2.12) is defined as the plate slenderness where the elastic buckling stress is equal to Fy(A-0.4)

W |

6 6 5 0 ( j K y ( J f y)
.(2 .6 1 ) _1_
'

t h

1.2

A -0.4 + J(A-0.4)2 + 4 ( F ) 2( ^ 2

J*y

As stated in the load factor design criteria and based on the above derivation, the equations for the maximum capacity o f flange plates are as fellows:

For

-^-2 < R j: (full yield o f the flange plate)

fb < F yA.

.(2 .6 2 )

69

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

For ^

> R2 : (elastic buckling o f the flange plate)

fh = 26.21 x 10 K

f t V

( f v) 2K ------------- ..................... 26.21 x 10 ( K .) '


a

(2 .6 3 )

if O 2

For R, < y < R , : (sinusoidal transition range is assumed)

'
u* V I 75

f
R2 -

fc.(2 .6 4 )

T C A - 0 . 4 1 - sin 2
k k

R2-R1

The above criterion is limited to a torsional shear stress (fv) less than 0.75(F / 73), if the torsional shear stress becomes dominant, conservative rules apply as follows:

- S< f v < S
ft, S M
and7 V ^ < R i

.(2 .6 5 )

70

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

* . T

Eq. 2.55

v/t

Fig. 2.12. Maximum Strength of Curved Bottom Flange Plate in Compression. (AASHTO 1993) [39]

2.7.2 Canadian H ighw ay B ridge D esign C ode (C H BD C ) (2000) [18]


The Canadian Highway Bridge Design Code (CHBDC) adopts AASHTO current strength equations that are based primarily on MacManus and Culver work [27] and limits the flange slenderness to class 3 or better.

2.7.3 Hanshin Guidelines


In addition to the AASHTO Guide Specification, there is another design specification for horizontally curved steel bridges. This specification was developed by the Hanshin expressway public corporation in Japan and requires that the stresses in the top flange o f a curved girder conform to the following interaction equation as discussed by Nakai and Yoo (1988) [61]:

71

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

(2 .66)
( CTba)c bao

Where,
crbc is the bending compressive stress,

a wc is the additional warping compressive stress, CTbao is upper limit o f the allowable compressive stress,

( a ba)c is the allowable compressive stress for lateral buckling.

Similar to the AASHTO specification, the Hanshin Guide relates the allowable lateral torsional buckling stress o f a curved girder to that o f corresponding straight girder using the following formula:

( b a ) c = V O ba) s

(2.67)

Where, T is a reduction factor; this reduction factor is defined as a function o f both the top flange slenderness ( a ) and the enclosed angle ( $ ) o f the unsupported portion o f the girder:

x p r = 1 -1.05V a(<D + 4.52(D2)

(2 .68)

Where

a and $ are subjected to the following constraints:

72

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

0.1 < a < 1.414


< 0.2

(2.69)

Where,

a is the top flange slenderness and it is defines as:

2k W p

(2.70)

K is a param eter that accounts for the ratio o f the web to flange area and is defined as:

(2.71)

Based on parametric studies of elastic buckling analyses, a girder will behave as curved if it satisfies the following range:

0.02a < < < 0.2rad

A curved girder has an effective length that is less than the unbraced length and is defined as:

leff= A L
Where, L= R 4 and

73

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

X = l - 1 . 9 7 0 1/3 + 4 .2 5 0 - 2 6 .3 0 3

(2.72)

I f $ < 0 .0 2

a , the flange behaves as straight one and A=1.0. If $ > 0 .2 , then the curvature is too large

and this procedure can not be used to calculate the strength. (dbaA that appears in Eq. 2.67 is the allowable compressive stress o f laterally unsupported straight girder. This allowable stress is a function o f the nominal yield stress and is determined by the following criteria: If L

/ b < c , then

or ifc < L / b < Co, then

Where,

v is a factor o f safety and is equal to 1.7. The parameters a, c and c 0 are functions o f the steel grade
as defined in Table 5.1 in reference [61] by Nakai and Yoo.

a u is the critical lateral torsional stress in the inelastic region as approximated by the Japanese
Specification for Highway Bridges and is defined as:

74

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 2

Litrature Review

gu

= cry [ 1 - 0 .4 1 2 ( a - 0 .2 ) ] ...................................................................... (2 .7 3 )

The above formula is valid where 0.2 < a < 1.414, if a is less than 0.2 then greater than 1.414, the strength is governed by elastic stability.

ctu

= o\ and if a is

75

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3 Curved Box Girder Bridge Field Test

3.1 Introduction

The design process o f curved composite bridges involves tracking the stresses in the girders during erection, construction and service loading stages. Generally, the construction o f curved steel bridges is more complex than the construction o f straight steel bridges, because curved bridges are typically less sta ble during erection than tangent bridges. Therefore, the designer should evaluate the bridge behavior dur ing the erection and provide a construction plan that includes the sequence o f girder erection and concrete deck placement and the locations of any temporary supports. During bridge erection and before the con crete deck hardens, the steel girders must carry all o f the dead and construction loads. The most significant o f which are those associated with concrete placement. Once the deck has hardened, the composite steelconcrete section carries additional loads, such as the wearing surface, sidewalks, barrier walls and traffic. For structural safety and serviceability, the designer estimates the stresses induced within a curved bridge girder as accurately as possible and assures that these stresses do not exceed the specified limit states. However, the designer might be skeptical about the level o f approximation that is used in his esti-

76

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

mate and m ay question how well these stresses compare with the stresses inferred by measured strains in real structure. The field testing o f a curved bridge during construction and service loading is examined here to answer this question and to provide the designer with more insight into the behavior o f such structures. In 1973 the 1-695 and 1-83 interchange, Ramp C viaduct located near Baltimore, Maryland was stud ied and analyzed both during construction and after completion o f the bridge (Buchanan, Yoo and Heins) [14]. There w ere two experimental testing programs performed on the bridge. The first one measured the response o f the bare steel girders to the placement o f the wet concrete and the second program measured the composite bridge section response to the service loads. The bridge was analyzed using finite difference programs, the analytical results were then compared with measured results and also compared with the val ues expected from the bridge design evaluation. According to this comparison, the study reported that the design was generally conservative. This chapter presents a study to re-examine the curved bridge behavior with a three-dimensional shell element model. Comparisons will be made between the finite element results with the measured results and the finite difference analytical results.

3.2 D escription o f the Field Test

3.2.1 Instrum entation


The superstructure o f Baltimore bridge (Buchanan, Yoo and Heins) [14] is composed of two large steel box girders with a composite concrete deck ten-inch thick. The unit bridge under study is continuous over three spans. The bridge consists o f five separate sections o f steel, spliced at four locations in the longitudi nal direction as shown in Fig. 3.1 and Fig. 3.2. Each section has the same cross sectional dimensions for both cells. The dimensions o f each o f the five different cross sections are given in Fig. 3.3. Corrugated sheets o f cold formed steel were used as a permanent form to support the concrete. Fig. 3.4 shows the cross section with reinforcing bars details. The concrete deck was placed with five separate pours as shown in Fig. 3.5. According to the report, the final three placements were made consecutively (i.e. considered as one pour)

77

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Fieis Test

4 BEARING, ABUTMENT

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

author. (Prof.Yoo, C.H.).

Reproduced with permission of the copyright owner. Further reproduction prohibited without perm ission.

Chapter 3____________________________________________C urved Box Girder Bridge Fiels Test

<t PtEH NO. I 4 FIELD SPLICE Bl" C /C MAX. MAX. it 30 XI is* FIELD SPLICES 17* c / a MAX.

n MER NO.2 ANO % BRS. - % PIELO SPLICES 24* C/C MAX. -----3 / 8 " WEB 8

w c /c

H.24-XI

W EB

E .2-W CfW
10 CEO

W EB 8

4* 3/a' S T S W F 20 U 106 l / t " X l / e A ft *06 V fc 33*- O 8.106 X 3 /4 ST I0WB27.5 f t 10 6 1 M # 3 /te 4 4 -0 . 1 0 6 *

table

o f d im e n s io n s

> * 0

V O

GIRDER
<L LEFT TOP FLANGE

A 70-63/4" 71-ltf
71-93/g

8
2 9 '- 2 'i5 /g

Li

L2

F
3*7%

G
8 6 -!%

L3
120-9%

R
1300.961

/e 99-9Hs 28-05> te 75'-23 75-9

28-OS/w 131'-3" 2 8 -3 V

IN SID E

<t RIGHT TOP FLANGE

29-5 ' % iOG-7*% 2 * 3 %


29-91/,e

1 3 2 -4 % 34-10'% 86*40% 121-9% ' 133-634 35-2% 134-7% 3 5 - 6 8 7 -7 %


!2 2 -I C f %

1311.981

< LLEFT TOP


FLANGE

101-67% 28-63/s 76^5% 2 8 -6 3 4 102-4% 28-9" 77 -l*4"

1323.481

OUTSIDE

t.

RIGHT TOP FLANGE

72-4 /,3 3 0 -0 "

2B -9"

8 8 '-4 '% 12340*%

1334.481

Fig. 3.2. Longitudinal View of the Baltimore Bridge.

Chapter 3

Curved Box Girder Bridge Field Test

_JL_

7.9'

28.5'

_75.6'_

28.1

86.6

100.H

Span A Wf

JL31.8l_ Span B

121.3

Span C

T]

4
tbf

Wbf

Section
1 2 3 4 5

tf (in.) 1.0 1.75 1.25 2.50 1.25

tw (in.) 0.375 0.375 0.375 0.500 0.375

tbf (in.) 0.500 1.000 0.750 1.375 1.000

Wf (in.) 24.0 30.0 24.0 30.0 24.0

w b f(in.) 106.5 106.5 106.5 106.5 106.5

(in -> 54.0 54.0 54.0 54.0 54.0

Fig. 3.3. Cross Sectional Dimension at Splice Locations (Baltimore Bridge).

80

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box G irder Bridge Fiels Test

a * **

<n

ac

>

* 3

It

3 I* 3
3 *

C M-

O Ui i5

**
3
S*

* ft 9*

81

Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Fig. 3.4. Typical Cross Section o f the Baltimore Bridge.

3 *

3 *

to

Chapter 3

Curved Box Girder Bridge Fiels Test

ka

J U

\ %
y

s A zn
2 2 u

r
\

Au \n
i \ L "1 o/
\

k.

v-

2 2

>-k 0

S S o z

J U

V.
a " S

4
s ^ < 5 ,0 v
-f
'

/
sr u

82

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Fig. 3.5. Concrete Placement Sequence.

Chapter 3

Curved Box Girder Bridge Field Test

In both th e dead and live load testing programs, the induced strains, deflections and rotations were measured. The m ajor portion of the test instrumentation was strain gages that were mounted on the inner surfaces o f the girder cells. The strain gages were SR-4 paperback, type A-3. The total time required to complete the concrete placement was four weeks. Therefore, it was impossible to have the strain reading equipment run continuously during this long period. Thus, an alternative was suggested to overcome this problem. All gages were zeroed prior to each pour and the equipment was turned off. The equipment was turned on after the completion of the pour and the strain data was recorded. The gages were placed at four specified locations in the longitudinal direction as shown in Fig. 3.6. Fewer gages were used at the section designated as DD-D because section DD-D would likely have a moment similar to that at section AA-A. The four locations for the strain gages in the longitudinal direction were selected as the places where the maximum positive and negative bending moments were expected to occur during the w et concrete placement. The locations AA-A, CC-C and DD-D were chosen to measure the strains resulting from large positive moments; section BB-B was located adjacent to an inte rior support to examine die strain due to the maximum negative moment. The strains were measured dur ing the final two concrete pours only. During construction, a number o f gage installations became defective and none o f the installed rosette gages worked properly. This prevented the estimation o f the shearing stresses. The deflections were mea sured during the construction phase o f testing at various points on the outside cell in the longitudinal direc tion using surveying equipment. Deflection measurements for the dead load test could not be m ade on the interior sides of each cell and on the exterior side o f the inside cell since there w as no scaffolding for what was considered to be the only access to the deflection locations along the girder. Therefore, the measure m ent of rotations was not possible. In the live load test, the top flange gages were placed on the underside o f the flange since the concrete deck covered the top side o f the flange. Federal Highway Administration (FHWA) deflectometers were used to measure the deflection in the live load test. A deflectometer is essen tially a short cantilever beam (12 inches long) that is attached to the bottom o f the girder.

83

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

AA-A

B B -B

CC-C

00-0

jE
100 .2

21
4j 6 ** H

"

"zr
63.4 4 3 1 .8 * 1213'

453'

;A

SH OTTT

Strain Gage Longitudenal Locations

Q F3

0F4

0 L1
EXTERIORSIDE

AA

OR I

84 "

2?

INSIDE

OUTSIDE

OFI

0F2

OPS

0F4

DD
3 _________________4

INSIDE

OUTSIDE

Uniaxial Gage x Rosette Gage * Same Gage Location for Section BB-B and CC-C, Except Gages OL1 and OR1 Which are Present Only at AA-A.

Fig. 3.6. Longitudenal and Transverse strain Gage Locations.

84

Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Near the fixed end o f the cantilever, two strain gages were mounted on the top and bottom surfaces of the cantilever beam. As the girder deflects, the resulting change in the strain is used to determine the actual deflection o f the girder based on predetermined calibration curves for the deflectometers. A deflectometer was placed near four-tenths of the span length from the end abutment in span A and another deflectometer was placed near mid span of span B. The Federal Highway Administration test vehicle, simulating an HS20-44 design loading was used to test the bridge under live load conditions. The actual load distribution for the vehicle was 10.250 kips, on the front axle, 32.500 kips on the drive axle and 31.750 kips on the trailer axle. The gross weight o f the vehicle was 74.500 kips. The vehicle traversed the bridge in four spec ified lanes as shown in Fig. 3.7.

5* -6

* *

L A M i2 f L A N K3

1C -0

----------- IDi.ALJ.2iD LA N E POSITION ----------A C TU A L LA K E POSITION

Fig. 3.7. Lane Locations for Live Load Test.

85

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3
3.2.2 A nalytical Procedure

Curved Box Girder Bridge Field Test

Two finite difference programs were used in the analysis, (CURSGL) can analyze a single curved girder while (CURSYS) can analyze a system of curved girders with a possible diaphragms interaction between each girder. CURSGL was used for the dead load test since the only interaction between the two cells was the steel corrugated sheeting that was assumed to have negligible effects. In these two programs, it was necessary to idealize the box girders as curved elements with section properties similar to those of the actual cells. The section properties are E, G, Ix, Iw and Kt. Where, E is modulus o f elasticity of steel (29500 ksi.), G is shear modulus o f elasticity o f steel (11200 ksi.), Ix is the moment o f inertia about the horizontal axis; in.4, Iw is the warping constant o f the cross section; in.6, Kt is the torsional constant o f the cross section; in.4. The spacing between the nodal points had to remain constant; this made the exact positioning o f the points o f field splices difficult. The total number of nodal points was selected in a way that the evenly spaced nodal points were positioned as close as possible to the actual location o f each o f the interior sup ports. The variations in the section properties were spaced linearly over ten nodal points for each splice as shown in Fig. 3.8. The designer added top lateral bracing to prevent excessive twist of the cross section during the casting of the concrete. The top lateral bracing effectively changes the box girder from an open to quasi-closed cell. In the dead load test, the cross section was m odified to account for the top lateral brac ing. This bracing was replaced with an equivalent plate o f constant thickness, teq; this plate would approx imate the stiffness o f the top lateral bracing. The value of teq was determined from an by Dabrowski [28] as follows: equation developed

86

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

eq

2A.E,sind>(cos<t>) ----1

Qj

(3 l )
'

The K-bracing and the longitudinal stiffeners were not considered in the finite difference analysis. To determine norm al stresses, and since actual gage locations did not coincide exactly with nodal points, the stresses were interpolated linearly between the nodal points at each side o f the actual gage location. In the live load test, the (CURSYS) program was used for the analysis. The composite section o f the bridge was idealized by two curved elements for the two cells connected by diaphragms that approximated the transverse stiffness o f the concrete deck. The longitudinal properties o f the bridge cross section were determined as follows: 1. 2. 3. The concrete deck was transformed into an equivalent area of steel. The open cell shape o f the box girder was modified to include this transformed area o f concrete. The height o f the web was increased to mid height o f the original ten-inch thick concrete deck to include the concrete additional area. 4. 5. The steel area of the flanges was then added to the transformed area o f the concrete. The total steel area was then distributed over the length between the top o f the webs at the new height.
6 . The combined area o f the steel was divided by this length to determine a new uniform thickness (ts).

Since the location of the centroidal axis o f the cross section was shifted up, the longitudinal stiffeners were considered in the live load analysis. The transverse properties o f the concrete deck were idealized by specifying diaphragms between the two girders in the (CURSYS) program.

87

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Value of Specified Parameter at Each Nodal Point

m* ** 4

i 4 i i l H

F 1

'

Actual Value of Specified Parameter to the Left Side of the Field Splice

Actual Point of Variation (Field Splice)

Actual Value of Specified Parameter to the Right Side of the Field Splice

Fig. 3.8. Typical Nodal Point Variation for Field Splice.

During the live load test, there were four distinct longitudinal positions o f the test vehicle that caused a maximum stress condition in each o f the four gage locations. The vehicle wheel loadings were specified as concentrated loads in (CURSYS). The concentrated loads could only be specified on or between the two girders, therefore, lanes one and four had to have a modified loading scheme as shown in Fig. 3.9. The loadings in these two lanes were taken as axle loads rather than wheel loads and were placed on the girder closest to the actual position o f the test vehicle, because for lanes one and four at least one o f the wheels of each axle was outside the area included and bounded by the two girders. There were sixteen loading points in total that produced maximum stresses for each o f the four gage locations in each o f the four transverse lanes. Load points for lanes 2 and 3 had six concentrated loads, while load points for lanes 1 and 4 had three concentrated loads representing each axle load. The positions

88

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

o f these load points were determined from the readings taken from the field test. The locations o f these load points are shown in Fig. 3.9.
s 10

U l

-M

a js 0) +> *- 4> S o 0>

_ _t

o W -P*

c C S

5 8

+ c HT3 C 0> S! m o <u C G

03 a> x : *w c +> o <d c U o <u

>

O J 4J TJ WT3 G * *0 ( S ' h O cd O Q*H


C O 0) H *H x : X -P *7-4 < S -rt O f-i r- <0 0> A J

os , o d) =5 A m

T3 c* o s>
fc

hJ 9 -

T C tft

5
40

c o <J u -S

sr>

bJ

to C

e o -S
03

o o

O fc
T3

J "3

03

i
89

.3> to

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3
3.3 Finite E lem en t analysis

Curved Box Girder Bridge Field Test

3.3.1 G eneral
As m entioned earlier in chapter two, there are many methods available for analyzing curved bridges, however, the finite element method is considered as the most powerful, versatile and flexible method. The finite element method is capable o f modeling curved bridges in a more realistic manner and it has been able to show the structural behavior within the elastic and plastic stages o f loading. In the current research, the various structural elements described previously are modeled using finite element discretization. A general description o f the finite element approach is presented next followed by background information pertaining to the finite element program ANSYS that was utilized throughout this study for the structural modeling and analysis of curved box bridges.

3.3.2 Finite E lem ent M ethod


The finite element method is a numerical procedure that allows the discretization o f the entire structure into a finite number o f elements. These elements are interconnected with each other by means o f certain points called nodes. In the case o f small displacements and linear material response, the stiffness matrix of each element is derived and the global stiffness matrix o f the entire structure can be formulated by assem bling the stiffness matrices o f all elements. The global stiffness matrix relates the nodal forces and dis placements; the relationship can be expressed by the following equilibrium equation in matrix form:

[K]{u} = { F } ............................................................................ (3 .2 )

Where, [K] is the global stiffness matrix assembled from the element stiffness matrices, {u} is the nodal displacement vector and {F} is the nodal load vector.

90

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

The above equation can be derived from the following basic relationships:

{u( x , y ) } = [Af]{u(e)} ................................................................ (3.3)

Where, {u} is the internal displacement vector of the element, {i/e)} is the node displacement vector and [N] is the interpolation function or shape function.

{e(x ,^ )} = | /> | ; ' : ..................................................................... (3.4)

Where, {} is the strain vector and [B] is the strain- displacement matrix.

{<r( x , y ) } = [ D] { z ( x , y ) } = [ B ] [ 5 ] { W } ................................(3.5)

Where, [D] is the stress- strain constitutive matrix. To obtain the finite element stiffness equations, the principle o f minimizing the total potential energy is used

91

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

s n = 5U - 8 W

0 ................................................................... (3.6)

Where, II is the total potential energy, U is the internal strain energy and W is the external energy.

U = i J { z } T{<3}dvol
vol =

\ \ [ B ? { u {e)} T[D}[B]{u{e)}dvol
vol

.(3.7)

{u{e)}T{ F } ................................................................................... (3.8)

l { u ie)} T[Kie)]{uie)} - {u(e)} { F } ...................................... (3.9)

From above, the element equilibrium equation and the element stiffness m atrix can be represented as

92

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

[K]{u} = { F } ................................................................................(3.10)

K = | [B]T[D][B]dvol ............................................................. (3.11)


vol

In a linear elastic analysis, loads are applied to the model and the response o f the structure is obtained directly, since the stiffness matrix in this case is independent of the displacements.

3.4 The Finite Elem ent Program : ANSY S

The finite element modeling and analysis performed in this study were done using a general purpose, multi-discipline finite element program, ANSYS [4], ANSYS is a commercial finite element program developed by Swanson Analysis Systems. The program is available for both PC and UNIX based systems and either could be used in the Civil Engineering Department at University o f Toronto. The analyses pre sented in this thesis were performed on a personal computer with 1.8 GHZ, Pentium 4 processor with 512 Mb o f RAM and 120 Gb hard drive and Windows 2000 operating system. Some o f the analyses were con ducted using ANSYS version 5.5 and other analyses were performed using the later version 6.1. ANSYS has an extensive library o f truss, beam, shell and solid elements. Shell elements were used to model the structural components o f the box girder bridges (webs, bottom flange, top flanges and the solid diaphragms), while truss and beam elements were used to model top bracing trusses and cross frames.

3.4.1 Shell Elements


ANSYS has a large library o f shell elements that have different capabilities. In this chapter, a compari son will be made between the analytical and the field measurements for incremental changes in the stresses

93

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

and the total accum ulated stresses in the girders. The structure will be assumed to behave elastically during each load increment. For linear elastic analysis, ANSYS has two shell elements to choose from; namely shell 63 (elastic shell) and shell 93. Below is a brief description o f each one: 1. Shell 63 (elastic shell): A four nodded element that has both bending and membrane capabilities.

The element has six degree o f freedom at each node, translations in the nodal X, Y, and Z directions and rotations about the nodal X, Y, and Z axes. Large deflection capabilities are included in the element. It is stated in ANSYS manual [4] that an assemblage o f this flat shell element can produce good results for a curved shell surface provided that each flat element does not extend over more than a 15 arc. Four integra tion points are used in the m id surface plane for the integration o f the mass and stiffness matrices. 2.

Shell 93: An eight nodded shell elements particularly suitable to model curved shells. The element

has six degrees o f freedom at each node, translations in the nodal X, Y and Z directions and rotations about the nodal X, Y and Z axes. The shape functions are quadratic in both in-plane directions. The element has plasticity, large deflection and large strain capabilities. Same as shell 63, four integration points are used in the plane o f the m id surface for the integration mass and stiffness matrices. For both elements and for lin ear materials, two integration points are used through the thickness direction. Cost and accuracy must be considered as primary aspects in the process o f choosing the suitable ele ment for the analysis. It is evident from the above description that shell 93 has a higher order shape func tion; this complexity in shape function transforms into an expected increase in the time needed to generate the global stiffness matrices. To help in choosing the appropriate element for the Baltimore bridge model ling presented in this chapter, both shell 63 and shell 93 will be used next in the modeling o f a simple elas tic experiment conducted by Fam and Turkatra [34],

3.5 M odel Study B y Fam and Turkstra [34]

Fam and Turkstra conducted a model study on a Plexiglas simply supported curved box girder. The model had a total span o f 80.11 inches and a radius o f curvature o f 51 inches to the centerline o f the girder

94

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

as shown in Fig. 3.11. This geometrical combination o f span length and radius o f curvature was considered as a relatively severe test to the analytical procedure. Two series o f tests were performed. The first series, called series A, had two solid end diaphragms and two pairs o f intermediate diaphragms. The second series, called series B, had only solid end diaphragms. The aim from these two configurations was to show the effect o f intermediate diaphragms on the overall stiffness o f the girder and on the distribution and magnitude o f the cross-sectional stresses. Many tests were perform ed using a variety of loading cases. The load case considered herein is a concentrated load over the exterior web at the girder mid span. For the current analysis, the transverse discretization of the cross section is shown in Fig. 3.10. Two elements were used to model the webs across their height, six ele ments were used for bottom flange and 10 for top flange. In the longitudinal direction, the discretization of the current m odel was determined by restricting the element aspect ratio to be less than two. Every element extended over a 2.5 arc, which is considerably less than the limit provided by the ANSYS manual [4] for shell 63. In the present study, two models were developed, one with shell 63 and the other with shell 93 for the same transverse and longitudinal discretization. Only h alf o f the girder was modeled to take advantage o f the boundary and loading symmetry (Fig. 3.10). The experimental and finite element deflections o f the current model for series A and B are compared in Figs. 3.12 and 3.13 respectively. The finite element results shown are for both shell 63 and shell 93. Generally, these finite element results show good agreement with the experimental deflection; also, it can be seen that the deflection produced by the shell 93 and the shell 63 models are almost identical. The influ ence of the intermediate diaphragms in series A is highly noticeable on the stiffness and the deflection val ues compared with series B that has no intermediate diaphragms. Figs. 3.14 and 3.15 show the experimental and analytical results o f the top flange outer surface tangential stress distribution at mid span for series A and B respectively. As expected, it can be noticed that the top flange element stresses in the region close to the concentrated load introduced some errors. This is due to the singularity presented by the application o f the concentrated load. It is also apparent that shell 93 is more sensitive to the local effect of

95

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

the concentrated load than shell 63. Figs. 3.16 and 3.17. show the tangential stresses o f the outer face of the bottom flange at mid span for both series. The results o f both FE models again are almost identical. The existence o f intermediate diaphragms in series A makes the stresses in the top and bottom flanges more uniformly distributed than in the case o f series B where no intermediate diaphragms exist. It can be concluded from the above results that using shell 93 has no advantage on accuracy as long as sufficient mesh refinement is used with shell 63. The following sections describe the finite element modeling o f the Baltimore field test using shell 63.

< * * * X,
X 'A =0
10 lbs

E=400000 psi v=0.36

Symmetrical Boundary at mid span

Fig. 3.10. The Current FE Model of Fam and Turkstra [34] Plexiglas Curved Box Girder.

96

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

R=51

All dimensions in inches

Fig. 3.11. Dimensions and Transverse Cross-Section Discretization of Fam and Turkstra [34] Model.

97

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

0.2 Exterior Web Deflection 0.16

c
S3

0.12

0.08 shell 63 0.04 Interior Web Deflection shell 93 experimental 50

>

10

20

30

40

Angle measured from support (Degrees)

Fig. 3.12. FE (Shell63, ShelI93) and Experimental Deflection at Mid-span for Fam and Turkstra [34] Series A Plexiglas Box Girder Bridge.
0.2
Exterior Web Deflection 0.16

c
0 0.12 =8

Q >

1 2 0.08 co o

shell 63 -a she II 93 Interior Web Deflection Experimental

> 0.04

10

20

30

40

50

Angle measured from support (degrees)

Fig. 3.13. FE (Shell63, Shell93) and Experimental Deflection at Mid-span for Fam and Turkstra [34] Series B Plexiglas Box Girder Bridge.

98

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

---------------- ,-------------0 -10 -5 _ 10 ) -20 -30 ? 5 10

1 8 s & t -10

a,

-40

top flange stress shell63 atop flange stress-shell93 -100 Lateral Distance along top flange (in.) experimental top flange stress

Fig. 3.14. FE (SheI193, Shell63) and Experimental Mid-Span Stresses in Top Flange of Fam and Turksrta [34] Series A Plexiglas Box Girder Bridge.

-10

10

shell63 top flange stress shell93 top flange stress

g
V)

in
w

Experimental top flange stress -io

-Z L +10

co

Lateral Distance along top flange (in.)

Fig. 3.15. FE (Shell93, Shell63) and Experimental Mid-Span Stresses in Top Flange ofFam and Turksrta [34] Series B Plexiglas Box Girder Bridge.

99

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

120

100

m a.
60 40 20
-|

bottom flange stress93 bottom flange stress63 experimental bottom flange stress

-10

-5

10

Lateral Distance along bottom flange (in.)

Fig. 3.16. FE (Shell93, Shell63) and Experimental Mid-Span Stresses in Bottom Flange of Fam and Turksrta [34] Series A Plexiglas Box Girder Bridge.

120

3.
bottom flange stress shell63
03

Bottom Flange Stress Shell93 e Experimental Bottom Flange Stress 5 10

-10

-5

Lateral Distance along bottom flange (in.)

Fig. 3.17. FE (ShelI93, Shell63) and Experimental Mid-Span Stresses in Bottom Flange of Fam and

Turksrta [34] Series B Plexiglas Box Girder Bridge.

100

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

3.6 Finite Element Modeling of the Baltimore Bridge 3.6.1 Non-Composite Bridge Modeling
The non-composite model discussed here simulates the construction process represented by three pours of the w et concrete. Fig. 3.18 shows the mesh discretization o f the non-composite girder model used in this study for the construction loading analysis. Four shell elements were used to model the web, two elements for each o f the top flanges, and four shell elements for the bottom flange. Shell elements were used as well to model longitudinal stiffeners and solid diaphragms at the supports location. The plate thick nesses and the material properties are required input for shell 63. The longitudinal discretization was cho sen to provide an aspect ratio of approximately one for each o f the web elements. The shell elements used were rectangular, while trapezoidal elements were used to model the solid diaphragms. Three dimensional truss and beam elements were used to model the top bracing truss and the cross frames. The following is a brief description o f these two elements:

Link 8 (3-D Spar): is a two-node, three dimensional truss element. It is a uniaxial tension-compression element with three degrees o f freedom at each node; translations in the nodal X, Y and Z directions. The element is a pin-jointed structure with no bending capabilities. Plasticity and large deflection capabili ties are included. The required inputs for this element are the material properties and cross-sectional area.

Beam 4 (3-D Elastic Beam): is a two-node, uniaxial element with tension, compression, torsion and
bending capabilities. The element has six degrees o f freedom at each node, translations in the nodal X, Y and Z directions and rotations about the nodal X, Y and Z axes. Large deflection capability is included. The required inputs for this element are the cross-sectional properties such as, the moment of inertia, the cross-sectional area and the torsional properties. As described in the preceding sections, the bridge consists o f five separate sections of steel spliced at four locations in the longitudinal direction as shown in Figs. 3.1 and 3.2. Each section has different plate thicknesses and a different top flange width (Fig. 3.3). In the finite difference model, the section properties were spaced linearly over ten nodal points for each splice as shown in Fig. 3.8. In the current fmite element

101

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

model and in contrast to the finite difference model, the splices and the intermediate supports locations were positioned exactly. The various plate thicknesses were input as real constants o f the corresponding shell elements, while the change of the top flange width was modeled by trapezoidal shell elements as shown in Fig. 3.18. The top bracing truss is an X-type bracing. The bracing members were modeled by truss elements. Since the truss elements are pin jointed at the nodes the X-bracing was modeled with one element (i.e. no joint was considered in the intersection point o f the two bracing members) to avoid geometrical instability that arises from joining multiple truss elements at a single joint. Beam elements were used to model the upper struts o f the K-ffames to avoid geometric instability. The diagonals and the lower strut o f the K-ffames were modeled using truss elements. The 5% super-elevation in the lateral direction o f the bridge was neglected in the model. The longitudinal stiffeners were modeled using shell elements. According to the bridge drawings some o f these stiffeners were structural tee sections and others were only steel plates as shown in Fig. 3.18. Due to the lack of information concerning the position o f the upper struts o f the K-ffames across the girder depth, the struts were arbitrarily connected to juncture between the top flange and the web. Each diagonal o f the K-frame was connected to the juncture between the web and the bottom flange and framed to the middle of the upper strut at the other end. During the construction process, each o f the two cells of the non-composite model would act independently since there is no connection between the cells, i.e. no external diaphragms specified, assuming that the steel corrugated sheeting forms have negligible connection effects. The report o f Ffeins et. al. [14] did not provide any information about the boundary conditions o f the tested bridge. It was assumed herein that each girder was supported by radial bearings. The vertical movements were constrained at the web and bottom flange juncture o f all supports. The normal practice is to have only one radially restrained bearing at the piers and abutments to reduce the reaction forces that re sult from the thermal loading. However, in the current study, the radial movement was prevented at all sup ports because the thermal loading was not considered in the analysis. Also, the tangential movement was

102

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

constrained in the intermediate pier No.2. All supports were modeled using zero displacement constraints.

Fig. 3.18. Typical F E Mesh for Baltimore Curved Box Girders

103

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

In ANSYS, to facilitate the radial and tangential movement restraint, the coordinate system o f the nodes at the supports were rotated to coincide with the global cylindrical coordinate system. During construction, the loads on the girders were composed o f the self weight o f the girder and the weight o f the w et concrete. As mentioned earlier, all strain gages in the field experiment were zeroed prior to each concrete pour. At subsequent pours the changes were recorded only for the particular pour. According to this process, only the incremental change in stress and defonnation would be considered (i.e. the self weight o f the girders would not be considered in the finite element analysis). On the other hand, the gravity load o f the concrete pour was applied at the middle o f each of the top flanges as a distributed load. This distributed load was estimated and proportioned based on the bridge drawings. The parapets were not considered since they were poured separately after the dead load test was completed. The thickness of the concrete slab was ten inches and the weight o f the reinforced concrete was taken as 145 pcf [14], Fig. 3.19. shows the distributed loads applied to the interior and exterior top flange o f both cells. The modulus of elasticity o f steel was set equal to 29500 ksi as used by Heins et.al.[14]. The first two pours were completed with a tim e interval o f about one week between the pours. The third pour was placed after a two-week interval, so the concrete o f the first two pours had enough time to gain strength. During the construction o f segment two, section AA-A located within segment one had some composite action between the steel section and the slab. This composite action made the section behave as a closed section that has more torsional stiffness, besides, the neutral axis was shifted towards the top flange. This shift tends to reduce the stresses in the top flanges. The vertical bending o f the girder induces a d d i tional forces in the top bracing members as concluded by Helwig et. al. [50]. Therefore, the reduction in the top flange stresses decreases the forces in the top bracing truss members. The same thing happened with section DD-D during the pouring o f segment three. The speed o f slab casting and the speed o f the concrete curing are factors that affect the degree of the composite action. It is stated by AASHTO LRFD 1994 [3] that permanent load that is applied before the slab has attained 75% o f fc shall be assumed to be carried by the steel section alone. Since the partial strength o f the concrete is unknown during pours 1 and 2, the

104

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

finite element modeling conservatively did not consider the partial hardening o f the slab. For this reason, the stresses in section AA-A during pour 2 and the stresses in section DD-D during pour 3 would be gener ally higher than the measured stresses.

3.6.2 C om posite Bridge m odeling


Fig. 3.20 shows a typical composite section o f the bridge model. The thickness o f the concrete slab is ten inches that is normally much larger than the steel girder plate thickness. This large thickness leads to a large concrete cross-sectional area which in turn leads to a large contribution to the moment o f inertia o f the composite section. The concrete slab in the current study was modeled by shell elements. These shell elements represent a two dimensional surface located at the mid plane o f the concrete deck. Therefore, the deck slab and the steel girder are considered as separate structural components in the finite element model. The nodes in the concrete deck were connected with the corresponding nodes in the middle of the top flanges using rigid constraint equations. The barrier walls were assumed to be o f uniform width and were also modeled with shell elements having real constants equal to the wall thickness. The shell elements modeled the mid surface o f the barrier wall. For live loading, a ten inches slab was used with a modular ratio equals to 8 . The three-dimensional model considered herein included the concrete deck over the entire length of the bridge as well as all the structural components used in the non-composite model such as the two cells, the K-frames, the longitudinal stiffeners, the solid diaphragms and the top bracing system. The same boundary conditions used in the non-composite model were used in the live loading analysis. There were sixteen loading points that produced maximum stresses for each o f the four instrumented sections in each of the four transverse lanes. The wheel loads for the truck were represented by 6 concentrated forces. The magnitude o f these forces and the spacing o f them were estimated based on the truck specification provided from reference [14], ANSYS has a language called ANSYS Parametric Design Language (APDL) which allows the user to build the model in term o f parameters. APDL contains a function that returns the node number closest to the global coordinate o f a specified point. Use o f this function was helpful for locating the nodes in the deck

105

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

slab that are closest to the position o f the six wheel loads of the deck.

0.1024 Kips/in.

0.1139 K ips/in.

0.1139 Kips/in.

0.1024 Kips/in.

!
4 -

Centre o Cuvature

Fig. 3.19. Distributed Loads Used to Simulate the Weight of Concrete Slab

106

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

V 5*

A* A'**

c s S >

\o 1
a A? i V O c x O An H V '

# e<
< 0^

AO0'

V xoO 'P

^V

G O '

\qV^

q n N

!&*
^ ' ,6 ^ce'

Chapter 3
3.7 C om parison o f R esults for C onstruction Loading

Curved Box Girder Bridge Field Test

3.7.1 D eflection
As m entioned earlier, the deflection was measured during the construction phase loading using survey ing equipment. Since there was no scaffolding, deflection measurements for the construction phase loading could not be m ade either on the interior girder or on the interior side of the exterior girder. Thus, the deflection measurements were restricted to the web-bottom flange juncture o f the exterior girder at various points in the longitudinal direction. There is no information available about the camber or the deflection due to self weight of the girder, therefore, the comparisons were made between the experi mental and analytical deflections due to the effect o f the concrete pours only. Also, there were no experi mental measurements for the rotations to substantiate the analytical rotations. Shown in Figs. 3.21 to 3.23 are comparisons o f the vertical deflections o f the web-bottom flange junc ture of the exterior cell under the concrete pours loading as obtained from the (CURSGL) finite difference program, from the field test and from the finite element model used in the current study. For a fair and con sistent comparison, the load considered in the current study was for wet concrete and reinforcing steel only as used in the (CURSGL) finite difference program. The web modeling could have a significant effect on the lateral flange moment and the girder deflection in curved girder bridges (Brockenbrough 1986) [13]. In reality, the lateral bending o f the flange is restrained by both the lateral bending stiffness o f the flange and the transverse bending stiffness o f the web. Therefore, it is necessary to model the transverse stiffness of the web with shell elements so that the web deformation effect is included in the analysis. In addition, the oretical and experimental studies (Hikosaka and Takami 1985) [51] revealed that a considerable increase in the deflections and stresses m ay arise from distortion in either curved or straight girders. The distortion was not considered in the finite difference analysis. In Figs. 3.21 and 3.22 both the ANSYS model and the CURSGL show good agreement with the measured field deflection. However, the finite difference results show a stiffer response than the one exhibited in the finite element model. This may be attributed to the web deformation effect and the distortion that were not considered in the form er model. The design deflec-

108

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

tions are available for the final pour and are plotted in Fig. 3.23. In Fig. 3.23, the finite difference model results for the final pour also show a stiffer response compared with the finite element model and the experimental results. The finite element model shows stiffer response than the measured response for pour III. The maximum difference between the finite element deflection and the measured deflection was 13%. The measured results were extracted from Heins et. al. report [14] by the digitization o f the plotted data; the digitization process is approximate in nature and may contribute to the above-mentioned difference. In addition, the use o f surveying equipment to measure the deflection during the construction phase m ay have encountered some error.

FEM Finite Difference Experimental

0.5

e
35

Span A

Span B

Span C

o O
-0.5

-2.5

Fig. 3.21. Comparison o f Vertical Deflections for Exterior Girder Under Concrete (Pour I) Load.

109

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

-j

Finite Difference Experimental

0.5
c Span A Span B Span C

g -0.5
Q

+ 3 O o

-2.5 J

Fig. 3.22. Comparison of Vertical Deflections for Exterior Girder Under Concrete (Pour II) load.

FEM Finite Difference 0.5


c Span A Span B Span C

Experimental Design

*3

C o

& -0.5

<3

-2.5

Fig. 3.23. Comparison of Vertical Deflections for Exterior Girder Under Concrete (Pour III) Load.

110

Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3
3.7.2 L ongitudinal Stress D istribution

Curved Box Girder Bridge Field Test

In this section, the girders self weight was not considered in the analysis. Thus, comparisons between analytical and field measured results are made only for the incremental changes in stress due to the three concrete pours. Field measurements are available for the final two pours only. The field results were extracted from Heins et. al. report [14] by the digitization o f the plotted data. The digitization process is approximate in nature and some errors result. In addition to these errors, the low percentage o f gages in working order in the dead load test does not give a clear image o f the normal stress distribution. The finite difference and the finite element normal stresses at each instrumented section under the effect of concrete pour I are compared in Figs. 3.24 to 3.27. The cross-sectional longitudinal stress distri bution determined using the finite difference method is composed o f the warping torsional stresses and the normal bending stresses. The report [14] indicated that the normal bending stresses were approximately 95% of the total normal stresses. During construction, the unsupported length o f the top flanges o f quasi-closed box girder between the cross frames can be subjected to lateral bending. The lateral bending can result from different sources in addition to torsional warping:

Distortional warping: Warping stresses caused by cross-sectional deformations. The horizontal truss system: A horizontal truss system is provided between the top flanges to
increase the torsional stiffness and reduce warping. This truss system is usually located at the top flange level where high vertical bending stresses exist. Due to compatibility, the truss must experience the same strain as the box girder in the axial direction. Accordingly, forces develop in the diagonal braces and struts, which then cause lateral bending in the top flange, even in the absence o f torsion. This effect was predicted using three-dimensional model by Helwig and Fan [50],

Web inclination: Because o f web inclination in the trapezoidal box girder, the loads o f the wet con
crete and formwork applied vertically on the top flange produce transverse components. These transverse components act as a uniformly distributed transverse load on the girder flanges.

Ill

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

All o f the above sources o f lateral bending contribute to the longitudinal stress distribution in the top flanges. The discrepancies in the longitudinal stresses o f the tips o f the top flanges between the finite dif ference and the finite element results are attributed to the lateral bending that was not considered in the finite difference results. However, it can be noticed that the normal bending stress represented by the aver age o f the two tip stresses o f the top flanges is very well correlated between the two solutions. Longitudinal stresses are distributed linearly across the web depth between mid top flange stress and the stress in the web-bottom flange juncture (not shown in the figures). Only the finite element longitudinal stresses in the web are shown. These are also distributed linearly across the web depth and the neutral axis appears to be approximately at mid depth of the box. A small decrease in the longitudinal stresses in the middle o f the bottom flanges is shown in Fig. 3.24. This reduction in the longitudinal stresses is attributed to the shear lag effect. Larger discrepancies generally occurred in sections with lower stress changes (Fig. 3.27), Shown in Figs. 3.28 to 3.31 are comparisons o f the longitudinal stresses for each instrumented section under the effect o f concrete pour II obtained from the present study, the finite difference program and the field measurements. The second pour was placed one week after the completion of pour I. Thus, the con crete of the first pour had enough time to gain some strength. Section AA-A is located within the first pour region and is thus expected to have some composite action between the steel girder and the concrete deck. The composite action causes the neutral axis o f the section to be shifted toward the top flange and the lon gitudinal stresses in the top flanges are reduced. This reduction is obvious in Fig. 3.28 for the measured stress as compared to the analytical stress. Although only few gages were in working order during the dead load test, the measurements from the gages that were functioning properly generally indicate a fair correlation with both the finite element and the finite difference solutions. Figs. 3.32 to 3.35 show comparisons o f the longitudinal stresses for each instrumented section under the effect o f concrete pour III obtained from the present study, the finite difference program, the field m ea surements and the design estimates (where available). The third pour was placed two weeks after the com-

112

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

pletion o f the second pour. Therefore, both section AA-A and section DD-D are expected to behave like closed sections due to the composite action between the concrete deck and the steel girder at these loca tions. Thus, the measured longitudinal stresses are less than the analytical longitudinal stresses in section AA-A. Design stresses are available for the final pour loading and they are shown in Figs 3.32 to 3.35. The report [14] indicated that the design was generally conservative as compared with the finite difference results. Figs 3.32 to 3.35 confirm the reports conclusion since the design stresses are higher than the mea sured and estimated stresses. Table 3.1 compares the maximum longitudinal stress in the bottom flange due to the final concrete pour obtained from the present study, the field measurements, the finite difference model with top lateral bracing, the finite difference program without top lateral bracing and the design pro cedure. The finite element results seem to be closer to the finite difference results with top lateral bracing and the measured results have a fair correlation with both solutions. This closeness in the results between the two analytical solutions indicates that Dabrowskis equation for the equivalent plate models the top lat eral bracing stiffness fairly well but at the same time using this method ignores the effect o f lateral bending on the top flanges caused by the forces in the diagonals o f the top truss system. It can be noticed from table 3.1 that the design stresses are higher than those measured and those deter mined analytically. The design o f the box girders did not include any consideration for the top bracing [14], however, the design stresses are still higher than those determined analytically using the finite difference model without considering top lateral bracing

113

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Interior Cell

Exterior Cell

7.78 7.47 7.33 7.33 7.41 7.62 7.82 7.62 7.57 7.62

7.87 7.77

Centre of Curvature Finite Element Finite Difference (Underlined)

Fig. 3.24. Longitudinal Stress Dead Load Test, Section AA-A, Pour I (ksi).

Interior Cell

Exterior Cell

-1.81
-1.83 -1.85 -1.89 -1.92 -1.96 -2.00 -1.96 -1.95 -1.93 -1.91

Centre of Curvature Finite Element Finite Difference (Underlined]

Fig. 3.25. Longitudinal Stress Dead Load Test, Section BB-B, Pour I (ksi).

114

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Interior Cell -0.55 ! -1.49 -1.27 -1.29 -1.30 -1.31 -1.31 -1.32

Exterior Cell

-1.34 -1.35 -1.35 -1.35

Centre of Curvature Finite Element Finite Difference (Underlined)

Fig. 3.26. Longitudinal Stress Dead Load Test, Section CC-C, Pour I (ksi).

-0.58 -0.69 -0.3 f

-0.54
-

0.6

-0.57 -0.69

0.68

-0.70 Interior Cell

-0.35
i-0 .ll

Exterior Cell

0.46 0.36 0.36 0.36 0 0.37 0.37 0.37 0.37

0.47 0.37

Centre of Curvature Finite Element Finite Difference (Underlined)

Fig. 3.27. Longitudinal Stress Dead Load Test, Section DD-D, Pour I (ksi).

115

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

> (-7.10)

Interior Cell

-4.54'

Exterior Cell

7.89

7.75

7.74 7.85 (5.98)

8.07

8.27

8.08

8.02

8.08

8.24

Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental)

Fig. 3.28. Longitudinal Stress Dead Load Test, Section AA-A, Pour II (ksi).
(1.03)

Interior Cell

-0.04

Exterior Cell

-0.55 -0.83 -0.85 -0.88 -0.91 -0.94 (-0.89) -0.96 -0.92 -0.90 -0.88 -0.85

Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental)

Fig. 3.29. Longitudinal Stress Dead Load Test, Section BB-B, Pour II (ksi).

116

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

|(2.56) j 3.90 f | 4.02 Interior Cell 3.93 4.17 4.04 4.09 3.94

(3.13) 4.04 4.16 Exterior Cell


1.21 11

(5.50) 4.05 4.30 '

T -3 .4 6 -3.06 -3.05 -3.06 -3.08 -3.09 (-2.96) (-5.00) -3.16 -3.16 -3.17 -3.20 -3.22

Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental)

Fig. 3.30. Longitudinal Stress Dead Load Test, Section CC-C, Pour II (ksi).
-7.60) -10.49
10.02

-10.08 -10.35

-10.05 -10.27

-10.75 -10.42 Exterior Cell -10.33 '-6.09

Interior Cell

6.09

5.99 6.02

6.11 (8.40)

6.43

6.28

6.27

6.32

Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental)

Fig. 3.31. Longitudinal Stress Dead Load Test, Section DD-D, Pour II (ksi).

117

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

(-3.40)
-

[-8.75]

(-9.00)

6.02

-6.24 -5.96 Interior Cell

[-8.30] -6.78 -6.32

-5.70
-

-6.17

6.12

- 0 . 22 \

2.65\ 5.84

/-3.31 ?-0.24 2.85 6.18

Exterior Cell

(3.15)

[8 .20 ]

5.91 | 5.72 [7.50] |


1(7.25)

5.65

5.70

5.84

(5.75) Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental) [Design]

Fig. 3.32. Longitudinal Stress Dead Load Test, Section AA-A, Pour HI (ksi).
15.00] [15.00]

Interior Cell

Exterior Cell

-4.47 -4.70 -4.75 -4.86 -4.98 -5.12 (-5.50) [-11.00] -5.17 -5.06 -5.01 -4.97 -4.93

[-11.00]

Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental) [Design]

Fig. 3.33. Longitudinal Stress Dead Load Test, Section BB-B, Pour H I (ksi).

118

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

[-6.30] -3.89 -3.87 Interior Cell

- 6 . 00 ]

-3.97 -3.69 Exterior Cell

3.19 2.97 2.98 (5.25) Centre of Curvature Finite Element Finite Difference (Underlined) (Experimental) [Design] 3.06 3.21 [6.50] 3.26 [5.50] 3.13 3.10 3.14 3.23

Fig. 3.34. Longitudinal Stress Dead Load Test, Section CC-C, Pour III (ksi).

1-13.20]

-13.20

Interior Cell

Exterior Cell

4.79

4.71

4.74

4.83

4.93 [8.00]

5.11

4.97

4.95

4.99

Centre of Curvature Finite Element Finite Difference (Underlined) [Design]

Fig. 3.35. Longitudinal Stress Dead Load Test, Section DD-D, Pour III (ksi).

119

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Table 3.1: C om parison o f Experim ental, A nalytical and Design Stresses D ue to Dead Load.

Max.Longitudinal Stress at Bottom Flange in (ksi), Determined By: Section Finite Difference with Bracing 6.25 -4.96 3.26 Finite Difference without Bracing 9.02 -6.66 4.27

Finite Element 5.91 -5.17 3.26

Experimental

Design

AA-A BB-B CC-C

7.70 -5.14 6.12

9.31 -11.51 7.06

3.8 C om parisons o f Results for the Live L oad Test

3.8.1 D eflection
Sixteen different loading conditions were considered in the analytical procedure. These loading condi tions were the ones that cause a maximum stress condition in each o f the four gage locations for the four lanes. In CURSYS, the finite difference program used to analyze the bridge for the live load testing, the concentrated loads that simulate vehicle wheel loadings could only be specified on or between the two girders; thus, the loading scheme for lanes one and four had to be modified. For lanes one and four, the loadings were applied as axle loads and placed on the girder closest to the actual position o f the test vehicle because at least one of the wheels o f each axle was outside the area surrounded by the two girders. This load modification might cause some approximation in the results. The report [14] indicates that the tor sional warping stresses in the live load test were on the order o f 0.1 percent o f the longitudinal bending stresses, therefore, only bending effect was considered in the analysis. For this reason, the bottom flange, for example, exhibited constant analytical deflection and longitudinal stress across its entire width. Figs. 3.36 to 3.39 display the deflection in the deflectometer position at span A for the test vehicle loading at load points 1 to 4 obtained from the finite element analysis (ANSYS), from the finite difference program (CURSYS) and from the field measurements. A s mentioned earlier, the finite difference deflection is con-

120

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

stant across the bottom flange width, while the finite element deflections and measured deflections vary across the bottom flange width. The finite element results show excellent correlation with the measured results. Centre of Curvature :

Interior Cell

Exterior Cell

c
o
g g

tj < D Q 0.2 0.3 Experimental Finite Element Finite Difference

0.2
0.3

Fig. 3.36. Live Load Deflection, Gage (1-4), Load Point 1.


Centre of Curvature I

Interior Cell
0 0
G

Exterior Cell

.2

a< o a 0.2

a < a

0.3 -----------Experimental Finite Element : Finite Difference

0.3

Fig. 3.37. Live Load Deflection, Gage (1-4), Load Point 2.

121

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0 0

Exterior Cell

| e o.i o O
*-P

S 3 O

< D
0.2 0.3 Experimental Finite Element Finite Difference

1 5 = 4 > Q 0.2 0.3

o < 0

Fig. 3.38. Live Load Deflection, Gage (1-4), Load Point 3.


Centre o f Curvature !

Interior Cell
0 0

Exterior Cell

a .2

C (D Q

c o 0.1 O < D
0> Q 0.2

0.3 Experimental Finite Element Finite Difference

0.3

Fig. 3.39. Live Load Deflection, Gage (1-4), Load Point 4.

122

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3
3.8.2 L on gitu d in al Stress D istribution

Curved Box Girder Bridge Field Test

Once the concrete slab develops its full strength, the cross section of the box girder becomes fully closed and fully composite. The presence of the concrete slab causes the neutral axis to move higher than its location for th e non-composite section. As indicated by the report [14], the longitudinal stiffener contri bution was m ore than 1% of the total moment o f inertia of the non-composite cross section. For this rea son, the longitudinal stiffener was considered in determining the moment o f inertia of the section in the finite difference analysis, contrary to the non-composite section case where the longitudinal stiffener was not considered in evaluating the moment of inertia o f the section. Also, the composite action resulted in relatively small top flange stresses. Therefore, in the live load test, only the longitudinal stresses in the bot tom flange o f the box girder are examined. In Figs. 3.40 to 3.55, the longitudinal stress distribution in the bottom flange o f the box girders obtained from the finite element analysis, from the finite difference analysis and from the field measure ments are plotted. As mentioned earlier, the finite difference analysis considers only the simple bending effect. Heins et. al. report [14] concluded that the warping stresses were negligible, therefore, the longitudi nal stress is constant across the bottom flange width. The bottom flange stress distribution obtained from the finite element analysis confirmed this conclusion. It can be seen that the finite element results have bet ter correlation with the measured results than do the finite difference results. The maximum longitudinal stresses in the bottom flange for each of the instrumented sections that were obtained from the present study, from the finite difference analysis and from the field measurements are tabulated in Table 3.2 for the live load test. The impact factor (I) applied to the analytical stress is based on the maximum average impact factor measured experimentally and is equal to 28.6% [14], The data in Table 3.2 indicates that the design stresses are much larger than the experimental and calculated stresses. The reason for these discrep ancies is that in the design process, the steel section over the interior supports was assumed to be non-com posite even though shear studs were provided in this region; while the analytical study assumed full

123

Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

composite action throughout the length. The low experimental and analytical live load stresses in section BB-B, close to the interior support, indicate that the section did exhibit significant composite action.
:

Centre of Curvature

Interior Cell
0 0

Exterior Cell

1
(A) <73

go

o -b C /3 2 3 Finite Difference - -------a Finite Element Experimental 3

Fig. 3.40. Longitudinal Stress Live Load Test, Section AA-A, Load Point 1.
Centre of Curvature

Interior Cell
0 0

Exterior Cell

CO

& o o < D b
<Z2

CO

< D -b

2
i

------------

Finite Difference

--- Finite Element a Experimental

Fig. 3.41. Longitudinal Stress Live Load Test, Section AA-A, Load Point 2.

124

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0 0

Exterior Cell

i I
1

Finite Difference Finite Element Experimental

Fig. 3.42. Longitudinal Stress Live Load Test, Section AA-A, Load Point 3.
Centre of Curvature

Interior Cell
0 0 0 1 1 0

Exterior Cell

'

Finite Difference Finite Element Experimental

Fig. 3.43. Longitudinal Stress Live Load Test, Section AA-A, Load Point 4.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0

Exterior Cell

i
1

&
(O -b

Finite Difference Finite Element a Experimental

Fig. 3.44. Longitudinal Stress Live Load Test, Section BB-B, Load Point 5.
Centre of Curvature i

Interior Cell
0 0

Exterior Cell

i
!

& < D s55 -2 -3 -----------Finite Difference Finite Element s Experimental


CO CO

Fig. 3.45. Longitudinal Stress Live Load Test, Section BB-B, Load Point 6.

126

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0 0

Exterior Cell

C O

oo

is on

Finite Difference Finite Element a Experimental

Fig. 3.46. Longitudinal Stress Live Load Test, Section BB-B, Load Point 7.
Centre of Curvature

Interior Cell
0

Exterior Cell 0

U
CO CO

CO CO

O ton _2

5o _2
-3 Finite Difference Finite Element s. Experimental

-3

Fig. 3.47. Longitudinal Stress Live Load Test, Section BB-B, Load Point 8.

127

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0 0

Exterior Cell

&
co
CD i-i CO

&
CO CO CD

(Z )

13 G O 2 3 Finite Difference -------------a Finite Element Experimental

Fig. 3.48. Longitudinal Stress Live Load Test, Section CC-C, Load Point 9.
Centre of Curvature

Interior Cell
0
C O

Exterior Cell
0

V j
CD

c-

<Z)

Finite Difference Finite Element Experimental

Fig. 3.49. Longitudinal Stress Live Load Test, Section CC-C, Load Point 10.

128

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre o f Curvature

Interior Cell
0 0

Exterior Cell

11 o o c o < D 05
4-*

1
CD

s-

is G O 2
i

Finite Difference Finite Element : Experimental

Fig. 3.50. Longitudinal Stress Live Load Test, Section CC-C, Load Point 11.
Centre of Curvature

Interior Cell
0 0

Exterior Cell

Finite Difference Finite Element k Experimental

Fig. 3.51. Longitudinal Stress Live Load Test, Section CC-C, Load P o in t 12.

129

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0

Exterior Cell

& C O 2

5-

GO

Finite Difference Finite Element


0

Experimental

Fig. 3.52. Longitudinal Stress Live Load Test, Section DD-D, Load Point 13.
Centre of Curvature I

Interior Cell
0 0

Exterior Cell

& < D
CO CO

GO

2
3

Finite Difference Finite Element Experimental

Fig. 3.53. Longitudinal Stress Live Load Test, Section DD-D, Load Point 14.

130

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Centre of Curvature

Interior Cell
0 0

Exterior Cell

1
i . .....i CO

co

CO CO

tCO

C Z )

2
3

Finite Difference Finite Element Experimental Fig. 3.54. Longitudinal Stress Live Load Test, Section DD-D, Load Point 15. Centre of Curvature
i

Interior Cell
0 0

Exterior Cell

--1
GO

L m J CO CO

U i

C/3

1) 5t

CO CO

2
3

3 Finite Difference Finite Element Experimental

Fig. 3.55. Longitudinal Stress Live Load Test, Section DD-D, Load Point 16.

131

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 3

Curved Box Girder Bridge Field Test

Table 3.2: Comparison of Experimental, Analytical and Design Stresses Due to Live Load+Impact Factor (28.6%).
Max.Longitudinal Stress at Bottom Flange in (ksi), Determined By: Section Finite Difference with Bracing 2.65 -1.05 2.07 Finite Difference without Bracing 2.65 -1.05 2.07

Finite Element 2.42 -0.96 1.94

Experimental

Design

AA-A BB-B CC-C

2.32 -0.76 1.83

6.86 -4.73 6.97

132

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4 Design Safety of Horizontally Curved Bath-tub Girder Bridges during Erection and Construction.

4.1 Introduction

The behavior o f horizontally curved girders is somewhat more complex than that o f straight girders as a result o f the interaction between prim aiy bending and torsion. It is well known that a curved composite box girder has a very large torsional stiffness and the top flanges are fully restrained; however, during transportation, erection and construction, the top flanges are braced only at the cross frame locations. Thus, they are susceptible to lateral buckling. Erection and construction o f horizontally curved steel bridges are generally more complex than that o f straight girder bridges and curved girder bridges tend to be more susceptible to stability problems. In 1974, ASCE-AASHTO Task Committee on Flexural M embers [5] reported the four most important major box girder construction failures as 1) Fourth Danube Bridge, Austria. 2) Milford Haven Road Bridge, United Kingdom. 3) West Gate Bridge, Australia. 4) Rhine River Bridge, Germany. The committee indicated that most box-girder bridge failures occurred during the erection or construction stages and that many design and construction deficiencies/ oversights contributed to the failures. The committee concluded that there was a lack o f understanding o f the margins of safety required during erection and construction o f box-girder bridges.

133

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

The AASHTO Guide Specifications for horizontally curved highway bridges (1993) [39] do not include provisions directly related to fabrication, erection and construction o f curved bridges. The National Cooperative Highway Research Program (NCHRP) project 12-38 (1999) [41] recommended completely revised guide specifications for horizontally curved bridges. The recommended specifications were approved in 2003 by AASHTO to represent the recent AASHTO Guide Specifications for horizontally curved highway bridges (2003) [40], The current Guide Specifications [40] include explicitly a constructibility limit state to ensure that the subsequent erection and construction stages are properly designed. The current Guide Specifications state that one scheme to construct the bridge within the constrains o f the site shall be shown on the design plans. The scheme shall include a sequence o f girder and deck placement and location of any necessary supports. The current Guide Specifications also include a detailed construction section that applies specifically to the fabrication, shipping and erection o f the steel superstructures and the placement o f the concrete deck. This section should be used by the contractor to prepare a construction plan. Few studies involving both analytical and field-testing components o f curved I and box girder bridges have been performed in the last three decades. Generally, these field tests involved tracking the loads and deformations in the girders during erection, construction and service loading. Field data extracted from these tests was utilized to examine the accuracy o f load and deformation estimations o f analytical tools (programs) that are commonly used to model such tests. The bridge behavior in m ost o f these analyses was assumed to be linear and elastic because it was expected that the bridge behaves linearly and elastically during erection, construction and service loading. A designer is faced with the question of the margin of safety or reliability that would be available over the erection and construction processes. In 2000, Helwig and Fan [50] conducted field studies on a new three span continuous curved box girder bridge in Houston, Texas. They indicated in their report that there were a number o f problems that occurred during girder erection and construction that raised the concerns o f the Texas Department of Transportation (TxDOT) engineers and the bridge contractor engineers. The construction plan was

134

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

reviewed by the project researchers using refined finite element models. The girder stress levels and the brace forces w ere examined as well. As a result o f this review, the initial plan o f concrete placement was modified and external frames between the girders at supports were retrofitted.

4.2 O bjectives

Detailed information about the box girders geometry, plate thicknesses, bracing members, K-Frames, external and internal diaphragms o f the Houston bridge were received from the research team at the University o f Houston. This specific bridge is used here as an example reflecting current fabrication and detailing practices and it is chosen for a detailed evaluation o f the structural safety/reliability during the construction process. The main objectives behind this evaluation and the process involved is to: Simulate the girder erection and the concrete slab placement sequence using comprehensive three-dimensional finite element models. Model the bridge by incorporating both material and geometric nonlinear behavior. Assess the simple factor o f safety for every erection step and for every construction stage. Observe the force and stress levels in the lateral top bracing members and cross-frame members at ultimate. Describe the modes o f failure o f every erection and construction stage. This evaluation can also be viewed as a demonstration o f the analytical investigation that can be employed as a secondary evaluation o f the details o f any proposed construction scheme.

4.3 Detailed Description o f the H ouston Bridge

The bridge under study, located in Houston, Texas is a three-span continuous unit. It is composed of two bath-tub girders with a composite concrete deck that is connected to the steel trapezoidal section through shear connectors. The general layout o f the bridge that illustrates the spans and the radius along the centerline is presented in Fig. 4.1. The cross section o f the completed bridge is shown in Fig. 4.2

135

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.,

7&rif

mQm

] - 188.1 ft

>{<

290.7 ft

>|<

" 179.7 flr- >|

I y

Section P Section N

Radius = 894.5 ft

Bent 17

Bent 20

BentXS
Back-to-Back Cross frames

Bent 19

Fig. 4.1. General Layout o f Houston Bridge. [50]

120.5

135' 9.5 Thick Concrete Slab

120.5

75.5

75.5"

Fig. 4.2. Cross-Section o f the Box Girder Bridge.[50]

136

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

The trapezoidal bath-tub girders were fabricated with two top flanges, one bottom flange and two inclined webs. They are o f uniform depth and width along the bridge length. The top flange was widened and thickened in the vicinity of the interior supports to account for the high negative moment in these regions; also, the thicknesses of the bottom flange and the webs were varied along the length o f the bridge. A single diagonal (SD) top flange lateral truss system was provided in the exterior spans, while an Xdiagonal lateral truss system was provided in the interior span as shown in Fig. 4.1. The top flange lateral truss was connected approximately 11 inches below the top flange. Interior solid diaphragms were provided at all supports and internal cross-frames were placed at approximately 20 ft. intervals, while the lateral bracing panels were approximately 10 ft. in length, (i.e. two lateral bracing panels extend between two consecutive cross frames). However, as an exception to this sequence, the cross frames at the splice locations were placed back-to-back with SD bracing even in the internal span as shown in Fig. 4.1. External solid diaphragms were provided in the interior supports, while exterior cross-frames drat connected the two girders were provided at external supports (Bents 17 and 20) (Fig. 4.1). For transportation and erection considerations, the internal and external girders were fabricated in seven segments that were spliced together during erection. The individual segments were numbered from 901 to 914. Two cranes were used to lift the girder segments into place and to hold them while the bolted splice assemblies were performed. Fig. 4.3 illustrates the erection sequence for the interior girder. The erection commenced from the external spans and continued inward with segments 907 and 908 becoming the last segments to be spliced for the interior and exterior girders respectively. The interior girder erection was accomplished through the following steps: 1) Segments 901 and 903 were erected and spliced together starting from bent 17 to form simply supported beams that cantilevered into the interior span. The erection procedure o f segments 911 and 913 followed the same procedure used in the erection o f segments 901 and 903, 2) Segments 905 and 909 were added to the cantilevered spans as shown in Fig. 4.3.,

137

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

3) The erection process was completed by dropping-in segment 907. The exterior girder was erected using a sim ilar sequence. The interior girder was instrumented at two locations on segments 901 and 903 (Fig. 4.1). These locations represent the negative and positive moment regions. Strain gages were installed in these two locations to m easure the strain developed at the top flanges, the bottom flanges, the longitudinal stiffeners and the webs. The struts and diagonals o f the top flange bracing system in the vicinity o f these two locations were also instrumented with strain gages at the middle o f their lengths to measure axial strains and to estimate the forces that developed in these members. The m odified construction plan o f the concrete bridge deck was divided into three stages as shown in Fig. 4.4. M etal deck forms were used as permanent forms to support the concrete during placement and curing. The strains in the interior girder and bracing members were monitored during the erection and construction loading sequence. Helwig and Fan [50] simulated the erection and the construction o f the bridge with an elastic and first order three-dimensional finite element model using ANSYS. M aterial behavior in all o f their analyses was assumed to be linear. The measured girder strains and the estimated brace forces were compared with the finite element solutions for various erection and construction steps. Their report concluded that the agreement between the finite element solutions and the field results for the erection o f segments 905 and 907 and the first stage o f concrete placement was reasonable, but the finite element solution appeared to overestimate the horizontal truss forces in the second and third stages of concrete placement. The permanent metal deck forms increased the torsional stiffness and enhanced the stability o f the girder during the construction. Also, the partially hardened concrete of the first stage provided some composite action with the steel girder that was not considered in the analysis. The research team attributed the discrepancies in the measured and estimated truss forces to the metal deck forms and the partially hardened concrete.

138

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 4

Design Safety of Curved Bath-tub Girder Bridges,

As m entioned earlier, detailed information about the box girders geometry, plate thicknesses, bracing members, K-Frames, external and internal diaphragms o f the Houston bridge were received from the research team at the University of Houston. ANSYS is used in the current finite element modeling. The erection of segments 905 and 907 and the three stages o f concrete placement were simulated using threedimensional finite element analysis. The geometrical nonlinearity was considered in the current analysis, while material behavior was assumed linear. The girder stresses and the horizontal truss forces obtained from the current analyses were within 10% in agreement with that obtained from Hilweg et. al. [50] analyses.
(a) Step 1. Erection of Segments 901, 903 and 911, 913

901

903

I 911

I 913

(b)

Step 2. Erection of Segments 905 and 909

0 i.;. ;:: ~ :..i ~ :i


905

cm
909

O'

(c)

Step 3. Erection of Segment 907

I d 907

ICT

Bent 17

Bent 18

Bent 19

Bent 20

Fig. 4.3. Erection Sequence of the Interior Girder. [50]

139

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 4

Design Safety of Curved Bath-tub Girder Bridges,

(aVStage 1: 71.4
<

72.0 >1

A
Bent 17 () Stage 2:

77
Bent 18

77
Bent 19

iiih A
Bent 20

A
Bent 17 (c) Stage 3:

7 7

77
Bent 19 164.3

A
Bent 20

Bent 18

174.7
^ ^

A
Bent 17 Bent 18
Fig. 4.4. Construction Sequence of the Concrete Slab. [50]

77
Bent 19

A
Bent 20

140

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

4.4 N onlinea r Finite Elem ent C apabilities o f ANSYS

As mentioned earlier, the main objective of the study presented in this chapter is to assess the load factors for the various erection and construction steps. A nonlinear static incremental technique has to be used to fulfill this goal. All finite element modeling and nonlinear analyses presented in this chapter were performed using ANSYS. Shell elements were used to model the webs, the bottom flange and the two top flanges o f the boxes, as well as all transverse/longitudinal stiffeners and the solid diaphragms. The top truss system and the cross frames were modeled using beam and truss elements. Beam 4 and Link 8 ANSYS elements described in the Chapter (3) were used to model the bracing and cross frame members. ANSYS has a large library of shell elements to choose from, each with different capabilities. In the previous chapter, it was expected that the bridge would behave linearly elastic during the field test and the levels o f stresses and deformations were predicted to be low. For these reasons, the list o f the shell elements to choose from was restricted to shell 93 (eight-node shell) and shell 63 (elastic shell). It was concluded from simulating the Fam and Turkstra experiment [34] that was modeled in Chapter (3) that using a refined mesh o f a 4-node shell produced satisfactory results and at the same time cut the cost and time of computation associated with a higher order element. For this reason, shell 93 (eight-node shell) was excluded from the list. Since the problem under investigation in here involves loading the structure up to failure, it is expected that the plated members could experience local instabilities and that the structure would behave nonlinearly. Thus, shell 63 (elastic shell) can not be used in the current model because the chosen shell element must have both material and geometric nonlinear capabilities. In finite element formulations, the stiffness equation o f the structure is based on the undeformed shape o f the structure, but when the structural displacements and/or strains become large, the structural equilibrium has to be based on the deformed configuration and the stiffness matrix for this case is dependent on the displacements.

141

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

ANSYS elem ent library has in addition to shell 93 (eight-node shell element) three other shell elements that could be used in modeling the nonlinear behavior of the structure. The following is a brief description o f these three shell elements:

Shell 43 (4-node plastic large strain shell): This element has six degrees o f freedom at each node
(translations in the nodal x, y, and z directions and rotations about the nodal x, y, and z axes). The element has plasticity, large deflection and large strain capabilities. It is indicated in ANSYS Manual [4] that this element is well suited for modeling thick shell structures.

Shell 143 (4-node plastic small strain shell): This element also has six degrees o f freedom at each node
(translations in the nodal x, y and z directions and rotations about the nodal x, y and z axes). The element has plasticity, large deflection and small strain capabilities. According to ANSYS elements manual, this element is well suited to nonlinear shell structures that are thin to moderately thick.

Shell 181: It is 4-node element with six degrees o f freedom at each node (three translations and three
rotations). The element is suitable for analyzing thin to moderately thick shell structures. The element has plasticity, large deflection and large strain capabilities and it is recommended by ANSYS manual for nonlinear applications with large strains. All o f the above elements use four integration points to integrate the stiffness and mass matrices. These points are located at mid surface of the shell. Two integration points are used in the through thickness direction for the shell with linear material properties while five points are used for nonlinear materials. According to the AASHTO Guide Specifications for horizontally curved girder bridges (2003)[40], it is possible for the top flange to be designed as a compact element; this would perm it the level o f strain in the material to go beyond the small strain theory. Therefore, shell 143 was excluded from the list of suitable elements to model this problem because it does not support large strain capabilities. According to ANSYS elements Manual [4], shell 181 is recommended over shell 43 in the situation where a large strain capability is needed because shell 181 has enhanced convergence capabilities. For this reason, shell 181

142

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

was the more suitable choice. In the following, a brief description o f the chosen element, shell 181, is presented.

4.4.1 Shell 181: Characteristics


Shell 181, as mentioned before, is a four-node element that uses bilinear displacement shape functions. Bilinear elem ents are attractive because they are simple and have only comer nodes. Unfortunately, however, they are too stiff in (in-plane) bending; this property is attributed to the presence o f the parasitic shear . W hen bilinear elements are subjected to pure bending, the shear strain is expected to be zero and the total strain energy stored in the element is that from normal strains only. However, because o f the bilinear shape functions, shear strain does exist and the strain energy resulting from shear strains will be added to the strain energy o f normal strains, therefore, the bending moment required for imposing the bending displacements is higher than the moment required if the shear strains are accurately estimated. This accounts for the stiffer response o f such elements in bending. The effect o f parasitic shear is significant when the element aspect ratio becomes large and the mesh is said to be locked. Shell 181 formulations avoid the problem o f membrane shear locking by either using reduced integration with hourglass modes (zero-energy modes) control or by using full integration with incompatible modes. Using the lower order quadrature rule reduced integration results in lower computation cost and softens the element, thus, countering the overly stiff behavior caused by parasitic shear. As mentioned earlier, four integration points are used to fully integrate the stiffness and mass matrices, while only one integration point per element in the center o f its surface is used in the reduced integration scheme. However, the number o f the integration points through the element thickness remains the same with two for linear and five for nonlinear material behavior. Using one integration point in the center o f the element (reduced integration) instead o f four points (full integration) leads to the inability o f the element to sense two bending modes o f deformation, these modes are zero-energy modes (hourglass modes); one of

143

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

them is associated with in-plane bending and the other is associated with out-of-plane bending. To control these zero-energy modes, ANSYS introduces hourglass stiffnesses to the element when reduced integration is used. These stiffnesses control the hourglass modes without stiffening the element modes that are already working well. Using the reduced integration, which is the default option in the program requires a m ore refined mesh compared with the mesh that can be used with the full integration option. Another alternative to relieve the constraints imposed by the parasitic shear is using the full integration scheme with compatible modes or bubble modes . This alternative adds internal degrees o f freedom to the elements. The presence o f these degrees o f freedom lets the shape functions accurately estimate the pure bending modes. Shell 43 does not support reduced integration; instead, the element bending performance can only be enhanced through the use o f the bubble modes options. For out-of-plane bending, shell 181 elements use the thick plate formulations that consider transverse shear deformations. When the shell thickness is small, the bilinear element formulation could lead to transverse shear locking. The transverse shear locking can be explained by considering the out-of-plane behavior o f an element as a plate. The stiffness matrix o f a plate element can be considered as being consisted o f a bending stiffness [Kb] and a transverse shear stiffness [Ks], The terms in [Kb] are proportional to U, while the terms in [Ks] are proportional to t, where t is the plate thickness [23], As the thickness gets smaller, [Ks] starts to act as a penalty matrix that yields deflections equal to zero, i.e. the computed deflections o f a very thin plate is almost zero and the mesh locks. This locking could be avoided by using reduced or selective integration in which fewer points o f integration are used to generate [Ks]. Bathe and Dvorkins [7] shear strain formulations were used in shell 181 to alleviate transverse shear locking rather than relying on reduced or selective integration schemes. They introduced the idea o f mixed interpolation for various strain components. In their formulations, the bending and the membrane strain components are calculated using the usual shape functions, however, the transverse shear strain are interpolated linearly. This approach means that the transverse shear strain is constant along one edge o f the

144

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 4

Design Safety of Curved Bath-tub Girder Bridges

element and varies linearly to zero at the opposite edge. This approach has proved to be efficient in removing the transverse shear locking phenomenon. A curved shell can be approximated by connecting flat elements with combined membrane and plate actions. Plate actions are presented by three degrees o f freedom per node, w, t?x, while membrane

actions in classical two dimensional formulations are represented by two degrees of freedom per node, u, v (in-plane displacements). Since most three-dimensional finite element formulations have six degrees of freedom per node, a sixth degree o f freedom called the drilling degree o f freedom is added, o9z, (the rotation about the normal to the shell mid surface). I f flat shell elements are connected to a certain node all happen to be coplanar, with no stiffness to resist this degree o f freedom, this could result in part o f the structures stiffness matrix corresponding to

i3z becoming singular. This singularity causes numerical

instabilities. To solve this problem, shell 181 uses a penalty method to relate this independent rotational degree o f freedom with in-plane displacements through the use o f a fictitious stiffness. ANSYS chooses an appropriate penalty stiffness by default

4.4.2 Nonlinear Analysis


The finite element equilibrium equation for a linear static analysis is formulated assuming that the structure undergoes small displacements and that the material response is linear. The assumption o f small ,displacement is considered reasonable if the deformed geometry of the structure is adequately defined by the undeformed shape. The assumption o f the material being linearly elastic, requires that the stress is proportioned to strain at all load levels and no pennanent deformations remains upon the removal o f the load. For the linear elastic formulation, the stiffness matrix [K] is independent o f the displacements. If the structure experiences large deflections or material yielding, this means that the above assumptions are no longer valid and a nonlinear analysis would have to be performed. The typical nonlinear problems are usually solved through a series o f successive linear approximations.

145

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4
4.4.3 G eom etric Nonlinearity:

Design Safety of Curved Bath-tub Girder Bridges,

In a nonlinear problem, the element stiffness matrix [Ke] is dependent on the displacements. The element stiffness matrix is ideally formulated as:

[Ke] = J[B ]T[D][B]dv


V

Where, [B] is the strain-displacement matrix, CD] is the stress-strain matrix. W hen the strain-displacement matrix [B] is not constant and is dependent on element displacements, the stiffness matrix becomes nonlinear and this nonlinearity is called geometric nonlinearity, while the material nonlinearity results from the stress-strain relationship [D] being nonlinear. Structural problem that exhibit geometric nonlinearitites can be classified as: 1) Large displacementsmall strain problems, where the displacements and rotations o f the fibers are large, but the fiber extensions and angle changes between fibers are small, 2) Large displacement-large strain problems, where fiber extensions and angle changes between them are large, fiber displacements and rotations may also be large [8 ]. The finite element formulation that accounts for geometric nonlinearities must be written with respect to the deformed geometry. ANSYS uses an element independent corotational (convected) procedure by Rankin and Brogan [64] for the treatment o f large displacements, in which the element coordinates follow the element as it deforms. This procedure is independent o f the specific finite element employed and thus the element developers are freed from the burden o f embedding the large deflection capability in the element formulations.

146

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4
4.4.4 M aterial Nonlinearity:

Design Safety of Curved Bath-tub Girder Bridges,

Nonlinear stress-strain relationships are a common source o f nonlinear structural behavior. Many factors can influence the materials stress-strain properties and these include: nonlinear elasticity, creep, swelling and plasticity. The phenomenon of interest that causes material nonlinearity in this study is plasticity. For metals, the uniaxial stress-strain curve shows the simplest description of how plasticity affects the relationship between stress and strain. The stress-strain relationship is linear up to a stress level known as the proportional limit. The material returns to its original geometry if the structure is unloaded. Beyond the yield point, which for all practical purposes is assumed to coincide with the proportional limit, the stress-strain relationship becomes nonlinear. Plastic behavior is characterized by non-recoverable strain when the structure is unloaded. Plasticity is a path dependent phenomenon and requires the load to be applied in a series o f small increments, so that the model follows the non-linear load response path as closely as possible. At any stage after initial yielding, the total strain is composed of elastic (recoverable) strain and plastic (unrecoverable) strain. For one-dimensional problems the material properties that are required to completely define the elasto-plastic behavior are obtained from a uniaxial tension test. However, for a general state o f three-dimensional stress, the material parameters required to define the elasto-plastic behavior include: 1) a stress-strain relationship before the onset of plastic deformation, 2 ) a yield surface, 3) a flow rule and 4) a strain hardening rule. For a general state o f stress, the yield criterion determines the stress level at which plastic deformation begins. For metals, the von Mises criterion is an adequate representation o f structural behavior. The von Mises criterion states that yielding occurs when the magnitude o f the equivalent octahedral shear stress,
cre, reaches a critical value given by:

147

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

After initial yielding, the material behavior will be partly elastic and partly plastic. The plastic strain increment is proportional to the stress gradient o f a quantity called the plastic potential. This relationship between the plastic strain and the stress gradient is also called a flow rule since it governs the plastic flow after yielding. The hardening rule defines how the yielding surface changes when further plastic deformation occurs. In the case o f an elastic perfectly plastic material, the yield surface has a fixed size and a fixed position. If the subsequent yield surfaces are uniform expansions o f the original yield surface without translation as shown in Fig. 4.5, the strain hardening is called isotropic strain hardening. Kinematic hardening includes the Baushinger effect where the total stress range from tension to compression is always equal to twice the initial yield stress. In this case, the subsequent surfaces preserve their shape and orientation but translate in the stress space as a rigid body, Fig. 4.5. ANSYS provides several options for describing plastic behavior. The multi-linear isotropic hardening (MISO) option was chosen in the present study. This option is not recommended for modeling structural steel subjected to cyclic loading, however, it is recommended for large strain analysis. The MISO option can contain up to 100 stress-strain data points. This option also uses the von-Mises yield criterion coupled with an isotropic work hardening model. To activate the geometric and material nonlinear capabilities for the elements that support them, the ANSYS command [NLGEOM, ON] must be issued.

148

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

O '2

j ;

_ S ubsequent yield surface, K inem atic hardening

/
i i '

< i i
...
. / ' / ' / / /

/ ' f

>

i '

In itial y ield surface

/ y
/

S ubsequent y ield surface, Isotropic hardening

Fig. 4.5. Isotropic and Kinematic Hardening.

4.4.5 Iterative Solutions


In static time-independent analyses, although the time does not have any effect, the program still uses time as a tracking parameter. In nonlinear static analyses, the load history is divided into load steps that are defined as a set o f loads applied over a given time span. Each load step is divided into sub-steps; these sub steps are used to apply the loads gradually so that an accurate solution can be obtained. Each sub-step is divided into equilibrium iterations that are defined as additional solutions calculated at a given sub-step for convergence purposes. The general finite element discretization process yields a set o f simultaneous equations. These equations are usually solved for the main variables (usually the degrees o f freedom). When geometric and

149

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

material nonlinearities exist, the equilibrium equation becomes nonlinear and can not be solved directly. An iterative solution technique is used to achieve a converged solution. ANSYS uses the Newton-Raphson method, an iterative process for solving nonlinear equations. The general nonlinear solution algorithm usually commences with the converged solution from the previous load increment, a load or displacement increment is then applied and the solution at the end o f this increment is sought. The Newton-Raphson method evaluates the out-of-balance load vector that is defined as the difference between the restoring forces (the loads corresponding to the element stresses) and the applied loads. Using this out-of-balance load, the program conducts a linear solution and checks for convergence. I f convergence does not occur, the out-of-balance load is recalculated, the tangent stiffness matrix is updated and the process is repeated until convergence takes place. So, the Newton-Raphson method can be defined as a technique for reducing the out-of-balance load vector until it is equal to zero or close to zero depending on the convergence criteria. When the stiffness matrix is updated at every iteration, the process is termed a full Newton-Raphson solution procedure. However, the modified Newton-Raphson solution procedure where the stiffness matrix is updated less frequently or the initial stiffness procedure where the stiffness matrix is not updated can alternatively he used. The modified and initial-stiffness Newton-Raphson procedures converge more slowly than the full Newton-Raphson procedure, but they require fewer matrix reformulations and inversions. The level o f nonlinearities of the modeled problem dictates the choice o f one option over the other. The higher the level o f nonlinearity the more frequently the tangent stiffness matrix needs to be updated. The full Newton-Raphson procedure was used in the current analyses. ANSYS uses the square root o f the sum o f the squares (SRSS) method for satisfying convergence. The SRSS for all element out-of-balance load are compared to the SRSS o f the applied load multiplied by a certain tolerance (default tolerance = 0 .001 ), if convergence is not satisfied, another iteration is performed. If convergence is not achieved within a specified number o f iterations (default o f 25 iterations) the problem either quits or continues to the next sub-step.

150

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

Force or displacement incrementing options for loading are available in the program. In the present study, the bridge was loaded by force increments that represent either the segments erection or wet concrete placement. By using the force incrementing loading option, the Newton-Raphson method is not able to track th e model response into the post peak range, so, an alternative scheme, the arc-length method is provided. T he arc-length method causes the Newton-Raphson equilibrium iterations to converge along an arc as shown in Fig. 4.6, and thus helps convergence even when the slope o f the load-deflection curve becomes zero or negative. The program calculates the reference arc-length radius from the load increment o f the first iteration o f the first sub-step using the following formula:

T o ta lL o a d r l ~ n
Substeps

Using more sub-steps results in a longer solution time. The solution starts with a guess o f the desired number o f sub-steps. If the arc-length analysis is not successful, the arc-length radius can be traced to be either too large or too small and then the reference arc-length radius can be adjusted by controlling the number o f sub-steps.

151

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

Spherical Arc

Converged Solutions

r t - The reference arc-length radius Fg - S u b seq u e n t arc-length radii

u
Newton-Raphson Method Arc-Length Method

Fig. 4.6. Traditional Newton-Raphson Method vs. Arc-Length Method Diagrammatic Descrip tion. [4]

4.5 Finite E lem ent M odeling o f H ouston Bridge

Based on the previous discussion, it was concluded that shell 181 is the more suitable choice for m odeling the present problem, therefore, shell 181 was used to model webs, bottom and top flanges, as well as longitudinal and transverse stiffeners and solid diaphragms. Trapezoidal elements were used to model the solid diaphragms and the top flange width transitions, while rectangular elements were used to model the remaining uniform plated members. To determine the appropriate transverse mesh density required for modeling the present problem, the erection o f segment 905 was simulated using two models with different transverse discretization. In the first model, four elements were used for the top flanges, eight elements for the webs and six elements for the bottom flange, while for the second model, two elements were used to model the top flanges, and four elements were used for the webs and the bottom flange. The mesh density in the longitudinal direction was chosen for both models by controlling the element aspect ratio to be less than 2. The effect o f the varied mesh density on girder stresses and axial forces in the bracing members was negligible, so, the transverse

152

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

discretization o f the second model was deemed adequate and was used in the erection and construction phases o f the bridge as pictured in Fig. 4.7. Three-dimensional truss elements (Link 8) were used to model the top lateral truss. In the X-type truss provided in the middle span, no joint was used at the intersection of the two members to avoid the geometrical instability that may be present from connecting multiple truss members to a single joint. External and internal K-frames were modeled using a combination o f truss and beam elements. Beam elements were used to model the upper struts to avoid geometrical instability, while truss elements were used to model the diagonals. Each diagonal was connected to the juncture between the web and the bottom flange and fram ed to the middle of the upper strut at the other end. At each support, the external and internal box girders were each supported by a pot bearing at the middle o f the bottom flange; tangential movement was only prevented at bent 18 (Fig. 4.1), while the radial m ovement was prevented at all bearings. All supports were modeled using zero displacement constraints. To model these radial and tangential m ovements restraints, the coordinate systems o f the nodes at the supports were rotated to coincide with the global cylindrical coordinate system. The self-weight o f the steel girders, diaphragms and the cross-frames were considered by specifying a mass density and an appropriate gravity constant. The erection o f segments 905 and 907 was simulated using the elements birth and death feature contained within the ANSYS program. This feature is useful in modeling the erection sequence o f the bridge that involves the deactivation (death) and reactivation (birth) o f the segments elements. The program kills the elements by multiplying their stiffness contribution in the global stiffness matrix by a small constant, and at the same time the contributions o f the killed elements to the load vector are eliminated. Then, the element contribution to the global stiffness matrix, mass matrix and load vector is recovered when it is reactivated. Using this feature, the segment erection is modeled by first deactivating its elements and then the elements are reactivated to account for its erection. Because the reactivated elements follow the deformed configuration o f the structure, the large deflection effect should be activated to ensure that the element stiffness is always calculated using its current global location.

153

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges

Fig. 4.7. Typical Finite Element Discretization for the Houston Bridge.

After successful deactivation and reactivation o f the segments elements and since the goal of this study is to assess the design safety for every erection phase, the total weight o f segments 905 and 907 were divided by the length o f the corresponding segment, and the resulting distributed loads were applied evenly to the top flanges as shown in Fig. 4.8. The modified slab construction plan is illustrated in Fig. 4.4. During construction, the loads applied to the box girder consisted o f the self-weight of the girder, gravity load o f the w et concrete, permanent metal

154

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

deck forms and miscellaneous construction loads. In the finite element model, all these loads were applied to the middle o f the top flanges as distributed linear loads. The total loads applied to the girders were proportioned between the top flanges based on the configuration o f the bridge deck. Fig. 4.9 shows the distributed loads that were used to simulate the weight o f the concrete slab that was estimated by the University o f Houston research team [50]. Unfortunately, specific information regarding the properties of the materials used in the Houston bridge was not available, so assumptions were made. Both elastic and plastic material properties were required. The following elastic material properties were assumed in the present study: 1. 2. Modulus o f elasticity, E= 29000 ksi. The density is required to include the self-weight through the creation o f the mass matrix, a value equal to 490 lb/ft3 was used. 3. A common value equal to 0.3 was used for the Poissons ratio. The plastic properties o f the structural steel was modeled using ANSYS multi-linear isotropic hardening plasticity model (MISO) where the stress-strain curve is represented by a number o f straight line approximations. These lines are entered though a table o f points. These points represent the yield point, the stress at the onset o f strain hardening and the ultimate stress and their corresponding strain values. With these three points, a tri-linear stress-strain curve to model the initial elastic response, yield plateau and the initial strain-hardening region can be constructed and shown in Fig. 4.10. The yield stress and ultimate stress of 50 ksi and 65 ksi were respectively assumed in the present study. With these modeling choices and property assumptions, the erection o f segments 905 and 907 and the construction process o f the concrete deck using this three-dimensional finite element model are discussed next.

155

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.,

(a) Erection of Segment 905: 1.062 kips/ft


I I * 1 M I

Segment 905 Bent 17 Bent 18

(b) Erection of Segment 907:

Mill

1.278 kips/ft

3 -L Tmrmf?
Bent 17 Bent 18 Segment 907 Bent 19 Bent 20

Fig. 4.8. Models Used to Simulate the Erection of Segments 905 and 907. [50]

Interior Side ^ 1.378 kips/ft

Exterior Side I 1.502 kips/ft

Center of Curvature

Fig. 4.9. Distributed Loads Used to Simulate the Weight of Concrete Slab. [50]

156

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

0.005

0.01

0.015 Strain

0.02

0.025

0.03

Fig. 4.10. Typical Modeling of Stress-Strain Curve Including Strain Hardening.

4.5.1 Erection of Segment 905


The erection o f segment 905 is illustrated in Fig. 4.11. It was assumed that segment 906 was already erected and spliced to the external girder. Segment 905 was lifted using two cranes, and the segment was bolted while the cranes held the segment. The first loading stage is based upon considering the self-weight o f the exterior and interior girders without segment 905; this was conducted by deactivating segment 905 elements. The load-vertical deflection curve for the loading sequence is illustrated in Fig. 4.12. The abscissa o f the curve represents the vertical deflection at the free end o f the interior girder. A t this stage o f loading, the free end deflected upwards (negative deflection). The second load stage simulated segment 905 erection. This was performed by reactivating the segment elements. It can be noticed that the deflection of the free end o f the interior girder at the time o f segment 905 erection was approximately equal to zero. At this level o f loading, the girder longitudinal stresses and the axial stresses in the bracing and

157

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

cross-frame m em bers were fairly low. For instance, the external top flange o f the exterior girder experienced a m aximum compressive longitudinal stress o f 7.8 ksi at the middle o f the span, while the maximum axial stress in the bracing and cross-frame members was about 2 ksi. As mentioned previously, the upper struts o f the cross frames were modeled with beam elements to avoid instability that results from connecting m ultiple truss members to one joint. It was found that these beam elements behave like axial elements w here their bending and torsional stiffnesses have little effect on the response at this level of loading. Since the goal o f the present study was to assess the design safety o f every erection and construction phase, the loading was continued until a limiting load was attained. The loads were applied to the bridge as linear distributed loads at the top flanges of segment 905. These distributed loads were calculated by dividing the total weight o f segment 905 by its length and were applied evenly to the top flanges. At a load equal to 7.9 times segment 905 load, the top flange in the transition width region o f the interior girder yielded in tension. At this level o f loading, the maximum axial tensile stress was 21 ksi in the bracing diagonal o f the panel near Bent 18. However, the axial stress at the modified members o f the exterior cross frame was 15 ksi. As the load was increased further, the bottom flange at a region before the start point of the longitudinal stiffener started to buckle. This behavior demonstrated the significant effect of the longitudinal stiffener on the buckling strength o f the bottom flange. The bridge reached its ultimate strength at a load equal to 11.2 times segment 905 load with a significant out-of-plane buckling deformation at the bottom flange. Fig. 4.13 illustrates the local buckling mode of failure of the bottom flange and the attached external and internal webs. The bottom flange buckled upwards while both webs buckled outwards. The large factor on the applied weight to cause failure at this erection phase was expected because o f the relatively low self-weight o f segment 905. To investigate the interactive effects and the distribution o f loads between the exterior and interior girders, a single girder model of the interior girder was built. The radial movement o f the girder web at bents 17 and 18 was prevented to simulate the external cross frame and solid diaphragm restraints as

158

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

shown in Fig. 4.14. The single girder model was subjected to the same loading sequence that was previously applied to the two-girder model. Fig. 4.12 illustrates the load-vertical deflection curve for the loading sequence. It can be seen that the behavior o f the single girder model is similar to the two-girder model and the ultimate load factor of the single m odel is close to the ultimate load factor o f the two-girder model. The load-deflection curve o f the two-girder model was shifted vertically and this vertical shift is equal to the self weight o f the exterior girder. This behavior indicates that there is only a little load distribution between the girders, and this is attributed to the high stiffness o f the external cross frame and solid diaphragm at Bents 17 and 18.

4.5.2 Erection of Segment 907


The erection o f segment 907 is illustrated in Fig. 4.3. Segment 907 was dropped in to complete the continuous interior girder. The loading sequence for the erection o f segment 907 followed the same pattern used in the erection o f segment 905. To support the ultimate analysis o f the two-girder model, large computer resources are required. Since it was concluded in the previous section that the exterior cross frames and solid diaphragms were stiff enough to prevent the rotation at the supports, only the interior girder model is considered in this and the following sections. The point at the middle o f segment 907 length experienced the maximum vertical deflection when the bridge attained its ultimate strength. Fig. 4.15 shows the load-vertical deflection history of segment 907 erection. Segment 907 erection was simulated in the same manner; the segments elements were killed and later reactivated. After erection, the top flanges in the middle span and in the transition width region experienced a maximum longitudinal compressive stress o f -8 ksi, this maximum stress was located in the region where the top flange has the maximum width-thickness ratio (b/t) o f 24. This slenderness value is higher than AASFITO Guide Specifications (2003) [40] limitation for non-composite sections, where the width to thickness ratio should not exceed 23.

159

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

Segment 906

Segment 905

Fig. 4.11. Erection of Segment 905.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

1200

Ultimate load=11.16xSeg.905

-x x---) -,xx.> xx-x


v - x X -X -X -X X

1000

x Ultimate j
load=11.25xSeg.905

800

~o
Q.
600/

400
Self-weight (Seg.905 elements activated) Self-weight (Seg.905 elements killed)

200 i
i

Interior Girder Model

- x - Two Girder Model

r------------------ 0 - + ---

1 ------------------------- 1 ------------------------- '----------------------- 1------------------------"---------------------

- r ...

-20

20

40

60

80

100

120

140

Vertical Deflection (in.)

Fig. 4.12. Applied Load-Vertical Deflection Curve for Segment 905 Erection.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

Undeformed Edges

Fig. 4.13. Deformed Shape of the Houston Bridge (Segment 905 Erection Phase) at Ultimate.

162

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

Interior Side

Exterior Side

Fig. 4.14. Analytical Application of Restraints at the Supports of the Single Interior Girder Model.

163

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges

2000

1600
Se If-weight (Seg.907 elements killed)

9.55xSeg.907 8.46xSeg.907 First Yielding

1200
T5

(0 O
800 BentZO

TS

" E L
<
a

400

Self-weight (Seg.907 element

oint of Max. deflection

10

20

30

40

50

60

Max. Vertical Deflection (in.)

Fig. 4.15. Applied Load-Vertical Deflection Curve for Segment 907 Erection.

At this stage o f loading, the maximum axial stress observed in the bracing and cross frame members was 3.7 ksi. This is viewed as a conservative value. The loading was continued by applying distributed loads on the top flanges o f segment 907. The total weight o f the segment was divided by its length to get the distributed load value. A t a load equal to 7.3 tim es segment 907 load, yielding started at the top flange in the same width transition region and the maximum axial stress observed in one o f the cross frame diagonals was 23 ksi. This yielding caused a slight reduction in stiffness as shown in Fig. 4.15. As the load was increased further, the top flange at the middle o f the interior span started to buckle locally at a load factor o f 8.46. The bridge attained its ultimate strength at a load factor of 9.55 with significant local

164

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

buckling deform ation at the middle o f the interior span in addition to local buckling o f the bottom flanges in the exterior spans that was commenced at this load level. Fig. 4.16 illustrates the local buckling mode of failure in the positive and negative moment regions. At the middle of the interior span, the top flanges at the width transition region were locally buckled and also the webs attached to them were buckled inwards. The mode o f failure o f the bottom flanges in the external spans was similar to the mode o f failure that was seen in segm ent 905 erection. The bottom flanges were buckled locally in the region ju st before the starting point o f the longitudinal stiffener where the slenderness ratio o f the bottom flange was the maximum; the webs also were buckled outward in a similar fashion to that observed in segment 905. In the preceding sections, it was shown that the erection of segments 905 and 907 did not involve a critical situation since the load factor or the design safety was found to be relatively high; this is attributed in part to the low self weight o f these particular segments compared with the total weight o f the bridge. The slab construction is discussed next where it is expected that this phase would involve a more critical case because o f the relatively high weight o f wet concrete.

4.5.3 Slab C on stru ction , Stage 1: Construction o f the concrete slab was divided into three stages as illustrated in Fig. 4.4. The distributed loads used to simulate the weight o f the wet concrete, reinforcement, metal deck forms and the construction workers and equipment are shown in Fig. 4.9. These were estimated by the University of Houston research team [50]. The in-plane shear stiffness of the metal forms m ay significantly increase the torsional stiffness o f the girders. However, in the current design methods, the bracing effects o f the metal forms are usually neglected, so, the metal forms were not considered in the finite element model. The first stage consisted o f applying the self-weight o f the bridge. By specifying a mass density and an appropriate gravity constant, the self-weight o f the bridge is accounted for using a mass matrix. The second load step consisted o f applying the construction loads as a distributed load on the top flanges; this estimated load was proportioned between the top flanges based on the bridge deck layout as shown in Fig. 4.9.

165

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

Undeformed Edges

Fig. 4.16. Deformed Shape of Houston Bridge (Segment 907 Erection Phase) at Ultimate.

166

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

At this level o f loading, the exterior top flange in the middle o f the loaded length between bents 17 and 18 experienced a maximum compressive longitudinal stress of 22 ksi compared to 8 ksi for the case of segment 907 erection. The maximum axial stress in the bracing and cross frame members observed at this stage of loading was 3.5 ksi. in compression. As the load was further increased, the top flange yielded at a load factor equal to 2.45 on the stage 1 load. Fig. 4.17 illustrates the load-vertical deflection curve o f the point that experienced the maximum deflection when the limiting load o f the bridge was attained. At a load factor equal to 3.26, the girder reached its ultimate strength with significant local lateral buckling between cross frames in the panel located in the middle o f the loaded portion between bents 17 and 18 as shown in Fig. 4.18. As noted in the previous section, loading the bridge with segment 907 weight produced a mode o f failure that involved local buckling in the top flange and webs in the middle o f the loaded portion. Flowever, loading the bridge with construction stage 1 load produced a different type o f failure that involved a local lateral buckling between the cross frames in the panel located in the middle o f the loaded portion. This dissimilarity in the mode o f failure was attributed to using a different type o f lateral truss system. As mentioned previously, the bracing system in the middle span was o f the X-type, while in the exterior spans the bracing system was o f the single diagonal type (SD-type). This difference in the mode of failure confirms the conclusion that had been reached by the research team at the University o f Houston [50] about the different behavior of these two types o f lateral bracing systems. They concluded that the lateral bending moment o f the top flange is m uch larger in the girder braced with SD-type system than the one braced with the X-type system; the wavelengths o f the lateral bending stresses are twice as large for the SD-type as for the X-type system. Top flange lateral bending in a girder with an X-type system is caused by the horizontal component from the applied load, while in the SD-type system, in addition to the lateral bending caused by o f the horizontal component from the applied load, most o f the lateral bending results from the large strut forces that result from the vertical bending o f the box girder.

167

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges

2500

Ult.Ioad=seIf-weight+3.26 x stage 1 load


2000

1st Yielding.

a 1500

1st concrete pour


Point of Max.defiection

71.4
a, 1000

72,0'

Bridge's self-weight
500

Bent 17

Bent 18

Bent 19

Bent 20

o i
0 2 4 6 8 10

12

14

16

Max. Vertical Deflection (in.)

Fig. 4.17. Applied Load-Vertical Deflection Curve for Slab Construction, Stage 1.

168

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

Undeformed Edges

Internal Girder

Exterior Top Flange

Fig. 4.18. Deformed Shape of the Houston Bridge (Slab Construction, Stage 1) at Ultimate.

169

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4
4.5.4 Slab C onstruction, Stage 2:

Design Safety of Curved Bath-tub Girder Bridges,

When the wet concrete o f stage 2 was placed, the concrete o f stage 1 was expected to have partial strength. Therefore, some composite action must exist between the girders and the slab o f the stage 1 segment. According to AASHTO LRFD 1994 [3], the permanent load that is applied before the slab has attained 75% o f fc shall be assumed to be carried by the steel section alone. Thus, for conservative results, the composite action o f the stage 1 segment was not considered in the finite element model o f slab construction stage 2. Neglecting this composite action softens the structure and results in a conservative estimate o f girder stresses. The self-weight of the girder and the construction loads o f stage 1 and stage 2 were applied sequentially. After applying stage 2 loads, the top flange experienced a maximum compressive longitudinal stress of 31 ksi in the width transition region at the middle o f the interior span. The maximum axial compressive stress observed in the bracing and cross frame members was equal to 12 ksi. The load-vertical deflection curve for the loading sequence is shown in Fig. 4.19. The top flange that had a slenderness ratio o f 21.6 in the width transition region yielded at a load equal to 1.78 times stage 2 load. As the load was increased further, the top flange and the webs in the same region started to buckle locally, this buckling caused a decrease in the stiffness o f the structure until the structure reached its ultimate strength at a load equal to 2.66 times stage 2 load. At ultimate, significant local buckling was developed in the top flanges and webs as shown in Fig. 4.20.

4.5.5 Slab Construction, Stage 3:


The same assumption adopted in the previous subsection was also considered here, i.e. the partial strength of the concrete cast in stages 1 and 2 was not considered when the stage 3 load was applied. The load-vertical deflection history o f the point that experienced maximum deflection when the ultimate strength o f the bridge was attained is shown in Fig. 4.21. The self-weight and the three stages o f construction loads were applied sequentially, then, stage 3 load was increased up to the girder failure. At the end o f the stage 3 loading, the top flange experienced a maximum compressive stress o f 25 ksi in the

170

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

middle o f the interior span and exactly in the width transition region where the flange had a minimum slenderness ratio o f 2 1 .6 .

3000

2500

Ult. Load=self-weight+stagel+2.66 X Stage 2 load

1st Yielding (1.78 X Stage 2 Load)


|

2000

-j

i i !

2nd concrete pour

O 1500

1st concrete pour 1000


i

500

;
| Bridge's self-weight
Bent 17 Bent 18

Bert 19
Point of Max. deflection

Bent 20

10

15

20

25

30

35

40

45

50

Max. Vertical Deflection (in.)

Fig. 4.19. Applied Load-Vertical Deflection Curve for Slab Construction, Stage 2.

171

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

B ent 17

Interior Girder

Undeformed Edges

Bent 20

172

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 4

Design Safety of Curved Bath-tub Girder Bridges,

Fig. 4.20. Deformed Shape of the Houston Bridge (Slab Construction, Stage2) at Ultimate.

At this level o f loading, the maximum axial compressive stress noticed in the bracing and cross frame members was 10.5 ksi. As the load was further increased, the top flange in the exterior span between bents 17 and 18 yielded at a load equal to 2.35 times stage 3 load. The ultimate strength of the girder was reached at a load factor equal to 3.11 o f the stage 3 load with local lateral buckling o f the unsupported length between the cross frames in a similar mode o f failure to that observed in the stage 1 loading (Fig. 4.22). As was expected, because o f the relatively high weight of the wet concrete and the other construction loads, the slab construction involved a more critical loading than did the steel girder erection process. It was determined that the critical load factor for the entire erection and construction processes was the one for slab construction stage 2 and it was equal to 1.78. This load factor m ay increase if the partially hardened concrete o f stage 1 and the stiffening effect o f the metal deck forms were considered in the finite element analysis. AASHTO Guide Specifications (2003) [40] state that a load factor o f 1.4 shall be applied to construction loads when computing actions for constructibility limit state. Even though the critical load factor of 1.78 that was determined in the current study is considered as a reasonable one when compared to the load factor specified by AASHTO, the girder after first yielding still has reserved strength where the girder was failed by local buckling at a load factor equal to 2 .66 .

173

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges.

5000

Ult. Load=self-weight+stage 1 load+stage 2 load+ 3.11 x stage 3 load 4500 J

4000

1st Yielding (2.35 x stage 3 load)

3500

3rd pour

Applied Load (kips)

3000

2500

2nd pour

:ooo
1st pour
.500 -

Bridge's self-weight

1000

tent t
500 Point of Max. deflection

Bert 19

Bert 20

-5

10

15

Max. Vertical Deflection (in.)

Fig. 4.21. Applied Load-Vertical Deflection for Slab Construction, Stage 3.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 4

Design Safety of Curved Bath-tub Girder Bridges,

SD-Type Bracing

Undeformed Edges

X-Type Bracing

SD-Type Bracing Ixterior Top Flange

Fig. 4.22. Deformed Shape of the Houston Bridge (Slab Construction, Stage 3) at Ultimate.

175

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5 Behavior of Horizontally Curved Bath-Tub Girder Bridges during Construction

5.1 Introduction

The clean lines and excellent torsional capacity make the composite horizontally curved bath-tub girder bridges with inclined webs a popular choice for curved alignments. However, during construction, the girder consists o f a quasi-closed section with a relatively low torsional stiffness. Bracing truss systems are installed to increase the torsional stiffness o f the steel section and internal cross frames and solid diaphragms are used to control the distortion of the cross section. Many methods to analyze horizontally curved steel box girder bridges have evolved. Some like the "plane grid" method that is widely used in the design offices require minimal modeling efforts. The modeling of the girders and external diaphragms considers only a two-dimensional grid system. Thus, the plane grid does not consider the cell like shape o f the actual cross section, but rather the properties o f the actual section (the moment o f inertia Ix and torsional constant Kt). The top lateral truss system is treated as a fictitious plate using the Equivalent Plate M ethod (EPM) developed by Kollbrunner and Basler

176

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges.

(1969)[54]. The grid method assumes that the girder cross section does not change shape (i.e. no distortion). Therefore, an adequate number o f cross frames is required to minimize distortional stresses. However, in recent years, new grid elements have been developed to account for torsional warping and cross-sectional distortion (Zhang and Lyons 1984)[80]. The web defonnation has a considerable effect on deflections and stresses in curved girders (Brockenbrough 1986)[13], but is not considered in this method. With advances in the computation technologies, the somewhat more descriptive finite element method is a powerful and versatile analytical tool because it imposes fewer limitations on the geometry, loads, boundary conditions and linear small deflection limits. However, the finite element method does require more effort on part o f the designer.

5.2 Objectives

To facilitate the finite element modeling effort for use by a designer, ANSYS Parametric Design Language (APDL) capabilities are used to develop an analysis/design tool for "Bath-Tub" style curved steel girder bridges. This tool is then used to evaluate the effects o f several important design variables on the response and behavior o f the girders during the construction phase. This study demonstrates the ability o f finite element modeling to assess the stiffness, serviceability performance and ultimate strength o f the curved bridge during construction.

5.3 Finite Elem ent M odel for C onstruction Staging

ANSYS was used to conduct this study. It contains a scripting language called ANSYS parametric design language (APDL) that allows the user to automate common tasks or even to build the whole model in terms o f parameters (variables). The ASCE task committee on horizontally curved steel box girder bridge conducted a comprehensive survey (Heins 1978) [47] on the details o f box girder bridges (straight and curved). This old survey showed that the typical cross section o f the steel-composite bridge in North America contains two bath-tub girders and this remains a popular configuration.

177

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

The box girder bridge model developed in this study also consists o f two steel girders o f trapezoidal cross section "Bath-Tub"; these girders are connected with each other by external solid diaphragms at support locations. The bridge cross section is shown in Fig. 5.1.

Fig. 5.1. Cross-Section of the Box Girder Bridge Model.

Shell elements were used to model the webs, the bottom flange, the top flanges and the external and internal solid diaphragms at the support locations, while truss elements were used to model the cross-frame and top bracing system. Four shell elements were used across the web depth and bottom flange width, while two shells were used to model the top flanges. The mesh discretization in the longitudinal direction was left to the user; element length was set as a parameter, but as a rule o f thumb the element length should be chosen to provide an aspect ratio approximately equal to one for each o f the web elements. The shell elements used to model webs and all flanges were rectangular, however, trapezoidal elements were used

178

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges.

for modeling solid diaphragms. The model that was built using (APDL) allows the designer to control the following parameters:
1 . Width (bt) and thickness (tj) o f the top flanges.

2. Width (B b) and thickness (tb) o f the bottom flange. 3. Web depth (h) and thickness (tw). 4. The slope o f the inclined web. 5. The distance between the center o f exterior top flange o f interior girder and the center o f interior top flange of exterior girder (C).
6 . Radius o f curvature (R).

7. Girder length (L).


8 . Cross-sectional area of cross frame diagonals (Acf), top bracing diagonals (Abr) and struts o f the

top bracing system (Aties). 9. Thickness o f internal and external solid diaphragms Oo t ).

10. Cross-frame spacing (x). 11. Elastic modulus and density o f steel. 12. Element length in the longitudinal direction (v). All geometry, boundary conditions, and loading conditions were modeled in the cylindrical coordinate system. Two models were built, one with simple span and the other with two continuous spans. In the continuous model, there is an option to install one longitudinal stiffener with (T) section in the bottom flange, the stiffener geometry and the starting and ending points o f the stiffener along the girder are also considered as parameters. The model also allows the designer to use different top and bottom flange thicknesses in four regions in the longitudinal direction. Trass elements were used for the internal (X type) cross frames. To avoid geometrical instability, no joint was used at the intersection o f the two diagonals. Each diagonal was connected to the web-top flange juncture at one side and to the web-bottom flange juncture at the other. The lateral top trass system used a

179

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges.

single diagonal with a lateral strut at each cross frame location. This means, in the current analysis/design tool, the top lateral panels were set at one-to-one relative to the cross frame spacing and the cross frames have to be equally spaced along the girder. Each girder was supported on radial bearings. In the simple span model, the vertical and radial movements w ere constrained at the web-bottom flange juncture o f both supports while the tangential movement w as only restrained in one. In the continuous model, the tangential movement was only restrained in the interior support. All boundary conditions were modeled by zero displacement constraints. The non-composite cross section o f the bridge was subjected to the self-weight o f the steel girders, diaphragms and cross frames by specifying a mass density and an appropriate gravity constant. The gravity load of the concrete deck and the other miscellaneous construction loads were distributed proportionally along the top flanges. In the simple span model, the whole concrete deck was applied in one stage, while in the continuous model, the concrete deck can be applied in four stages as shown in Fig.5.2. Where, Cj is the arc-length o f the bridge along its centerline from the east abutment to the end o f stage 1, C 3 is the arc-length o f the bridge along its centerline from the east abutment to the end o f stage 3, C 3 is the arc-length o f the bridge along its centerline from the east abutment to the end o f stage 4, C 4 is the arc-length o f the bridge along its centerline from the east abutment to the end o f stage 2.

c2
Cj
S ta g e d --------T
Interior Pier
C3 C4

St

Fig. 5.2. Construction Staging in FE Continuous Model.

180

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5
5.4 Param etric Study

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

The analysis/design tool developed and described in the previous section was used to evaluate the effects o f several important design variables on the response and behavior of the girders during the construction phase. The bridge model chosen for the parametric study consists o f two simple span bath tub girders as shown in Fig. 5.3. Cross-sectional area o f cross-frame diagonals, top bracing diagonals and lateral struts w ere 1360 mm , 5050 mm and 3060 mm respectively.
0 0 0

1 n n cc\

...

.._

j \ T T

380

-- ------- 3 3 6 0 ------- -- ---------- 3 0 0 0 ------- --- 1 7 0 0 / r 2 5 __________ T y _ J T A . ......................... . . . J L T U 19

1450 _L _
-

215

20
1

2275----

2275

All Dimensions in mm.

Fig. 5.3. Cross-Section of the B ridge M odel.

181

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

Typical m aterial properties and densities were used; in the first phase, linearly elastic small deformation behavior was assumed. Fig. 5.4 illustrates the mesh discretization o f the non-composite bridge model as used for the analyses o f the construction loading.

Fig. 5.4. Finite Element Discretization, Non-composite Model.

182

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

The param eters that are considered in this study include: cross-frame spacing, span length, girder depth, top flange width, web inclination and degree o f curvature. The deflection and bending stress o f the curved girder were normalized with the deflection and bending stress o f the corresponding straight girder having a length equal to the arc length of the curved one and with the same cross-sectional dimensions. During construction, the top flanges o f a quasi-closed curved box girder would be subjected to lateral bending. The lateral bending has several contributing sources that are listed here in addition to the torsional warping:

Distortional warping: Warping stresses caused by cross section deformations. The horizontal truss system: Horizontal truss system is provided between the top flanges to increase
the torsional stiffness and reduce warping. This truss system is usually located at the top flange level where high vertical bending stresses exist. Due to compatibility, the truss m ust experience the same strain as the box girder in the axial direction. Accordingly, forces develop in the diagonal braces and struts, which again cause lateral bending in the top flange, even in the absence of torsion. This effect is predicted using threedimensional m odel by Helwig and Fan [50].

Web inclination: Because o f web inclination in the trapezoidal box girder, the load o f the wet con
crete and formwork that is applied vertically on the top flange produces a transverse component. These transverse components act as a uniformly distributed transverse load on the girder flanges. In the present study, the "warping stress" at the top flanges at mid span caused by the various sources mentioned above is presented as a fraction o f the bending stress at that location.

5.4.1 Span Length


Figs. 5.5 and 5.6 illustrate the effects o f span length on the deflection and on the bending stress respectively in the exterior top flange o f the exterior girder for several radii o f curvature. The arc-length o f the bridge along its centerline was varied with different radii o f curvature to investigate its influence on deflection and bending stress o f the curved girder at mid span. The cross-frame spacing was chosen to be equal to 3 m.; it had been determined that this spacing restrains the warping in the top flange and keeps the

183

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

ratio o f w arping to bending stress (crw/ a b) within the allowable limits ( o w/o'b< 0.5) permitted by the AASHTO G uide Specifications (2003)[40], It was found that the span length as an isolated parameter does not have significant effect on the deflection ratio (Acurved/Astraight) nor on the bending stress ratio ( c curved/ a straight). However, the effect is amplified by increasing the degree of curvature (L/R).

5.4.2 C ross-Fram e Spacing


To investigate the effect o f internal cross-frame spacing on the deflection ratio and warping-bending stress ratio o f the top flange, the span length o f the model was chosen to be 24 m. and the number o f cross frames was varied with different degree o f curvature. Fig. 5.7 shows the effects of the number o f cross frame intervals on the deflection ratio (Acurved/A strajgjlt). The deflection was stabilized with four intervals (cross-frame spacing o f 8 m.), while Fig. 5.8 shows the need for more cross-frames to stabilize the warping to bending stress ratio within the acceptable limits indicated by the AASHTO Guide Specifications (2003). The noteworthy influence o f the degree o f curvature (L/R) on the warping-bending ratio is also evident in Fig. 5.8.

5.4.3 G irder Depth


The effects o f the depth o f curved bath-tub girder on the deflection, bending stress and warping stress was examined. The arc length along the bridge centerline was 24 m with cross-frame spacing equal to 4 m. The bridge model was analyzed using girder depth o f 800 mm, 1000 mm, 1200 mm and 1450 mm, these values give span to depth ratio o f 30, 24, 20 and 16.5 respectively. According to the ASCE survey (Heins 1978) [47] on simple span bridges, the maximum span to depth ratio was 30 with an average o f 23. Figs. 5. 9 and 5.10 illustrate the effect o f the varying depth on the deflection and bending stress ratios respectively. It was found that the variation in girder depth does not have significant effect on deflection and bending stress of curved girders as compared to that o f a straight girder having the same length and cross sectional dimensions, however, the girder depth does influence the warping-bending ratio as shown in Fig. 5.11.

184

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges

2.4 2.2 2 1curved


1straight 1. 8
1.6
-

R=180 m R=120 m R-90 m . -x - - R=80 m . -X - R=60 m . - - R=48 m .


-

1.4 1.2

18

20

22

24

26

28

30

32

34

36

S p a n L e n g t h (m .)

Fig. 5.5. Effect of Span Length on Deflection Ratio.

1.4
-

------- R=180 m . hi R=120 m . -------R=90 m . x - - - R=8Q m . - x R=60 m . R=48 m .


-X

1.3 . -x
-

curved straight

1.2

X-

-X

18

20

22

24

26

28

30

32

34

36

S p a n L e n g t h (m .)

Fig. 5.6. Effect of Span Length on Bending Stress Ratio.

185

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

L /R = 0 1 3 3 * L /R = 0 .2

3.5

-a L /R = 0 .3

L/R=0.4 2.5
k curved kstraight

L/R=0.5

0.5

No. o f C ro s s -F ra m e In terva ls

Fig. 5.7. Effect of Cross-Frame Intervals on Deflection Ratio.

5.5 5 4.5 4 3.5 3

L/R=0.1 33

* L/ R = 0 .4 * L /R = 0.5

(7

2.5

2 1.5 1 0.5
0

-0 .5 No. o f C r o s s - F r a m e I n t e r v a l s I n t e r v a l s

Fig. 5.8. Effect of Cross-Frame Intervals on Warping-Bending Stress Ratio.

186

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges.

Increasing girder depth tends to decrease cross frame stiffness, which in turn increases the distortional longitudinal stress. Nonetheless, this increase is considered as minor compared to the effect o f the cross frame spacing.

5.4.4 W eb Inclination:
Trapezoidal box girders (with inclined webs) result in a narrower bottom flange; and offer several advantages over a rectangular cross-section. The narrow flange provides plate stockiness in the compression area o f a continuous bridge, and these bath-tub girders are more aesthetically pleasing. The AASHTO Guide Specifications (2003) indicates that the web inclination o f the webs relative to a plane normal to the bottom flange should not exceed 1 to 4. The bridge used in the web inclination investigation was 24 m. long with cross-frame spacing equal to 4 m. and the girder was 1450 mm deep. Figs. 5.12 and 5.13 show the effect o f web inclination angle on the deflection and bending stress ratios. As in the case of girder depth, it seems that the web inclination does not have considerable effect on the deflection and bending stress ratios. The web inclination introduces transverse forces components at the flanges that create lateral bending in the top flanges. For this reason, the effect o f web inclination angle on the warping to bending stress ratio was evident in Fig. 5.14. However, within the AASHTO limitation (inclination angle not to exceed 14.5), the effect can be considered insignificant.

5.4.5 Top Flange W idth:


Bridge models with varying flange widths and curvatures were used to investigate the effect o f top flange width on the deflection, bending stress and warping to bending stress ratios. The effect o f top flange width on deflection and bending stress ratios is illustrated in Figs. 5.15 and 5.16 respectively. It is apparent that the top flange width does not affect the deflection and bending stress in a curved system compared to those in a straight system, while it has a more prominent influence on the warping to bending stress ratio as shown in Fig. 5.17. Increasing top flange width causes a significant reduction in the warping stresses for all levels o f curvatures. Increasing the top flange width also increases the moment of inertia with respect to the web, and thus decreases the stresses caused by lateral bending.

187

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

1.7
1.6

1.5 1.4 1
cu rv e d s tr a ig h t

-# LI R=0. 2
HU LI R ^ O ^

1.3 1.2
1.1
1

800

1000

1200

1400

1600

G i r d e r D e p t h (mm. )

Fig. 5.9. Effect of Girder Depth on Deflection Ratio.

1.2

1.18 4 1.16

curved straight

1.14
1.1 2 1.1
-

-L /R = 0 .2 - U R = 0 .4

1.08 J 1.06 800 1000 1200 1400 1600 G ird e r D e p th (m m .)

Fig. 5.10. Effect of Girder Depth on Bending Stress Ratio.

0. 7

0.6
0. 5

LI R=0.4
8 00

1000

1200

1400

1600

G i r d e r D e p t h ( mm. )

Fig. 5.11. Effect of Girder Depth on Warping-Bending Stress Ratio.

188

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges.

1 .5 1 1 ---------------------- 1 1 (| ^ c u rv e d ^ s tra ig h t 0 5 0 0 4 8 12 16

ij|

# # LI R = 0 .2 ... M..... LI R = 0 . 4

20

24

28

Web I n c l i n a t i o n Angle ( D e g ree s )

Fig. 5.12. Effect of Web Inclination on Deflection Ratio.

1.4 1.2

1
0.8
^ c u rv e d s tra ig h t 0 .6 0.4

U R=0. 2
-L/R=0.4

0. 2 H

0
4 8 12 16 20 24 28 Web In c lin atio n Angie ( Degrees)

Fig. 5.13. Effect of Web Inclination on Bending Stress Ratio.

0.8
0.6 0.4 L / R=0. 2 L /R = 0 .4

12

16

20

24

28

Web I n c lin atio n Angle ( Degrees)

Fig. 5.14. Effect of Web Inclination on Warping-Bending Stress Ratio.

189

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

1.5

curved Astrai!ght
L/R=0.133 0 .5 L/R=0.2

-* U R=0.4
200 250 300 350 400 450 500

T op Fla n ge Width (mm.)

Fig. 5.15. E ffect o f Top Flange W idth on Deflection Ratio.

1.4
1.2 1

0.8 cu rv ed 0.6

straight

0.4
0 .2

L /R =0.133 L /R =0.2 L /R = 0.4

0
200

250

300

350

400

450

500

T o p F la n g e W idth (m m .)

Fig. 5.16. Effect o f Top Flange Width on Bending Stress Ratio.

T op Fla n ge W id th (mm.)

Fig. 5.17. Effect o f Top Flange Width on W arping-Bending Stress Ratio.

190

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

As a summ ary o f the various parametric effects on the curved bath-tub systems, it was concluded that span length, degree of curvature and cross-frame spacing have considerable effect on deflection and bending stresses in the curved system as compared to the corresponding straight one. Also, the degree o f curvature, cross-frame spacing and top flange width have a significant influence on the warping to bending stress ratio. To accentuate the above conclusions, the influence o f these parameters on the ultimate behavior o f curved bath-tub girder bridges is evaluated next. Both the material and the geometric nonlinearity of the structure are considered.

5.5 Param etric Effects on Ultim ate Behavior

Improved numerical methods, such as the finite element method, and the increase in the available computational resources, enable us to perform incremental stress-strain analyses and study the effects o f different parameters on the buckling behavior and the ultimate strength o f structures with complex geometry like the curved bath-tub girder bridges. Reliable and computationally efficient shell elements with material and geometrical nonlinear capabilities provide the necessary tools for performing these tasks. Analytical studies (Simpson 2000 [68 ], Sennah 1998 [66 ]) have used the nonlinear finite element modeling that led to successful comparisons o f experimental and analytical load-deformation behavior and ultimate strength magnitudes for both shear and flexure. A large library o f shell elements is available in ANSYS and other finite element programs; they posses broadly different analytical capabilities. In the parametric study presented earlier it was expected that the bridge would behave linearly, so, the element used was shell 63 (elastic 4-noded shell element). However, shell 181 is used in the current ultimate behavior investigation while using the same cross-sectional dimensions of the earlier model. This element is 4-noded element with 6 degrees of freedom per node and has large strain, large deflection and enhanced convergence capabilities. M aterial stress-strain relations are characterized with a multi-linear isotropic hardening material option. An elastic perfectly plastic stress-

191

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

strain curve w as assumed and the material properties for the steel consisted o f a density o f 7850 kg/m3, a modulus o f elasticity o f 200000 MPa, a Poisson's ratio o f 0.30 and a yield stress of 345 MPa. ANSYS adopts the Newton-Raphson approach for solution o f nonlinear problems. Force or displacement loading options are available. To trace the post-buckling response of the structure, the displacement loading option is the suitable one, however, the use o f this option is not possible because the loading here represents either the self weight o f the girder or the w et concrete load or both o f them. In the force incremental loading, the Newton-Raphson method is not able to track unloading (post-buckling response), therefore, an alternative scheme, the arc-length method is provided to trace the response past the limit load. The arc-length method causes the Newton-Raphson method equilibrium iterations to converge along an arc, and thus helps convergence even when the slope o f load-deflection curve becomes zero or negative. However, the arc-length method requires several trials to adjust the reference arc length radius (using No. o f Substeps) to obtain a solution at the limit load. Otherwise, it might more convenient to use the Newton-Raphson iteration with automatic bisection to the load steps if the goal is only to determine the values o f nonlinear buckling loads. In the current investigation, the arc-length method was used to trace the post-buckling response. With particular reference to the effects o f span length, top flange width, cross-frame spacing and degree o f curvature, the ultimate behavior o f the bridge is discussed next.

5.5.1 Span Length:


The length o f the bridge along its centerline was varied with different radii o f curvature. To isolate the span length effect, the cross-frame spacing o f 3 m was kept constant. The load applied to the bridge, including its self-weight, versus the vertical deflection history o f the point experiencing the maximum vertical deflection when the bridge attains its ultimate strength is shown in Fig. 5.18. These plots indicate that the ultimate strength and the stiffness o f the bridge were decreased considerably by the increase in the degree o f the curvature. Fig. 5.19 shows the effect o f span length on the ultimate strength ratio (PgtraiglV Pcurved) for various degrees o f curvature. The figure emphasizes the conclusion that was also reached in the

192

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

18000 16000

.
14000 12000

L = 18 m.

straight R=18G m. R=120 m.

L= 24 m. s 10000 -

R=60 m.

>-

L= 36 m

<

Q. fi.

6000 4000

100

200

300

400

500

600

700

Vertical Deflection (mm)


Fig. 5.18. Span Length Effect on Load-Deflection Curves.

straight

-* R=60 m .

18

20

22

24

26

28

30

32

34

36

Span L e n g t h ( m .)
Fig. 5 .1 9 . Span Length Effect on Ultimate Strength Ratio.

193

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridge;

elastic parametric study, where the degree o f curvature was found to have more effect on the curved system behavior than the span length.

5.5.2 Top Flange Width:


The ultimate behavior of the bridge was investigated using bridge models with various top flange widths. The bridge used in this investigation was 24 m long with a cross-frame spacing o f 4 m. The applied load versus the vertical deflection curves o f straight models and curved models with various degrees of curvature are displayed in Fig. 5.20. The figure shows the effect of the top flange width on the ultimate strength and stiffness o f the models. It can be seen that the top flange width has a considerable effect on the ultimate strength and stiffness o f the models. Doubling the top flange width from 240 mm to 480 mm, almost doubles the ultimate strength o f the straight girder bridges, however, this increase in top flange width has more impact on the strength o f curved bridges.

5.5.3 Cross-Fram e Spacing:


It was found, in the earlier elastic parametric study, that the internal cross frame spacing as a parameter has a significant effect on stabilizing the deflection o f curved bridges and on reducing the warping stresses in their top flanges. Here, a bridge model with a centerline length o f 24 m is used to investigate the effect o f the internal cross frame spacing on its ultimate strength and stiffness. The number o f cross frames was varied for bridge models with different degrees o f curvature. The load-deflection curves for bridge models cross frames are exhibited in Fig. 5.21. For the straight model, it can be seen that the models with 3(6 m. interval), 5(4 m. interval), and 7(3 m. interval) internal cross frames have fairly similar stiffnesses and their load-deflection curves start to bifurcate at a load close to their ultimate strength. This is a confirmation to our earlier conclusion that the girder deflections stabilize quite fast with fewer cross frames, while the ultimate strength o f the girder is significantly affected by the number o f the cross frames. With the presence o f curvature, this behavior exists to a small extent except that the curvature induces more flexibility to the structure and provides a need for more cross frames (compared to the straight girder case) to stabilize the deflection. For instance, as mentioned above that the straight girder bridges with 3, 5, and 7

194

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 2

Litrature Review

E E B
o
B oe
CM

VP

5=

< D
(1 )

O li o r 0)

>
o o o o o o o o o
CO

o Tf

E B

< /)

o & . 13 3

C O C O T C M

e*

o o o
5

o o o
N

o o o
p

o o o
CO

o o o
to

(M M ) p e o i payddy

195

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Fig. 5.20. T op Flange Width Effect o n Load -Vertical Deflection Curve.

o o

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

cross fram es have almost the same stiffness, while in the highly curved bridge with L/R=0.4, the model with 3 cross frames has significantly lower stiffness than those with 5 and 7 cross frames. The ultimate strength of the curved bridge with 3 cross frames is also reduced significantly compared to those with 5 or 7 cross frames.

5.6 Modes o f Failure

The lateral buckling behavior o f multiple curved box girder bridges can be classified into two distinct types: overall lateral buckling and local lateral buckling of the girder portion between the cross frames. The modes o f failure o f the bridge models used in the current investigation were examined by observing the deformed shape at ultimate load. The deformed shape o f the curved bath-tub models show that the overall lateral buckling is not an important issue, however, the dominant mode o f failure was the local lateral buckling o f the external top flange o f the external girder between the cross frames o f the panels close to mid span. A bridge with a degree of curvature equal to (L/R=0.2) and a cross frame spacing o f 4m, was chosen as a typical model to examine mode o f failure. Fig. 5.22 shows the load-vertical deflection curve for this bridge. The bridge initially behaved linearly, and as the load increased, the external top flange o f the exterior girder at the central region o f panels close to m id span started to lateraly buckle. At a load equal to about 4 times the construction load, the inside tip o f the external top flange o f the exterior girder yielded in compression; this yielding caused a slight decrease in the stiffness as shown in Fig. 5.22. As the load was increased further, the lateral displacement at the middle of the panels increased significantly and the maximum lateral displacement alternated between the middle o f the panels located to the left and to the right of mid span. At ultimate load, (equal to 4.62 times the construction load), the lateral displacement at the middle o f the panel right to mid span was 47 mm. Fig. 5.23 displays the local lateral buckling deformation of the external top flange o f the external girder for panels near mid span.

196

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

12000

Straight
10000

L/R = 0.2

8000 6000

4000

2000

100
12000

100 200 200 VERTICAL DEFLECTION (mm)

300

400

10000

L/R = 0.133

UR = 0.4

8000 J 6000 a. 4000

2000

100

200

100

200

300

400

VERTICAL DEFLECTION (mm)

Fig. 5.21. Cross-Frame Spacing Effect on Load-Vertical Deflection Curves.

197

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

12000 Significant Local Lateral Buckling First Yielding


8000

10000

8
"O Q > o.
<

6000

4000

2000

L=24 m. R=120m . L/R=0.20 Cross-Frame Spacing=4 m.

50

100

150

200

250

300

Vertical Deflection (mm.)

Fig. 5.22. Load-Vertical Deflection Curve for Model with (L/R=0.2)

198

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

.,
. v w , un :" t n " r G " the External vveu ,v a . s ig . 5.13. Bocal ^ ea lB < = U iSOt

r fo rV ad st< e a r M i4 S ' ,an'

199

, W edv<ho u , p e ^ -

.
f the copVr'9w 0VJ duced vJilP peonissioo

^ rth e tre p v o d o c lW P '0

C hapter 5

Behavior of Horizontally Curved Bath-Tub Grider Bridges,

5.7 Sum m ary and Conclusions

In the present study, the benefits o f the finite element method were realized through the use of the ANSYS param etric design language (APDL) capabilities to develop an analysis/design tool for "BathTub" style curved steel girder bridges. The model was built in terms o f parameters that represent the bridge and girders geometry, cross-sectional area o f cross frames and top bracing truss members, thickness of internal and external solid diaphragms, cross frames spacing and construction sequence loading scheme. This analysis/design tool was then used to evaluate the effects o f several important design variables on the response and behavior o f the girders during the construction phase. Design parameters included cross frames spacing, span length, top flange width and degree o f curvature. Also, the influence o f these parameters on the ultimate behavior o f curved box girder bridges was evaluated by considering both material and geometric nonlinearities o f the structure. The parametric study demonstrated the ability o f FE modeling to assess stiffness, serviceability performance and ultimate strength o f the bridge at the state of construction. It was concluded that the parameters that most significantly effect the ultimate behavior of the structure were the degree o f the curvature and the cross frames spacing. The type o f failure that dominates the ultimate behavior during construction is the local lateral buckling o f the external girder between the cross frames.

200

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 6 Summary and Conclusions

Horizontally curved girder bridges are considered as a practical and aesthetic solution for the severe geometric restrictions in highly congested urban areas or construction sites. For a curved highway align ment o f moderate span, two types o f steel cross sections are currently being used: an open cross section that usually consists o f a number o f steel I-girders interconnected by cross frames with a composite con crete deck slab. The other section that is increasingly in use is a closed section consisting o f one or more separate box girders. Because o f the high torsional stiffness o f the closed box cross section, the torsional moment induced by the curvature can be resisted with less transverse bracing than in the case o f the Ishaped girders. The behavior o f the box girder system is more complex than that o f the I-girder system. The majority o f the experimental and analytical research efforts have focused on I-girder systems, and thus, their behav ior is relatively well understood. However, for the box girder systems, a significant need for more research to establish sound design criteria especially in the construction phase was identified. With the increase o f speed and availability o f computational resources, significant additional research and development are required to bridge the transition to the use o f advanced analytical techniques in the

201

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 6

Sum m ay & Conclusions

design and construction processes. The finite element analysis method and through ANSYS (a commer cially available software) is used to investigate the behavior o f horizontally curved bath-tub girder bridges during construction. The research reported in this thesis is aimed to demonstrate the effectiveness of the finite element method in the evaluation o f the stiffness, the buckling behavior and the ultimate strength of such structures during all phases o f steel construction and to provide a useful advanced application for design office implementation. The review o f the literature presented in chapter 2 commenced with a general description o f the behav ior of curved box girders with emphases on the behavior during construction. This was followed with a dis cussion about research related to girder erection and construction and the different methods that are commonly used in the static analysis o f curved box girder bridges. Many experimental studies have been conducted to investigate the curved steel plate girder while only few studies related to similar bridges using box girders. These few experimental studies were divided into field test experiments and model test exper iments and were reviewed in chapter 2. Much experimental and analytical information concerning nonlin ear behavior to collapse is found for straight box girders, but far less is available for horizontally curved box girders; the few available studies were reviewed. Finally, a review o f the design aids, specifications and standards for curved box girder bridges was presented at the close o f chapter 2 . In chapter 3, to compare the correlation o f the field test measured results with the estimated results and to provide the designer with more insight into the behavior o f the curved bath-tub girder bridges, the field testing during construction and service loading o f a curved box girder bridge located near Baltimore, Maryland was re-examined using linear elastic three dimensional finite elem ent modeling. Comparisons were made between the finite element results, finite difference results and the field test measurements. In chapter 4, the finite element method was used to develop a nonlinear m odel for the structural evalu ation during the erection and the construction o f a three span continuous curved box girder bridge in Hous ton, Texas. This bridge was chosen as an example that reflects current fabrication and detailing practices.

202

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & Conclusions

The level o f safety and reliability o f the bridge that would be available at various stages of the erection and construction processes was evaluated by detailed nonlinear finite element modeling. ANSYS Parametric Design Language (APDL) was used in chapter 5 to develop an analysis/design tool for the bath-tub style curved steel girder bridges during construction. The objective of the use o f this tool is to employ advanced analytical techniques such as the finite element method in the analysis and design process and thus make this method more popular and easy to use by a designer. This tool was then used to evaluate the influence o f several important design parameters on the response o f the girders during the construction phase. Subsequently, this tool was used to assess the influence o f selected design variables on the buckling behavior and ultimate strength of horizontally curved bath-tub girder bridges during con struction. A summary o f the major results, observations and conclusions obtained from the current study is pre sented in section 6.1 and the principal achievements o f the current research are outlined in section 6 .2 . Finally, recommendations for future work are proposed in section 6.3.

6.1 Sum m ary o f R esults, O bservations and Conclusions.

This summaiy is divided into four sections, based on the main topics o f each chapter in the thesis; therefore, section 6.1.1 is related to the general observations made about previous literature. In section
6 . 1.2 , main conclusions pertaining to the element choice in the finite element m odel and the correlation of

the finite element results with the measured and finite difference results are presented. This is followed by section 6.1.3 that primarily deals with the nonlinear finite element m odeling and the evaluation o f the m ar gin o f safety/reliability available for every erection and construction step o f the bridge. Finally, section 6.1.4 states the main conclusions related to the effects o f important design variables on the elastic and ulti mate behavior of horizontally curved bath-tub girder bridges during construction using the analysis, design tool and procedure that were developed or used in this study.

203

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & Conclusions

6.1.1 L iterature Review


A review o f the available literature revealed that m any experimental studies have been conducted to investigate curved steel I-girder systems but far fewer are available for curved box-girder systems. Most of the box-girder experimental studies involve testing o f scaled models and a high percentage of these tests deal with only the elastic behavior o f the girder. Few experiments exist that demonstrate the buckling behavior and ultim ate strength characteristics. The US Federal Highway Administration (FHWA) in col laboration w ith HDR Engineering, Georgia Institute o f Technology and the Auburn University initiated a comprehensive experimental program involving the testing of full-scale, simply supported curved steel Igirder bridge in Virginia to arrive at a complete and thorough understanding o f the I-girder system behav ior during construction. However, to the authors knowledge, no full scale testing of the box girder systems is available and more experimental data related to the ultimate behavior o f the box girder system, espe cially during the construction phase is needed. Also, no attempt has been reported on experimentation of composite box girder bridges to investigate the steel trough and the concrete slab deck behavior at failure, except the unsuccessful attempt in 1985; this w as explained in detail in chapter 2, Daniels [29], It was apparent from the literature review that most o f the available design aids involve empirical design equations that are usually based on parametric studies. Specifically, no attempt has been reported to use a general purpose finite element program such as ANSYS as an analysis and design aid representing the local and global behavior for bath-tub girder bridges. The review also revealed that in addition to the AASHTO Guide Specifications 92003)[40], there is only one other known design specification for horizontally curved steel bridges worldwide and that speci fication is the Hanshin Guidelines, Japan [41], The Canadian Highway Bridge Design Code (CHBDC)(2000)[ 18] adopts AASHTO current design equations that are based primarily on the work from The Consortium o f University Research Team (CURT) project that started at the beginning o f the seven ties.

204

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & Conclusions

6.1.2 L inear E lastic M odeling o f Curved Bath-Tub G irder Bridge Structures


The field testing o f a curved bridge located near Baltimore, M aryland was performed during construc tion and service loading. The bridge had been previously analyzed using two-dimensional finite difference programs. The bridge was re-examined in the current study using a comprehensive three-dimensional finite element model. Two ANSYS shell elements were found suitable for linear elastic analysis, namely, shell 63 (4-node element) and shell 93 ( 8 -node element). To help in choosing the appropriate element for the Baltimore bridge modeling, shell 63 and shell 93 were used to model simple elastic experiments conducted by Fam and Turkstra [34] before the larger modeling was attempted.

6.1.2.1. Fam and Turkstra E xperim ents


The large curvature combined with the presence o f cross section distortion o f the models tested in these experiments were considered as relatively severe tests o f the analytical procedure to produce accurate results. It is for this reason that these experiments were modeled. Generally, the longitudinal stresses and the deflections obtained from the finite element modeling o f the Fam and Turkstra experiments correlated well with their experimental results but this correlation varied with the magnitude o f the results. Larger errors generally occurred with the smaller increments. The differences of the finite element deflections with respect to the experimental deflections ranged from 1% to 11%, while the differences o f the lon gitudinal stresses in the top and bottom flanges obtained from the finite element model with respect to the experimental stresses ranged from 1% to 8 %. The deflections and the longitudinal stresses at the top and bottom flanges o f shell 63 and shell 93 models were almost identical. Thus, it was concluded that using the quadratic shell element (shell 93) has no advantages on accuracy as long as the linear shell (shell 63) extends over 2.5 arc or less in the longitu dinal direction and the more refined the mesh the better the finite elem ent accuracy obtained.

205

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & C onclusions

Both analytical and experimental results showed that the existence o f intermediate diaphragms make the stresses in the top and bottom flanges more uniformly distributed than does the case when no interme diate diaphragms exist.

6.I.2.2. Baltimore Bridge Field Test


Shell 63 w as therefore used to model the Baltimore bridge field test. Comparisons were m ade between the finite elem ent results with the measured results and the finite difference results. The main observations concerning the differences between the finite element modeling and the finite difference m odeling o f the Baltimore bridge are discussed next followed by comparisons between the finite element results, finite dif ference results and measured results. Finite Element and Finite Difference Modeling In the finite difference model, the exact positioning o f the longitudinal cross section variations and intermediate supports was difficult, while in the finite element model, the splices and intermediate loca tions were positioned exactly. The position o f the top flange width transition in the finite elem ent model was also located precisely using trapezoidal shell elements. Heins et.al. [14] estimated that the longitudinal stiffener contribution to the total moment o f inertia of the non-composite section was less than 1%, thus, the longitudinal stiffeners were not considered in their solution. However, the stiffeners were considered in the composite section because o f the shifting o f the centroidal axis. In the current finite element modeling, the longitudinal stiffeners were modeled using shell elements in both non-composite and composite models. In the finite difference model, the top flange truss system was replaced with an equivalent plate o f con stant thickness, while the top flange bracing members were modeled with beam elements and truss ele ments in the finite element model. In the finite difference model, one also assumes no change in the shape o f the girder cross section (i.e. no distortion) and the total longitudinal stress consists o f the longitudinal bending stress and the torsional warping stress. The corrugated steel deck forms were not considered for either model.

206

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & C onclusions

In the finite difference program, the vehicle wheel loading could only be specified on or between the two girders, thus, the vehicle loadings were taken as axle loads rather than wheel loads in lanes where at least one wheel of each axle was outside the area bounded by the two girders, while in the finite element model, the six wheel loads o f the test vehicle were applied to the closest nodes in the slab deck with the help o f ANSYS Parametric Design Language (APDL). C om parisons of Results In the dead load test, correlation between measured deflections with the deflections obtained from the finite element and the finite difference analyses was somewhat difficult and dependent on the deflection value at a specific section. As would be expected, larger errors as a percentage o f the total deflection value occur with lower deflection values. Besides, the measured results were extracted from the report [14] by the digitization o f the plotted data; the digitization process is approximate in nature and some errors result. However, the finite difference model did show stiffer response than did the finite element model. This dif ference in deflection ranged between 1% to 13% and it is attributed to the web deformation effect that is not considered in the two-dimensional modeling. In the dead load test, the average top flange longitudinal bending stress agreed well between the finite element and the finite difference solutions, however, the longitudinal stresses at the tip of the top flanges did not correlate because the lateral bending o f the top flange was not considered in the finite difference solution. The concrete that was placed earlier had enough time to gain some strength and this partial strength likely provided some composite action with the steel girder. Since this partial strength was not considered in the estimated stresses, the analytical solutions tended to overestimate the stresses in sections located in the regions o f the partially set concrete In the live load test, generally, the finite element results had better correlation with the measured results than did the finite difference results. This is attributed largely to the approximate idealization o f the transverse stiffness o f the concrete deck that was explained in more detail in chapter 3.

207

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & Conclusions

The composite concrete slab caused the neutral axis to shift upward to be within the depth o f the top flange or even higher; this resulted in relatively small top flange stresses. For this reason, only the bottom flange stresses were compared. The bottom flange stress distribution obtained from the finite element anal ysis confirmed the report [14] conclusion about the warping stress; the report [14] indicated that the warp ing stresses were negligible. The design stresses were much larger than the measured and estimated stresses because the steel sec tion in the interior support region was assumed to be non-composite even though shear studs were pro vided there. However, full composite action throughout the length was assumed in the analytical solutions. The low experimental and analytical live load stresses in the section close to the interior support appeared to validate the analytical assumption. In general, the two-dimensional analysis produced reasonable results compared to the finite analysis results. However, during construction, the lateral bending stresses o f the top flanges caused by the lateral component o f the applied loading and the lateral stresses caused by the axial forces in the struts that result from the vertical bending of the girder as developed by Helwig and Fan [50] should be added to the verti cal bending stresses and the torsional warping stresses o f the top flanges that are already estimated from the two dimensional analysis. The additional forces induced in the diagonals and the struts of the top truss system because of the vertical bending o f the box girder as developed by H elwig and Fan [50] should also be considered in the two dimensional analysis.

6.1.3 Design Safety o f H orizontally Curved Bath-Tub G irder B ridges during Erection and Construction.
A comprehensive nonlinear three-dimensional finite element model o f a three span continuous curved box girder bridge was developed in chapter 4. The main objective o f the study was to evaluate the level of safety/reliability during the construction process. Since the safety evaluation involves loading the structure up to failure, the element chosen to model such behavior must necessarily possess large displacement,

208

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Chapter 6

Sum m ay & Conclusions

large strain an d non-linear material capabilities to model all aspects o f eventual failure. ANSYS element shell 181 that has all these capabilities was found to be a suitable choice for modeling this problem. Force or displacem ent controlled loading options are available in ANSYS. Although the displacement controlled loading option is more suitable for tracing the post-buckling response o f the structure, the use of this option w as not possible in the current study because the loading here either represents the segments gravity load or the wet concrete load. In the force controlled loading, the arc-length technique was used to trace the response past the limit or buckling points. The arc-length method requires several trials to adjust the reference arc length radius before the appropriate results can be obtained. A particular lateral bracing systems (SD type versus X-Type) m ay influence the mode of failure. The lateral bending moment o f the top flange in a girder braced with SD type system was found to be much larger than the girder braced with X-type system and influenced the failure for these proportions. Thus, the mode o f failure observed in the girder braced with SD type system involved local lateral buckling o f the top flange between the cross frames, while the top flange of the girder braced with X-type system failed by local buckling. The mode o f failure that involved local buckling o f the top flange took place in the transition width region where the flange slenderness was equal to 24. This slenderness value is higher than the maximum suggested value o f 23 in AASHTO. The non-compact flange, though behaved as usual, sustained a stress equal to the yield stress before it failed by local buckling. The slab construction involves a more critical loading case than does the steel girder erection process because of the relatively high weight o f the wet concrete. The load factor determined for slab construction stage 2 was the critical factor for the entire construction process and it was equal to 1.78. The partially hardened concrete o f stage 1 and the effect o f the metal deck forms that were not considered in the finite element modeling may increase this factor, however, taking into account the resistance factor (1.78*0.9 =
1.6 ), this load factor is considered as a reasonable one when compared to AASHTO load factor o f 1.4 that

should be applied to the construction loads in the constructibility limit state, although some conservatism

209

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & Conclusions

still exists. T his analysis produced the critical failure mode and a load factor on the design load that veri fied the adequacy o f the girder with respect to structural safety during this stage o f construction.

6.1.4 Behavior o f H orizontally Curved Bath-Tub G irder Bridges during C onstruction


In chapter 5, the benefits o f the finite element method are realized through the use o f ANSYS Para metric Design Language (APDL) capabilities to develop an analysis/design tool for bath-tub style curved steel girder bridges. Input data is supplied to ANSYS by the user through the input file that is listed in appendix I. A fter the execution o f the input file, the user can visualize the analysis results using ANSYS post-processor. This tool was then used to evaluate the effects o f several important design parameters on the response and behavior o f the girders during the construction phase. Based on this parametric study, sev eral observations and conclusion were made: It was concluded that the span length as an isolated parameter does not have significant effect on: 1) the deflections and bending stresses o f the curved girder as compared to that o f a straight girder having the same length and cross-sectional dimensions, 2 ) on the stiffness and ultimate strength o f the curved girder, however, the effect is amplified with the increase in the degree o f curvature (L/R). It was found that the cross frame spacing has significant effect on the deflection o f the curved girder as compared to that o f a straight girder having the same length and cross-sectional dimensions (deflection ratio); cross frame spacing o f 6 m. was found adequate to keep the deflection ratio constant, however, the influence o f increasing the cross frame spacing beyond this value is significant on the deflection ratio. Also the bending-warping stress ratio o f the curved girder is significantly influenced by the cross frame spacing. It was found that there is a need for more cross frames (smaller cross frame spacing) to keep this ratio within the allowable limit specified by AASHTO Guide specifications [40]; cross frame spacing o f 4 m. or less was needed. The cross frame spacing also has considerable influence on the stiffness and the ultimate strength o f the curved girder bridges. However, it was found that reducing the cross frame spacing beyond a certain limit (4 m. in the current study) has only a slight influence on the stiffness and the ultimate strength o f the girder.

210

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & C onclusions

The effect o f the web inclination angle on the warping/bending stress ratio is insignificant, if it is kept within the AASHTO limitation (inclination angle not to exceed 14.5 with the vertical). It was found that increasing the top flange width caused a significant reduction in the warping stresses and a considerable increase in the ultimate strength as well. On the other hand, increasing the top flange width and keeping the same thickness leads to an increase in the slenderness ratio and this makes the top flange more susceptible to a local buckling. Thus, the designer should compromise these two effects in the design and keep in mind that once the slab deck gains its strength, the top flange effect is minimal. Within the full practical parameter ranges that were considered in this study, it was found that the over all lateral buckling is not an important issue. The type o f failure that dominates the ultimate behavior dur ing construction is the local lateral buckling o f the external girder between the cross frames.

6.2 Principal A chievem ents o f the C urrent Study

1.

Preparing a comprehensive summary o f curved girder research with emphasis on the research directly related to curved box girders; this summaiy was preceded by a comprehensive description o f the gen eral behavior o f the curved box girder bridges.

2.

The designer skepticism about the level o f approximation encountered in various analytical methods was answered; an elastic three-dimensional finite element model was developed. This led to successful serviceability level comparisons between the measured and the finite element deflections and strains. Also, successful comparisons between the two-dimensional modeling trends that are widely used and the advanced finite element modeling were made.

3.

Determining the level o f safety and reliability that would be available at various stages o f the erection and construction processes o f a bridge was demonstrated. The procedure that was used to assess the load factor for the construction sequence loading in the current study can be considered as a perfor mance based design approach, in which the estimated load factor must satisfy the load factor specified

211

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & C onclusions

in the standard and the required structural members performance was identified: they were shown to attain a stress equal to the yield stress before their failure by buckling. 4. An analysis/design tool for bath-tub style curved steel girder bridges during construction was devel oped and demonstrated. This tool can be used as a simple analysis tool; the structure is loaded with the factored construction load and then the stresses, forces and deflections that are obtained from the anal ysis are compared with the specified limits o f the standards. The tool also can be used as a design tool including a performance evaluation for yielding and/or stability. The structure is loaded up to failure and the failure modes are observed; specific failure modes can be easily altered or even avoided by modifying the construction plan or retrofitting the dimensions of a specific member.

6.3 Future Research

There are m any areas related to the behavior o f bath-tub girder bridges that require further study. The analytical/design tool that was developed in the current research should be extended to include the com posite steel box girder behavior during service loading. Parametric studies should be performed using the suggested tool to investigate, independently, the effects o f important design variables such as the radius o f curvature, girder spacing and overhang width on the composite bridge behavior. This analytical study should be supported by experimental data on the composite bridge behavior during service loading and at ultimate strength. The analytical/design tool should also be extended to include other types o f top truss system and other types o f cross frame. In addition, the tool should provide the designer the freedom to insert the cross frames in any pattern relative to the top lateral panels. Although the bath tub girder analysis, as presented here follows the recommendations for modeling that were presented in earlier research on multiple plate girder bridges, a comprehensive experimental pro gram involving the testing o f a full scale curved bath-tub girder bridge similar to the one conducted by the

212

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

C hapter 6

Sum m ay & Conclusions

U.S. Departm ent o f transportation, Federal Highway Administration is required to verify the significant analytical findings from advanced finite element modeling. Particular attention should be given to the influence o f the analytical modeling o f boundary/support conditions on the ultimate behavior o f the curved bath-tub girder bridges.

213

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

References

References
1. Abdel-Sayed, G. 1973 Curved Webs Under Combined Shear and Normal Stresses,ASCE, Journal of the Structural Division, Vol. 99, No.(ST3), March 1973:511-525. 2. American Association of State Highway and Transportation Officials (AASHTO), 1996, Stan dard Specifications for Highway Bridges, Washington, D.C. 3. American Association of State Highway and Transportation Officials (AASHTO), 1994, LRFD Bridge Design Specifications, SI Units, First Edition, Washington, D.C. 4. ANSYS Element Reference, 001084, Tenth Edition, SAS IP Inc. 5. ASCE-AASHTO Task Committee on Flexural Member, Steel Box-Girder Bridges-Ultimate Strength Consideration, ASCE, Journal of the Structural Division, Dec. 1974, Vol. 100, ST( 12):243 3-2447. 6. Aslam, M. and Godden, W. 1975 Model Studies of Multi cell Curved Box-Girder Bridges, ASCE, Journal of the Engineering Mechanics Division, Vol.(101), No.(EM3), June 1975. 7. Bathe,K.J. and Dvorkin, E.N. (1986), A formulation of General Shell Elements- The Use of Mixed Interpolation of Tensorial Components, International Journal for Numerical Methods in Engineering, V.22, pp. 697-722. 8. Bathe, K.J.,Finite Element Procedures in Engineering Analysis, Prentice-Hall Inc., Engle wood Cliffs, New jersey, 1982. 9. Bradford, M.A. and Wong, T.C. 1992 Local Buckling of Composite Box-Girders Under Neg ative Bending, The Structural Engineer, Vol.(70), No.(21), Nov. 1992. 10. Branco, F.A. and Green, R. 1985 Composite Box-Girder Bridge Behavior During Construc tion, ASCE, Journal of Structural Engineering, V ol.Ill, No.(3), March 1985.

214

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

R eferences

11. Brennan, P.J. 1974 Horizontally Curved Highway Bridges, Analysis and Structural Testing of a Multiple Configuration Small Scale Horizontal Curved Highway Bridge, Department of Civil Engineering, Syracuse University, Syracuse, N.Y., Dec. 1974. 12. Brennan, P.J. and Mandel, J.A. Analysis of Small Scale Structure of Ramp P Over Interstate 1-291 Springfield, Massachusetts, Through Three-Dimensional Mathematical Method and Structural Testing, Research Report, Department of Civil Engineering, Syracuse University, Syracuse, N.Y., May 1974. 13. Brockenbrough, R.L., "Distribution Factors for Curved I-Girder Bridges", Journal of Struc tural Engineering, Vol. 112, No. 10, October, 1986:2200-2214. 14. Buchanan, J.D., Yoo,C.H. and Heins,C.P. 1973 Field Study of a Curved Box Beam Bridge, Interim Report No.59, University of Maryland at College Park, Maryland. 15. Canadian Institute of Steel Construction (CISC) workshop, Steel Bridges Design, Fabrica tion and Construction Based on Canadian Highway Bridges Design Code S6-2000, April,
2001 .

16. Chapman, C., Dowling, P.J., Lim, P.,T. and Billington, C.J. The Structural Behavior of Steel and Concrete Box-Girder Bridges, the Structural Engineer, Vol.(49), No.3, March 1971. 17. Chavel, B.W. and Earls, C.J. 2002, Stability of Curved Steel I-Girder Bridge Components During Erection, Proceedings of the Annual Stability Conference, SSRC:75-93. 18. Canadian Highway Bridge Design Code, CHBDC-S6,(2000). 19. Chen, B.S. and Yura, J.A. 2003 Metal Deck Bracing Systems for Trapezoidal Steel BoxGirder Bridges, Proceedings of Annual Technical Session and Meeting of Structural Stability Research Council, April 2003, Baltimore, Maryland: 175-193.

215

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

References

20. C heung, Y.K. an d C heung, M .S. 1971 A n aly sis o f C urved B o x -G ird er B ridges B y Finite Strip M e th o d , In tern atio n al A ssociation o f B ridges and S tructural E n g in eerin g (IA B SE ), Vol.31-1. 21. C heung, M .S . and F oo, S.H. 1995 D esign o f H orizontally C urved C om posite B ox-G irder B ridges: A S im plified A p proach, C anadian jo u rn a l o f C ivil E ngineering, 22:93-105. 22. Chu, K .H . and P injarkar, S.G. 1971 A n aly sis o f H orizontally C urved B ox-G irder B rid g es , A S C E , Jo u rn a l o f the Structural D ivision, Vol.97, N o.S T IO , O ct. 1971. 23 . C ook, R o b ert, M alkus, D avid, and Plesha, M ic h a e l,C oncept and A p p licatio n s o f F inite E le m ent A n a ly sis , 4th E dition, Jo h n W iley & Sons, N e w York, 2002. 24. C orrado, J.A . a n d Yen, B.T. 1973 F ailure T ests o f R ectan g u lar M o d el Steel B ox G irders , A S C E , Jou rn al o f the Structural D ivision, V ol.99, N o .(S T 7), July 1973. 25. C ulver, C.G. an d M ozer, J.D. Stability o f C u rv ed B ox G ird er , C o n so rtiu m o f U niversity R esearch T eam s, R e p o rt N o .B l, C arn eg ie-M ello n U niversity, P ittsburgh, P a, Sep. 1970. 26. Culver, C.G. a n d M ozer, J.D . S tability o f C u rv ed B ox G ird er , C o n so rtiu m o f U niversity R esearch Team s, R ep o rt N o.B 2, C arn eg ie-M ello n U niversity, P ittsburgh, P a, July 1971. 27. C ulver., C. and M cM anus, P., In stabiltiy o f H o rizo n tally C u rv ed P late G irders: L ateral B u ck ling o f C urved P late G irders , R ep o rt on a p ro je c t sponsored by T he P en n sy lv an ia D ep art m ent o f T ransportation. D ep artm en t o f C ivil E n g in eerin g C arn eg ie-M ello n U niversity, 1971. 28. D abrow ski, R. (1968) C urved T hin-W alled G irders, T heory and A n a ly sis , C em ent and C on crete A ssociation, L ondon. 29. D aniels, J.H. 1985 U ltim ate S trength Test o f a H o rizo n tally C u rv ed C om posite B o x G irder , F ritz E ngineering L ab R ep o rt N o .454.3, L eh ig h U niversity, B eth leh em , Pennsylvania.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

References

30. D av id so n , J.S ., K eller, M .A . and Yoo, C .H ., 1996 C ross F ram e S pacing an d P aram etric E ffects in H o rizo n tally C urved I-G irder B rid g es, A S C E , Journal o f S tructural E ngineering, Vol. 122, N o .(9 ): 1089-1096 31. D ogaki, M ., M ikam i, I., Y onezaw a, H. and O zaw a, K. (1979) F u rth er Test o f the C urved G irder W ith O rthotropic Steel Plate D eck , T echnical R ep o rt N o .20, K ansai U niversity, O saka, Jap an . 32. Evans, H .R . and A l-R ifaie, W.N. 1975 A n E x p erim en tal a n d T heoretical In v estig atio n o f the B eh avior o f B o x G irders C urved in P lan , P roceedings o f the In stitu tio n o f C ivil E ngineers, Part 2, 59, June 1975:323-352. 33. Fam , A. a n d T urkstra, C., 1975 A F inite elem en t S chem e fo r B ox B ridge A n a ly sis , C o m p u t ers and S tructures, V ol.(5), P ergam on Press: 179-186. 34. Fam , A . a n d T urkstra, C. 1976 M odel Study o f H o rizo n tally C urved B o x -G ird er , A S C E , Journal o f S tructural D ivision, Vol.(102), N o .S T 5 , M ay 1976. 35. Fountain, R .S . and M attock, A .H . C om posite Steel -C oncrete M ulti-B ox G irder B rid g es , P roceedings o f the C anadian Structural E n g in eerin g C onference, Toronto, 1968:19-30. 36. Fu,C .C . and H su, Y.T. 1995 The D ev elo p m en t o f a n Im p ro v ed C urvilinear T hin-W alled V la sov E lem ent , C om puters and Structures, Vol.(54), N o .l , 1995:147-159. 37. G aluta, E .M . and C heung, M .S. 1995 C o m b in ed B o u n d a ry E lem en t and F inite E lem ent A nalysis o f C om posite B ox G irder B rid g es , C o m p u ters a n d S tructures, Vol.(57), N o .3:427437. 38. G uide S pecifications fo r the D esign o f H o rizo n ta lly C u rv ed H ig h w ay B rid g es , A m erican A ssociation o f State, H ig hw ay and T ran sp o rtatio n O fficials, Inc., W ashington D .C ., 1976.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

R eferences

39. G uide S pecifications for the D esign o f H o rizo n tally C urved G irder B rid g es, P u b lish ed by the A m e ric an A ssociation o f State H ig h w ay and T ransportation O fficials, 1993. 40. G uide S pecifications for the D esig n o f H o rizo n tally C urved G irder B rid g es , P u b lish ed by the A m e ric an A ssociation o f State H ig hw ay and T ransportation O fficials, 2003. 41. H all, D .H ., G rubb, M .A . and Yoo, C.H . Im p ro v ed D esig n Specifications for H orizo n tally C urved Steel G irder H ighw ay B rid g es , R ep o rt 424, 1999. 42. H eins, C.P. and O leinik, J.C . 1976 C urved B o x B eam B ridge A n aly sis , C om puters and Structures, V ol.( 6 ), Pergam on P ress 1976:65-73. 43. H eins,C .P. and Sheu, F.H. 1982 D esig n /A n aly sis o f C urved B o x G irder B rid g es , C o m p u t ers an d Structures, Vol.(15), N o .3, 1982:241-258. 44. H eins, C.P., B onakdarpour, B. an d B ell L .C . M u lti cell C urved G irder M odel S tudies A S C E , Journal o f the S tructural D ivision, V ol.(98), N o.(S T S ), A pril 1972. 45. H eins, C.P. and L ee, W .H. 1981 C urved B o x G irder B rid g e:F ield Test , A S C E , Jou rn al o f the Structural D ivision, V ol.(107), N o .(S T 2 ):3 17-327. 46. H eins,C.P. and H um phreys, R .S ., 1979 B en d in g and T orsion Interaction o f B o x -G ird ers, A S C E , Journal o f the Structural D ivision, Vol. 105, N o .(S T 5 ), M ay 1979:891-904. 47. H eins, C.P. 1978 B o x G irder B ridge D esign- State o f the A rt , A m erican Institute o f Steel C onstruction, E n gineering Journal, F o u rth Q uarter, 1978. 48. H eins, C.P. 1983 Steel B ox G irder B rid g es-D e sig n G uides and M eth o d s, A m erican Institute o f Steel C onstruction, E ngineering Journal, 1983. 49. H eins, C.P. and Jin, J.O ., 1984, L ive L o ad D istrib u tio n on B raced C urved I-G ird er , Journal o f Structural E ngineering, Vol. 110, N o .3, M a rc h 1984:523-530.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

References

50. H elw ig, T. and Fan, Z. 2000 F iled and C om putation Studies o f Steel T rapezoidal B ox-G irder B rid g es , R esearch R ep o rt 1395-3, T he U niv ersity o f H ouston, Texas. 51. H ikosaka, H., and T akam i, K. (1985). F o rm u latio n o f D istortional b e h av io r o f T hin-W alled C urved B eam w ith O pen C ross S ectio n . P ro ceed in g o f Japan S ociety o f C ivil E ngineers, S tructural E ngineering/ E arthquake E gg., Vol. 2, N o. 1, A pril 1985. 52. Jetann, C .A ., H elw ig, T.A. and L ow ery, R. 2002 L ateral B racing o f B ridge G irders B y P e r m an en t M etal D eck F o rm s , P ro ceed in g s o f th e A nnual Stability C onference, SSR C , A pril 2002, Seattle, W ashington:291-309. 53. K issane,R .J. and B eal, D .B . 1975 F ield Testing o f H orizontally C urved Steel G irder B rid g es , F ederal H ig h w ay A d m in istratio n , R esearch R eport 27, M ay 1975. 54. K ollbrunner, F.A., and B asler,K .,T orsion in Structures , Springer-V erlag, B erlin, W est G er m any, 1969. 55. K orol, R .M ., T him m hardy, E.G. and C heung, M .S. 1988 A n E x p erim en tal In v estig atio n o f the E ffects o f Im perfection on the S trength o f Steel B ox G irders , C an ad ian Journal o f C ivil E ngineering, Vol. 15, N o .(3): 443-4 4 9 , June 1988. 56. L ee,S .C ., Yoo,C.H. and L ee,D .S. E lastic B u ckling o f H orizontally C u rv ed Steel B o x G irder F langes U nder C o m pression and S h ear , P ro ceed in g s o f the A n n u al S tability C onference, SSR C , A pril 2002, Seattle, W ash in g to n :4 9 9 -5 14. 57. M cD onald, R .E ., C hen, Y.S., Y ilm az, C. and Yen, B.T. 1976 O pen Steel B o x S ections W ith Top L ateral B racin g . A S C E , Jo u rn al o f th e Structural D ivision, Jan. 1976, Vol. 102, ST (1):3549. 58. M cm anus, P.F., et.al., H o rizo n tally C u rv ed G irders- State o f the A rt , Jo u rn al o f the S truc tural D ivision, A S C E , M ay, 1969.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

References

59. M eyer, C. and Scordelis, A .C ., 1971 A n aly sis o f C urved F o ld ed P late Structures , A SC E , Journal o f the Structural D ivision, ST (10), O ctober. 60. N ak ai,H .an d H eins, C.P. 1977 A n aly sis C riteria for C urved B rid g es , A S C E , Journal o f the Structural D ivision, Vol.(103), N o .(S T 7 ), Ju ly 1977. 61. N akai, H. and Yoo, C.H. (1988), A n aly sis and D esign o f C urved Steel B rid g es , M cG raw H ill B o o k Co. 62. N g , S.F., C heung, M .S. and H ach em , H .M . 1993 Study o f a C u rv ed C ontinuous C om posite B o x G irder B rid g e , C anadian Jou rn al o f C ivil E ngineering, 20, 107-119. 63. O leinik, J.C . and H eins, C.P. 1975 D iap h rag m s for C urved B o x B e a m B rid g es , A S C E , Jour nal o f the Structural D ivision, Vol.(101), N o .(S T lO ), O ct. 1975. 64. R ankin, C.C. and B rogan, F.A., A n E lem en t In dependent C o ro tatio n al A p p ro ach for the T reatm ent o f L arge R o tatio n s , Jo u rn al o f P ressure Vessel T echnology , Vol. 108, M ay 1986, pp. 165-174. 65. Scordelis, A .C . (1960) A m atrix F o rm u latio n o f the F o ld ed P late E q u a tio n s , A S C E , Journal o f the Structural D ivision, St(10), O ctober.
6 6 . Sennah, K halid. M ., "L oad d istrib u tio n and dynam ic characteristics o f cu rv ed co m posite c o n

crete deck-steel cellu lar bridges", P h.D . T hesis, U niversity o f W indsor, 1998. 67. Siddiqui, A .H . and N g, S.F. 1988 E ffect o f D iaphragm s on Stress R e d u c tio n in B o x G irder B ridge S ections , C anadian Jo u rn a l o f C ivil E ngineering, 15, 1988:127-135.
6 8 . Sim pson, M ichael. D. "A nalytical In v estig a tio n o f C urved Steel G ird er B ehavior", Ph.D . T h e

sis, U niversity o f Toronto, 2000.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

References

69. S isodiya, R .C ., C heung, Y.K. and G hali, A . F inite E lem ent A n aly sis o f Skew, C urved B oxG irder B rid g e , International A sso ciatio n o f B ridges and Structural E n gineering (IA B SE ),3ii: 191-199. 70. T him m hardy, E.G. and K orol, R .M ., 1988, G eom etric Im perfections and T olerances fo r Steel B ox G ird er B ridges , C anadian Journal o f C ivil E ngineering, Vol. 15, 1988: 437-442. 71. T him m hardy, E.G. 1991 N o n lin ear A n aly sis o f B uckling B ehavior o f Steel B ox G irder C om p o nents , C om puters and Structures, Vol. 40, N o.2:469-476. 72. Tung, D .H . and Fountain, R .S., 1970 A p p roxim ate T orsional an alysis o f C u rv ed B ox G irders B y the M /R M ethod , A m erican In stitu te o f Steel C onstruction E n g in eerin g Journal, Vol.7, N o .3, Ju ly 1970. 73. V lasov, V.Z. (1961) T h in W alled E lastic B e a m s, 2nd E dition, N a tio n a l Science F oundation, W ashington D .C. 74. W right, R .N ., A bdel-S am ad,S .R . and R obinson, A .R . 1968 B eam s on E lastic F oundation A nalogy fo r A nalysis o f B ox G irder , A S C E , Journal o f the S tructural D ivision, Vol.(94), N o.(S T 7): 1719-1743. 75. Yabuki, T., A rizum i, Y. and V innakota, S. 1995 S trength o f T hin-W alled B o x G irder B ridges C urved in P lan , A S C E , Journ al o f S tructural E ngineering, V ol.1121, N o.(5):907-914. 76. Yasunori, A ., H am ada, S. and O shiro, T. 1988 B ehavior Study o f C u rv ed C om posite B oxG irders, A S C E , Journal o f S tructural E ng in eerin g , Vol. 114, N o .11, N ov. 1988. 77. Y onezaw a, FL, M ikam i, I., K am atsu, Y. an d D ogaki, M . (1978) T est o f a C u rv ed G irder W ith O rthotropic Steel Plate D eck , T echnical R ep o rt N o .19, K ansai U niversity, O saka, Japan: 115125.

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

R eferences

78. Yoo, C. H . and L ittrel, P.C. 1986, C ross B racing E ffects in C urved Stringer B rid g es, A S C E , Journal o f S tructural E ngineering, Vol. 112, N o.(9):2127-2140. 79. Z hang, J., P aram etric Studies o f Etorizontally C urved B ox-G irders U n d er D ead L oads (Effects o f E x ternal and Internal B racing) , M .A .S c. T hesis, A u b u rn U niversity, A uburn, A la bam a, A u g . 1996. 80. Z hang, S .H . an d L yons, L.P.R. 1983 A T hin-W alled B ox B eam F inite E lem en t for C urved B ridge A n a ly sis , C om puters and Structures, V ol.(18), N o .6:1035-1046. 81. Z urieck A ., N aq u ib , R .,Y adlow sky, J.M ., C u rv ed Steel B ridge R esearch R ep o rt In terim . R ep o rt I: S y n th e sis , U .S . D epartm ent o f T ransportation F ederal H ig hw ay A dm inistration. P u b licatio n N o . FH W A -R D -93-129, D ec. 1994.

222

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

APPENDIX I

ANSYS Input File for Two Spans Continuous Curved Bath tub Girder Bridge During Construction.

223

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Details concerning the bridge cross-sectional dimensions and the design parameters were explained in chapter 5. Input data is supplied to ANSYS by the user through the input text file that is listed below. The user should create a new directory and copy the input file in the new directory. After running ANSYS, and from its Graphical User Interface (GUI), click on File menu, Read Input From....; a dialogue box will appear; search for the input text file and choose it. The file generates a finite element mesh, element connectivity, material properties and applies constraints and loads. This tool (input text file) can be used as a simple analysis tool, the structure is loaded with factored construction loads; after execution, the user can visualize the analysis results using ANSYS post-processor. The stresses, forces and deflections that are obtained from the analysis are compared then with the specified limits o f the standards. Keep in mind that F d l, Fd2, Fd3, and Fd4 are load/node, so, the load (N/ mm.) must be multiplied by the length o f the shell elem ent in the longitudinal direction (v) to obtain the load per node. The tool can also be used as a design tool including a performance evaluation for yielding and/or stability. The structure is loaded up to failure by performing an incremental stress-strain analysis and the failure modes are then observed; specific failure mode can be easily altered or even avoided by modifying the construction plan or retrofitting the dimensions o f specific member; also the estimated load factor must satisfy the load factor specified in the standards. If the buckling analysis option is activated in the input file, the constructions will be applied incrementally through the use o f other input files called: stage 1, stage2, stage3 or stage4. The text file stagel is listed at the end o f this appendix as an example o f loading file. As was explained earlier in chapter 5, Newton-Raphson method that is adopted by ANSYS is not able to trace unloading, therefore, the arc-length method is provided. I f the goal o f the analysis is to trace the post-buckling response o f the structure, the arc-length method must be activated through ANSYS command arclen that is shown in sta g e l list below. Several trials are required to adjust the reference arc length radius to obtain a solution at the limit load. This should be done by changing the number o f sub-

224

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

steps (Nsubst ANSYS command) as shown in stage 1 list. Otherwise, it might more convenient to use the standard Newton-Raphson method iterations with automatic bisection to determine values o f nonlinear buckling loads. In this case, the arclen ANSYS command must be deactivated (erased) from stagel list. It is usually good practice to graph the load-deflection curve (using ANSYS post26 processor) for every analysis for evaluation.

225

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

ITfais input file generates the horizontally curved bath-tub bridge during construction
ex=200000 density=7.85e-9 tw=19 tbl=14 tb2=20 tb3=25 ttl=20 tt2=25 tt3=35 Abr=5050 !modulus of elasticity of steel (mpa) !steel's density (M g/mm*3) !Web Thickness(mm.)
IBottom Flange Thickness(mm.)(further to the internal support)

!Bottom Flange Thickness(mm.) !Bottom Flange Thickness(mm.)(closer to the internal support) !Top Flange Thickness(mm.)(further to the internal support) !Top Flange Thickness(mm.) !Top Flange Thickness(mm.)(closer to the internal support) ICross Sectional Area of Top Bracing Member diagonal(mm.)(top bracing of !single diagonal type)

Acf=1360 Aties=3060 tDh=12

ICross Sectional Area of one Cross Frame Daigonal(mm.)(cross frames of X-type) ICross Sectional Area of Lateral Ties(mm.)(struts) ISolid Diaphragm Thickness(mm.) !(extrnal and internal solid diaphragms in every support)

h=1450
slope=4

IWeb Vertical Height(mm.) IWeb Slope IBottom Flange Width(mm.) IDistance Between Mid Top Flanges of Each Girder(mm.) ITop Flange Width(mm.) IRadius of Curvature(mm.) IDistance Between Mid Exterior Top Flange of Interior Girder and Mid Interior ITop Flange of Exterior Girder(mm.)

Bb=2275 bl=3000 bt=380 R=120000 c=3360

L=72000

ITotal Bridge Length(mm.)

226

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Lsl=36000 Ls2=36000 xl=24500 x2=32000 x3=24500 x4=32000 x=3000 v^SOO

IFirst Span Length (east span)(mm.) !Second Span Length (west span)(mm.) !Distance From the East Bent to the 1st Change in Top and Bottom Flange Thickness IDistance From the East Bent to the 2nd Change in Top and Bottom Flange Thickness IDistance From the West Bent to the 1st Change in Top and Bottom Flange Thickness IDistance From the West Bent to the 2nd Change in Top and Bottom Flange Thickness ICross Frames Intervals(mm.) IDescritization in the longitudinal direction.)

111111111111111111Longitudinal Stiffener (Tee Section)!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! longstif='yes' !If Longitudinal Stiffener Exist Say 'Yes',otherwise Say 'No' d=177 F=205 Ll=18000 L2=54000 tfl 16.8 twl=9.4 !Longitudinal Stiffener Depth(mm.) ITop Flange Width of Longitudinal Stiffener(mm.) IDistance from The East Bent to the Starting Point of Longitudinal Stiffener(mm.) IDistance from The East Bent to the End Point of Longitudinal Stiffener(mm.) ITop Flange Thickness of Longitudinal Stiffener(mm.) !Web Thickness of Longitudinal Stiffener (mm.)

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ILoading!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Simple-analysis= yes !For simple analysis, Say yes, otherwise No Buckling-analysis=yes !For buckling analysis, say yes, otherwise N o
se!fw t='no'

!For the Effect of the Steel Girder Self weight Say'Yes',otherwise 'No' !For the Effect o f Stagel Deck Construction+Girder Self weight Say'Yes' !For the Effect o f Stagel+2 Deck Construction+Girder Selfweight Say'Yes' !For the Effect of Stagel+2+3 Deck Construction+Girder Selfweight Say'Yes' !For the Effect of Stagel+2+3+4 Deck Construction+Girder Selfweight Say'Yes'

stagel='no' stage2='no' stage3='no' stage4='yes'

227

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

cl=0 c2=0 c3=0 c4=L Fdl=11319 Fd2=10272 Fd3=9988 Fd4=10469

!See Concrete Deck Placement Sequence(mm.) Fig. 5.2 !See Concrete Deck Placement Sequence(mm.) Fig. 5.2 !See Concrete Deck Placement Sequence(mm.) Fig. 5.2 !See Concrete Deck Placement Sequence(mm.) Fig. 5.2
!Load/N ode-M id E x te rio r Top Flange of Exterior Girder(N/mm.)

!Load/Node-Mid Interior Top Flange of Exterior Girder(N/mm.) !Load/Node-Mid Exterior Top Flange of Interior Girder(N/mm.) !Load/Node-Mid Interior Top Flange of Interior Girder(N/mm.) !construction Load(wet concrete slab and haunches;steel reb a rs;fo rm w o rk
! ;w alkw ay live load on cantilever brackets)(N/500mm)(use suitable

Iload factors and load combinations)

!!!!!!!!!!!!End of parameters input!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!End of parameters input!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! nmmmu


I*

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

/NOPR /PMETH,OFF KEYW,PR_SET,1 KEYW,PR_STRUC, 1 KEYW,PR_THERM,0 KEYW,PR_FLUID,0 KEYW,PR_ELMAQO KEYW,MAGNOD,0 KEYW,MAGEDG,0

228

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

KEYW ,MAGHFE,0 KEYW,MAGELC,Q KEYW,PR_MULTI,0 KEYW,PR_CFD,0 /GO


!*

/COM, /COM ,Preferences for GUI filtering have been set to display: /COM, Structural
!*

/PREP7 ET,1,SHELL181 KEYOPT,1,3,1 et,2 ,link 8

Ful1 integration for Shell 181

!non-linear material properties


m p,ex,l,ex mp,nuxy, 1,0.3 mp,dens, 1,density tb,m iso,l,l,4 tbtemp,0.0 tbpt,defi,345/ex,345 tbpt,defi,0.1,345*l .1

Ireal conatants(thickness of plates)


r,l,ttl

229

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

r,2 ,tt2 r,3,tt3

r,5,tw r,6 ,tbl r,7,tb2 r,8,tb3

r,10,Abr r,ll,A c f r, 12,tfl r,13,twl r,14,tdh r,15,tslab r,16,Aties pi=acos(-l) theta=(v* 180)/(r*pi)

Inodes

k, 1,r-c/2,0 W PSTYLE1 wprot,0,0,180 wprot,0,-90,0 kplot

230

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

KWPAVE,

CSWPLA, 11,1,1,1, n,l,r-c/ 2 , 0,0 n,5,r-c/2-h/slope,0,-h fill, 1,5 n,9,r-c/2-bl+h/slope,0,-h fill,5,9 n,13,r-c/2-bl,0,0 fill,9,13 n, 14,r-c/2+bt/2,0,0 n,15,r-c/2-bt/2,0,0 n, 16,r-c/2-b 1+bt/ 2 ,0,0 n, 17,r-c/2-b 1-bt/2 ,0,0 n,18,r+c/2+bl,0,0 n,2 2 ,r+c/2 +bl-li/slope, 0 ,-h fill, 18,22 n,26,r+c/2+h/slope,0,-h fill,22,26 n,30,r+c/2,0,0 fill,26,30 n ,3 1,r+c/2 +bl+ bt/ 2 , 0,0 n,32,r+c/2+bl-bt/2,0,0 n,3 3 ,r+c/2+bt/2,0,0 n,34,r+c/2-bt/2,0,0

231

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

ngen,(L/v)+1,34,1,34,1,0,theta,0

lEIements !top flanges


type,l real, 1 m at,l

en, 1,1,14,48,3 5 en,2,1,3 5,49,15 en,3,13,16,50,47 en,4,13,47,51,17 en,5,l 8,31,65,52 en,6,18,52,66,32 en,7,30,33,67,64 en,8,30,64,68,34

egen,L/v,34,1,8,1

!modify real constants for top flange elements


*if,x 2 ,gt,xl,then nsel, s,loc,y,theta* (x 1/v),theta*Ls 1/ (v) esln,s, 1 emodif,all,real ,2 *endif

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

*if,Lsl,gt,x2,then nsel,s,loc,y,theta*(x2/v),theta*Lsl/(v)
esln,s,l

emodif,all,real,3 *endif

*if,x4,gt,x3,then nsel, s,loc,y,theta* ((L-x3 )/v),theta* ((L-x4)/v) esln,s,l emodif,all,real ,2 *endif *if,Ls2,gt,x4,then nsel,s,loc,y,theta*((L-x4)/v),theta*((L-Ls l )/v) esln,s,l emodif,all,real,3 *endif

allsel

!web elem ents


real, 5 en,3501,1,2,36,35 egen,4,1,3501 en,3505,9,10,44,43

233

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

egen,4,1,3505 en,3509,18,19,53,52 egen,4,1,3 509 en,3513,26,27,61,60 egen,4,l,3513 egen, L/v,34,3501,3516

IBottom flange
real ,6 en,10001,6,5,39,40 egen,4 , 1,10001 en,10005,23,22,56,57 egen,4,1,10005 egen,L/v,34,10001,10008

!modify real constants for Bottom flange elements


*if,x2 ,gt,xl,then nsel,s,loc,z,-h nsel,r,loc,y,theta*(xl/v),theta*Lsl/v esln,s,l emodif,all,real,7 *endif *if,Lsl,gt,x2,then nsel,s,loc,z,-h

234

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

nsel,r,loc,y,theta*(x2/v),theta*Lsl/(v) esln,s,l emodif,all,real ,8 *endif

*if,x4,gt,x3,then nsel,s,loc,z,-h nsel,r,loc,y,theta*((L-x3)/v),theta*((L-x4)/v) esln,s,l emodif,all,real,7 *endif

*if,Ls2,gtx4,lhen nsel,s,loc,z,-h nsel,r,loc,y,theta*((L-x4)/v),theta*((L-Ls 1)/v) esln,s, 1 emodif,all,real,8 *endif allsel

!Top B racing SD1 ! interior girder

235

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

type ,2 *do,i,0,(L/(2*X))-l n 1=node(r-c/2 , 0 +theta *2 *(x/v)* i,0 ) n 2 =node(r-c/ 2 -b 1,theta*(x/v)+theta* 2 *(x/v)*i,0 ) real, 10 en,14001+i,nl,n2 *enddo

*do,i,0,(L/(2*X))-l n 1=node(r-c/2 -b 1,theta* (x/v)+theta* 2 *(x/v)*i,0 ) n 2 =node(r-c/ 2 ,theta* 2 *(x/v)+theta* 2 *(x/v)*i,0 ) real, 10 en, 14002+(L/(2*X))-1+i,n 1,n2 *enddo

! exterior girder
*do,i,0,(L/(2*X))-1 n 1=node(r+c/2 +b 1, 0 +theta* 2 *(x/v)* i,0 ) n 2 =node(r+c/2 ,theta*(x/v)+theta* 2 *(x/v)*i,0 ) real, 10 en, 14003+(L/X)-2+i,n 1,n2 *enddo

*do,i,0,(L/(2*X))-1 nl=node(r+c/ 2 ,theta*(x/v)+theta* 2 *(x/v)*i,0 )

236

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n2 =node(r+c/ 2 + b 1,theta* 2 *(x/v)+theta* 2 *(x/v)*i,0 ) real, 10 en, 14004+(((L/(2 *X ))-1)*3 )+i,n 1,n2 *enddo

IL ateral ties (Struts) *do,i,0,(L/x)-2 nl=node(r-c/ 2 ,theta*(x/v)+theta*(x/v)*i,0 ) n 2 =node(r-c/ 2 -b 1,theta* (x/v)+theta* (x/v) *i, 0 ) real, 16 en, 14005+(((L/(2*X))-1)*4)+i,n 1,n2 *enddo

*do,i,0,(L/x)-2 nl=node(r+c/ 2 ,theta*(x/v)+theta*(x/v)*i,0 ) n 2 =node(r+c/ 2 +bl,theta*(x/v)+theta*(x/v)*i, 0 ) real, 16 en, 14006+(((L/(2*X))-1)*4)+(L/x)-2+i,n 1,n2 *enddo

! Cross Fram es
*do,i,0,(L/x)-2 nl=node(r-c/ 2 ,theta*(x/v)+theta*(x/v)*i, 0 )

237

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n2=node(r-c/2-bl+fa/slope,theta*(x/v)+theta*(x/v)*i,-h) real, 11 en, 14007+(((L/(2*X)> 1)*4)+((L/x)-2)*2+i,nl,n2 *enddo

*do,i,0,(L/x)-2 n3=node(r-c/2-b 1,theta*(x/v)+theta*(x/v)*i,0) n4=node(r-c/2-h/slope,theta*(x/v)+theta*(x/v)*i,-h) real, 11 en, 14008+(((L/(2*X))-1)*4)+((L/x)-2)*3+i,n3,n4 *enddo

*do,i,0,(L/x)-2 nl=node(r+c/2+bl,theta*(x/v)+theta*(x/v)*i,0) n2=node(r+c/2+h/slope,theta*(x/v)+theta*(x/v)*i,-h) en, 14009+(((L/(2 *X ))-1)*4)+((L/x)-2)*4+i,n 1,n2 *enddo

*do,i,0,(L/x)-2 n3=node(r+c/2,theta*(x/v)+theta*(x/v)*i,0) n4=node(r+c/2+bl-h/slope,theta*(x/v)+theta*(x/v)*i,-h) real, 11 en, 14010+(((L/(2 *X))-1)*4)+((L/x)-2)*5+i,n3,n4 *enddo

238

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

longitudinal stiffener *if,longstif,EQ,'yes',then Inodes interior girder n, 14001 ,r-c/2-b l/2+f/2,theta*(L 1/v),-h+d
11,14002,r-c/2-b 1/2,theta*(L 1/v),-h+d

n, 14003 ,r-c/2-bl/2-f/2,theta*(Ll/v),-h+d ngen,((L2-L 1)/v)+l ,3,14001,14003,1,0,theta,0

! lognitudinal stiffener elements interior girder type,l real, 12 en, 14501,14001,14004,14005,14002 en, 14502,14002,14005,14006,14003 egen,(L2-L 1)/v,3,145 01,14502

*do,i,0,((L2-Ll)/v)-l nl=node(r-c/2-bl/2,theta*(Ll/v)+theta*i,-h+d) n2=node(r-c/2-b 1/2,theta*((L l/v)+1)+theta*i,-h+d) n3=node(r-c/2-b 1/2,theta*((L 1/v)+ 1)+theta*i,-h) n4=node(r-c/2-b 1/2,theta*(L 1/v)+theta*i,-h) real, 13 en, 15001 +i,n 1,n2,n3 ,n4 *enddo

239

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

llognitudinal stiffener exterior girder Inodes exterior girder n, 14251 ,r+c/2+b 1/2+f/2,theta*(L 1/v),-h+d n, 14252,r+c/2+b 1/2,theta*(L l/v),-fa+d n, 14253,r+c/2+b 1/2-f/2,theta* (L 1/v),-h+d ngen,((L2-L 1)/v)+1,3,14251,14253,1,0,theta,0

! longitudinal stiffener elements real, 12 en,15501,14251,14254,14255,14252 en ,15502,14252,14255,14256,14253 egen,(L2-Ll)/v,3,15501,15502

*do,i,0,((L2-Ll)/v)-l n 1=node(r+c/2+b 1/2,theta* (L 1/v)+theta*i,-h+d) n2=node(r+c/2+b 1/2,theta* ((L l/v )+ 1)+theta* i,-h+d) n3 =node(r+c/2+b 1/2,theta* ((L 1/v )+ 1)+theta* i,-h) n4=node(r+c/2+bl/2,theta*(Ll/v)+theta*i,-h) real, 13 en, 16001 +i,n 1,n2,n3,n4 *enddo *endif

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

ICreate Diaphragm s ! At E ast Bent ! NODES

fill,1,13,3,14501,4 fill,14501,6,3,14502,1 fill, 14505,7,3,14506,1 611,14509,8,3,14510,1

Ilnternal Bent

fill,(((Lsl/(v))+l)*34)-33,(((Lsl/(v))+l)*34)-21,3,14513,4

fill, 14513,(((Lsl/(v))+l)*34)-28,3,14514,1 fill, 14517,(((Lsl/(v))+l)*34)-27,3,14518,1


fill,14521,(((Lsl/(v))+l)*34)-26,3,14522,1

! West Bent

fill,(((L/(v))+l)*34)-33,(((L/(v))+l)*34)-21,3,14525,4 fill, 14525, (((L/(v))+l)*34)-28,3,14526,1 fill, 14529,(((L/(v))+1)*34)-27,3,l 4530,1 fill,14533,(((L/(v))+l)*34)-26,3,14534,1

lElements !East Bent

241

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

*do,i,0,3 nl=node(r-c/2-(h/(slope*4))*i,0,-(h/4)*i) n2=node(r-c/2-(h/(slope*4))-(h/(slope*4))*i,0,-(h/4)-(h/4)*i) n3=node(r-c/2-bl/4-(((h/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)-(h/4)*i) n4=node(r-c/2-b l/4-((((h/slope)+(bb/4))-(b 1/4))/4)*i,0,-(h/4)*i)

type,l real, 14 en,16301+i,nl,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r-c/2-bl/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)*i) n2=node(r-c/2-bl/4-(((h/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-h/4-(h/4)*i) n3=node(r-c/2-(bl/4)*2,0,-h/4-(h/4)*i) n4=node(r-c/2-(bl/4)*2,0,-(h/4)*i) type,l real, 14 en, 16305+i,n 1,n2,n3,n4 *enddo

*do,1,0,3 nl=node(r-c/2-(bl/4)*3+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)*i) n2=node(r-c/2-(bl/4)*3+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-h/4-(h/4)*i)

242

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n3 =node(r-c/2-(b 1/4) *2,0,-h/4-(h/4)* i) n4=node(r-c/2-(bl/4)*2,0,-(h/4)*i) type,l real, 14 en, 163 09+i,n4,n3 ,n2,n 1 *enddo

*do,i,0,3 nl=node(r-c/2-(bl/4)*4+(h/slope/4)*i,0,-(h/4)*i) n2=node(r-c/2-(bl/4)*4+(h/slope/4)+(h/slope/4)*i,0,-h/4-(h/4)*i) n3=node(r-c/2-(bl/4)*3+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-h/4-(h/4)*i) n4=node(r-c/2-(b 1/4) *3+((((h/slope)+(bb/4))-(b 1/4))/4)* i,0,-(h/4)*i) type,l real, 14 en, 16313+i,n4,n3 ,n2,n 1 *enddo

lElements llnternal Bent *do,i,0,3 nl=node(r-c/2-(lV(slope*4))*i,theta*(Lsl/(v)),-(h/4)*i) n2=node(r-c/2-(h/(slope*4))-(lr/(slope*4))*i,theta*(Lsl/(v)),-(h/4)-(h/4)*i) n3=node(r-c/2-bl/4-(((h/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)(h/4)*i) n4=node(r-c/2-bl/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)*i)

243

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

type, 1 real, 14 en, 16317+i,n 1,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r-c/2-bl/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)*i) n2=node(r-c/2-b 1/4-(((h/slope)+(bb/4))-(b 1/4))/4-((((h/slope)+(bb/4))-(b 1/ 4))/4) *i ,theta*(Ls 1/(v)),-h/4-(h/ 4)*i) n3=node(r-c/2-(bl/4)*2,theta*(Lsl/(v)),-h/4-(h/4)*i) n4=node(r-c/2-(bl/4)*2,theta*(Lsl/(v)),-(h/4)*i) type,l real, 14 en, 16321 +i,n 1,n2,n3 ,n4 *enddo

*do,i,0,3 nl=node(r-c/2-(bl/4)*3+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(li/4)*i) n2=node(r-c/2-(bl/4)*3+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),h/4-(h/4) *i) n3=node(r-c/2-(bl/4)*2,theta*(Lsl/(v)),-h/4-(h/4)*i) n4=node(r-c/2-(bl/4)*2,theta*(Lsl/(v)),-(h/4)*i) type,l real, 14 en, 16325+i,n4,n3 ,n2,n 1

244

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

*enddo

*do,i,0,3 n 1=node(r-c/2-(b 1/4)*4+(h/slope/4)*i,theta*(Ls 1/ (v)),-(h/4)* i) n2=node(r-c/2-(bl/4)*4+(h/slope/4)+(h/slope/4)*i,theta*(Lsl/(v)),-h/4-(h/4)*i) n3=node(r-c/2-(bl/4)*3+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),h/4-(h/4)*i) n4=node(r-c/2-(bl/4)*3+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)*i) type,l real, 14 en,16329+i,n4,n3,n2,nl *enddo

!EIements !West Bent *do,i,0,3 nl=node(r-c/2-(h/(slope*4))*i,theta*(L/(v)),-(h/4)*i) n2=node(r-c/2-(h/(slope*4))-(h/(slope*4))*i,theta*(L/(v)),-(h/4)-(h/4)*i) n3=node(r-c/2-bl/4-(((h/slope)+(bb/4))-(bl/4))/4-((((hyslope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)-(li/ 4)*i) n4=node(r-c/2-bl/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i) type,! real, 14 en, 1633 3+i,n 1,n2,n3 ,n4 *enddo

245

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

*do,i,0,3

nl=node(r-c/2-bl/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i)
n2=node(r-c/2-bl/4-(((h/s]ope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-h/4-(h/
4 )* i)

n3=node(r-c/2-(bl/4)*2,theta*(L/(v)),-h/4-(h/4)*i) n4=node(r-c/2-(bl/4)*2,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 1633 7+i,n 1,n2,n3 ,n4 *enddo

*do,i,0,3 nl=node(r-c/2-(bl/4)*3+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i) n2=node(r-c/2-(bl/4)*3+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-h/4(h/4)*i) n3 =node(r-c/2-(b 1/4)* 2,theta* (L/(v)),-h/4-(h/4)* i) n4=node(r-c/2-(bl/4)*2,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 16341 +i,n4,n3 ,n2,n 1 *enddo

*do,i,0,3 nl=node(r-c/2-(bl/4)*4+(h/slope/4)*i,theta*(L/(v)),-(h/4)*i)

246

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n2=node(r-c/2-(bl/4)*4+(h/slope/4)+(h/slope/4)*i,theta*(L/(v)),-h/4-(h/4)*i)
n3=node(r-c/2-(bl/4)*3+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-h/4(h/4)*i) n4=node(r-c/2-(bl/4)*3+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 16345+i,n4,n3 ,n2,n 1 *enddo

!!!!!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!Exterior girder ! NODES ! East Bent fill,18,30,3,14537,4 fill,14537,23,3,14538,1 fill,14541,24,3,14542,1 fill,14545,25,3,14546,1

Ilnternal Bent

fill,(((Ls 1/(v))+l)*34)-16,(((Ls l/(v ))+ 1)*34)-4,3,14549,4 fill, 14549,(((Ls 1/(v))+1)*34)-11,3,14550,1 fill,14553,(((Lsl/(v))+l)*34)-10,3,14554,1 fill,1455 7,(((Ls 1/(v))+ 1)*34)-9,3,14558,1

247

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

!West B ent

fill,(((L/(v))+1) *3 4)-16,(((L/(v))+1)*34)-4,3,14561,4 fill,14561 ,(((L/(v))+1)*34)-11,3,145 62,1 fill,14565,(((L/(v))+l)*34)-10,3,14566,1 fill,14569,(((L/(v))+l)*34)-9,3,14570,1

JElements !East Bent

*do,i,0,3 nl=node(r+c/2+bl-(h/(slope*4))*i,0,-(h/4)*i) n2=node(r+c/2+bl-(h/(slope*4))-(h/(slope*4))*i,0,-(h/4)-(h/4)*i) n3=node(r+c/2+.75 *b 1-(((h/slope)+(bb/4))-(b 1/4))/4-((((h/slope)+(bb/4))-(b l/4))/4)* i,0,-(h/4)-(h/4)* i) n4=node(r+c/2+.75*bl-((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)*i) type,l real, 14 en, 16349+i,nl ,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r+c/2+.75*bl-((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)*i) n2=node(r+c/2+.75*bl-(((h/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-h/4-(h/4)*i) n3=node(r+c/2+(bl/2),0,-h/4-(h/4)*i) n4=node(r+c/2+(b l/2),0,-(h/4)*i)

248

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

type,l real, 14

en,16353+i,nl,n2,n3,n4
*enddo

*do,i,0,3 nl=node(r+c/2+(bl/4)+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)*i) n2=node(r+c/2+(bl/4)+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-h/4-(h/4)*i) n3 =node(r+c/2+(b 1/2),0,-h/4-(h/4)*i) n4=node(r+c/2+(b 1/2),0,-(h/4)*i)

type,l
real, 14 en, 16357+i,n4,n3 ,n2,n 1 *enddo

*do,i,0,3 nl=node(r+c/2+(h/slope/4)*i,0,-(h/4)*i) n2=node(r+c/2+(h/slope/4)+(h/slope/4)*i,0,-h/4-(h/4)*i) n3=node(r+c/2+(bl/4)+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-h/4-(h/4)*i) n4=node(r+c/2+(bl/4)+((((h/slope)+(bb/4))-(bl/4))/4)*i,0,-(h/4)*i)

type,l
real, 14 en, 16361 +i,n4,n3 ,n2,n 1 *enddo

249

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

!Elements Ilnternal Bent *do,i,0,3 n 1=node(r+c/2+b 1-(h/(slope* 4))* i,theta* (Ls 1/(v)),-(h/4) * i) n2=node(r+c/2+bl-(h/(slope*4))-(h/(slope*4))*i,theta*(Lsl/(v)),-(h/4)-(h/4)*i) n3=node(r+c/2+.75*bl-(((h/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/ 4)-(h/4)*i) n4=node(r+c/2+.75*bl-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)*i) type,l real, 14 en, 16365+i,nl ,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r+c/2+.75*bl-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)*i) n2=node(r+c/2+.75*bl-(((li/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-h/ 4-(h/4)*i) n3=node(r+c/2+(bl/4)*2,theta*(Lsl/(v)),-h/4-(h/4)*i) n4=node(r+c/2+(b 1/4)*2,theta*(Ls 1/(v)),-(h/4)* i) type,l real, 14 en, 16369+i,n 1,n2 ,n3 ,n4 *enddo

*do,i,0,3

250

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n 1=node(r+c/2+(b 1/4)+((((h/slope)+(bb/4))-(b 1/4))/4)*i,theta* (Lsl /(v)),-(h/4)*i)


n2=node(r+c/2+(b 1/ 4)+(((h/slope)+(bb/4))-(b 1/4))/4+((((h/slope)+(bb/4))-(b 1/4))/4)*i,theta* (L sl/ (v)),-h/ 4-(h/4)*i) n3=node(r+c/2+(bl/4)*2,theta*(Lsl/(v)),-h/4-(h/4)*i) n4=node(r+c/2+(b 1/ 4)*2,theta* (Ls 1/(v)),-(h/4) *i) type,l real, 14 en,16373+i,n4,n3,n2,nl *enddo

*do,i,0,3 n 1=node(r+c/2+(h/slope/4)* i,theta*(Ls 1/(v)),-(h/4)*i) n2=node(r+c/2+(h/slope/4)+(h/slope/4)* i,theta* (Ls 1/ (v)),-h/4-(h/4)*i) n3=node(r+c/2+(bl/4)+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-h/ 4-(h/4)*i) n4=node(r+c/2+(bl/4)+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(Lsl/(v)),-(h/4)*i) type, 1 real, 14 en,16377+i,4,n3,n2,nl *enddo

!Elements !West B ent *do,i,0,3 n 1=node(r+c/2+b 1-(h/(slope*4))*i,theta* (L/(v)),-(h/4) *i)

251

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n2=node(r+c/2+bl-(h/(slope*4))-(h/(slope*4))*i,theta*(L/(v)),-(h/4)-(h/4)*i)
n3=node(r+c/2+.75 *b 1-(((h/slope)+(bb/4))-(bl/4))/4-((((h/slope)+(bb/4))-(b 1/4))/4)*i,theta*(L/(v)),-(h/4)(h/4)*i)
n 4 = n o d e ( r + c / 2 + .7 5 *b 1-((((h/slope)+(bb/4))-(b 1/4))/4) *i,theta* (L/(v)),-(h/4)* i )

type,l real, 14 en, 16381 +i,n 1,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r+c/2+.75*bl-((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i) n2=node(r+c/2+.75 *b 1-(((h/slope)+(bb/4))-(b 1/4))/4-((((h/slope)+(bb/4))-(b 1/4))/4)*i,theta*(L/(v)),-h/4(h/4)*i) n3 =node(r+c/2+(b 1/4) *2,theta* (L/(v)),-h/4-(h/4) *i) n4=node(r+c/2+(b 1/4) *2,theta* (L/ (v)),-(h/4) *i) type,l real, 14 en, 163 85+i,n 1,n2,n3 ,n4 *enddo

*do,i,0,3 nl=node(r+c/2+(bl/4)+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i) n2=node(r+c/2+(bl/4)+(((h/slope)+(bb/4))-(bl/4))/4+((((h/sIope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-h/4(h/4)*i) n3=node(r+c/2+(bl/4)*2,theta*(L/(v)),-h/4-(h/4)*i)

252

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n4=node(r+c/2+(bl/4)*2,theta*(L/(v)),-(h/4)*i)
type,l real, 14 en,16389+i,n4,n3,n2,nl *enddo

*do,i,0,3 n 1=node(r+c/2+(h/slope/4)*i,theta*(L/(v)),-(h/4)* i) n2=node(r+c/2+(h/slope/4)+(h/slope/4)*i,theta*(L/(v)),-h/4-(h/4)*i) n3=node(r+c/2+(bl/4)+(((h/slope)+(bb/4))-(bl/4))/4+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-h/4(h/4)*i) n4=node(r+c/2+(bl/4)+((((h/slope)+(bb/4))-(bl/4))/4)*i,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 16393+i,n4,n3,n2,n 1 *enddo

!!!!!!!!!!!!!!!!!!!!!!!!!!!External Solid Diaphragms!!!!!!!!!!! !East Bent solid diaphragm !nodes fill, 1,30,3,14580,5 fill,5,26,3,14584,5 fill, 14580,14584 fill,14585,14589 fill, 14590,14594

253

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

!Internal B ent !nodes fill,(((Lsl/(v))+l)*34)-33,(((Lsl/(v))+l)*34)-4,3,14595,5 fill,(((Lsl/(v))+l)*34)429,(((Lsl/(v))+l)*34)-8,3,14599,5 fill, 14595,14599 fill, 14600,14604 fill,14605,14609

iWest Bent inodes fill,(((L/(v))+l)*34)-33,(((L/(v))+l)*34)-4,3,14610,5 fill,(((L/(v))+l)*34)-29,(((L/(v))+l)*34)-8,3,14614,5 fill,14610,14614 fill,14615,14619 fill,14620,14624

iE ast Bent ielements *do,i,0,3 nl=node(r+C/2+(h/(slope*4))*i,theta*Q,-(h/4)*i) n2=node(r+c/2+h/(4*slope)+(h/(slope*4))*i,theta*0,-h/4-(h/4)*i) n3=node(r+c/2-c/4+h/(8*slope)+(h/(8*slope))*i,theta*0,-h/4-(h/4)*i) n4=node(r+c/2-c/4+(h/(8*slope))*i,theta*0,-(h/4)*i) type,l

254

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

real, 14
en,16397+i,nl ,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r+c/2-c/4+h/(8*slope)+(h/(8*slope))*i,theta*0,-li/4-(h/4)*i) n2=node(r+c/2-c/ 4+(h/(8* slope)) *i,theta*0,-(h/4) *i) n3=node(r,theta*0,-h/4-(h/4)*i) n4=node(r,theta*0,-(h/4)*i) type,l real, 14 en, 16401 +i,n2,n 1,n3 ,n4 *enddo

*do,i,0,3 n 1=node(r-c/4+(h / (8 * slope))* i,theta* 0,-(h/4)*i) n2=node(r-c/4+h/(8*slope)+(h/(8*slope))*i,theta*0,-h/4-(h/4)*i) n3=node(r,theta*0,-h/4-(h/4)*i) n4=node(r,theta*0,-(h/4)*i)

type,l
real, 14 en, 16405+i,n4,n3,n2,n 1 *enddo

*do,i,0,3

255

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

nl=node(r-c/4-(h/(8*slope))*i,theta*0,-(h/4)*i) n2=node(r-c/4-h/(8*slope)-(h/(8*slope))*i,theta*0,-h/4-(h/4)*i) n3=node(r-c/2-(h/(4*slope))*i,theta*0,-(h/4)*i) n4=node(r-c/2-(h/(4*slope))-(h/(4*slope))*i,theta*0,-h/4-(h/4)*i) type,l real, 14 en, 16409+i,n 1,n2,n4,n3 *enddo

!Internal Bent lelements *do,i,0,3 n 1=node(r+C/2+(h/(slope *4)) *i,theta*(Ls 1/ (v)),-(h/4)*i) n2=node(r+c/2+li/(4*slope)+(h/(slope*4))*i,theta*(Lsl/(v)),-h/4-(h/4)*i) n3=node(r+c/2-c/4+h/(8*slope)+(h/(8* slope))*i,theta*(Ls l/(v)),-h/4-(h/4)*i) n4=node(r+c/2-c/4+(h/(8*slope))*i,theta*(Lsl/(v)),-(h/4)*i) type,l real, 14 en, 16413+i,n 1,n2,n3 ,n4 *enddo

*do,i,0,3 nl=node(r+c/2-c/4+h/(8*slope)+(h/(8*slope))*i,theta*(Lsl/(v)),-h/4-(h/4)*i) n2=node(r+c/2-c/4+(h/(8*slope))*i,theta*(Lsl/(v)),-(h/4)*i) n3=node(r,theta*(Lsl/(v)),-h/4-(h/4)*i)

256

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n4=node(r,theta*(Ls 1/(v)),-(h/4) *i)


type, 1 real, 14 en, 16417+i,n2,n 1,n3 ,n4 *enddo

*do,i,0,3 nl=node(r-c/4+(h/(8*slope))*i,theta*(Lsl/(v)),-(h/4)*i) n2=node(r-c/4+h/(8 *slope)+(h/( 8 *slope))* i,theta* (Lsl/(v)),-h/4-(h/4)*i) n3=node(r,theta*(Lsl/(v)),-h/4-(h/4)*i) n4=node(r,theta*(Lsl/(v)),-(h/4)*i) type,l real, 14 en, 16421 +i,n4,n3 ,n2,n 1 *enddo

*do,i,0,3 n 1=node(r-c/4-(h/(8 *slope))* i,theta*(Ls l/(v)),-(h/4)*i) n2=node(r-c/4-h/(8*slope)-(h/(8*slope))*i,theta*(Lsl/(v)),-h/4-(h/4)*i) n3=node(r-c/2-(h/(4 *slope))*i,theta*(Ls 1/(v)),-(h/4)* i) n4=node(r-c/2-(h/(4*slope))-(h/(4*slope))*i,theta*(Lsl/(v)),-h/4-(h/4)*i) type,l real, 14 en, 16425+i,nl ,n2,n4,n3 *enddo

257

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

!West Bent !elements *do,i,0,3 nl=node(r+C/2+(h/(slope*4))*i,theta*(L/(v)),-(h/4)*i) n2=node(r+c/2+h/(4*slope)+(h/(slope*4))*i,theta*(L/(v)),-h/4-(h/4)*i) n3=node(r+c/2-c/4+h/(8 *slope)+(h/(8*slope))*i,theta*(L/(v)),-h/4-(h/4)*i) n4=node(r+c/2-c/4+(h/(8*slope))*i,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 16429+i,nl ,n2,n3,n4 *enddo

*do,i,0,3 nl=node(r+c/2-c/4+h/(8*slope)+(h/(8*slope))*i,theta*(L/(v)),-h/4-(h/4)*i) n2=node(r+c/2-c/4+(h/(8*slope))*i,theta*(L/(v)),-(h/4)*i) n3 =node(r,theta* (L/(v)),-h/4-(h/4)* i) n4=node(r,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 1643 3+i,n2,n 1,n3,n4 *enddo

*do,i,0,3 nl=node(r-c/4+(h/(8*slope))*i,theta*(L/(v)),-(h/4)*i)

258

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

n2=node(r-c/4+h/(8*slope)+(h/(8*slope))*i,theta*(L/(v)),-h/4-(h/4)*i)
n3=node(r,theta*(L/(v)),-h/4-(h/4)*i) n4=node(r,theta*(L/(v)),-(h/4)*i) type,l real, 14 en, 16437+i,n4,n3 ,n2,n 1 *enddo

*do,i,0,3 n 1=node(r-c/4-(h/(8 *slope))*i,theta*(L/(v)),-(h/4)*i) n2=node(r-c/4-h/(8*slope)-(h/(8*slope))*i,theta*(L/(v)),-h/4-(h/4)*i) n3=node(r-c/2-(h/(4 *slope)) *i,theta* (L/(v)),-(h/4)* i) n4=node(r-c/2-(h/(4*slope))-(h/(4*slope))*i,theta*(L/(v)),-h/4-(h/4)*i) type,l real, 14 en, 16441 +i,nl ,n2,n4,n3 *enddo

!!!!!!!!!!!!!!!!!!!!!!Boundary Conditions!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! nrotat,5 nrotat,9 nrotat,22 nrotat,26 nrotat, 34*((Lsl/(v))+1)-29

259

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

nrotat,34*((Ls 1/(v))+l)-25 nrotat,3 4 *((Ls 1/(v))+1)-12 nrotat,3 4 *((Ls 1/(v))+1)- 8 nrotat,34*((L/'(v))+1)-29 nrotat,3 4 *((L/(v))+1)-2 5 nrotat, 34 *((L/(v))+1)-12 nrotat,34*((L/(v))+1)-8

d,5,ux,0uz d,9,ux,0uz d,22 ,ux,0 uz d,26,ux,0uz d,3 4 *((L/(v))+1)- 8 ,ux, 0uy,uz d,34*((L/(v))+l)-12,ux,0,,uy,uz d,34*((L/(v))+l)-25,ux,0uy,uz d,34*((L/(v))+l)-29,ux,0uy,uz d,34*((Lsl/(v))+l)-8,ux,0uy,uz d,34*((L sl/(v))+l)-12,ux,0,,uy,uz d,34*((Lsl/(v))+l)-25,ux,0uy,uz d,34 *((Ls 1/(v))+1)-29,ux,0uy,uz

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!Loading!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!!! lSelf-w eight!!!!!!!!!!!!!!!!!!!!! *if,simple-analysis,EQ,yes,then

260

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

*if,selfwt,EQ,''yes', then /SOLU ANTYPE, STAT,new SOLCONTROL,ON OUTRES,ALL,all Acel,0,9800,0 tim e,l NSUBST, 1,100,1 NCNV, 1 allsel fcum,add save solve save finish *endif

!!!!!!Construction Load!!!!!!!!! !!!!!!!!!! Self-weigfat+Stagel *if, stage 1,Eq,'yes',then /SOLU ANTYPE,STAT,new SOLCONTROL,ON OUTRES,ALL,all

261

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Acel,0,9800,0 tim e,l NSUBST, 1,100,1 NCNV,1 allsel fcum,add f,l,fy,-Fd3,,(((cl/(v))+l)*34)-33 ,34 f,13,fy,-Fd4 (((cl/(v))+l)*34)-21 ,34 f,18,fy,-Fdl (((cl/(v))+l)*34)-16 ,34 f,30,fy,-Fd2 (((cl/(v ))+ 1)*34)-4 ,34 save solve save finish *endif

!!!!!!!!!! Self-weight+Stagel+stage2 *if, stage2,Eq,'yes',then /SOLU ANTYPE, STAT,new SOLCONTROL,ON

OUTRES,ALL,all A cel,0,9800,0 tim e,l

262

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

NSUBST, 1,100,1 NCNV,1 allsel fcum,add f,l,fy,-Fd3(((cl/(v))+l)*34)-33 ,34 f,13,fy,-Fd4 (((cl/(v))+l)*34)-21 ,34 f,18,fy,-Fdl (((cl/(v))+l)*34)-16 ,34 f,30,fy,-Fd2 (((c 1/(v))+1)*34)-4 ,34

f,(((c3/(v))+1)*34)-33,fy,-Fd3(((c4/(v))+1)*34)-33 ,34 f,(((c3/(v))+l)*34)-21,fy,-Fd4 (((c4/(v))+l)*34)-21 ,34 f,(((c3/(v))+1)*34)-16,fy,-Fd 1 (((c4/(v))+l)*34)-16 ,34 f,(((c3/(v))+l)*34)-4 ,fy,-Fd2 (((c4/(v))+l)*34)-4 ,34 save solve save finish *endif

!!!!!!!!!! Self-weight+stagel+stage2+stage3
*if,stage3,Eq,'yes',then /SOLU ANTYPE,STAT,new SOLCONTROL,ON OUTRES,ALL,all

263

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Acel,0,9800,0 tim e,l NSUBST, 1,100,1 NCNV,1 allsel fcum,add f,l,fy,-Fd3(((cl/(v))+l)*34)-33 ,34 f,13,fy,-Fd4 (((cl/(v))+l)*34)-21 ,34 f,18,fy,-Fdl (((cl/(v))+l)*34)-16 ,34 f,30,fy,-Fd2 (((cl/(v))+l)*34)-4 ,34

f,(((c3/(v))+l)*34)-33,fy,-Fd3(((c 4 /(v))+1)*34)-33 ,34 f,(((c3/(v))+l)*34)-21,fy,-Fd4 (((c4/(v))+l)*34)-21 ,34 f,(((c3/(v))+1)*34)-16,fy,-Fd 1 (((c4/(v))+l)*34)-16 ,34 f,(((c3/(v))+1)*34)-4 ,fy,-Fd2 (((c4/(v))+l)*34)-4 ,34

f,(((cl/(v))+l)*34)-33,fy,-Fd3(((c2/(v))+l)*34)-33 ,34 f,(((c 1/(v))+1)*34)-21,fy,-Fd4 (((c2/(v))+l)*34)-21 ,34 f,(((cl/(v))+l)*34)-16,fy,-Fdl (((c2/(v))+l)*34)-16 ,34 f,(((cl/(v))+l)*34)-4 ,fy,-Fd2 (((c2/(v))+l)*34)-4 ,34

save solve save finish

264

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

*endif

!!!!!!!!!! Self-weight+Total Construction Load


*if,stage4,Eq,'yes',then /SOLU ANTYPE,STAT,new SOLCONTROL,ON OUTRES,ALL,alI Acel,0,9800,0 time,l NSUBST, 1,100,1 NCNV, 1 allsel fcum,add f ,l 5 fy,-Fd3(((c4/(v))+l)*34)-33 ,34 f,13,fy,-Fd4 (((c4/(v))+1)*34)-21 ,34 f,18,fy,-Fdl (((c4/(v))+1)*34)-16 ,34 f,30,fy,-Fd2 (((c4/(v))+l)*34)-4 ,34 save solve save finish *endif *endif

265

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

!!!!!!!!!!!!!!!!!!!Buckling analysis loading!!!!!!!!!!!!!!!! *if,buckling-analysis,EQ,yes,then !!!!!!!!!!!!Construction Load

!!!!!!!!!JFirst stage+Self W eight!!!!!!!!!!!!


*if, stage 1,Eq,'yes',then /input,stage 1 *endif

!!!!!! Stagel+Stage2+Self W eight!!!!!!!!!!!!! *if, stage2,Eq,'yes' ,then /input, stage 2 *endif

!!!!!!! Stagel+Stage2+Stage3+Self Weight *if,stage3 ,Eq,'yes',then /input,stage3 *endif

!!!!!!!!Total construction Load(stagesl+2+3+4) plus Self Weight *if,stage4,Eq,'yes',then /input,stage4 *endif *endif

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!I!!!I!I!IIStagel Text File List!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! /SOLU

ANTYPE,STAT,new NLGEOM ,ON SOLCONTROL,ON OUTRES,ALL,all nropt,full acel,0,9800,0 TIME,1 NSUBST, 1,100,1 NCNV,1 save solve

i=0 f,l,fy,-Fd3(((cl/(v))+l)*34)-33,34 f,13,fy,-Fd4 (((cl/(v))+l)*34)-21,34 f,18,fy,-Fdl (((cl/(v))+ l)*34)-16,34 f,30,fy,-Fd2 (((cl/(v))+l)*34)-4,34 allsel fcum,add OUTRES,ALL,all TIME,2+i NSUBST, 10,100,1

267

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

NCNV,1 SAVE SOLVE SAVE i=l f,l,fy,-Fd3,,(((cl/(v))+l)*34)-33,34 f, 13,fy,-Fd4 (((c 1/(v))+1)*34)-21,34 f j l 8 ,fy,-Fdl (((cl/(v))+l)*34)-16,34 f,30,fy,-Fd2 (((cl/(v))+l)*34)-4,34 allsel fcum,add arclen,on ----------------------------------------------------- Arc-length activation command

OUTRES,ALL,all Max. number of sub-steps TIME,2+i NSUBST, 10,100,1 NCNV,1 SAVE SOLVE SAVE i=2 f,l,fy,-Fd3(((cl/(v))+l)*34)-33,34 f,13,fy,-Fd4 (((cl/(v))+l)*34)-21,34 f,18,fy,-Fdl (((cl/(v))+l)*34)-16,34 f,30,fy,-Fd2 (((cl/(v))+l)*34)-4,34 allsel Number o f sub-steps Min. number o f sub-steps

268

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

fcum,add arclen,on OUTRES,ALL,all TIME,2+i NSUBST, 10,100,1 NCNV,1 SAVE SOLVE SAVE i=3 f,l,fy,-Fd3(((cl/(v))+l)*34)-33,34 f, 13,fy,-Fd4 (((cl/(v))+ 1)*34)-21,34 f,18,fy,-Fdl (((cl/(v))+l)*34)-16,34 f,30,fy,-Fd2 (((cl/(v))+l)*34)-4,34 allsel fcum,add arclen,on OUTRES,ALL,all TIME,2+i NSUBST, 10,100,1 NCNV, 1 SAVE SOLVE SAVE finish

269

R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.

Вам также может понравиться