Beruflich Dokumente
Kultur Dokumente
Nikolaos Kostis
Master of Science
in Structural Engineering,
This current research thesis report is a partial fulfilment of my Master of Science degree requirements. It is done
as the last part of my degree in Structural Engineering at Delft University of Technology, the Netherlands. The
subject of the research is the buckling behaviour of welded steel tubes used in combi-walls both empty and filled
with sand.
The current research report acts as complementary to the experiments performed by the SBRCURnet
investigation programme which was executed in 2013 and had as a goal to study the effect of the sand-fill on the
moment and the ovalization of the combi-wall steel tubes. Specifically, the intention of this research is to
develop a FEM model which describes the 4 point bending test experiments done on both empty and sand-filled
tubes.
Admittedly, this research could not be fulfilled without the help of certain individuals. First of all, my committee
members who helped me enormously along this tiresome but very fascinating academic journey. Their
comments and guidance had shown me my mistakes and the correct path to reach to the end. Especially, Mr.
S.H.J. van Es for his vast support, his insightful and detailed guidance. He was always there to answer any
questions I had over both the conceptual part and the actual building of the FEM model.
Finally, I should not neglect to recognise the support I received from my friends and family during the whole
period of the Master programme. I hope this research will contribute to anyone who wills to focus on the future
research of the sand effect on the bending behaviour of the sand-filled steel tubes.
Nikolaos Kostis
Delft, May 2016
i
ii
Contents
Preface ......................................................................................................................................................... i
Contents .................................................................................................................................................... iii
List of figures ............................................................................................................................................. v
List of tables ............................................................................................................................................... x
Nomenclature ........................................................................................................................................... xi
Acronyms.................................................................................................................................................. xii
Abstract ................................................................................................................................................... xiii
1. Introduction ..................................................................................................................................... 1
1.1 Combined walls ...................................................................................................................................................... 1
1.2 Design rules overview ........................................................................................................................................... 2
1.3 Stress based design vs Strain based design ........................................................................................................ 3
1.4 Previous research on the effect of sand-fill in the steel tubes ........................................................................ 4
1.5 Current situation and CUR investigation objective .......................................................................................... 5
1.6 Thesis objective and research............................................................................................................................... 5
2. Experimental testing ...................................................................................................................... 7
2.1 Purpose and requirements of the experimental test programme ................................................................... 7
2.2 Test design and setup description ....................................................................................................................... 7
2.3 Material properties ............................................................................................................................................... 13
2.4 Definition of Curvature ...................................................................................................................................... 17
3. Analytical formulas ........................................................................................................................ 19
3.1 Moment-curvature diagram analytical approach ............................................................................................. 19
3.2 Moment-curvature diagram without ovalization and residual stresses ........................................................ 19
3.3 Moment-curvature diagram with ovalization effect ....................................................................................... 20
3.4 Moment-curvature diagram with residual stresses effect .............................................................................. 22
3.5 Analytical moment-curvature diagrams ............................................................................................................ 23
3.6 Comments ............................................................................................................................................................. 25
4. FEM model .................................................................................................................................... 26
4.1 General .................................................................................................................................................................. 26
4.2 FEM History ......................................................................................................................................................... 26
4.3 FEA in the current project ................................................................................................................................. 26
4.4 General description of the slice numerical model .......................................................................................... 27
4.5 Material models .................................................................................................................................................... 28
4.6 Residual stresses ................................................................................................................................................... 30
4.7 Boundary conditions, constraint, geometry, load introduction and contact properties ........................... 32
4.8 Element types ....................................................................................................................................................... 33
4.9 Mesh size refinement study ................................................................................................................................ 36
4.10 Solution technique and output parameters ...................................................................................................... 39
4.11 Full scale model description ............................................................................................................................... 40
4.12 Load introduction, Boundary conditions and Geometry of the full scale model...................................... 41
4.13 Solution procedures ............................................................................................................................................. 42
4.14 Element type ......................................................................................................................................................... 42
4.15 Number of Integration Points in the shell elements ...................................................................................... 43
4.16 Mesh refinement .................................................................................................................................................. 43
4.17 Elastic buckling analysis ...................................................................................................................................... 44
5. Model validation ............................................................................................................................ 50
5.1 Comparison of the moment-curvature results of the slice model ............................................................... 50
5.2 Conclusions for the slice model moment-curvature diagrams ..................................................................... 70
5.3 Comparison of the ovalization results for the slice model ............................................................................ 72
5.4 Conclusions for the ovalization results ............................................................................................................ 73
5.5 Stresses in the sand-fill ........................................................................................................................................ 75
5.6 Comparison of the full scale model results...................................................................................................... 85
5.7 Conclusions of the full scale model results.................................................................................................... 104
iii
6. Parametric study of the sand-fill properties with the slice model ............................................... 108
6.1 Results of the variation of the parameter Esand ............................................................................................. 109
6.2 Discussion and conclusions for the Esand parameter .................................................................................... 112
6.3 Results of the variation of the parameter friction angle φ ........................................................................... 113
6.4 Discussion and conclusions for the φ parameter .......................................................................................... 116
6.5 Results of the variation of the parameter dilation angle ψ........................................................................... 117
6.6 Discussion and conclusions for the ψ parameter .......................................................................................... 120
6.7 Results of the variation of the parameter cohesion c ................................................................................... 121
6.8 Discussion and conclusions for the c parameter .......................................................................................... 124
6.9 Results of the variation of the parameter Poisson’s ratio ν ......................................................................... 125
6.10Discussion and conclusions for the ν parameter .......................................................................................... 128
6.11Examining the effect of each parameter on the moment and the ovalization ......................................... 129
7. Conclusions and Recommendations ........................................................................................... 130
7.1 Conclusions ......................................................................................................................................................... 130
7.2 Recommendations ............................................................................................................................................. 132
APPENDICES ........................................................................................................................................ 133
Appendix A ABAQUS contact model ..................................................................................................... 135
Appendix B Photographs of failure mode shapes................................................................................... 137
Appendix C Model implementation in ABAQUS ................................................................................... 144
Bibliography ............................................................................................................................................ 153
iv
List of figures
Figure 1: Combined wall [22] ................................................................................................................................................. 1
Figure 2: Combined wall types [7] ......................................................................................................................................... 1
Figure 3: Basis for Stress based design approach [15] ........................................................................................................ 3
Figure 4: Basis for Strain based design approach [15] ........................................................................................................ 3
Figure 5: Tube bending ........................................................................................................................................................... 4
Figure 6: Load-Displacement curves of concrete and sand-filled steel tubes [16] ........................................................ 4
Figure 7: Setup of the experimental test [17]. ...................................................................................................................... 7
Figure 8: Setup of the test with the position of the displacement gauges [17]. .............................................................. 8
Figure 9: Test tube and end pieces drawing with exaggerated deformation [17]. .......................................................... 9
Figure 10: End-piece drawing [17]....................................................................................................................................... 10
Figure 11: Cross-section view of end-piece with end plate [17]. .................................................................................... 10
Figure 12: Pictures from the test execution, the jack system and the guidance profiles [17] ..................................... 11
Figure 13: Photographs of typical buckling shapes in early (top) and in advanced (bottom) stage of empty (left)
and filled (right) tubes [17] .................................................................................................................................................... 12
Figure 14: Type B steel tube stress-strain diagram [17].................................................................................................... 13
Figure 15: Type D steel tube stress-strain diagram [17] ................................................................................................... 14
Figure 16: Sieve curve of the sand-fill [17] ......................................................................................................................... 15
Figure 17: Curvature definition ............................................................................................................................................ 17
Figure 18: View of the buckled tube ................................................................................................................................... 18
Figure 19: Stress distribution in cross-section in bending and moment-curvature diagram without influences
taken into account [12] .......................................................................................................................................................... 19
Figure 20: Ovalization of the steel tube during bending [19] .......................................................................................... 20
Figure 21: Ovalization forces [11] ....................................................................................................................................... 20
Figure 22: Yield surface [12] ................................................................................................................................................. 20
Figure 23: Comparison of the moment-curvature diagrams when ovalization is taken into account and when not
................................................................................................................................................................................................... 21
Figure 24: Comparison of moment-curvature diagrams with the effect of ovalization and residual stresses ......... 22
Figure 25: Analytical curves and test results for tube A................................................................................................... 23
Figure 26: Analytical curves and test results for tube B ................................................................................................... 23
Figure 27: Analytical curves and test results for tube C ................................................................................................... 24
Figure 28: Analytical curves and test results for tube D .................................................................................................. 24
Figure 29: Undeformed and Deformed ovalization model ............................................................................................. 27
Figure 30: True plastic and engineering stress-strain diagrams ....................................................................................... 28
Figure 31: Stress-Strain diagram for tube C ....................................................................................................................... 29
Figure 32: Mohr-Coulomb Plasticity model for the sand-fill [1] .................................................................................... 30
Figure 33: Numerical model and basic geometric characteristics [20]. .......................................................................... 30
Figure 34: Residual stress distribution ................................................................................................................................ 31
Figure 35: Boundary conditions of the FEM model ........................................................................................................ 32
Figure 36: Numbering of the nodes and the integration points of a shell element in ABAQUS [1] ....................... 33
Figure 37: Moment-curvature diagram of type A empty tube ........................................................................................ 33
Figure 38: Ovalization-curvature diagram of type A empty tube ................................................................................... 34
Figure 39: Numbering of the nodes and the integration points of a solid element in ABAQUS [1] ....................... 35
Figure 40: Moment-curvature diagram of type A sand-filled tube ................................................................................. 35
Figure 41: Ovalization-curvature diagram of type A sand-filled tube............................................................................ 35
Figure 42: Moment-curvature diagram of type A empty tube for mesh size sensitivity analysis of the shell
element ..................................................................................................................................................................................... 36
Figure 43: Ovalization-curvature diagram of type A empty tube for mesh size sensitivity analysis of the shell
element ..................................................................................................................................................................................... 36
Figure 44: Mesh refinement study for empty tube A of the shell element ................................................................... 37
Figure 45: Moment-curvature diagram of type A empty tube for mesh size sensitivity analysis of the solid
element ..................................................................................................................................................................................... 37
v
Figure 46: Ovalization-curvature diagram of type A empty tube for mesh size sensitivity analysis of the solid
element ..................................................................................................................................................................................... 38
Figure 47: Mesh refinement study for the sand-filled tube A of the solid element ..................................................... 38
Figure 48: Geometry of the full scale model ..................................................................................................................... 40
Figure 49: Boundary Conditions of the full scale model ................................................................................................. 41
Figure 50: Boundary Conditions of the left edge of the model ...................................................................................... 41
Figure 51: Element type effect on the moment-curvature for Tube A ......................................................................... 42
Figure 52: Effect of the integration point number on the moment and curvature ..................................................... 43
Figure 53: Mesh size effect for empty tube A of the shell element S4R ....................................................................... 44
Figure 54: First 6 buckling modes of Tube B .................................................................................................................... 44
Figure 55: Mode 1 of Tube B ............................................................................................................................................... 45
Figure 56: Mode 2 of Tube B ............................................................................................................................................... 45
Figure 57: Mode 3 of Tube B ............................................................................................................................................... 45
Figure 58: Mode 4 of Tube B ............................................................................................................................................... 46
Figure 59: Mode 5 of Tube B ............................................................................................................................................... 46
Figure 60: Mode 6 of Tube B ............................................................................................................................................... 46
Figure 61: Tube B buckle (empty tube) of Mode 1(left) and Mode 2(right)................................................................. 47
Figure 62: Tube B buckle (empty tube) - Test................................................................................................................... 47
Figure 63: Tube B buckle (sand-filled tube) of Mode 1 ................................................................................................... 48
Figure 64: Buckle of a sand-filled tube - Test .................................................................................................................... 48
Figure 65: Imperfection height in the buckling mode ...................................................................................................... 49
Figure 66: Steel tube surface after the input of the mode in the Static analysis ........................................................... 49
Figure 67: Moment-Curvature diagram for tube A3 (empty) ......................................................................................... 51
Figure 68: Moment-Curvature diagram for tube A4 (empty) ......................................................................................... 51
Figure 69: Moment-Curvature diagram for tube A6 (empty) ......................................................................................... 51
Figure 70: Moment-Curvature diagram for tube A (empty)............................................................................................ 52
Figure 71: Moment-Curvature diagram for tube B2 (empty) .......................................................................................... 53
Figure 72: Moment-Curvature diagram for tube B5 (empty) .......................................................................................... 53
Figure 73: Moment-Curvature diagram for tube B6 (empty) .......................................................................................... 53
Figure 74: Moment - Curvature diagram for tube B (empty).......................................................................................... 54
Figure 75: Moment-Curvature diagram for tube C2 (empty) .......................................................................................... 55
Figure 76: Moment-Curvature diagram for tube C3 (empty) .......................................................................................... 55
Figure 77: Moment-Curvature diagram for tube C6 (empty) .......................................................................................... 55
Figure 78: Moment - Curvature diagram for tube C (empty).......................................................................................... 56
Figure 79: Moment-Curvature diagram for tube D1 (empty) ......................................................................................... 57
Figure 80: Moment-Curvature diagram for tube D2 (empty) ......................................................................................... 57
Figure 81: Moment-Curvature diagram for tube D3 (empty) ......................................................................................... 57
Figure 82: Moment - Curvature diagram for tube D (empty) ......................................................................................... 58
Figure 83: Moment-Curvature diagram for tube A1 (sand-filled) .................................................................................. 59
Figure 84: Moment-Curvature diagram for tube A2 (sand-filled) .................................................................................. 59
Figure 85: Moment-Curvature diagram for tube A5 (sand-filled) .................................................................................. 59
Figure 86: Moment - Curvature diagram for tube A (sand-filled) .................................................................................. 60
Figure 87: Moment - Curvature diagram for tube B1 (sand-filled) ................................................................................ 61
Figure 88: Moment - Curvature diagram for tube B3 (sand-filled) ................................................................................ 61
Figure 89: Moment - Curvature diagram for tube B4 (sand-filled) ................................................................................ 61
Figure 90: Moment - Curvature diagram for tube B (sand-filled) .................................................................................. 62
Figure 91: Moment - Curvature diagram for tube C1 (sand-filled) ................................................................................ 63
Figure 92: Moment - Curvature diagram for tube C4 (sand-filled) ................................................................................ 63
Figure 93: Moment - Curvature diagram for tube C5 (sand-filled) ................................................................................ 63
Figure 94: Moment - Curvature diagram for tube C (sand-filled) .................................................................................. 64
Figure 95: Moment - Curvature diagram for tube D4 (sand-filled) ............................................................................... 65
Figure 96: Moment - Curvature diagram for tube D5 (sand-filled) ............................................................................... 65
Figure 97: Moment - Curvature diagram for tube D6 (sand-filled) ............................................................................... 65
Figure 98: Moment - Curvature diagram for tube D (sand-filled).................................................................................. 66
Figure 99: Moment - Curvature diagram for tube A (empty and sand-filled) .............................................................. 68
Figure 100: Moment - Curvature diagram for tube B (empty and sand-filled) ............................................................ 68
vi
Figure 101: Moment - Curvature diagram for tube C (empty and sand-filled) ............................................................ 69
Figure 102: Moment - Curvature diagram for tube D (empty and sand-filled)............................................................ 69
Figure 103: Ovalization comparison between measurements and ABAQUS analyses for empty tubes. ................ 72
Figure 104: Ovalization comparison between measurements and ABAQUS analyses for soil-filled tubes. ........... 72
Figure 105: Ovalization comparison between measurements and ABAQUS analyses for soil-filled tubes with
different sand properties. ...................................................................................................................................................... 74
Figure 106: Sections chosen for the sand stress depiction .............................................................................................. 75
Figure 107: S11 (σxx) of Sections 1 and 2............................................................................................................................ 76
Figure 108: S11 (σxx) of Sections 3 and 4............................................................................................................................ 76
Figure 109: S22 (σyy) of Sections 1 and 2 ............................................................................................................................ 76
Figure 110: S22 (σyy) of Sections 3 and 4 ............................................................................................................................ 77
Figure 111: S33 (σzz) of Sections 1 and 2 ............................................................................................................................ 77
Figure 112: S33 (σzz) of Sections 3 and 4 ............................................................................................................................ 77
Figure 113: Node location used to the p-q diagrams ....................................................................................................... 80
Figure 114: Top node p-q diagram ...................................................................................................................................... 81
Figure 115: Mid top node p-q diagram ............................................................................................................................... 81
Figure 116: Central node p-q diagram ................................................................................................................................ 81
Figure 117: Mid bottom node p-q diagram ........................................................................................................................ 82
Figure 118: Bottom node p-q diagram................................................................................................................................ 82
Figure 119: Left node p-q diagram ...................................................................................................................................... 82
Figure 120: Mid left node p-q diagram ............................................................................................................................... 83
Figure 121: Mid right node p-q diagram............................................................................................................................. 83
Figure 122: Right node p-q diagram.................................................................................................................................... 83
Figure 123: Moment-Curvature diagrams of Tube A of the full scale model (empty). .............................................. 86
Figure 124: Ovalization-Curvature diagrams of Tube A of the full scale model (empty). ......................................... 86
Figure 125: Moment-Curvature diagrams of Tube B of the full scale model (empty). ............................................... 87
Figure 126: Ovalization-Curvature diagrams of Tube B of the full scale model (empty). ......................................... 87
Figure 127: Moment-Curvature diagrams of Tube C of the full scale model (empty)................................................ 88
Figure 128: Ovalization-Curvature diagrams of Tube C of the full scale model (empty). ......................................... 88
Figure 129: Moment-Curvature diagrams of Tube D of the full scale model (empty). .............................................. 89
Figure 130: Ovalization-Curvature diagrams of Tube D of the full scale model (empty). ......................................... 89
Figure 131: Moment-Curvature diagrams of Tube A of the full scale model (sand-filled). ....................................... 90
Figure 132: Ovalization-Curvature diagrams of Tube A of the full scale model (sand-filled)................................... 90
Figure 133: Moment-Curvature diagrams of Tube B of the full scale model (sand-filled). ....................................... 91
Figure 134: Ovalization-Curvature diagrams of Tube B of the full scale model (sand-filled). .................................. 91
Figure 135: Moment-Curvature diagrams of Tube C of the full scale model (sand-filled). ....................................... 92
Figure 136: Ovalization-Curvature diagrams of Tube C of the full scale model (sand-filled). .................................. 92
Figure 137: Moment-Curvature diagrams of Tube D of the full scale model (sand-filled)........................................ 93
Figure 138 Ovalization-Curvature diagrams of Tube D of the full scale model (sand-filled). .................................. 93
Figure 139: Moment-Curvature diagrams for empty and sand-filled Tube A for different Imp/Thick percentages.
................................................................................................................................................................................................... 94
Figure 140: Ovalization-Curvature diagrams for empty and sand-filled Tube A for different Imp/Thick
percentages. ............................................................................................................................................................................. 94
Figure 141: Moment-Curvature diagrams for empty and sand-filled Tube A for different Imp/Thick percentages.
................................................................................................................................................................................................... 95
Figure 142: Moment-Curvature diagrams for empty and sand-filled Tube B for different Imp/Thick percentages.
................................................................................................................................................................................................... 96
Figure 143: Ovalization-Curvature diagrams for empty and sand-filled Tube B for different Imp/Thick
percentages. ............................................................................................................................................................................. 96
Figure 144: Moment-Curvature diagrams for empty and sand-filled Tube B for different Imp/Thick percentages.
................................................................................................................................................................................................... 97
Figure 145: Moment-Curvature diagrams for empty and sand-filled Tube C for different Imp/Thick percentages.
................................................................................................................................................................................................... 98
Figure 146: Ovalization-Curvature diagrams for empty and sand-filled Tube C for different Imp/Thick
percentages. ............................................................................................................................................................................. 98
vii
Figure 147: Moment-Curvature diagrams for empty and sand-filled Tube C for different Imp/Thick percentages.
................................................................................................................................................................................................... 99
Figure 148: Moment-Curvature diagrams for empty and sand-filled Tube D for different Imp/Thick percentages.
................................................................................................................................................................................................. 100
Figure 149: Ovalization-Curvature diagrams for empty and sand-filled Tube D for different Imp/Thick
percentages. ........................................................................................................................................................................... 100
Figure 150: Moment-Curvature diagrams for empty and sand-filled Tube D for different Imp/Thick percentages.
................................................................................................................................................................................................. 101
Figure 151: Moment-Curvature diagrams for sand-filled Tube A for different Imp/Thick percentages and sand-
fill properties. ........................................................................................................................................................................ 102
Figure 152: Ovalization-Curvature diagrams for sand-filled Tube A for different Imp/Thick percentages and
sand-fill properties................................................................................................................................................................ 102
Figure 153: Moment-Curvature diagrams for sand-filled Tube D for different Imp/Thick percentages and sand-
fill properties. ........................................................................................................................................................................ 103
Figure 154: Ovalization-Curvature diagrams for sand-filled Tube D for different Imp/Thick percentages and
sand-fill properties................................................................................................................................................................ 103
Figure 155: Mmax/Mpl - D/t for all Tube types ................................................................................................................ 107
Figure 156: crit. Curvature - D/t for all Tube types ....................................................................................................... 107
Figure 157: Ovalization/Diameter - D/t for all Tube types ......................................................................................... 107
Figure 158: Moment-Curvature diagram of Tube 1 for different Esand ....................................................................... 109
Figure 159: Ovalization-Curvature diagram of Tube 1 for different Esand .................................................................. 109
Figure 160: Moment-Curvature diagram of Tube 2 for different Esand ....................................................................... 109
Figure 161: Ovalization-Curvature diagram of Tube 2 for different Esand .................................................................. 110
Figure 162: Moment-Curvature diagram of Tube 3 for different Esand ....................................................................... 110
Figure 163: Ovalization-Curvature diagram of Tube 3 for different Esand .................................................................. 110
Figure 164: Moment-Curvature diagram of Tube 4 for different Esand ....................................................................... 111
Figure 165: Ovalization-Curvature diagram of Tube 4 for different Esand .................................................................. 111
Figure 166: Effect of Esand on the Moment of the Tubes.............................................................................................. 111
Figure 167: Effect of Esand on the Ovalization of the Tubes ........................................................................................ 112
Figure 168: Moment-Curvature diagram of Tube 1 for different friction angles ...................................................... 113
Figure 169: Ovalization-Curvature diagram of Tube 1 for different friction angles ................................................. 113
Figure 170: Moment-Curvature diagram of Tube 2 for different friction angles ...................................................... 113
Figure 171: Ovalization-Curvature diagram of Tube 2 for different friction angles ................................................. 114
Figure 172: Moment-Curvature diagram of Tube 3 for different friction angles ...................................................... 114
Figure 173: Ovalization-Curvature diagram of Tube 3 for different friction angles ................................................. 114
Figure 174: Moment-Curvature diagram of Tube 4 for different friction angles ...................................................... 115
Figure 175: Ovalization-Curvature diagram of Tube 4 for different friction angles ................................................. 115
Figure 176: Effect of friction angle on the Moment of the Tubes .............................................................................. 115
Figure 177: Effect of friction angle on the Ovalization of the Tubes ......................................................................... 116
Figure 178: Moment-Curvature diagram of Tube 1 for different dilation angles ...................................................... 117
Figure 179: Ovalization-Curvature diagram of Tube 1 for different dilation angles................................................. 117
Figure 180: Moment-Curvature diagram of Tube 2 for different dilation angles ...................................................... 117
Figure 181: Ovalization-Curvature diagram of Tube 2 for different dilation angles................................................. 118
Figure 182: Moment-Curvature diagram of Tube 3 for different dilation angles ...................................................... 118
Figure 183: Ovalization-Curvature diagram of Tube 3 for different dilation angles................................................. 118
Figure 184: Moment-Curvature diagram of Tube 4 for different dilation angles ...................................................... 119
Figure 185: Ovalization-Curvature diagram of Tube 4 for different dilation angles................................................. 119
Figure 186: Effect of dilation angle on the Moment of the Tubes .............................................................................. 119
Figure 187: Effect of dilation angle on the Ovalization of the Tubes ......................................................................... 120
Figure 188: Moment-Curvature diagram of Tube 1 for different cohesion values ................................................... 121
Figure 189: Ovalization-Curvature diagram of Tube 1 for different cohesion values .............................................. 121
Figure 190: Moment-Curvature diagram of Tube 2 for different cohesion values ................................................... 121
Figure 191: Ovalization-Curvature diagram of Tube 2 for different cohesion values .............................................. 122
Figure 192: Moment-Curvature diagram of Tube 3 for different cohesion values ................................................... 122
Figure 193: Ovalization-Curvature diagram of Tube 3 for different cohesion values .............................................. 122
Figure 194: Moment-Curvature diagram of Tube 4 for different cohesion values ................................................... 123
viii
Figure 195: Ovalization-Curvature diagram of Tube 4 for different cohesion values .............................................. 123
Figure 196: Effect of cohesion on the Moment of the Tubes ...................................................................................... 123
Figure 197: Effect of cohesion on the Ovalization of the Tubes................................................................................. 124
Figure 198: Moment-Curvature diagram of Tube 1 for different Poisson’s ratio values ......................................... 125
Figure 199: Ovalization-Curvature diagram of Tube 1 for different Poisson’s ratio values .................................... 125
Figure 200: Moment-Curvature diagram of Tube 2 for different Poisson’s ratio values ......................................... 125
Figure 201: Ovalization-Curvature diagram of Tube 2 for different Poisson’s ratio values .................................... 126
Figure 202: Moment-Curvature diagram of Tube 3 for different Poisson’s ratio values ......................................... 126
Figure 203: Ovalization-Curvature diagram of Tube 3 for different Poisson’s ratio values .................................... 126
Figure 204: Moment-Curvature diagram of Tube 4 for different Poisson’s ratio values ......................................... 127
Figure 205: Ovalization-Curvature diagram of Tube 4 for different Poisson’s ratio values .................................... 127
Figure 206: Effect of Poisson’s ratio on the Moment of the Tubes ............................................................................ 127
Figure 207: Effect of Poisson’s ratio on the Ovalization of the Tubes....................................................................... 128
Figure 208: Pressure-overclosure relationship of “Hard” Contact .............................................................................. 135
Figure 209: Stick region for the basic Coulomb friction model ................................................................................... 135
ix
List of tables
Table 1: Tubular cross-section classification according to EN1993-1-1 ......................................................................... 2
Table 2: Test specimen properties ....................................................................................................................................... 11
Table 3: Steel material information ..................................................................................................................................... 14
Table 4: Grain analysis in percentage .................................................................................................................................. 14
Table 5: Sample mass and moisture content of the sand-fill [17] ................................................................................... 15
Table 6: Sand volume and steel specimen weight [17] ..................................................................................................... 15
Table 7: Measured sand weight [17] .................................................................................................................................... 16
Table 8: Measured dimensions for each tube .................................................................................................................... 16
Table 9: Stress-Strain values for tube A .............................................................................................................................. 28
Table 10: True plastic stress-strain values for type A tube .............................................................................................. 28
Table 11: Stress-Strain values for tube C ............................................................................................................................ 29
Table 12: True plastic stress-strain values for type C tube .............................................................................................. 29
Table 13: Residual stress distribution at the integration points ...................................................................................... 31
Table 14: Number of elements, nodes and DoFs of the empty tube type A for different mesh sizes .................... 37
Table 15: Number of elements, nodes and DoFs of the filled tube type A for different mesh sizes of the solid
element ..................................................................................................................................................................................... 38
Table 16: Effect of the element type on moment and curvature in local buckling ..................................................... 42
Table 17: Number of elements, nodes and DoFs of the empty tube type A for different mesh sizes .................... 43
Table 18: Bending moment results from experimental testing ....................................................................................... 50
Table 19: Model error between the measured and the calculated bending moments ................................................. 67
Table 20: Ovalization values from measurements and from ABAQUS analyses for tubes A and B ....................... 73
Table 21: Ovalization values from measurements and from ABAQUS analyses for tubes C and D ...................... 73
Table 22: Ovalization values from measurements and from ABAQUS analyses for more appropriate sand
properties ................................................................................................................................................................................. 73
Table 23: Sand-fill and Steel Tube total axial reaction forces in the Sections and moment capacity contribution.78
Table 24: Sand volume change and longitudinal strain at section axis .......................................................................... 79
Table 25: Sand and Steel total axial reaction forces in the Sections and moment capacity contribution for Esand =
1000MPa. ................................................................................................................................................................................. 79
Table 26: Imperfection factors for all tube types .............................................................................................................. 85
Table 27: Results for empty Tube A ................................................................................................................................... 86
Table 28: Results for empty Tube B.................................................................................................................................... 87
Table 29: Results for empty Tube C ................................................................................................................................... 88
Table 30: Results for empty Tube D ................................................................................................................................... 89
Table 31: Results for sand-filled Tube A ............................................................................................................................ 90
Table 32: Results for sand-filled Tube B ............................................................................................................................ 91
Table 33: Results for sand-filled Tube C ............................................................................................................................ 92
Table 34: Results for sand-filled Tube D ........................................................................................................................... 93
Table 35: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube A 95
Table 36: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube B.97
Table 37: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube C.99
Table 38: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube D.
................................................................................................................................................................................................. 101
Table 39: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube A.
................................................................................................................................................................................................. 102
Table 40: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube D.
................................................................................................................................................................................................. 103
Table 41: Examined properties and their values for the parametrical study .............................................................. 108
Table 42: Moment and Ovalization changes in percentage for different parameter values ..................................... 129
x
Nomenclature
a - Half horizontal ovalization of the tube [mm]
b - Half vertical ovalization of the tube [mm]
c - Cohesion [kPa]
D - Diameter [mm]
dM - Change in moment [Nmm]
dm - Change in plate moment [Nmm/mm]
E - Young’s modulus [MPa]
Esand - Young’s modulus of sand [MPa]
Esteel - Young’s modulus of steel [MPa]
fu - Ultimate stress [MPa]
fy - Actual yield stress [MPa]
g - Plate moments and plate normal forces effect
h - Reduction of the section modulus factor
I - Moment of inertia [mm4]
L - Initial length [mm]
Lbuckle - Buckle length [mm]
Lf - Final length [mm]
M - Moment [Nmm]
Me - Elastic moment [Nmm]
Mm - Maximum moment [Nmm]
Mp - Plastic moment [Nmm]
mp - Plate plastic moment [Nmm/mm]
my - Plate moment [Nmm/mm]
np - Plate plastic normal force [Nmm/mm]
ny - Plate normal force [Nmm/mm]
p - Equivalent effective pressure stress [MPa]
q - Equivalent deviatoric stress [MPa]
Reh - Yield strength to product standards [MPa]
Rp0,2% - The 0.2% offset yield strength [MPa]
Rt0,5% - The 0.5% extension under load yield strength [MPa]
r - Radius [mm]
S - Deviatoric stress [MPa]
S11, S22, S33 - ABAQUS stresses in x, y, z directions [MPa]
s - Arc length [mm]
t - Thickness [mm]
w - Imperfection size [mm]
y - Distance from neutral axis [mm]
α - Elastic ovalization [mm]
αres - Curvature factor due to residual stresses
δ - Mode amplitude [mm]
δα - Change in ovalization [mm]
δκ - Change in curvature [1/mm]
ε - Strain / Coefficient depending on fy
εc - Strain capacity
εd - Strain demand
εnom - Nominal (engineering) strain
εtrue - True strain
εpl,true - True plastic strain
εu - Ultimate strain
εy - Yield strain
θ - Arc angle / Plastification angle [rad]
θbuckle - Angle at the buckle location [o ]
xi
κ - Curvature [1/mm]
κcrit - Critical curvature [1/mm]
κel - Elastic curvature [1/mm]
κpl - Plastic curvature [1/mm]
κres - Curvature due to residual stresses [1/mm]
ν - Poisson ratio
ρ - Radius of curvature [mm]
σ - Normal stress [MPa]
σnom - Nominal (engineering) stress [MPa]
σtrue - True stress [MPa]
σu - Ultimate stress [MPa]
σ1, σ2, σ3 - Principal stresses [MPa]
τ - Shear stress [MPa]
τmax - Shear stress limit [MPa]
φ - Friction angle [o ]
φi - Angular rotation at the edges [rad]
ψ - Dilation angle / Slope of yield surface [o ]
Acronyms
3D - Three dimensional
DoF - Degree of Freedom
SBD - Strain Based Design
xii
Abstract
Bending behaviour of steel tubes has been examined in many research projects through the years, especially for
empty ones. Also the infill and pressure effect in the tubes has been elaborately investigated but not much has
been done on the effect that the sand infill has on the bending of the tubes. Combi-wall tubes are filled with
sand during their installation on site not on purpose, since the sand enters through the lower open edge of the
tube. The presence of the sand helps in achieving better post-yielding behaviour and local buckling capacity
(failure at higher curvature), and also smaller ovalization of the steel tube cross-section. One of the advantages of
the sand infill of combi-wall tubes is that it is free of costs and present at the application site. In order to have a
better view of this favourable effect, the SBRCURnet organization performed an investigation programme,
which included 4-point bending tests on empty and sand-filled tubes.
In the current thesis, a numerical model of the experiments is designed which is verified and compared to the
experimental test results and to analytical formulas. The numerical model is built with the ABAQUS software
and consists of two approaches (models) on the issue. The first consists of a “slice” of the cross-section and it
produces the moment-curvature and ovalization-curvature diagrams but it cannot predict the local buckling of
the tube. The second is created in order to simulate experiment setting. It is a full scale model and it represents
the mid-section of the set-up which has a constant bending moment. From this, the moment-curvature and the
ovalization-curvature diagrams and the local buckling failure can be derived. The numerical curves and results are
compared to curves derived from analytical formulas taken from literature research and to the results taken from
the experimental tests. Additionally, the slice model is used to perform a parametrical study of the sand
properties. By this, the influence of each sand property on the moment and ovalization can be examined.
Generally, the final conclusion is that the numerical moment-curvature curves are a very good fit to the ones
derived from the analytical formulas and they also have a good fit to the results of the experiment except for a
very few of them. Furthermore, it can be clearly seen that the favourable effect of the sand is depicted in the
numerical results as well as in the experimental ones. The sand infill contributes mainly to the post-yielding
behaviour and the local buckling capacity since the buckle occurs in a later curvature value compared to the
empty tubes. Also, the buckling shape of a sand-filled tube differs for the buckling shape of an empty tube due
to the presence of the sand, which averts the inward local buckling and allows only an outward buckle to form.
Additionally, by preventing the excessive ovalization of an empty steel tube, it can contribute to the bending
moment capacity especially for the higher curvature values. In the curvature values close after the yielding of the
tube, the contribution in the moment capacity is significantly smaller but nevertheless obvious.
To sum up, filling the combi-wall steel tubes is an easy and very cheap manner in order to enhance their local
buckling capacity.
xiii
xiv
Chapter 1 - Introduction
1. Introduction
The major advantage of the combined walls is that they are economical solutions due to the fact that they have
much higher capacity than sheet piling walls and also because the tubular piles fabrication is cost effective. The
tubular piles are longitudinally or spirally welded.
In this report the local buckling behaviour of the tubular primary elements filled with sand will be examined
through the comparison between the results of experiments performed by SBRCURnet investigation programme
[17] and the FEM analyses performed with the ABAQUS software.
1
Chapter 1 - Introduction
For D/(t*ε2) >90 the cross-section is categorized as class 4 and has to be checked for local buckling since local
buckling takes place before the attainment of the yielding moment. This formula shows that except for the D/t
ratio, also the steel grade is significant for the local buckling behaviour. In the past Dutch regulations, local
buckling was not taken into account for piles with cross-section ratio D/t≤90, but equations (1) and (2) depict
that for higher than 235 MPa grade steel piles, local buckling shall occur for D/t<90. In Table 1 the tubular
cross-section classification according to EN1993-1-1 [4] is depicted.
Class 4 tubular cross-sections are designed according to EN1993-1-6 [5] in which the local buckling check is
affected by axial, shear, flexural and hoop stress and also the asymmetrical radial loads effect has to be
considered.
As other research has shown [8], [9], the current EN1993-1-6 [5] rules lead to more uneconomic designs. The
EN1993-5 [7] mentions that the shell design has to be done in accordance with the EN1993-1-6 [5] for the
strength and local buckling calculations. As mentioned before, the EN1993-1-6 [5] local buckling verification
calculations lead to conservative designs and the main reason for this is that a local stress buckling criterion is
applied. By applying this criterion, all the parameters that influence the local buckling behaviour are taken into
account by applying a reduction on the local buckling stress. This approach has a very restrictive effect on the
tubes, in which a part of the shell yields before local buckling.
Therefore, a different approach is followed which is called Strain Based Design (SBD) and it takes into account
the influential parameters, such as ovalization, geometrical imperfections, load introduction, by introducing a
reduction to the critical curvature/critical strain. The tubular piles of the combi-walls buckle after reaching the
elastic limit and therefore local buckling behaviour depends on critical strain and critical curvature rather than
critical stress [17].
2
Chapter 1 - Introduction
For thicker walled shells (such as tubes and pipes), the stress based buckling criterion, based on imperfections,
gives a too large reduction of the design stress and the design bending moment capacity. The local stress
buckling criterion is appropriate for thin-walled structures such as tanks and silos but not for thicker walled
shells such as tubes [8]. In Figure 3 the concept of the stress based design approach is depicted.
The SBD encloses both the strain demand (applied strain) and the strain capacity (strain limit) and it permits a
more effective use of the longitudinal strain capacity.
The fundamental criterion in SBD is the comparison of the strain demand (εd) to the strain capacity (εc):
𝜀𝑑 ≤ 𝜀𝑐 (3)
3
Chapter 1 - Introduction
In general, all the design codes refer on steel tubes without taking into account a possible filling in them which
has a favourable effect. The EN1993-5 [7] though, in paragraph 5.5.4 suggests that a tube with concrete fill or
filled with “high grade compacted, non-cohesive material”, could be considered to be an effective way to avoid
the phenomenon of local buckling at the locations where concentrated loads are applied. It also states that in
order to use the full cross-sectional resistance of the tube, the EN1992, EN1993 and EN1994 should be
considered, which implies that the cross-section should be considered as composite. Still though, it does not
suggest clearly what should be done in the case of combining a steel tube with sand-fill.
A research study, published in 2009 by European Commission in the context of the ETIB programme with
subject on the enhanced economy of tubular piles by improved local buckling design [14], dealt also with the
effect of the sand-fill on the steel tubes. Actually, the research aimed on the sand continuum both outside the
tubular pile and inside it. FEM analyses, analytical methods and small scale tests were implemented but not large
or full scale tests. As a final conclusion of the study, it can be said that it indicates a small but favourable effect
of the sand-fill on the bending capacity of the piles. But in general the conclusions are not presented clearly and
are rather complex.
Another study, implemented by Patch et al. [16], performed thin-walled reduced scale bending tests along with
FEM analyses. The tests included a comparison between sand-filled and concrete filled tubes. The sand-filled
tube exhibited several load peaks, which are associated with the phases of the initial or post-critical buckling
deformations. It can be seen that the local buckling of the sand-filled tube still happens, but the post buckling
behaviour is seriously improved by the sand-fill. After local buckling, a large part of the plastic resistance is
maintained during a significant range of increasing curvature (Figure 6).
4
Chapter 1 - Introduction
According to the test report of the local buckling experiments, the questions that needed to be researched, in
order to provide safe and economic designs for tubular piles were:
What is the effect of the sand-fill of the piles on the occurrence of local buckling?
To what extend does the sand-fill prevent or limit ovalization?
What is the effect of the sand-fill on the post-buckling behaviour?
Could the presence of the sand-fill be integrated in a more advanced design approach?
These questions are referring to tubes that are spirally welded with average to rather large out-of-roundness and
welding tolerances. The CUR investigation programme purpose was to provide:
An agreed design method for tubular piles, suitable for common practice and recommendation in the
Handbook Quay Walls.
Verification and evidence for the hypothesis that the failure behaviour of (sand-filled) tubular piles
justifies a more economic design rule than in the present EN1993-1-6 [5].
The European scientific community and construction industry are informed of the uneconomic design rules
issue and have initiated a research programme named COMBITUBE. This focused also on addressing the local
buckling of the tubular piles in combi-walls, but it did not include investigating on the effect of sand-fills and it
gave less focus on the post-buckling behaviour. For that reason, this experimental programme focusing on the
above mentioned questions was considered an invaluable contribution to the programme.
Recently, studies that have been implemented in the TU Delft, which have already developed a finite element
model in ABAQUS ([8], [19], [20]), simulating a pipeline tube. These studies were dealing with the local buckling
behaviour of spirally welded tubes taking into account the material imperfections, the residual stresses and the
effect of the strain based design on the economy and safety of the cross-section selection. They included FEM
modelling, analytical solutions as well as experimental investigation, comparing the results of every method to
each other. The current research is based on the work of this research, since the steel tube model in ABAQUS is
implemented taking them as a guide.
This model is going to simulate the 4 point bending tests carried out in the CUR programme. In order to achieve
this, several aspects have to be taken into account:
The material properties of the steel tube and the sand-filling.
The dimensions of the tubes used in the experiments.
The sand behaviour.
The interaction behaviour between the two materials.
5
Chapter 1 - Introduction
6
Chapter 2 - Experimental testing
2. Experimental testing
The critical strains were calculated from the measurement of the elongation of the jack at the failure. The
definition of failure during the testing was considered to be the moment just before a sudden drop of more than
10% in the tube’s bending resistance.
The experimental testing pile sizes range from 508 to 711mm whereas the diameters of the combi-wall piles used
in reality range between 900 to 1600mm and therefore the scale factor of the piles is from 1.8 to 2.4. The length
of the tubes in the experiments was 11m. The sand particle size was selected to be very fine and the scale effect
factor to be 2 which stayed within the band width of the sieve curve. As a result, no scaling laws were applied to
the tests [17].
The load was applied with the use of hydraulic jacks and a deformation controlled procedure was followed when
the yield and local buckling area was reached. By applying small load steps, a compromise between
force/pressure controlled load application and a deformation controlled load application was created. The
pressure in the jacks dropped due to volume increase when a new load step leads to progressive deformation,
then the force dropped in order to lead to a deformation controlled load application [17]. The test setup is
depicted in Figure 7.
The steel tubes were located in the middle of the setup and end plates were placed at the openings of the tube in
order to form a closed volume for the sand-fill. The test tubes were connected with bolted joints between end
7
Chapter 2 - Experimental testing
plates. According to the purpose of the test programme, the local buckling would have to occur inside the
undisturbed central area of the tube (at least at a distance of approx. 2D from the end plates). In order to
prevent the end plates from disturbing the local buckling of the tube and the ovalization near the vicinity of the
end plates because the ovalization is prevented at the location of the end plates, the bending moment in the
central zone was increased. For the sand-filled tubes this was achieved with the sand-fill, which increased the
bending moment by roughly 15% to 20%. For the empty tubes it was achieved by applying imposed loads in the
form of concrete blocks and bags filled with sand.
The setup of the experimental system is secured in the sideward stability by double chains in a downward triangle
shape at each anchored position. This allowed the torsional movement and deformation of the steel tube prior to
or during the buckling process and it was implemented with the aim of permitting the tube to freely adopt its
‘weakest’ position [17].
Before the implementation of the test some data were documented and also some data and results were recorded
during the execution. The data which was documented before the test was:
Wall thickness (every 0.5 m, 4 sides)
Out-of-roundness of the cross-section (every 0.5 m, over two axis)
Shape of the welds (average plus exceptions)
Yield limit of the steel tubes
σ-ε diagram of 2 of the 4 steel tubes
Weld dimples (high lows of spiral welds and circumferential welds)
Eccentricity in connection of tube sections
Overall out-of-straightness of the tube
Exact weights of steel components, sand-fill and applied extra weights
Figure 8: Setup of the test with the position of the displacement gauges [17].
The connection of the test steel tubes to the end-pieces was achieved with the end plates using bolts. Even
though there were no deformations of the end plates observed, the end plate thicknesses were designed on basis
of the theory of plasticity and taking into account the prying forces. Due to the differences in the tube diameters,
8
Chapter 2 - Experimental testing
the bolted connection was rather complex. The end-piece was connected to the holding-down chains and to the
jacks. The lateral stability of the test arrangement was ensured by setting the holding-down chains in an angled
position an in the same time permitting a small lateral and torsional movement. The importance of this
‘freedom’ was considered high because the tube could buckle a bit side wards and possibly combined with
torsion. That was the reason why that especially the empty tubes had a twisted appearance.
Figure 9: Test tube and end pieces drawing with exaggerated deformation [17].
The lateral freedom was limited by guide columns which were installed in the form of H-shapes with 50mm
spacing between the end pieces and the guide column profiles. In the vertical upward movement of the tube the
end-piece leaned on one of the guidance profiles, which created friction forces. Also the axial displacements
were recorded an in some cases considerable axial deformation was noticed. Some additional guidance was
installed which provided axial restraint at the one side because after the end of the test, the end pieces did not
return back on their temporary support but aside of it.
9
Chapter 2 - Experimental testing
Each test lasted typically 4 to 6 hours. The jacking procedure had 3 phases:
Phase 1 jacking up to lifting from the temporary supports
Phase 2 jacking up to the beginning of chain readings (sometimes a bit of an unstable phase; jacking one
side first and the other side after that did work well)
Phase 3 jacking up to predicted safe levels
The total jacking procedure lasted 30 to 45 minutes and the typical duration of the 3rd phase was 15 to 20
minutes.
10
Chapter 2 - Experimental testing
Figure 12: Pictures from the test execution, the jack system and the guidance profiles [17]
At the beginning of the experimental testing the end-pieces were sitting on temporary supports and after the
jacking procedure had started, the tube was lifted and the chain forces started to act. Next, the jacks lifted the
tube, which lead to the jack forces being activated on the experimental system and finally by increasing the load
the chain forces were reactivated. The readings of chain forces were regarded to be the recordings with the most
accuracy [17].
The four types of the steel tubes are presented in the Table 2. In total 24 (4x6) tests were implemented.
The steel tubes used in the tests were supplied by ArcelorMittal Projects Netherlands B.V. The fourth test tube
(D) was a special production for the tests whereas the rest of them were dependent on the availability of the
stock material. The length of each tube was 11 meters.
11
Chapter 2 - Experimental testing
In Figure 13 photographs from typical buckling shapes are depicted. It can be seen that due to the radial support
that the sand contributes, the tube cannot buckle inwards and therefore only an outward buckle can be formed.
On the other hand, the empty tubes buckled inwards creating a folding that reduced the internal lever arm which
reduces the axial capacity of the steel tube.
Figure 13: Photographs of typical buckling shapes in early (top) and in advanced (bottom) stage of empty (left) and filled (right) tubes [17]
12
Chapter 2 - Experimental testing
Initially, for the steel mechanical properties there were tests conducted in all four types of tubes.
Type A
Material certificates of the type A tubes report values of ReH and Rt0,5% of 307.3 MPa (four tests, hardly any
variation). For the tubes A3 and A4 the testing in Scielab gave as result, Rp0,2% = 305 and 314 MPa, respectively.
No stress-strain diagram is available for this tube type. Therefore, only two out of the six tubes were tested for
the derivation of the material properties and the tests were performed only for the longitudinal direction.
Type B
For the tubes used in this set of tests only Schielab tests were carried out. The Rt0,5% value had an average of
488MPa and varied from 458 to 508 MPa. The full stress-strain diagrams are available. Also tests of samples in
the circumferential direction are available.
Type C
Material certificates of the type C tubes report values of ReH and Rt0,5% of 311.8 MPa (six tests, hardly any
variation). For the tubes C2 and C3 the testing gave as result, Rp0,2% = 317 and 329 MPa, respectively. No stress-
strain diagram is available for this tube type.
Type D
Material certificates report 49 readings of Reh, with an average of 554 MPa. The variation is quite high, but the
peak value of Reh is less important than the 0.2% and 0.5% values. For the Rp0,2% there were two readings, 509
and 548 MPa in the certificates. Schielab tests on two samples gave 493 and 505 MPa for Rp0,2%. The 548 MPa
value has been neglected in the analysis.
13
Chapter 2 - Experimental testing
Next, the sand-fill properties were based on requirements for an average grain size of 150 micrometres, which is
considered to represent the ‘Maasvlakte’ sand [17]. The sieve curve of the sand is depicted in Figure 16. The sand
used for this project would be ideally oven-dried. The moisture content of the sand-fill was measured by
obtaining samples from different bags [17]. In Table 5 the moisture content of the samples is presented.
14
Chapter 2 - Experimental testing
The sand-fill specific mass was tested in the laboratory and the average bulk specific mass of the sand was found
to be 1483 kg/m3 and the specific mass at maximum achievable compaction to be 1595 kg/m3.
In order to achieve the requirement of identical sand-fill properties for all tests, a sand-fill process procedure was
implemented and it included the following steps [17]:
Mass of empty tube (still having some excess length) with one end-plate was measured
Mass of gravity-filled tube with excess length and with one end-plate was measured
Filled tube was vibrated in vertical position and the remainder of the volume was gravity-filled
Two half-circular thin closure plates were placed of top of the sand-fill (50 mm) below the end plate
position). The tube was cut on the right length and the end plates were welded.
Mass of the ready filled tube was measured.
The final derived sand sill specific mass was 1630 kg/m3. By observation it can be concluded that the specific
masses measured during the first fill, before vibrating the sample, showed a strong correlation with the diameter,
which implied that an arching effect might have been present and it disappeared after vibration.
15
Chapter 2 - Experimental testing
By assuming 2700kg/m3 unit weight of the particles, the fully saturated soil would have had a mass of 1974
kg/m3 (19.4kN/m3). According to the NEN 9997-1+C1:2012 standard soil classification table, the mechanical
properties for a clean sand are: φ=32.50 and the characteristic Young’s modulus Esand=30MPa. The Young’s
modulus value though, when the results of the tests were processed, was derived to be different [17].
There were also measurements of the dimensions of every tube and they are presented in the Table 8 below.
16
Chapter 2 - Experimental testing
𝑀 = 𝐸𝐼𝜅𝑒𝑙 (4)
𝜅𝑒𝑙 = 1⁄𝜌 (5)
𝑠 = 𝜌𝜃 (6)
𝜃 = 𝜑1 + 𝜑2 (7)
where:
M is the moment
E is the modulus of elasticity
I is the moment of inertia
κelastic is the elastic curvature
ρ is the radius of curvature
s is the arc length
θ is the angle of the arc
φi is the angular rotation at the edges
In Figure 17 the variable y is the distance from the neutral axis to any other radius in the cross-section.
Therefore, the radius of the curvature at any y is (ρ-y) and the final length at any y is:
𝐿𝑓 = (𝜌 − 𝑦)𝜃 (9)
17
Chapter 2 - Experimental testing
After the yielding, the beam enters the plastic range and the rotation will localize in the buckling location. The
curvature of the beam due to the rotation in the local buckling location is:
𝜃𝑏𝑢𝑐𝑘𝑙𝑒
𝜅= (11)
𝐿𝑏𝑢𝑐𝑘𝑙𝑒
In an ideally plastic section, the beam will bend as in Figure 18 and the rotation will be equal to:
𝜃𝑏𝑢𝑐𝑘𝑙𝑒 = 𝜑1 + 𝜑2 (12)
If the approximation that the rotation can be distributed over the entire tube length, is followed, then:
𝜑1 + 𝜑2
𝜅= (14)
𝐿
18
Chapter 3 - Analytical formulas
3. Analytical formulas
In Figure 19 the stress distribution in the cross-section is depicted when it is in the elastic-plastic phase. The
moment is normalized by dividing it with the elastic moment and the curvature by dividing with the elastic
curvature.
1.4
normalized moment
1.2
1
M/Me
0.8
0.6
0.4
0.2
0
0 2 4 6
normalized curvature k/ke
Figure 19: Stress distribution in cross-section in bending and moment-curvature diagram without influences taken into account [12]
19
Chapter 3 - Analytical formulas
There are elastic and elastic-plastic behaviour models. The elastic behaviour model proposed by Reissner and
Weitnitschke and the elastic-plastic proposed by Gresnigt [11] are presented next.
In the elastic-plastic ovalization behaviour by Gresnigt [11], [12], the normality principle can be applied for the
full plastic cross-section in order to determine the relation between the plate bending moments, ovalization and
curvature. The plastic moment capacity decreases with the increase of the plate bending moments. This
procedure has a step wise iterative approach, δα, since the slope ψ of the yield surface is not a constant.
𝑑𝑀⁄𝑀𝑝
𝜓= (25)
𝑑𝑚𝑦 ⁄𝑚𝑦
where my is the plate moment and
𝑟3
𝛿𝛼 = − (2𝜓 ∗ 𝛿𝜅) (26)
𝑡
In Figure 22 yield surface is depicted.
20
Chapter 3 - Analytical formulas
Due to the complexity of this approach a different approximate equation, which has as an origin equation 14, has
been suggested for practical applications which it is has equation 14 as an origin. Therefore, for the elastic-plastic
part the ovalization is given by:
𝑟5 𝜀𝑦
𝑎𝑝 = 𝜅1,5 ∗ 𝜅𝑒0.5 ∗ 2 𝑤𝑖𝑡ℎ𝜅𝑒 = (27)
𝑡 𝑟
The bending moment capacity Mm of the cross-section is reduced from the Mp due to the ovalization which
causes plate moments. The Mm is given by:
𝑀𝑚 = ℎ ∗ 𝑔 ∗ 𝑀𝑝 (28)
where h is the reduction in the section modulus and g the plate moments and plate normal forces effect.
2𝑎
ℎ =1− (29)
3𝑟
𝑐1 𝑐2
𝑔= + (30)
6 3
where:
2 2
𝑛𝑦 𝑚𝑦 𝑛𝑦
𝑐1 = √4 − 3 ( ) − 2√3 𝑎𝑛𝑑𝑐2 = √4 − 3 ( ) (31)
𝑛𝑝 𝑚𝑝 𝑛𝑝
𝑀𝑚 𝜅
𝑛𝑦 = 0.2 ≤ 𝑛𝑝 = 𝑡𝑓𝑦 𝑎𝑛𝑑𝑚𝑦 = 0.071𝑀𝑚 𝜅 ≤ 𝑚𝑝 = 0.25𝑡 2 𝑓𝑦 (32)
𝑟
At the first step it can be assumed that Mm=Mp. The section moment M assuming elastic-plastic stress strain
relation is:
𝜃
𝑀 = 𝑀𝑚 ∗ 0.5 ( + 𝑐𝑜𝑠𝜃) (33)
𝑠𝑖𝑛𝜃
𝜃 = arcsin(𝜅𝑒 /𝜅) (34)
1.2
normalized moment M/Me
0.8
No ovalization, no residul stress effect
0.6 Ovalization effect
0.4
0.2
0
0 1 2 3 4 5
normalized curvature κ/κe
Figure 23: Comparison of the moment-curvature diagrams when ovalization is taken into account and when not
21
Chapter 3 - Analytical formulas
In Figure 24 the moment-curvature diagrams of the effects of ovalization and residual stresses on the tube are
shown.
1.4
1.2
normalized moment M/Me
1.0
0.8
No ovalization, no residual stresses effect
0.6 Ovalization effect
0.0
0 1 2 3 4 5
normalized curvature κ/κe
Figure 24: Comparison of moment-curvature diagrams with the effect of ovalization and residual stresses
The above formulas have been derived through tests on spirally welded tubes and FEM simulations that were
validated to these tests. Therefore, these formulas can be used for spirally welded tubes, but they have not been
validated for other type of welded tubes. In this research study, though, they are used for all the tube types.
22
Chapter 3 - Analytical formulas
4.0E+08
Moment - Curvature for Tube A
3.0E+08
Analytical no oval., resid.
Analytical with ovalization
Moment (Nmm)
1.0E+09
Moment - Curvature for Tube B
9.0E+08
8.0E+08
7.0E+08
Moment (Nmm)
23
Chapter 3 - Analytical formulas
8.0E+08
Moment - Curvature for Tube C
7.0E+08
6.0E+08
Moment (Nmm)
5.0E+08
Analytical no oval., resid.
Analytical with ovalization
4.0E+08
Analytical with residual
Analytical with oval., resid.
3.0E+08
C2 - Empty - Test
C3 - Empty - Test
2.0E+08
C6 - Empty - Test
C1 - Filled - Test
1.0E+08
C4 - Filled - Test
C5 - Filled - Test
0.0E+00
0.0E+00 4.0E-06 8.0E-06 1.2E-05 1.6E-05 2.0E-05
Curvature κ (1/mm)
Figure 27: Analytical curves and test results for tube C
1.6E+09
Moment - Curvature for Tube D
1.4E+09
1.2E+09
Moment (Nmm)
1.0E+09
Analytical no oval., resid.
Analytical with ovalization
8.0E+08
Analytical with residual
Analytical with oval., resid.
6.0E+08
D1 - Empty - Test
D2 - Empty - Test
4.0E+08
D3 - Empty - Test
D4 - Filled - Test
2.0E+08
D5 - Filled - Test
D6 - Filled - Test
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 28: Analytical curves and test results for tube D
24
Chapter 3 - Analytical formulas
3.6 Comments
From Figure 25, Figure 26, Figure 27 and Figure 28, the comparison between the analytical curves shows that
the ovalization and the residual stresses are having significant effect on the moment capacity of the steel tube.
Initially, the comparison between the curve with no effects taken into account and the curve with the effect of
the ovalization in it, shows that the geometrical non-linearity has an effect in the post yield behaviour of the steel
tube. After yield the moment follows a stable decreasing path, because of the change in the radius due to
bending of the initially circular steel tube.
In the case of adding the residual stresses effect to the ovalization effect, the moment-curvature diagram starts to
deviate from both the other curves at the half of the elastic moment Me according to this analytical formula. This
happens because the residual stresses are already in the steel material since before the start of the loading due to
the forming process of the tube and therefore it reaches to yielding sooner than in the case of this effect is not
taken into account. This is the analytical curve which is used as a comparison to the empty tube results.
In the case of the residual stresses effect alone without the ovalization, the moment-curvature diagram starts to
deviate at the half of the elastic moment Me. This happens, as mentioned previously, due to the residual stresses
of the steel material. The yielding moment value is smaller than the value without the residual stresses taken into
account. After yielding, the curve tends to coincide with the curve without the effects of ovalization and residual
stresses. The analytical curve with the residual stresses is used as a comparison to the sand-filled tube results.
Finally, by comparing the analytical curves to the test results, it can be seen that in some cases the curves are a
good fit to the tests and in other cases (tube A for empty tubes and tube B) a significant difference is noted. It
should be mentioned that the test results of the empty tubes should be compared to the curves with the
ovalization and the residual stresses curves and the filled tube test results with the curves that only the residual
stresses effect is taken into account, since as it will be seen later, these curves are a good fit with the numerical
curves of the filled tubes.
25
Chapter 4 - FEM model
4. FEM model
4.1 General
The Finite Element Method (FEM) is a numerical technique which is used to find approximate solutions with
the help of Partial Differential Equations. This method makes use of the concept of the subdivision of the whole
problem to smaller, simpler parts, which are called finite elements. It includes methods for connecting many
simple element equations in many small subparts in order to approximate a more complex equation for a larger
domain. The subdivision of large complex domain into many simpler parts has several advantages such as the
accurate representation of a complex geometry, the inclusion of dissimilar material properties, making the
representation of the total solution easier, capturing of local effects, etc. The FEM approximates the solution by
transforming the Partial Differential Equations into algebraic ones.
The practical application of the Finite Element Method (FEM) is known as Finite Element Analysis (FEA),
which is a computational tool used in engineering projects in order to perform analyses on complex problems.
These complex problems are physical systems with their physical behaviour (e.g. Euler-Bernoulli beam equation,
Navier-Stokes equations, etc.) being expressed in either Partial Differential Equations or integral equations [24].
This technological breakthrough gave the opportunity to the engineers and the researchers to carry out very
complex and costly projects, which have serious safety and reliability impact on the society. However, the FEA is
not to be used without caution since an error is very probable. Critical view and verification are the tools to
check the results of every outcome from the FEA.
This research project is going to compare the results between the experimental tests which were performed by
SBRCURnet and TU Delft and the results from the FEA model in ABAQUS. Therefore, the FEA model is
going to be loaded in bending and the moment-curvature diagram and the ovalization of the cross-section is
going to be derived. The designing of the model includes several nonlinearities which have to be implemented as
close to the reality as possible. In general, the sources of nonlinearity are:
Material nonlinearity (Materials with not complete linear stress-strain curve)
Geometry nonlinearity (Change of shape cause nonlinear changes in the stiffness in large deformations)
Boundary Condition nonlinearity (Components in contact with each other produce disproportionate
changes in deformation)
Loading Condition nonlinearity (Loading changes over time)
The numerical models that are implemented in this project are two. The first is developed in order to simulate
the behaviour of the cross-section of the tubes and represents a very thin segment of the tube and the sand-fill.
26
Chapter 4 - FEM model
The second one is the full scale model of the tube and the sand-fill, which will try to simulate the experimental
testing and compare the results of the two methods.
This model was used to perform a comparison with the experimental tests for the bending moment and the
ovalization and for a parametrical study which investigates the sensitivity of the tube (mainly the sand-filled) to
various changes of its properties.
The horizontal ovalization can be given by the horizontal displacement of point A and the curvature is derived
by:
𝜃
𝜅= (37)
𝐿
Where θ is the rotation of the model applied by the Reference Point and L is the depth of the cross-section
which here was selected to be 10mm.
27
Chapter 4 - FEM model
According to these requirements the chosen values for the stress-strain diagram of tube type A are:
Table 9: Stress-Strain values for tube A
Strain Stress (MPa)
0 0
0.00146 307.0
0.025 307.0
0.20 430.0
The resulting strain hardening modulus is Esh≈E/300 (it is low but the strain hardening onset value is not
reached). The calculated true plastic stress-strain values for the type A tube are presented in Table 10 and in
Figure 30 the engineering and the true stress-strain diagrams are depicted.
Table 10: True plastic stress-strain values for type A tube
Strain Stress (MPa)
0 307.0
0.02319 314.67
0.1798 516.0
Tube A Steel
550
500
450
400
Stress (MPa)
350
300
Engineering stress-strain
250
True stress-strain
200
150
100
0 0.05 0.1 0.15 0.2
Strain
Figure 30: True plastic and engineering stress-strain diagrams
28
Chapter 4 - FEM model
Accordingly to the type A tubes, the chosen values for the stress-strain diagram of tube type C are:
The strain hardening Young’s modulus is Esh≈E/300 (it is low but the strain hardening onset value is not
reached). The calculated true plastic stress-strain values for the type C tube are presented in Table 12 and in
Figure 31 the engineering and the true stress-strain diagrams are depicted.
Tube C Steel
550
500
450
400
Stress (MPa)
350
300
Engineering stress-strain
250
True stress-strain
200
150
100
0 0.05 0.1 0.15 0.2
Strain (%)
The properties of the sand-fill are inputted as mentioned previously. ABAQUS has the ability to offer various
material models. For the analyses implemented in this research study the Mohr-Coulomb material model of
ABAQUS is chosen for the sand which is considered as linearly elastic-perfectly plastic idealized material
behaviour. This material model allows plastic deformation when the material element yields but it does not take
into consideration the effect of stress history.
The Mohr-Coulomb criterion is the most commonly used failure criterion in the geotechnical engineering and is
defined as:
𝜏 = 𝑐 − 𝜎 ∗ 𝑡𝑎𝑛𝜑 (41)
where
σ is the normal stress (negative in compression in ABAQUS)
τ is the shear strength
c is the cohesion
φ is the friction angle
29
Chapter 4 - FEM model
For the Mohr-Coulomb Plasticity model the input parameters used are:
Usually sand has zero cohesion (e.g. a saturated silty clay and a saturated clay have cohesion values 10~20kPa
[23]), but in order for the numerical analysis to achieve convergence, a small value is demanded.
Vasilikis et al. [20] have modelled the forming process of the steel coil to a tubular cross-section for spirally
welded tubes in ABAQUS. The goal was to predict the stresses which develop in the cold bending
manufacturing process of the steel coil. The resulting stresses are applied as initial residual stresses [20].
30
Chapter 4 - FEM model
The calculated residual stresses are normalized by the yield stresses fy across the tube thickness and then they can
be used for every spirally welded tube by multiplying by the yield stress of each tube. Some of the used tubes are
longitudinally welded and formed by different process but nevertheless the current residual stress distribution is
going to be used. In the case of the longitudinally welded tubular sections, the residual stresses are expected to
be higher in the hoop direction and lower in the axial direction compared to the spirally welded tubes [19].
In the next figure the residual stress distribution across the thickness of the tube is depicted, which was derived
by using 15 integration points through the thickness.
0.4
0.2
0
-0.2
-0.4
Normalized Axial Stress
-0.6
Normalized Hoop Stress
-0.8
-1
0 0.2 0.4 0.6 0.8 1
Normalized thickness
The integration points are numbered from the inside to the outside of the tubular section. In the next table the
calculated residual stresses for all tube types are presented.
31
Chapter 4 - FEM model
The geometry of the model takes into account the full cross-section but its depth is only 10mm, considering to
be exactly at the middle of the tube’s span where the moment is constant. Therefore, this model examines what
it is happening at the middle of the tube’s span.
The load introduction is implemented by applying a rotation at the Reference Point and through this to the face
of the cross-section model.
The contact between the steel tube and the sand-fill has to be defined in the design of the model. The surface-to-
surface contact of the ABAQUS interaction models is used, in which a master and a slave surface have to be
defined. The contact properties used to simulate the interaction between the two materials are the Normal
Behaviour and the Tangential Behaviour options. In the Normal Behaviour a “Hard” Contact is selected with
allowing the separation after contact and in the Tangential Behaviour a friction coefficient of 0.39 is used. The
0.39 coefficient is derived from the following calculations:
𝜑 = 32,5𝑜
2
32,5𝑜 ∗ = 21,67𝑜
3
𝑓𝑟𝑖𝑐𝑡𝑖𝑜𝑛𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 = tan(21,67𝑜 ) = 0,39
More information about the contact mechanism of ABAQUS can be found in Appendix A.
32
Chapter 4 - FEM model
For the steel tube the S4R shell element is chosen because it accounts for finite membrane strains and arbitrarily
large rotations; therefore, it is suitable for large-strain analysis. The element is a 4-node, 6 DoF, general-purpose
shell, reduced integration with hourglass control and finite membrane strains. In Figure 36 the element’s
numbering of nodes and integration points is presented and compared to the 4-node fully integrated and 8-node
reduced integration elements. In this analysis the shell elements are chosen to have 15 integration points in the
though thickness direction.
Figure 36: Numbering of the nodes and the integration points of a shell element in ABAQUS [1]
In order to make a selection over the reduced or full integration element or if a 2nd order element should be
used, a comparison analysis is implemented and the results are the moment-curvature diagram and the
ovalization.
3.50E+08
3.00E+08
Moment (Nmm)
2.50E+08
2.00E+08
1.50E+08
Reduced integration
1.00E+08
Full integration
5.00E+07 Reduced int., 2nd order accuracy
0.00E+00
0.00E+00 2.00E-06 4.00E-06 6.00E-06 8.00E-06 1.00E-05
Curvature (1/mm)
Figure 37: Moment-curvature diagram of type A empty tube
33
Chapter 4 - FEM model
Ovalization
4.5
Reduced integration
4
Full integration
3.5
Reduced integr., 2nd order accuracy
Ovalization (mm)
3
2.5
2
1.5
1
0.5
0
0.00E+00 2.00E-06 4.00E-06 6.00E-06 8.00E-06 1.00E-05
Curvature (1/mm)
From Figure 37 and Figure 38, it can be seen that there is no difference in the results and therefore the 1st order
S4R element will be chosen due to its computational efficiency. The mesh size in this analysis is 20mm.
For the solid element selection of the sand-fill, there are some recommendations in the ABAQUS Analysis
manual:
Make all elements as “well shaped” as possible to improve convergence and accuracy.
If possible, use hexahedral elements in three-dimensional analyses since they give the best results for the
minimum cost.
For linear and “smooth” nonlinear problems use reduced-integration, second-order elements if possible.
Use second-order, fully integrated elements close to stress concentrations to capture the severe gradients
in these regions. However, avoid these elements in regions of finite strain if the material response is
nearly incompressible.
Use first-order quadrilateral or hexahedral elements or the modified triangular and tetrahedral elements
for problems involving contact or large distortions. If the mesh distortion is severe, use reduced-
integration, first-order elements.
If the problem involves bending and large distortions, use a fine mesh of first-order, reduced-integration
elements.
For the sand-fill the C3D8 solid element is chosen. The element is an 8-node linear brick with 3 DoFs. In Figure
39 the element’s numbering of nodes and integration points is presented and compared to the 8-node reduced
integrated and the 20-node elements.
34
Chapter 4 - FEM model
Figure 39: Numbering of the nodes and the integration points of a solid element in ABAQUS [1]
Again, in order to make a selection over the reduced or full integration element or if a 2nd order element should
be used, a comparison analysis is implemented and the results are the moment-curvature diagram and the
ovalization.
3.00E+08
2.50E+08
2.00E+08
1.50E+08
Reduced integr., 2nd order
1.00E+08 Full integration
5.00E+07 Reduced integration
0.00E+00
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04
Curvature (1/mm)
Figure 40: Moment-curvature diagram of type A sand-filled tube
Ovalization
12
10
Ovalization (mm)
6
Reduced integr., 2nd order
4
Full integration
Reduced integration
2
0
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04
Curvature (1/mm)
35
Chapter 4 - FEM model
From the previous two diagrams (Figure 41, Figure 42) it can be seen that the moment-curvature diagrams are
the same for every case but there is a minor difference at the ovalization-curvature diagrams between the full
integration 1st order element and the other two cases. The full integration element completed the analysis in
fewer steps and time than the other two. Due to the computational efficiency, the ABAQUS manual
recommendations and because of some mesh size issues of the reduced integration 1st order element which will
be discussed in a next paragraph, the full integration 1st order C3D8 element is selected.
3.50E+08
3.00E+08
Moment (Nmm)
2.50E+08
2.00E+08
1.50E+08
10mm
1.00E+08 20mm
5.00E+07 40mm
60mm
0.00E+00
0.00E+00 2.00E-06 4.00E-06 6.00E-06 8.00E-06 1.00E-05
Curvature (1/mm)
Figure 42: Moment-curvature diagram of type A empty tube for mesh size sensitivity analysis of the shell element
Ovalization
5
4
Ovalization (mm)
2 10mm
20mm
40mm
1
60mm
0
0.00E+00 2.00E-06 4.00E-06 6.00E-06 8.00E-06 1.00E-05
Curvature (1/mm)
Figure 43: Ovalization-curvature diagram of type A empty tube for mesh size sensitivity analysis of the shell element
36
Chapter 4 - FEM model
Table 14: Number of elements, nodes and DoFs of the empty tube type A for different mesh sizes
Nominal shell element Number of Number of nodes DoF
mesh size (mm) elements
10 316 474 2844
20 240 160 1440
40 80 120 720
60 52 78 468
6%
5% Mmax
Error vs 10mm mesh (%)
4% Max Ovalization
3%
2%
60mm
1% 40mm
20mm 10mm
0%
0 500 1000 1500 2000 2500 3000
-1%
DoF
Figure 44: Mesh refinement study for empty tube A of the shell element
From Figure 42, Figure 43, Figure 44 and Table 14, it can be concluded that the mesh size is not playing an
important role on the bending moment estimation but it has more significant effect on the ovalization
estimation. It can be seen that there is a logarithmic decreasing trend on both results. As a conclusion a mesh
size of 20mm is selected for the shell elements of the steel tube.
The mesh size refinement study for the sand-fill is presented in the next diagrams and table. The study is
implemented to the complete model with the shell element mesh size for the steel tube being 20mm.
4.00E+08
3.50E+08
Moment (Nmm)
3.00E+08
2.50E+08
2.00E+08
1.50E+08
1.00E+08 20mm
30mm
5.00E+07 40mm
60mm
0.00E+00
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04
Curvature (1/mm)
Figure 45: Moment-curvature diagram of type A empty tube for mesh size sensitivity analysis of the solid element
37
Chapter 4 - FEM model
Ovalization
12
10
Ovalization (mm)
6
20mm
30mm
4
40mm
60mm
2
0
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04
Curvature (1/mm)
Figure 46: Ovalization-curvature diagram of type A empty tube for mesh size sensitivity analysis of the solid element
Table 15: Number of elements, nodes and DoFs of the filled tube type A for different mesh sizes of the solid element
Nominal solid element Number of Number of nodes DoF
mesh size (mm) elements
20 1280 2043 6849
30 704 1137 4131
40 440 723 2889
60 312 513 2259
1.40%
1.20% Mmax
Error vs 20mm mesh (%)
0.80%
0.60%
0.40%
0.20%
60mm 40mm
30mm 20mm
0.00%
2000 3000 4000 5000 6000 7000
DoF
Figure 47: Mesh refinement study for the sand-filled tube A of the solid element
From Figure 45, Figure 46, Figure 47 and Table 15, it can be concluded that the mesh size is not playing an
important role for both the bending moment estimation and the ovalization estimation. It is clear that changes in
the mesh size have a greater impact on the ovalization estimation but the difference is really small –around 1%.
As a conclusion a mesh size of 30mm is selected for the solid elements of the sand-fill.
38
Chapter 4 - FEM model
Another issue of the mesh size effect on the solid elements of the sand-fill is that there were mesh convergence
problems when the reduced integration C3D8R elements were used along with a very fine mesh. When the mesh
was coarser these convergence problems of the C3D8R elements ceased to exist. This is one of the reasons that
this type of integration scheme for the hexahedral solid element of ABAQUS is not selected for the current
study.
Finally, the subject of the time completion of every case was not really an issue since all the analyses are
completed within a few minutes (1 to 3 minutes depending on the mesh refinement).
The main outputs of the FEM analyses are the ones which are going to be compared to the experimental results.
These are (notated by the output names in ABAQUS):
The horizontal displacement in the centre of the steel tube (U1).
The moment of the Reference Point (CM1).
The rotation of the Reference Point (UR1).
The moment and the rotation of the Reference Point are used to derive the moment-curvature diagrams.
Curvature is calculated by converting the rotation with the use of equation (37). The horizontal displacement is
used to derive as the ovalization of the tube. Except for these output, the stresses of the soil in different
increments are also presented.
39
Chapter 4 - FEM model
The numerical model represents the mid-section of the experimental test set whose length is 11000 mm. Two
rotations are applied to the two ends of the tube to simulate the constant moment in the middle part which is a
result of the four point bending test set. Since the ovalization of the cross-section cannot happen at both ends of
the model due to the end plate connection of the tube, the coupling of the tube’s edges to the Reference Points
is possible. In the case in which this kind of end plate connection was not implemented at the experimental set
up, then special care for the ovalization at the edges of the middle part of the tube should have been taken.
As in the case of the slice model, the residual stress distribution is also inputted in the full scale model. Since
though, no other imperfection has been measured (welding locations, coil connections, characteristic hump
heights), only the imperfection of the characteristic humps will be inputted in the form of a buckling mode. The
buckling modes are derived by an elastic buckling analysis of the steel tube which has as a result the translational
deformations of the elements’ nodes.
Next, the deformations can be applied to the bending analysis of the tube, using a scaling factor according to the
hump height. Usually, the first buckling mode is used since its form is closer to the desired buckling result which
is the formation of the local buckle in the middle of the tube.
In the next parts of this chapter, the details of the full scale model are going to be presented except for the
parameters that are same as in the slice model. These are the material properties, the residual stresses, the contact
properties and the output parameters.
Also, a more refined mesh of the tube and the sand infill is chosen for the middle part of the model (6000mm of
the total 11000mm). This is done in order to reduce the total amount of elements and as a consequence of the
DoFs. Since the local buckle is expected to happen in the middle because of the mode shape then there a more
refined mesh is needed.
40
Chapter 4 - FEM model
Since only the middle part of the tube is modelled which has constant moment, the basic geometry of the model
is depicted in Figure 49. At both ends a Reference Point is coupled to the edges of the steel tube and the sand-
fill. The RPs are located at the center of the tube cross-section. On these RPs, the rotation is applied and trough
the RPs to the tube edges since they are kinematically coupled to them. At the right edge support of the tube, the
translations in x,y, and z directions and the rotations around y and z axes are restrained, whereas at the left edge
support, the translations in y and z and the rotations around y and z axes are restrained.
In Figure 50, the boundary conditions of the left edge of the model are depicted. The blue arrows are the
rotation constraints and the orange the translation constraints.
41
Chapter 4 - FEM model
The solution procedures which are used in the current analysis are the Static General and the Static Riks. Initially
the Static General is used in order to import the residual stresses and the imperfections in the steel tube and after
the Static Riks is used in order to calculate the local buckling and the post-buckling equilibrium path. This
method is used for cases where loading is proportional which means where the load magnitudes are governed by
a single scalar parameter [1]. Sometimes though, this method is not easily converged past the buckling moment
for very smooth equilibrium paths or for longer tubes [19].
4.0E+08
3.5E+08
3.0E+08
Moment (Nmm)
2.5E+08
2.0E+08
1.5E+08 S4R
S4
1.0E+08
5.0E+07
0.0E+00
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05
Curvature (1/mm)
Figure 51: Element type effect on the moment-curvature for Tube A
Table 16: Effect of the element type on moment and curvature in local buckling
Element type Mmax (Nmm) κcrit (1/mm)
S4R 334x106 8.834x10-6
S4 335x106 8.887x10-6
From Figure 51 and Table 16, it is concluded that the difference in using these two elements type is really
insignificant -less than 1% in both cases. Therefore, the S4R element will be used here, same as in the slice
model.
For the solid elements of the sand-fill a similar comparison was conducted between the C3D8 and C3D8R but
the C3D8R elements had a convergence issue even during the first increments of the analysis and for this reason
the C3D8 element is chosen.
42
Chapter 4 - FEM model
120%
100%
Error percentage vs 17 IPs
40%
20%
0%
4 6 8 10 12 14 16 18
Number of Integration Points
Figure 52: Effect of the integration point number on the moment and curvature
The number of the integration points of the shell element of the steel tube has a significant effect on the results.
It can be seen that more than 9 points have no difference on the results. Finally, 15 points are selected to be used
as input because the residual stress distribution was based on this amount of integration points.
Table 17: Number of elements, nodes and DoFs of the empty tube type A for different mesh sizes
Nominal shell element Number of Number of nodes DoF
mesh size (mm) elements
5 102720 102800 616800
10 54720 54800 328800
20 30720 30800 184800
40 18720 18800 112800
60 14720 14800 88800
The shell element used is the S4R with 15 integration points through thickness and multiplication factor 1 for
the first mode.
43
Chapter 4 - FEM model
6% 60mm
3%
2% 40mm
1%
20mm 10mm 5mm
0%
0 100000 200000 300000 400000 500000 600000 700000
Number Degrees of Freedom
Figure 53: Mesh size effect for empty tube A of the shell element S4R
From the Figure 53 it can be concluded that the error percentage is really insignificant when the mesh size gets
20mm and smaller. The critical curvature is more affected from the mesh size compared to the maximum
moment, especially for the 60mm and 40mm mesh sizes. The mesh size effect follows an exponential pattern
which decreases as the number of DoFs is getting higher. Taking into account also the time efficiency, the 10mm
mesh size is selected since the error percentage is really trivial compared to the 5mm mesh size.
Usually one of first two buckling modes is used as input but it is not always the case since it mainly depends on
the shape of the imperfection. Which brings up another issue, that of the element size of the tube. In order for
the buckling mode to be well shaped in the results, the element size should be small enough to be able to match
the mode’s shape.
In the next figures (Figure 55 - Figure 60), the first 6 buckling modes of Tube B are presented. The buckling
modes are occurring in pairs of sinusoidal (antisymmetric) and cosinusoidal (symmetric) (Figure 54). In this
research only the mode 1 or 2 are going to be used in the analyses, factored accordingly for the desired
imperfection.
44
Chapter 4 - FEM model
45
Chapter 4 - FEM model
46
Chapter 4 - FEM model
In Figure 61, the buckle of Tube B is depicted when Mode 1 and Mode 2 of the elastic buckling analysis with a
multiplication factor of 1 are inputted. The buckle is formed symmetrically with respect of the longitudinal plane
of the steel tube. It can be seen that the buckle shape is identical with only insignificant differences. In Figure 62,
the buckle photo of the test experiment for empty Tube B is depicted. The shape of the buckle of the numerical
model and of the test is the same.
Figure 61: Tube B buckle (empty tube) of Mode 1(left) and Mode 2(right)
47
Chapter 4 - FEM model
In Figure 63 and Figure 64, it can be seen that the failure shape of the numerical model is the same as the shape
which happened at the local buckling of the sand-filled tubes of the experimental testing. The shape of the
buckle of the numerical model and of the test is the same. Here, it is clear that the sand infill prevents the steel
tube from shaping an inward buckle and as a result it only buckles outwards. The sand infill creates an inner strut
which supports the top side of the tube and prevents excessive ovalization. This gives the filled tube better
curvature capacity and drives the forming of the local buckling at later curvature stage.
48
Chapter 4 - FEM model
The imperfection shape is either sinusoidal or cosinusoidal and the total imperfection size is w=2δ. Since
δ=1mm which is calculated in the buckling analysis, then the total imperfection w=2mm. Therefore, when a
certain imperfection height, w, should be inputted, then the multiplication factor is δ=w/2. In Figure 65, a clear
view of the modal height pattern is depicted.
δ
W
In Figure 66 it can be seen that at least 7 elements are considered per half-wavelength in order for the wrinkling
pattern to be described accurately. The deformation scale factor of the figure is 10.
Figure 66: Steel tube surface after the input of the mode in the Static analysis
49
Chapter 5 - Model validation
5. Model validation
The bending moment of the tests was caused by the self-weight and the ballast when the jacks lifted the tube
from the temporary supports and added to that is the moment caused by the chain forces. No partial factor was
used to the bending moments. The values presented in the previous table were corrected in order to remove the
axial contribution of the sand-fill, since the compressive stresses of the sand added strength to the bending
capacity and increased the bending moment resistance.
In the tests, the shell buckling of the sand-filled tube was delayed by the sand-fill which provided radial support
to the tube and when the buckle occurred, it was formed as an outward buckle. The sand-fill had also a
contribution to the bending resistance of the filled tube. Next, the results of the bending moments from the
experimental tests are presented in Table 18.
Below the Moment – Curvature diagrams are presented from the four types of tubes that were used in the
experimental tests for empty and filled tubes. The curves derived from the ABAQUS analyses and the failure
points of the tests are plotted and compared first for the empty tubes and then for the soil filled ones.
Additionally, the analytical curve for every tube type is plotted with the effects of ovalization and the residual
stresses taken into account.
The numerical analyses of every tube whose dimensional details (diameter and wall thickness) had been
measured, are presented and compared to the respective test values.
There is also the moment-curvature curve of tube A3 measured and presented here.
50
Chapter 5 - Model validation
Tube A empty
3.0E+08
Moment (Nmm)
2.0E+08
Analytical with oval., resid.
A3 - Empty - Test
1.0E+08
A3 - Empty - Test curve
A3 - Empty - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05
Curvature κ (1/mm)
Figure 67: Moment-Curvature diagram for tube A3 (empty)
3.0E+08
Moment (Nmm)
2.0E+08
3.0E+08
Moment (Nmm)
2.0E+08
51
Chapter 5 - Model validation
3.0E+08
Moment (Nmm)
2.0E+08
Empty tube - ABAQUS - 1.15fy
A3 - Empty - Test
A3 - Empty - Test curve
A3 - Empty - ABAQUS
1.0E+08
A4 - Empty - Test
A4 - Empty - ABAQUS
A6 - Empty - Test
A6 - Empty - ABAQUS
0.0E+00
0.00E+00 3.00E-06 6.00E-06 9.00E-06 1.20E-05
curvature κ (1/mm)
Figure 70: Moment-Curvature diagram for tube A (empty)
52
Chapter 5 - Model validation
Tube B empty
7.0E+08
6.0E+08
5.0E+08
4.0E+08
3.0E+08 Analytical with oval., resid.
2.0E+08 B2 - Empty - Test
1.0E+08 B2 - Empty - ABAQUS
0.0E+00
0.0E+00 1.0E-05 2.0E-05 3.0E-05 4.0E-05 5.0E-05
Curvature κ (1/mm)
Figure 71: Moment-Curvature diagram for tube B2 (empty)
7.0E+08
6.0E+08
5.0E+08
4.0E+08
3.0E+08
Analytical with oval., resid.
2.0E+08
B5 - Empty - Test
1.0E+08
B5 - Empty - ABAQUS
0.0E+00
0.0E+00 1.0E-05 2.0E-05 3.0E-05 4.0E-05 5.0E-05
Curvature κ (1/mm)
Figure 72: Moment-Curvature diagram for tube B5 (empty)
7.0E+08
6.0E+08
5.0E+08
4.0E+08
3.0E+08 Analytical with oval., resid.
2.0E+08 B6 - Empty - Test
1.0E+08
B6 - Empty - ABAQUS
0.0E+00
0.0E+00 1.0E-05 2.0E-05 3.0E-05 4.0E-05 5.0E-05
Curvature κ (1/mm)
Figure 73: Moment-Curvature diagram for tube B6 (empty)
53
Chapter 5 - Model validation
9.00E+08
8.00E+08
7.00E+08
Moment (Nmm)
6.00E+08
B2 - Empty - Test
5.00E+08
B2 - Empty - ABAQUS
4.00E+08
B5 - Empty - Test
3.00E+08
B5 - Empty - ABAQUS
2.00E+08
B6 - Empty - Test
1.00E+08
B6 - Empty - ABAQUS
0.00E+00
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05
curvature κ (1/mm)
Figure 74: Moment - Curvature diagram for tube B (empty)
54
Chapter 5 - Model validation
Tube C empty
5.0E+08
4.0E+08
3.0E+08
Analytical with oval., resid.
2.0E+08
C2 - Empty - Test
1.0E+08
C2 - Empty - ABAQUS
0.0E+00
0.0E+00 4.0E-06 8.0E-06 1.2E-05
Curvature κ (1/mm)
Figure 75: Moment-Curvature diagram for tube C2 (empty)
5.0E+08
4.0E+08
3.0E+08
Analytical with oval., resid.
2.0E+08
C3 - Empty - Test
1.0E+08
C3 - Empty - ABAQUS
0.0E+00
0.0E+00 4.0E-06 8.0E-06 1.2E-05
Curvature κ (1/mm)
Figure 76: Moment-Curvature diagram for tube C3 (empty)
5.0E+08
4.0E+08
3.0E+08
2.0E+08 Analytical with oval., resid.
1.0E+08 C6 - Empty - Test
C6 - Empty - ABAQUS
0.0E+00
0.0E+00 4.0E-06 8.0E-06 1.2E-05
Curvature κ (1/mm)
Figure 77: Moment-Curvature diagram for tube C6 (empty)
55
Chapter 5 - Model validation
7.00E+08
Moment - Curvature for Tube C
6.00E+08
5.00E+08
Moment (Nmm)
4.00E+08
C2 - Empty - Test
56
Chapter 5 - Model validation
Tube D empty
1.0E+09
8.0E+08
6.0E+08
Analytical with oval., resid.
4.0E+08
D1 - Empty - Test
2.0E+08
D1 - Empty - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06
Curvature κ (1/mm)
Figure 79: Moment-Curvature diagram for tube D1 (empty)
1.0E+09
8.0E+08
6.0E+08
4.0E+08 Analytical with oval., resid.
D2 - Empty - Test
2.0E+08
D2 - Empty - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06
Curvature κ (1/mm)
Figure 80: Moment-Curvature diagram for tube D2 (empty)
1.0E+09
8.0E+08
6.0E+08
4.0E+08 Analytical with oval., resid.
D3 - Empty - Test
2.0E+08
D3 - Empty - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06
Curvature κ (1/mm)
Figure 81: Moment-Curvature diagram for tube D3 (empty)
57
Chapter 5 - Model validation
1.40E+09
Moment - Curvature for Tube D
1.20E+09
1.00E+09
Moment (Nmm)
8.00E+08
D1 - Empty - Test
6.00E+08
D1 - Empty - ABAQUS
58
Chapter 5 - Model validation
Tube A sand-filled
3.0E+08
Moment (Nmm)
2.0E+08
4.0E+08
Moment - Curvature for Tube A2
3.0E+08
Moment (Nmm)
2.0E+08
4.0E+08
Moment - Curvature for Tube A5
3.0E+08
Moment (Nmm)
2.0E+08
59
Chapter 5 - Model validation
3.0E+08
Moment (Nmm)
2.0E+08
60
Chapter 5 - Model validation
Tube B sand-filled
7.0E+08
6.0E+08
5.0E+08
4.0E+08
3.0E+08
B1 - Filled - Analytical
2.0E+08
B1 - Filled - Test
1.0E+08
B1 - Filled - ABAQUS
0.0E+00
0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05
Curvature κ (1/mm)
Figure 87: Moment - Curvature diagram for tube B1 (sand-filled)
1.0E+09
Moment - Curvature for Tube B3
9.0E+08
8.0E+08
7.0E+08
Moment (Nmm)
6.0E+08
5.0E+08
4.0E+08
3.0E+08
B3 - Filled - Analytical
2.0E+08
B3 - Filled - Test
1.0E+08
B3 - Filled - ABAQUS
0.0E+00
0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05
Curvature κ (1/mm)
Figure 88: Moment - Curvature diagram for tube B3 (sand-filled)
7.0E+08
6.0E+08
5.0E+08
4.0E+08
3.0E+08
B4 - Filled - Analytical
2.0E+08
B4 - Filled - Test
1.0E+08
B4 - Filled - ABAQUS
0.0E+00
0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05
Curvature κ (1/mm)
Figure 89: Moment - Curvature diagram for tube B4 (sand-filled)
61
Chapter 5 - Model validation
9.0E+08
8.0E+08
7.0E+08
Moment (Nmm)
6.0E+08
5.0E+08
B1 - Filled - Test
4.0E+08
B1 - Filled - ABAQUS
3.0E+08 B3 - Filled - Test
2.0E+08 B3 - Filled - ABAQUS
62
Chapter 5 - Model validation
Tube C sand-filled
7.0E+08
Moment - Curvature for Tube C1
6.0E+08
5.0E+08
Moment (Nmm)
4.0E+08
3.0E+08
2.0E+08
C1 - Filled - Analytical
1.0E+08 C1 - Filled - Test
C1 - Filled - Abaqus
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 1.8E-05
Curvature κ (1/mm)
Figure 91: Moment - Curvature diagram for tube C1 (sand-filled)
6.0E+08
5.0E+08
Moment (Nmm)
4.0E+08
3.0E+08
6.0E+08
5.0E+08
Moment (Nmm)
4.0E+08
3.0E+08
2.0E+08
C5 - Filled - Analytical
1.0E+08 C5 - Filled - Test
C5 - Filled - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 1.8E-05
Curvature κ (1/mm)
Figure 93: Moment - Curvature diagram for tube C5 (sand-filled)
63
Chapter 5 - Model validation
6.0E+08
5.0E+08
Moment (Nmm)
4.0E+08
3.0E+08
C1 - Filled - Test
C1 - Filled - Abaqus
2.0E+08
C4 - Filled - Test
C4 - Filled - ABAQUS
1.0E+08
C5 - Filled - Test
C5 - Filled - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 1.8E-05
Curvature κ (1/mm)
Figure 94: Moment - Curvature diagram for tube C (sand-filled)
64
Chapter 5 - Model validation
Tube D sand-filled
1.0E+09
8.0E+08
6.0E+08
4.0E+08 D4 - Filled - Analytical
2.0E+08 D4 - Filled - Test
D4 - Filled - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 95: Moment - Curvature diagram for tube D4 (sand-filled)
1.0E+09
8.0E+08
6.0E+08
Analytical with residual
4.0E+08
Sand-filled tube (residual) - ABAQUS
2.0E+08
D5 - Filled - Test
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 96: Moment - Curvature diagram for tube D5 (sand-filled)
1.0E+09
8.0E+08
6.0E+08
4.0E+08 D6 - Filled - Anaytical
2.0E+08 D6 - Filled - Test
D6 - Filled - ABAQUS
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 97: Moment - Curvature diagram for tube D6 (sand-filled)
65
Chapter 5 - Model validation
1.6E+09
Moment - Curvature for Tube D
1.4E+09
1.2E+09
Moment (Nmm)
1.0E+09
8.0E+08
66
Chapter 5 - Model validation
The curvature values of the experimental tests were derived by the critical strains measured during the testing. It
can be seen from the diagrams that the measured bending moments before failure are close to the curve derived
by the numerical analyses. In Table 19 the model error percentage between them is presented. Except for two
measurements - one for an empty tube and one for a soil-filled tube - in tube D which have a difference more
than 10%, all the other comparisons are around 10% and less.
The MABAQUS is the moment of the numerical moment-curvature diagram corresponding to the critical curvature
value of the experimental test.
Table 19: Model error between the measured and the calculated bending moments
Tube A
Test ID A1 A2 A3 A4 A5 A6
fill Sand Sand Empty Empty Sand Empty
Mtest (kNm) 353.0 354.3 361.7 393.6 382.8 378.3
Critical strain 0.003395 0,002609 0.001880 0.002415 0.002339 0.002716
Curvature 1.336Ε-05 1.03Ε-05 7.403E-06 9.507Ε-06 9.21Ε-06 1.069Ε-05
MABAQUS (kNm) 375.95 366.88 317.5 342.1 361.08 350.2
Model error (%) -6.51 -3.54 12.23 13.09 5.67 7.43
Tube B
Test ID B1 B2 B3 B4 B5 B6
fill Sand Empty Sand Sand Empty Empty
Mtest (kNm) 946.8 891.9 994.4 946.4 888 824
Critical strain 0.019313 0.011333 0.014846 0.015693 0.011518 0.008456
Curvature 7.60Ε-05 4.462Ε-05 5.84Ε-05 6.18Ε-05 4.534Ε-05 3.329Ε-05
MABAQUS (kNm) 924.83 839.09 918.81 920.72 837.41 853.46
Model error (%) 2.32 5.92 7.60 2.71 5.65 -3.55
Tube C
Test ID C1 C2 C3 C4 C5 C6
fill Sand Empty Empty Sand Sand Empty
Mtest (kNm) 668.4 596 597 681.4 660.7 661
Critical strain 0.004260 0.002621 0.002849 0.004647 0.005029 0.0032751
Curvature 1.396Ε-05 8.593Ε-06 9.34Ε-06 1.52Ε-05 1.65Ε-05 1.07Ε-05
MABAQUS (kNm) 685.37 638.75 648.42 690.05 693.56 662.14
Model error (%) -2.54 -7.17 -8.68 -1.27 -4.97 -0.22
Tube D
Test ID D1 D2 D3 D4 D5 D6
fill Empty Empty Empty Sand Sand Sand
Mtest (kNm) 1228 1045 1194 1378.0 1119.4 1391.1
Critical strain 0.002444 0.002717 0.002196 0.003197 0.004079 0.004658
Curvature 6.87E-06 7.641E-06 6.177E-06 8.99E-06 1.15E-05 1.31E-05
MABAQUS (kNm) 1151.14 1227.83 1044.03 1252.98 1335.70 1371.23
Model error (%) 6.40 -17.30 12.61 9.07 -19.32 1.43
Next, the moment curvature diagrams of every tube type depicting both the empty tubes and the sand-filled
ones, are presented, in order to have a clear image of how the sand-fill affects the bending behaviour of the tube.
67
Chapter 5 - Model validation
3.00E+08
Moment (Nmm)
2.00E+08
Empty tube (residual stresses) - ABAQUS
A3 - Empty - Test
A4 - Empty - Test
A6 - Empty - Test
1.00E+08
Sand-filled tube (residual stresses) -ABAQUS
A1- Filled - Test
A2 - Filled - Test
A5 - Filled - Test
0.00E+00
0.00E+00 3.00E-06 6.00E-06 9.00E-06 1.20E-05 1.50E-05
curvature κ (1/mm)
Figure 99: Moment - Curvature diagram for tube A (empty and sand-filled)
9.00E+08
8.00E+08
7.00E+08
Moment (Nmm)
6.00E+08
5.00E+08
Empty tube (residual) - ABAQUS
4.00E+08 B2 - Empty - Test
B5 - Empty - Test
3.00E+08 B6 - Empty - Test
Sand-filled (residual) - ABAQUS
2.00E+08
B1 - Filled - Test
1.00E+08 B3 - Filled - Test
B4 - Filled - Test
0.00E+00
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05
curvature κ (1/mm)
Figure 100: Moment - Curvature diagram for tube B (empty and sand-filled)
68
Chapter 5 - Model validation
7.00E+08
6.00E+08
Moment (Nmm)
5.00E+08
Empty tube (residual) - ABAQUS
4.00E+08 C2 - Empty - Test
C3 - Empty - Test
3.00E+08 C6 - Empty - Test
Sand-filled tube (residual) - ABAQUS
2.00E+08
C1 - Filled - Test
C4 - Filled - Test
1.00E+08
C5 - Filled - Test
0.00E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
curvature κ (1/mm)
Figure 101: Moment - Curvature diagram for tube C (empty and sand-filled)
1.40E+09
1.20E+09
Moment (Nmm)
1.00E+09
Empty tube (residual) - ABAQUS
69
Chapter 5 - Model validation
For the sand-filled tube A (Figure 83-Figure 86), the comparison of the test values and the numerical curves
show a good fit except for one of the measured test values (A5). This may be happening due to the residual
stress distribution and the material properties of the tube not being accurately represented in the numerical
model or there might be an error in the measurements. The comparison between the numerical and analytical
curves show a perfect fit. This also happening due to the stress-strain behaviour of the Tube A steel material
since it has been considered to act as bilinear with a yield plateau and a strain hardening zone which is not
reached, at least at the low curvature values, which are examined here. This is in agreement with the bilinear
material model but with no strain hardening taken into account of the analytical formula.
Tube B
For the empty tube B (Figure 71-Figure 74), the comparison of the numerical and test results shows a really good
fit. Also the analytical curve is a close fit until the yielding. After the yielding the numerical and the analytical
have a small difference due to the steel stress-strain diagram approach, since for the analytical formula the stress-
strain follows a bilinear behaviour.
For the sand-filled tube B (Figure 87-Figure 90), the comparison of the test values and the numerical curves
show a good fit except for one of the measured test values (B3). The comparison between the numerical and
analytical curves shows that until the yielding stage they fit perfectly but after yielding they have some difference.
This can be attributed to the different stress-strain shape used in these two approaches, since the analytical
formula considers a bilinear material behaviour, whereas the material stress-strain diagram of Tube B, there is
not a yielding plateau.
Tube C
For the empty tube C (Figure 75-Figure 78), the test results are close to the numerical and the analytical curves.
For the sand-filled tube C (Figure 91-Figure 94), the comparison of the test values and the numerical curves
show a good fit. The comparison of the numerical and analytical curves shows a perfect fit.
Tube D
For the empty tube D (Figure 79-Figure 82), two of the test values show a difference in the moment
measurement, whereas the other one is close to the numerical results. This may be happening due to the residual
stress distribution and the material properties of the tube not being accurately represented in the numerical
model. The analytical curve, though, coincides to the numerical ones.
For the sand-filled tube D (Figure 95-Figure 98), the comparison of the test values and the numerical curves
show a good fit except for one of the measured test values (D5) which is also really far from the other measured
ones. The comparison between the numerical and analytical curves shows an almost perfect fit.
General
In general, the bending moment estimation from the numerical analyses is similar to the values measured from
the tests. The only dissonance are two test values which have a difference more than 10% (17% for D2 and 19%
for D5). This could be a measurement problem or a scatter in the properties of the materials since these values
are also not close to the rest of the test for the same tube.
70
Chapter 5 - Model validation
The sources of these small differences also may come from the fact that this numerical model is just a cross-
sectional model and not a full scale model of the test in which also the local buckling failure can be calculated. A
parametrical study of the sand’s properties will be conducted in order to have a clear view of the effect that each
one of them has on the model behaviour.
The comparison of the analytical formulas to the numerical and measured ones also shows a good fit. It can be
concluded that the tube types A and C curves were closer to the analytical ones because of their material input in
the analysis. Since the analytical formulas account for an elastic-perfectly plastic material behaviour, these two
tubes are closer to the analytical because of the plateau right after yielding in the stress-strain diagram. Of course
after the plateau, it follows a strain hardening zone but in this case the rotation is not big enough to reach at this
point.
Comparing the empty tubes to the filled ones, it is clear that the sand-fill provides radial support to the steel tube
which helps for the tube to have a better post-yielding behaviour. In the numerical and analytical models, it can
be seen that the sand-fill contributes to both the post-yielding behaviour and the bending moment capacity since
the filled tubes exhibit higher moment values. Also in the numerical and analytical models the moment-rotation
diagram continues without taking the local buckling and the drop in the moment into account, but the test
results show where that has happened for the real tubes. The contribution of the sand to the moment capacity
and the local buckling can be seen also in the test results since the tube fails reaching higher moments and
curvatures.
71
Chapter 5 - Model validation
Ovalization
20
Empty tubes
A
B
15 C
Measured ovalization α (mm)
y = 0.5251x
10 R² = 0.3225
0
0 2 4 6 8 10 12 14 16 18 20
Calculated ovalization - ABAQUS (mm)
Figure 103: Ovalization comparison between measurements and ABAQUS analyses for empty tubes.
Ovalization
14 Sand-filled tubes
A
12 B
C
D
Measured ovalization α (mm)
10
4
y = 0.2016x
R² = 0.8562
2
0
0 2 4 6 8 10 12 14
Calculated ovalization - ABAQUS (mm)
Figure 104: Ovalization comparison between measurements and ABAQUS analyses for soil-filled tubes.
72
Chapter 5 - Model validation
Table 20: Ovalization values from measurements and from ABAQUS analyses for tubes A and B
Tube A B
Meas. oval. (mm) 0.90 0.95 1.06 8.42 6.58 5.94
empty Abaqus ov. (mm) 2.58 3.52 4.13 18.58 19.06 11.67
Difference (%) 185.6 270.2 288.6 120.8 189.9 96.4
Meas. oval. (mm) 0.72 0.40 0.18 3.20 2.27 2.12
sand Abaqus ov. (mm) 1.85 1.50 1.31 13.10 10.51 10.99
Difference (%) 157.7 274.0 625.7 309.3 363.9 418.3
Table 21: Ovalization values from measurements and from ABAQUS analyses for tubes C and D
Tube C D
Meas. oval. (mm) 2.09 1.60 1.06 9.27 10.46 7.93
empty Abaqus ov. (mm) 4.71 5.07 5.84 7.44 8.94 6.22
Difference (%) 125.8 216.6 450.8 -19.7 -14.5 -21.5
Meas. oval. (mm) 0.24 0.60 0.33 0.20 0.53 0.25
sand Abaqus ov.(mm) 2.74 2.92 3.16 3.57 3.42 5.30
Difference (%) 1051.7 387.3 846.8 1682.7 552.3 1998.7
One of the sources of this difference is the lack of reliable data input for the sand-fill properties. Another reason
is that probably this numerical model cannot predict the ovalization values close enough to the real ones.
By changing the Esand from 30MPa to 120MPa, φ from 32.50 to 330 and ψ from 2.50 to 30, which are in the
normal value range of a typical sand soil, then the ABAQUS ovalization values for the sand-filled tubes are
closer to the measured ones. Since that also the empty tubes have a difference between the measured and the
calculated values, then the new trendline of the soil filled tubes ovalization diagram should be closer to it. In
Table 22 and Figure 105 the results of this change are depicted.
Table 22: Ovalization values from measurements and from ABAQUS analyses for more appropriate sand properties
Tube A B
Meas. oval. (mm) 0.72 0.40 0.18 3.20 2.27 2.12
Abaqus ov. (mm) 0.77 0.58 0.51 5.92 4.60 4.85
Difference (%) 7.4 45.6 181.1 85.0 102.9 128.7
Tube C D
Meas. oval. (mm) 0.24 0.60 0.33 0.20 0.53 0.25
Abaqus ov. (mm) 1.20 1.28 1.35 1.04 1.57 1.78
Difference (%) 405.2 112.9 303.9 420.0 199.0 605.0
73
Chapter 5 - Model validation
Ovalization
7
Sand-filled tubes
A
6 B
C
D
Measured ovalization α (mm)
1 y = 0.4703x
R² = 0.9099
0
0 1 2 3 4 5 6 7
Calculated ovalization - ABAQUS (mm)
Figure 105: Ovalization comparison between measurements and ABAQUS analyses for soil-filled tubes with different sand properties.
From Figure 105, it can be seen that the change in sand-fill’s material properties makes the ovalization numerical
results smaller, since the difference of the initial numerical results to the measured ones was significant. After
this change, the difference is smaller, but since that also the empty tubes ovalization results do not much with
each other - measured and calculated - a set of sand property values which both give better ovalization results
and lay within the normal values range is tested and gives an outcome close to the empty tube trendline.
The moment change due to the sand-fill properties is not examined since as it can be seen in Chapter 6, in which
a parametric study over the sand-fill’s material properties is conducted, the change in the moment is very small
and it only can be seen in high curvature values.
74
Chapter 5 - Model validation
4.00E+08
3.50E+08
3.00E+08
Series1
Moment (Nmm)
2.50E+08
Section 1
2.00E+08
Section 2
1.50E+08 Section 3
1.00E+08 Section 4
0.00E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05 2.50E-05
Curvature (1/mm)
Figure 106: Sections chosen for the sand stress depiction
The name of the stresses depicts their direction and it is 1 for X-axis, 2 for Y-axis and 3 for Z-axis, which means
that the S11 stress is the σxx. The values in the legend are in MPa and the tensional stresses are positive. The
tensional stresses are depicted with white colour, in order to give a good view of their distribution in the cross-
sectional stress pattern.
Checking the area where the failure in the experiments is happening then Section 4 is far after the failure. The
failure values of the tube A are located around the 10-5 curvature value of the moment-curvature diagram.
75
Chapter 5 - Model validation
76
Chapter 5 - Model validation
77
Chapter 5 - Model validation
From the S22 plots, it can be seen that in the middle a supporting strut to the steel tube is forming and at the
sides small areas with tensional stresses are noticed since the tube is ovalizing and the supressed sand tends to
move side wards to fill up the free space. These areas are really small with small tensional stresses. The forming
strut is the main reason why the moment-curvature curve does not start to drop once the top moment has been
reached, since it offers a support to the tube’s top and bottom parts and makes the ovalization smaller.
The S11 plots show that at the beginning the middle part of the sand-fill is in tension and the top and bottom
parts in compression and the stresses are significantly lower when compared to the next sections. Later there are
almost nowhere tensional stresses, also at the places that tension would be expected (left and right sides). This
happens due to the confinement of the sand from the steel tube. Due to the ovalization and the plastification of
the sand, the sand tends to expand sidewards but the presence of the tube confines it and prevents it from
moving freely thus the compressional stresses of the sand-fill at the sides.
The S33 plots show that at the beginning, half of the sand-fill is in tension due to the imposed rotation but there
are already developing some plastic strains. These tensional values are smaller than the compressional ones and
close to the cohesion value which shows that the plastification of the sand-fill has already started. When the
rotation and the compressional stresses become bigger, then the tensional stresses become really small due to the
plastification.
In Table 23 the axial reaction forces on the sand-fill and the steel tube and the moment capacity contribution of
each material are presented. It is clear from these values that the sand-fill contribution in the axial direction is
very small compared to the steel tube contribution. This contribution becomes even less for the tension since the
sand has a very small cohesion value and in reality cannot take any significant tensional force (for numerical
analysis reasons there must be at least a small cohesion value). Due to this phenomenon and additionally due to
the plastification of the sand, the contribution in tension in the axial direction becomes smaller and smaller as the
loading of the cross-section continues. On the other hand, the compressional axial force is getting higher but as
was mentioned before its contribution compared to steel is very small and this is the reason why the neutral axis
of the cross-section does not move significantly from its original position. Finally, this can be seen also in the
moment capacity contribution of each material to the applied rotation. Almost all the applied rotation is taken by
the steel tube whereas the sand does not contribute at all (99.95% to 0.05%).
Table 23: Sand-fill and Steel Tube total axial reaction forces in the Sections and moment capacity contribution.
Moment capacity
Sand-fill Steel Tube
contribution
Incr.
Tension Compression Tension Compression Sand Steel
(N) (N) (N) (N) (%) (%)
Section 1
84.9 320.5 190619.0 190383.4 0.07 99.93
(0.0144)
Section 2
7.5 538.8 647725.0 647193.7 0.03 99.97
(0.0500)
Section 3
1.1 1762.1 944141.0 942380.0 0.05 99.95
(0.1025)
Section 4
0.2 3670.7 1082102.8 1078432.2 0.10 99.90
(0.2025)
The initial volume is 1,945,817mm3. It can be seen in Table 24 that the volume change of the sand is almost zero
but nevertheless there is a small reduction. This shows that the whole sand volume is in general in compression
just as the plots in Figure 107-Figure 112 depict.
78
Chapter 5 - Model validation
Table 24: Sand volume change and longitudinal strain at section axis
Volume change Strain ε at
Volume Change section axis
(mm3) (%) (x10-6)
Section 1 1,945,782.88 -0.0018 0.31
Section 2 1,945,822.88 0.0003 3.41
Section 3 1,945,758.13 -0.003 -1.37
Section 4 1,945,583.25 -0.012 11.77
As an investigation for the behaviour of the sand-fill in the axial direction of the tube, a numerical analysis is
conducted with Esand equal to 1000MPa. This will give a better insight of the sand behaviour in the axial direction
and it is done only for research reasons since such a value for the sand-infill is unrealistic. The results as in Table
23 are derived and presented in Table 25.
Table 25: Sand and Steel total axial reaction forces in the Sections and moment capacity contribution for E sand = 1000MPa.
Moment capacity
Sand-fill Steel Tube
contribution
Incr.
Tension Compression Tension Compression Sand Steel
(N) (N) (N) (N) (%) (%)
Section 1
16.0 1245.7 193390.1 192160.4 0.19 99.81
(0.01454)
Section 2
0.0 5403.0 612875.5 607472.5 0.23 99.77
(0.04664)
Section 3
0.0 12027.5 959958.1 947930.6 0.35 99.65
(0.10664)
Section 4
0.0 23424.9 1093323.8 1069898.9 0.62 99.38
(0.20726)
Due to the reason that the step increments of these two analyses do not coincide precisely, there is a small
difference between the total force of each section compared to the respective one of the other analysis. The
increment closer to the initial analysis with Esand = 30MPa is chosen from the analysis with Esand = 1000MPa.
It can be concluded that the increase in the sand’s Young’s modulus, makes the moment capacity contribution of
the sand-fill higher but nevertheless still very small compared to the contribution of the steel tube. The
compressional forces are becoming significantly higher but the tensional ones are becoming smaller and equal to
zero very fast. This means that the sand-fill plastificates very early at the application of loading because the
stresses in compression are very big.
79
Chapter 5 - Model validation
Next, the p-q diagrams of the soil in different nodes is going to be presented. Here p is the equivalent effective
pressure stress defined by:
1 𝜎1 + 𝜎2 + 𝜎3
𝑝 = − 𝑡𝑟𝑎𝑐𝑒(𝜎) = − , (42)
3 3
and q is the equivalent deviatoric stress (the Mises equivalent stress) defined by:
is the deviatoric stress. The previous were taken from the ABAQUS Analysis manual.
The p-q diagram shows the stress trajectory followed in the solution. In the next figure the location of the nodes
is depicted. The positive stresses on the graphs are compressional.
80
Chapter 5 - Model validation
Top node:
81
Chapter 5 - Model validation
82
Chapter 5 - Model validation
83
Chapter 5 - Model validation
The graphs of Figure 114-Figure 122 confirm that apart from some small parts at the edges -right and left edges
and only for few increments-, in general the sand-fill is in compression because the sand can mainly support in
compression and not in tension. The tensional stresses that occur are really small and happen during the first
increments when the sand-fill is partly elastic, since the plastification of the sand-fill happens in the early
increments of the analyses.
Additionally, the stresses in the right and left edge are really low compared to the stresses of the other nodes
(almost 5kPa at the edges to 30-100 kPa at the other nodes for the equivalent deviatoric stress q). This is
happening due to the ovalization of the tube since at the right and left edges the tube moves outwards and at the
top and bottom edges it compresses.
84
Chapter 5 - Model validation
For Tube D, only the 4% and 20% analyses are implemented since due to lack of time, the choice of
investigating the change of the sand properties and how the local buckling failure is affected by it, was followed.
Therefore, the sand-fill properties are changed to the ones chosen in Chapter 5.4 (Esnnd = 120MPa, φ = 330, ψ =
30) and some analyses are performed for 4% and 20% imperfection/thickness ratios for tube types A and D.
Next, the moment-curvature and ovalization-curvature diagrams of each tube type are presented for different
multiplication factors of the applied mode. The maximum moment is considered to be the buckling moment
which happens just before the drop in the diagram. At this point, the curvature is considered to be the critical
one.
Initially, the empty tube diagrams are presented and after these the sand-filled ones. Finally, the diagrams from
both the empty and the sand-filled are compared to each other.
85
Chapter 5 - Model validation
TUBE A Empty
Tube A has D/t=102 and fy=307MPa and it is spirally welded. The moment-curvature and ovalization-curvature
diagrams are depicted in Figure 123 and in Figure 124 respectively.
A6 - Empty - Test
Empty - Imp/Thick 20%
2.0E+08 Empty - Imp/Thick 16%
Empty - Imp/Thick 12%
Empty - Imp/Thick 8%
1.0E+08 Empty - Imp/Thick 4%
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 123: Moment-Curvature diagrams of Tube A of the full scale model (empty).
0
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 124: Ovalization-Curvature diagrams of Tube A of the full scale model (empty).
86
Chapter 5 - Model validation
TUBE B Empty
Tube B has D/t=72 and fy=488MPa and it is longitudinally welded. The moment-curvature and ovalization-
curvature diagrams are depicted in Figure 125 and in Figure 126 respectively.
12 Empty - Imp/Thick 8%
Empty - Imp/Thick 4%
10
8
6
4
2
0
0.00E+00 1.00E-05 2.00E-05 3.00E-05 4.00E-05 5.00E-05
Curvature κ (1/mm)
Figure 126: Ovalization-Curvature diagrams of Tube B of the full scale model (empty).
87
Chapter 5 - Model validation
TUBE C Empty
Tube C has D/t=97 and fy=312MPa and it is spirally welded. The moment-curvature and ovalization-curvature
diagrams are depicted in Figure 127 and in Figure 128 respectively.
7.0E+08
Moment - Curvature for Tube C
6.0E+08
5.0E+08
Moment (Nmm)
7
Ovalization - Curvature for Tube C
Empty - Slice model
Test
6 Empty - Imp/Thick 20%
Empty - Imp/Thick 16%
5 Empty - Imp/Thick 12%
Ovalization (mm)
Empty - Imp/Thick 8%
Empty - Imp/Thick 4%
4
0
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05
Curvature κ (1/mm)
Figure 128: Ovalization-Curvature diagrams of Tube C of the full scale model (empty).
88
Chapter 5 - Model validation
TUBE D Empty
Tube D has D/t=119 and fy=503MPa and it is longitudinally and circumferentially welded. The moment-
curvature and ovalization-curvature diagrams are depicted in Figure 129 and in Figure 130 respectively.
4.0E+08
2.0E+08
0.0E+00
0.0E+00 1.0E-06 2.0E-06 3.0E-06 4.0E-06 5.0E-06 6.0E-06 7.0E-06 8.0E-06 9.0E-06 1.0E-05 1.1E-05
Curvature κ (1/mm)
Figure 129: Moment-Curvature diagrams of Tube D of the full scale model (empty).
Empty - Imp/Thick 4%
8
0
0.0E+00 1.0E-06 2.0E-06 3.0E-06 4.0E-06 5.0E-06 6.0E-06 7.0E-06 8.0E-06 9.0E-06 1.0E-05 1.1E-05
Curvature κ (1/mm)
Figure 130: Ovalization-Curvature diagrams of Tube D of the full scale model (empty).
89
Chapter 5 - Model validation
TUBE A Sand-filled
Tube A has D/t=102 and fy=307 MPa and it is spirally welded. The moment-curvature and ovalization-
curvature diagrams are depicted in Figure 131 and in Figure 132 respectively.
3.0E+08
Moment (Nmm)
Sand-filled - Imp/Thick 8%
Sand-filled - Imp/Thick 4%
1
0.5
0
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 132: Ovalization-Curvature diagrams of Tube A of the full scale model (sand-filled).
90
Chapter 5 - Model validation
TUBE B Sand-filled
Tube B has D/t=72 and fy=488 MPa and it is longitudinally welded. The moment-curvature and ovalization-
curvature diagrams are depicted in Figure 133 and in Figure 134 respectively.
1.0E+09
Moment - Curvature for Tube B
9.0E+08
8.0E+08
Analytical with residual
7.0E+08 Sand-filled (residual) - ABAQUS
Moment (Nmm)
0
0.0E+00 1.0E-05 2.0E-05 3.0E-05 4.0E-05 5.0E-05 6.0E-05 7.0E-05 8.0E-05
Curvature κ (1/mm)
Figure 134: Ovalization-Curvature diagrams of Tube B of the full scale model (sand-filled).
91
Chapter 5 - Model validation
TUBE C Sand-filled
Tube C has D/t=97 and fy=312 MPa and it is spirally welded. The moment-curvature and ovalization-curvature
diagrams are depicted in Figure 135 and in Figure 136 respectively.
6.0E+08
5.0E+08
Moment (Nmm)
4.0
Ovalization - Curvature for Tube C
Sand-filled - Test
3.5 Sand-filled - Strip model
Sand-filled - Imp/Thick 20%
3.0 Sand-filled - Imp/Thick 16%
Sand-filled - Imp/Thick 12%
Ovalization (mm)
1.5
1.0
0.5
0.0
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05 1.6E-05 1.8E-05
Curvature κ (1/mm)
Figure 136: Ovalization-Curvature diagrams of Tube C of the full scale model (sand-filled).
92
Chapter 5 - Model validation
TUBE D Sand-filled
Tube D has D/t=119 and fy=503 MPa and it is spirally welded. The moment-curvature and ovalization-
curvature diagrams are depicted in Figure 137 and in Figure 138 respectively.
1.2E+09
Moment (Nmm)
9.0E+08
0
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05 1.6E-05
Curvature κ (1/mm)
Figure 138 Ovalization-Curvature diagrams of Tube D of the full scale model (sand-filled).
93
Chapter 5 - Model validation
3.0E+08
Moment (Nmm)
0
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 140: Ovalization-Curvature diagrams for empty and sand-filled Tube A for different Imp/Thick percentages.
94
Chapter 5 - Model validation
3.0E+08
Moment (Nmm)
3.0E+08
Moment (Nmm)
2.0E+08 2.0E+08
1.0E+08 1.0E+08
Empty - Imp/Thick 4% Empty - Imp/Thick 8%
Sand-filled - Imp/Thick 4% Sand-filled - Imp/Thick 8%
0.0E+00 0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm) Curvature κ (1/mm)
3.0E+08
Moment (Nmm)
3.0E+08
Moment (Nmm)
2.0E+08 2.0E+08
1.0E+08 1.0E+08
Empty - Imp/Thick 12% Empty - Imp/Thick 16%
Sand-filled - Imp/Thick 12% Sand-filled - Imp/Thick 16%
0.0E+00 0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm) Curvature κ (1/mm)
3.0E+08
Moment (Nmm)
2.0E+08
Table 35: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube A
Imp Mmax (kNm) κcrit (1/mm)*10-6 Ovalization (mm)
/ Diff. Diff. Diff.
Empty Sand-fill Empty Sand-fill Empty Sand-fill
Thick (%) (%) (%)
4% 363.4 367.9 1.24 13.90 14.13 1.65 5.46 1.92 -64.84
8% 358.3 363.1 1.34 12.20 12.73 4.34 4.74 1.81 -61.81
12% 354.5 360.1 1.58 11.35 11.99 5.64 4.35 1.65 -62.07
16% 350.8 357.8 2.00 10.67 11.61 8.81 4.20 1.75 -58.33
20% 347.6 355.6 2.30 10.20 11.22 10.00 4.16 1.67 -59.86
95
Chapter 5 - Model validation
9.0E+08
8.0E+08
7.0E+08
Moment (Nmm)
6.0E+08
Empty - Imp/Thick 20%
5.0E+08
Empty - Imp/Thick 16%
4.0E+08 Empty - Imp/Thick 12%
Empty - Imp/Thick 8%
3.0E+08 Empty - Imp/Thick 4%
Sand-filled - Imp/Thick 20%
2.0E+08 Sand-filled - Imp/Thick 16%
Sand-filled - Imp/Thick 12%
1.0E+08 Sand-filled - Imp/Thick 8%
Sand-filled - Imp/Thick 4%
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 1.8E-05 2.1E-05 2.4E-05 2.7E-05 3.0E-05 3.3E-05
Curvature κ (1/mm)
Figure 142: Moment-Curvature diagrams for empty and sand-filled Tube B for different Imp/Thick percentages.
0
0.00E+00 3.00E-06 6.00E-06 9.00E-06 1.20E-05 1.50E-05 1.80E-05 2.10E-05 2.40E-05 2.70E-05 3.00E-05 3.30E-05
Curvature κ (1/mm)
Figure 143: Ovalization-Curvature diagrams for empty and sand-filled Tube B for different Imp/Thick percentages.
96
Chapter 5 - Model validation
8.0E+08 8.0E+08
Moment (Nmm)
Moment (Nmm)
6.0E+08 6.0E+08
4.0E+08
4.0E+08
Empty - Imp/Thick 8%
Empty - Imp/Thick 4% 2.0E+08
2.0E+08
Sand-filled - Imp/Thick 8%
Sand-filled - Imp/Thick 4% 0.0E+00
0.0E+00
0.0E+00 6.0E-06 1.2E-05 1.8E-05 2.4E-05 3.0E-05
0.0E+00 8.0E-06 1.6E-05 2.4E-05 3.2E-05
Curvature κ (1/mm) Curvature κ (1/mm)
8.0E+08 8.0E+08
Moment (Nmm)
Moment (Nmm)
6.0E+08 6.0E+08
4.0E+08 4.0E+08
Empty - Imp/Thick 12% Empty - Imp/Thick 16%
2.0E+08 2.0E+08
Sand-filled - Imp/Thick 12% Sand-filled - Imp/Thick 16%
0.0E+00 0.0E+00
0.0E+00 6.0E-06 1.2E-05 1.8E-05 2.4E-05 3.0E-05 0.0E+00 6.0E-06 1.2E-05 1.8E-05 2.4E-05 3.0E-05
Curvature κ (1/mm) Curvature κ (1/mm)
8.0E+08
Moment (Nmm)
6.0E+08
4.0E+08
Empty - Imp/Thick 20%
2.0E+08
Sand-filled - Imp/Thick 20%
0.0E+00
0.0E+00 6.0E-06 1.2E-05 1.8E-05 2.4E-05 3.0E-05
Curvature κ (1/mm)
Figure 144: Moment-Curvature diagrams for empty and sand-filled Tube B for different Imp/Thick percentages.
Table 36: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube B.
Imp Mmax (kNm) κcrit (1/mm)*10-6 Ovalization (mm)
/ Diff. Diff. Diff.
Empty Sand-fill Empty Sand-fill Empty Sand-fill
Thick (%) (%) (%)
4% 854.6 879.5 2.91 28.54 32.12 12.54 9.93 6.15 -38.07
8% 842.5 866.0 2.79 25.30 28.35 12.06 8.79 5.62 -36.06
12% 830.1 854.7 2.96 22.99 26.07 13.40 8.03 5.35 -33.37
16% 820.1 845.8 3.13 21.46 24.56 14.45 7.62 5.22 -31.50
20% 810.4 836.5 3.22 20.17 23.19 14.97 7.37 5.17 -29.85
97
Chapter 5 - Model validation
2.0E+08
1.0E+08
0.0E+00
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05
Curvature κ (1/mm)
Figure 145: Moment-Curvature diagrams for empty and sand-filled Tube C for different Imp/Thick percentages.
0
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05
Curvature κ (1/mm)
Figure 146: Ovalization-Curvature diagrams for empty and sand-filled Tube C for different Imp/Thick percentages.
98
Chapter 5 - Model validation
7.0E+08
Moment - Curvature - C - Imp/Thick 4% 7.0E+08
Moment - Curvature - C - Imp/Thick 8%
6.0E+08 6.0E+08
Moment (Nmm)
Moment (Nmm)
5.0E+08 5.0E+08
4.0E+08 4.0E+08
3.0E+08 3.0E+08
2.0E+08 2.0E+08
Empty - Imp/Thick 4% Empty - Imp/Thick 8%
1.0E+08 1.0E+08
Sand-filled - Imp/Thick 4% Sand-filled - Imp/Thick 8%
0.0E+00 0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm) Curvature κ (1/mm)
Moment (Nmm)
5.0E+08 5.0E+08
4.0E+08 4.0E+08
3.0E+08 3.0E+08
2.0E+08 2.0E+08
Empty - Imp/Thick 12% Empty - Imp/Thick 16%
1.0E+08 1.0E+08
Sand-filled - Imp/Thick 12% Sand-filled - Imp/Thick 16%
0.0E+00 0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05 0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm) Curvature κ (1/mm)
7.0E+08
Moment - Curvature - C - Imp/Thick 20%
6.0E+08
Moment (Nmm)
5.0E+08
4.0E+08
3.0E+08
2.0E+08
Empty - Imp/Thick 20%
1.0E+08
Sand-filled - Imp/Thick 20%
0.0E+00
0.0E+00 3.0E-06 6.0E-06 9.0E-06 1.2E-05 1.5E-05
Curvature κ (1/mm)
Figure 147: Moment-Curvature diagrams for empty and sand-filled Tube C for different Imp/Thick percentages.
Table 37: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube C.
Imp Mmax (kNm) κcrit (1/mm)*10-6 Ovalization (mm)
/ Diff. Diff. Diff.
Empty Sand-fill Empty Sand-fill Empty Sand-fill
Thick (%) (%) (%)
4% 672.5 682.3 1.46 12.04 12.82 6.48 6.03 2.49 -58.71
8% 663.8 674.7 1.64 10.63 11.54 8.56 5.19 2.19 -57.80
12% 657.4 664.3 1.05 9.94 10.22 2.82 4.95 1.94 -60.81
16% 652.0 663.0 1.69 9.47 10.13 6.97 4.86 1.98 -59.26
20% 646.0 659.5 2.09 9.00 9.91 10.11 4.82 2.06 -57.26
99
Chapter 5 - Model validation
Sand-filled - Imp/Thick 4%
9.0E+08
6.0E+08
3.0E+08
0.0E+00
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05 1.6E-05
Curvature κ (1/mm)
Figure 148: Moment-Curvature diagrams for empty and sand-filled Tube D for different Imp/Thick percentages.
Sand-filled - Imp/Thick 4%
8
0
0.0E+00 2.0E-06 4.0E-06 6.0E-06 8.0E-06 1.0E-05 1.2E-05 1.4E-05 1.6E-05
Curvature κ (1/mm)
Figure 149: Ovalization-Curvature diagrams for empty and sand-filled Tube D for different Imp/Thick percentages.
100
Chapter 5 - Model validation
1.5E+09
Moment - Curvature - D - Imp/Thick 4% Moment - Curvature - D - Imp/Thick 20%
1.5E+09
1.2E+09 1.2E+09
Moment (Nmm)
Moment (Nmm)
9.0E+08 9.0E+08
6.0E+08 6.0E+08
Empty - Imp/Thick 4%
3.0E+08 Empty - Imp/Thick 20%
3.0E+08
Sand-filled - Imp/Thick 4%
0.0E+00 Sand-filled - Imp/Thick 20%
0.0E+00
0.0E+00 4.0E-06 8.0E-06 1.2E-05 1.6E-05
0.0E+00 4.0E-06 8.0E-06 1.2E-05 1.6E-05
Curvature κ (1/mm) Curvature κ (1/mm)
Figure 150: Moment-Curvature diagrams for empty and sand-filled Tube D for different Imp/Thick percentages.
Table 38: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube D.
Imp Mmax (kNm) κcrit (1/mm)*10-6 Ovalization (mm)
/ Diff. Diff. Diff.
Empty Sand-fill Empty Sand-fill Empty Sand-fill
Thick (%) (%) (%)
4% 1272.0 1401.4 10.17 10.11 15.17 50.05 14.65 5.28 -63.96
8% 1254.5 - - 9.67 - - 14.82 - -
12% 1234.0 - - 9.22 - - 14.76 - -
16% 1217.7 - - 8.94 - - 14.67 - -
20% 1205.4 1343.4 11.45 8.77 12.17 38.77 14.77 5.18 -64.93
101
Chapter 5 - Model validation
Here, the moment-curvature and ovalization-curvature diagrams for Tube A are presented for different sand-fill
properties (Esnnd = 120MPa, φ = 330, ψ = 30) and Imperfection/Thickness percentages (4% and 20%).
3.0E+08 3.0E+08
Moment (Nmm)
Moment (Nmm)
2.0E+08 2.0E+08
1.0E+08 1.0E+08
Sand-filled - Imp/Thick 4% Sand-filled - Imp/Thick 20%
Sand-filled - Imp/Thick 4% - Esand = 120MPa Sand-filled - Imp/Thick 20% - Esand=120MPa
0.0E+00 0.0E+00
0.0E+00 6.0E-06 1.2E-05 1.8E-05 0.0E+00 6.0E-06 1.2E-05 1.8E-05
Curvature κ (1/mm) Curvature κ (1/mm)
Figure 151: Moment-Curvature diagrams for sand-filled Tube A for different Imp/Thick percentages and sand-fill properties.
1.00
0.00
0.0E+00 4.0E-06 8.0E-06 1.2E-05 1.6E-05
Curvature κ (1/mm)
Figure 152: Ovalization-Curvature diagrams for sand-filled Tube A for different Imp/Thick percentages and sand-fill properties.
Table 39: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube A.
Imp Mmax (kNm) κcrit (1/mm)*10-6 Ovalization (mm)
/ Esand = Esand = Diff. Esand = Esand = Diff. Esand = Esand = Diff.
Thick 30 MPa 120 MPa (%) 30 MPa 120 MPa (%) 30 MPa 120 MPa (%)
4% 367.9 372.4 1.22 14.13 15.41 9.06 1.92 0.61 -68.23
20% 355.6 360.9 1.49 11.22 12.20 8.73 1.67 0.46 -72.46
102
Chapter 5 - Model validation
Here, the moment-curvature and ovalization-curvature diagrams for Tube D are presented for different sand-fill
properties (Esnnd = 120MPa, φ = 330, ψ = 30) and Imperfection/Thickness percentages (4% and 20%).
1.2E+09 1.2E+09
Moment (Nmm)
Moment (Nmm)
9.0E+08 9.0E+08
6.0E+08 6.0E+08
Sand-filled - Imp/Thick 4% Sand-filled - Imp/Thick 20%
3.0E+08 3.0E+08
Sand-filled - Imp/Thick 4% - Es=120MPa Sand-filed - Imp/Thick 20% - Es=120MPa
0.0E+00 0.0E+00
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05 0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature κ (1/mm) Curvature κ (1/mm)
Figure 153: Moment-Curvature diagrams for sand-filled Tube D for different Imp/Thick percentages and sand-fill properties.
3.00
2.00
1.00
0.00
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature κ (1/mm)
Figure 154: Ovalization-Curvature diagrams for sand-filled Tube D for different Imp/Thick percentages and sand-fill properties.
Table 40: Comparison of Mmax, κcrit and Ovalization results from ABAQUS for Empty and Sand-filled Tube D.
Imp Mmax (kNm) κcrit (1/mm)*10-6 Ovalization (mm)
/ Esand = Esand = Diff. Esand = Esand = Diff. Esand = Esand = Diff.
Thick 30 MPa 120 MPa (%) 30 MPa 120 MPa (%) 30 MPa 120 MPa (%)
4% 1401.4 1447.8 3.31 15.17 18.17 19.78 5.28 1.29 -75.57
20% 1343.4 1398.8 4.12 12.17 14.72 20.95 5.18 0.97 -81.27
103
Chapter 5 - Model validation
Initially it can be clearly seen that moment-curvature curves of the full scale model of Tube A are perfectly fitted
to the one of the slice model both for empty and for sand-filled tubes. Also the comparison of the full scale
model curves gives a perfect fit to the analytical formula curve which takes into account the ovalization and the
residual stresses for the empty tube and only the residual stresses for the sand-filled tube.
The advantage of the full scale model is that when local buckling happens then the moment drops. This is
something that neither the current analytical formulas nor the slice model can simulate. It also can be seen that
for different imperfection value input, then the drop happens in a different stage.
Comparing the full scale curves of the empty tube to the test result failures (Figure 123), it can be seen that there
is a difference in the maximum moments. Empty Tube A seems to have a certain difference in the steel material
properties that are used as input for the analyses and the real one. The comparison of the failure between the test
results to the numerical results and to the analytical formula curve shows this claim clearly and it also indicates
that the yield stress value given as data is smaller than the real yield stress. As it was shown before, in the slice
model curve for Tube A (Figure 70), increasing the yield stress value by 15%, a good fit of the numerical curve
to the test failure points is achieved.
Comparing the full scale curves of the sand-filled tube to the test result failures (Figure 131), it is derived that
they have a good fit regarding the maximum moments, except for one value (tube A5).
The comparison of the ovalization-curvature curves show that the full scale model curves are close to the slice
model curve, but after the local buckle they deviate from it. But even in the case of the full scale model, as in the
slice model, there is a significant difference between the numerical values and the measured ones.
Finally, comparing the empty and sand-filled curves and the Mmax, κcrit and Ovalization values derived from
numerical analyses (Figure 139, Figure 140, Figure 141 and Table 35), some useful conclusions can be extracted.
It can be noted that the sand-fill’s contribution to the moment capacity is really small (1.2~2.3%) but the
contribution is higher as the imperfection/thickness percentage increases. For the critical curvature values, it can
be observed that the sand-fill’s contribution is higher compared to its contribution to the maximum moment
(1.6~10%) and again it makes bigger impact at the higher imperfection/thickness percentages. Lastly, the
ovalization reduction of the steel tube due to the presence of the sand-fill is significant (60~65%). This is the
reason why the previous two parameters have favourable changes in their values. The sand-fill prevents the
excessive ovalization of the steel tube and as a result mainly the critical curvature and afterwards the maximum
moment depict an improvement in their values. In the ovalization case, the main contribution seems to be at the
lower percentages of imperfection/thickness ratios.
TUBE B
As in the case of Tube A, Tube B full scale moment-curvature diagrams are fitted perfectly to the slice model
curve both for empty and for sand-filled tubes. The comparison of the full scale model curves gives a perfect fit
to the analytical formula curve which takes into account the ovalization and the residual stresses for the empty
tube and only the residual stresses for the sand-filled tube, until the yield area. After the yield, the curves deviate
from each other but not significantly. This happens also due to the stress-strain curve used as input for the
numerical analysis, since the analytical formula considers bilinear stress-strain behaviour for the steel material,
whereas Tube B steel stress-strain diagram input for the numerical model is not bilinear.
Comparing the full scale curves of the empty tube to the test result failures (Figure 125), it can be seen that there
is a big difference at the failure points. The failure points of the tests are happening for curvature values of
around 3.3*10-5 and 4.5*10-5, whereas the closest numerical one is happening for curvature value less than 3.0E-
05. This difference is a very significant and therefore since no additional data are provided, not much can be said
over the comparison of these. The only comparison that can be done is between the experimental results and the
slice model moment-curvature curve, which has been implemented in a different chapter (chapter 5.1)
104
Chapter 5 - Model validation
Comparing the full scale curves of the sand-filled tube to the test result failures (Figure 133), the same
conclusion as for the empty tube is derived.
The comparison of the ovalization-curvature diagrams show that the full scale model curves are a close match to
the slice model curve, until before the local buckle. But even in the case of the full scale model, as in the slice
model, there is a significant difference between the numerical values and the measured ones.
Finally, comparing the empty and sand-filled curves and the Mmax, κcrit and Ovalization values derived from
numerical analyses (Figure 142, Figure 143, Figure 144 and Table 36), some useful conclusions can be extracted.
It can be noted that the sand-fill’s contribution to the moment capacity is small (2.9~3.2%) but the contribution
is higher as the imperfection/thickness percentage increases. For the critical curvature values, it can be observed
that the sand-fill’s contribution is higher compared to its contribution to the maximum moment (12.5~15%) and
again it makes bigger impact at the higher imperfection/thickness percentages. Lastly, the ovalization reduction
of the steel tube due to the presence of the sand-fill is significant (30~38%). The sand-fill prevents the excessive
ovalization of the steel tube and as a result mainly the critical curvature and afterwards the maximum moment
depict an improvement in their values. In the ovalization case, the main contribution seems to be at the lower
percentages of imperfection/thickness ratios.
TUBE C
As in the previous two tubes, Tube C full scale moment-curvature diagrams are fitted perfectly to the slice model
curve both for empty and for sand-filled tubes. The comparison of the full scale model curves gives a perfect fit
to the analytical formula curve which takes into account the ovalization and the residual stresses for the empty
tube and only the residual stresses for the sand-filled tube, until the yield area.
Comparing the full scale curves of the empty tube to the test result failures (Figure 127), it can be seen that there
is a perfect fit for one of the test failures to the Imp/Thick 8% curve. The other two failure points of the tests
are happening for Imp/Thick percentages close to 16% and 20% (1.0mm and 1.26mm imperfection values
respectively), but their moment values are a little smaller than the ones derived by the full scale models.
Therefore, it can be said that there is a perfect match for one of the failures and close match for the other two.
Comparing the full scale curves of the sand-filled tube to the test result failures (Figure 135), it can be seen that
there is a difference in the failure points. The failure points of the numerical model are happening in lower
curvature values than the experimental tests. This might be happening due to the material inconsistency of the
sand-fill.
The comparison of the ovalization-curvature diagrams show that the full scale model curves are a close match to
the slice model curve, until before the local buckle. But even in the case of the full scale model, as in the slice
model, there is a significant difference between the numerical values and the measured ones.
Finally, comparing the empty and sand-filled curves and the Mmax, κcrit and Ovalization values derived from
numerical analyses (Figure 145, Figure 146, Figure 147 and Table 37), some useful conclusions can be extracted.
It can be noted that the sand-fill’s contribution to the moment capacity is small (1.5~2.1%) but the contribution
is higher as the imperfection/thickness percentage increases, except for the case of Imp/Thick 12%. For the
critical curvature values, it can be observed that the sand-fill’s contribution is higher compared to its contribution
to the maximum moment (6.5~10.1%) and again it makes bigger impact at the higher imperfection/thickness
percentages (again except for the case of Imp/Thick 12%). Lastly, the ovalization reduction of the steel tube due
to the presence of the sand-fill is significant (57.3~60.8%). The sand-fill prevents the excessive ovalization of the
steel tube and as a result mainly the critical curvature and afterwards the maximum moment depict an
improvement in their values. In the ovalization case, the main contribution seems to be at the lower percentages
of imperfection/thickness ratios.
105
Chapter 5 - Model validation
TUBE D
As in the previous tube cases, Tube D full scale moment-curvature curves are fitted perfectly to the slice model
curve both for empty and for sand-filled tubes. The comparison of the full scale model curves gives a perfect fit
to the analytical formula curve which takes into account the ovalization and the residual stresses for the empty
tube and only the residual stresses for the sand-filled tube, until the yield area.
Comparing the full scale curves of the empty tubes to the test result failures (Figure 129), it can be seen that
there is a difference between the analytical and numerical curves to the test failure points. There seems to be a
difference in the steel material properties even between the test results. The analytical and numerical curves are
located between the test failure points with a scatter. Also the test critical curvatures values depict an important
difference to the ones from the numerical model. The factor values from Table 26 give curves that fail after the
test failures. This could be due to the lack of measured imperfection data and the details of the welding. Maybe
the welding of tube played a more significant role in this case.
Comparing the full scale curves of the sand-filled tube to the test result failures (Figure 137), it can be seen that
there is a difference in the failure points. Two of the failure points of the experiments are happening in lower
curvature values than the numerical ones. Also two of the maximum moments of the experimental tests have
significant difference both compared to the numerical ones and to each other. This probably is happening due to
the material inconsistencies of the steel tube.
The comparison of the ovalization-curvature diagrams show that the full scale model curves are a close match to
the slice model curve, until before the local buckle. For the empty tubes the experimental ovalization values at
failure are close to the diagrams of the numerical models, but the numerical ovalization values at failure are
higher and also happening at higher curvature values. This phenomenon, though, is not noticed at the sand-filled
tubes in which the numerical ovalization values are significantly higher than the measured ones.
Finally, comparing the empty and sand-filled curves and the Mmax, κcrit and Ovalization values derived from
numerical analyses (Figure 148, Figure 149, Figure 150 and Table 38), some useful conclusions can be extracted.
It can be noted that the sand-fill’s contribution to the moment capacity is rather high if compared to the other
tube types (10.2~11.5%) and the contribution gets higher as the imperfection/thickness percentage increases.
For the critical curvature values, it can be observed that the sand-fill’s contribution is higher compared to its
contribution to the maximum moment (38~50%) but the biggest impact is not for the low Imp/Thick ration as
it is noted for the previous tube types. Lastly, the ovalization reduction of the steel tube due to the presence of
the sand-fill is significant (~64%). The sand-fill prevents the excessive ovalization of the steel tube and as a result
mainly the critical curvature and afterwards the maximum moment depict an improvement in their values. In the
ovalization case, the main contribution seems to be at the lower percentages of imperfection/thickness ratios.
General
Figure 151-Figure 154 and Table 39-Table 40 show the difference in the maximum moment, the critical
curvature and the critical ovalization for Tube A and Tube D, when different sand-fill properties are inputted
(Esnnd = 120MPa, φ = 330, ψ = 30). The maximum moment increases by 3%~4%, the critical curvature increases
by 20% and the critical ovalization reduces by 75%~80%. Also, the ovalization-curvature diagrams are located
close to the experimental failure points. Therefore, by improving the sand-fill properties the overall bending
moment behaviour of the filled steel tube is enhanced.
In Figure 155, Figure 156 and Figure 157, the maximum moment, the critical curvature and the critical
ovalization numerical values, in respect with D/t ratio, are presented, normalized by dividing them by the plastic
moment, the κi = t/(D-t)2 quantity and the tube diameter respectively.
It can be seen that the normalized moment values are declining as the D/t ratio increases for tube types with the
same or close yield stress values (S235 and X60, X65). This is normal since the slenderer the cross-section, the
lower the maximum moment in respect with the plastic moment. Also, it is noted that the sand-filled tubes reach
a higher Mmax/Mpl ratio compared to the empty ones due to the presence of the sand infill.
106
Chapter 5 - Model validation
For the κcrit/κi, the conclusion is that the D/t ratio has a very small effect on the critical curvature of cross-
sections with the same (S235) or close (X60, X65) yield stress. The main effect on the critical curvature comes
from the difference on the yield stress since cross-sections with lower yield stress seem to have lower κcrit/κi
ratios.
Finally, the ovalization/diameter to the D/t figure shows that the the D/t ratio has an effect on the ovalization
of cross-section but not a significant one. Again, it can be seen that the difference on the yield stress has a higher
impact on the ovalization/diameter values.
1.05
1.00
y = -0.0009x + 1.0173
Mmax / Mpl
0.95
S235
X60,X65
Tube A - Empty
0.90 Tube A - Sand Filled
Tube B - Empty
Tube B - Sand-filled
0.85 Tube C - Empty
Tube C - Sand-filled y = -0.0026x + 1.1556
Tube D - Empty
Tube D - Sand-filled
0.80
70 80 90 100 110 120
D/t
Figure 155: Mmax/Mpl - D/t for all Tube types
S235
1.3 X60, X65
Tube A - Empty
Tube A - Sand Filled
Tube B - Empty
1.1 Tube B - Sand-filled
Tube C - Empty
Tube C - Sand-filled
y = -0.0004x + 0.919 Tube D - Empty
κcrit / κi
0.7
y = -0.0009x + 0.6939
0.5
70 80 90 100 110 120
D/t
Figure 156: crit. Curvature - D/t for all Tube types
2.1
Ovalization/Diameter (x10^-2)
1.5
S235
1.2 X60, X65
Tube A - Empty
Tube A - Sand Filled
0.9 Tube B - Empty
Tube B - Sand-filled
Tube C - Empty
0.6 Tube C - Sand-filled y = 0.0051x + 0.1002
Tube D - Empty
Tube D - Sand-filled
0.3
70 80 90 100 110 120
D/t
Figure 157: Ovalization/Diameter - D/t for all Tube types
107
Chapter 6 - Parametric study of the sand-fill properties with the slice model
The sand parameters which are being investigated are the Young’s modulus Esand, the friction angle φ, the
dilation angle ψ, Poisson’s ratio ν and the cohesion c. These parameters are investigated for different geometrical
and material properties of the steel tube such as for different thickness and yield stress. By this, the D/(tε2) ratio
changes which is desirable because it covers cross-sections with different slenderness. The parameters of the
steel tube are selected to represent a wide range of the D/(tε2) ratio. The steel tube which is used is Tube C.
Table 41: Examined properties and their values for the parametrical study
D t fy
Tube
(mm) (mm)
D/t
(MPa)
ε2 D/tε2 Esand (MPa) φ (degrees) ψ (degrees)
1 610 6.3 97 312 0.75 128.6
2 610 4.5 136 312 0.75 180.0
30 100 300 1000 30 35 40 0 5 10
3 610 3 203 312 0.75 270.0
4 610 6.3 97 488 0.48 201.1
By changing these parameters, the effect on the moment-curvature and the ovalization-curvature diagrams is
examined. The model used for this parametric study is the slice model, since it gives results very fast compared
to the full scale model, which takes at least one day of analysis run, with the current means, to give results.
For every parameter the values of the normalized moment and the normalized ovalization in curvature 10-5 are
compared, since the failure of Tube 1/C happens around that curvature value. The moment and the ovalization
are normalized to the plastic moment and the radius of the tube respectively.
108
Chapter 6 - Parametric study of the sand-fill properties with the slice model
5.0E+08
E=30MPa
4.0E+08 E=100MPa
3.0E+08 E=300MPa
E=1000MPa
2.0E+08
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 158: Moment-Curvature diagram of Tube 1 for different Esand
C - Test
2.5
2.0
1.5
1.0
0.5
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 159: Ovalization-Curvature diagram of Tube 1 for different Esand
5.0E+08
Moment (Nmm)
4.0E+08
E=30MPa
3.0E+08 E=100MPa
E=300MPa
2.0E+08 E=1000MPa
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 160: Moment-Curvature diagram of Tube 2 for different Esand
109
Chapter 6 - Parametric study of the sand-fill properties with the slice model
2.5E+08 E=30MPa
2.0E+08 E=100MPa
E=300MPa
1.5E+08 E=1000MPa
1.0E+08
5.0E+07
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 162: Moment-Curvature diagram of Tube 3 for different Esand
E=1000MPa
2.5
2.0
1.5
1.0
0.5
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 163: Ovalization-Curvature diagram of Tube 3 for different Esand
110
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.0E+09
Moment (Nmm)
8.0E+08
E=30MPa
6.0E+08 E=100MPa
E=300MPa
4.0E+08 E=1000MPa
2.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 164: Moment-Curvature diagram of Tube 4 for different Esand
4.0 E=1000MPa
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 165: Ovalization-Curvature diagram of Tube 4 for different Esand
1.00 Tube 1
Tube 2
Tube 3 y = 0.0003x + 0.9671
0.98
Tube 4
y = 0.0002x + 0.9632
0.94
0.92
y = 0.0003x + 0.9231
0.90
0 5 10 15 20 25 30 35
Normalized Esand E/30
Figure 166: Effect of Esand on the Moment of the Tubes
111
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.60
Tube 1
Tube 2
0.80
y = 1.0478x-0.673
0.40
y = 1.0768x-0.878
y = 0.879x-1.089
0.00
0 5 10 15 20 25 30 35
Normalized Esand E/30
Figure 167: Effect of Esand on the Ovalization of the Tubes
In Figure 166 and Figure 167 the normalized moment and normalized ovalization trendlines of each Tube are
shown. It is clear that the moment of each tube is not seriously affected by the Esand change (e.g. 0.961 to 0.970
for Tube 1 and 0.919 to 0.932 for Tube 4) but it is affected by the change in the D/(tε2) ratio and even more by
the material change. For the tubes 2, 3 as the slenderness ratio increases, the normalized moment also increases,
compared to Tube 1 with the difference being really small though. For Tube 4, the slenderness ratio also
increases but the normalized moment decreases compared to Tube 1. Therefore, it can be seen that the
slenderness change due to the material properties of the steel has a more significant effect on reaching the plastic
moment value.
On the other hand, the ovalization is affected significantly by the change of Esand and of slenderness ratio. For
the tubes 2, 3 as the slenderness ratio increases, the ovalization decreases, compared to Tube 1. For the Tube 4
the slenderness ratio also increases but the ovalization increases compared to Tube 1. As in the case of the
moment, this happens due to the increase of the yield stress of the steel tube.
The ovalization follows a power trendline whereas the moment follows a linear trendline. The difference
between the ovalization of the different tubes gets smaller as the normalized Esand value gets bigger.
Finally, it can be seen that the increase of slenderness ratio gives different results for the moment and the
ovalization for changes of different parameters of the slenderness.
112
Chapter 6 - Parametric study of the sand-fill properties with the slice model
5.0E+08
φ=30.0
4.0E+08 φ=32.5
3.0E+08 φ=35.0
φ=40.0
2.0E+08
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 168: Moment-Curvature diagram of Tube 1 for different friction angles
φ=40.0
2.5
2.0
1.5
1.0
0.5
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 169: Ovalization-Curvature diagram of Tube 1 for different friction angles
5.0E+08
Moment (Nmm)
4.0E+08
φ=30.0
3.0E+08
φ=35.0
2.0E+08 φ=40.0
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 170: Moment-Curvature diagram of Tube 2 for different friction angles
113
Chapter 6 - Parametric study of the sand-fill properties with the slice model
2.5E+08
φ=30.0
2.0E+08
φ=35.0
1.5E+08
φ=40.0
1.0E+08
5.0E+07
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 172: Moment-Curvature diagram of Tube 3 for different friction angles
2.5
2.0
1.5
1.0
0.5
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 173: Ovalization-Curvature diagram of Tube 3 for different friction angles
114
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.0E+09
Moment (Nmm)
8.0E+08
6.0E+08
4.0E+08
φ=30.0
2.0E+08 φ=35.0
φ=40.0
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 174: Moment-Curvature diagram of Tube 4 for different friction angles
4.0
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 175: Ovalization-Curvature diagram of Tube 4 for different friction angles
1.00
Tube 1
Tube 2
0.98 Tube 3
Tube 4 y = 0.0003x + 0.9549
0.96
M / Mpl
0.90
30 32 34 36 38 40
Friction angle φ (degrees)
Figure 176: Effect of friction angle on the Moment of the Tubes
115
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.80
y = -0.0133x + 1.9015
y = -0.0142x + 1.4334
1.00
y = -0.0234x + 1.6941
0.60
Tube 1
Tube 2
Tube 3 y = -0.0343x + 1.8207
Tube 4
0.20
30 32 34 36 38 40
Friction angle φ (degrees)
Figure 177: Effect of friction angle on the Ovalization of the Tubes
In Figure 176 and Figure 177 the normalized moment and normalized ovalization trendlines of each Tube are
shown. It is clear that the moment of each tube is not seriously affected by the friction angle change (e.g. 0.961
to 0.963 for Tube 1 and 0.919 to 0.922 for Tube 4) but it is affected by the change in the D/(tε2) ratio and even
more by the material change. For the tubes 2, 3 as the slenderness ratio increases, the normalized moment also
increases, compared to Tube 1 with the difference being really small though. For Tube 4, the slenderness ratio
also increases but the normalized moment decreases compared to Tube 1. Therefore, it can be seen that the
slenderness change due to the material properties of the steel has a more significant effect on reaching the plastic
moment value.
On the other hand, the ovalization is affected notably by the change of the friction angle, φ, and of slenderness
ratio. For the tubes 2, 3 as the slenderness ratio increases, the ovalization decreases, compared to Tube 1. For the
Tube 4 the slenderness ratio also increases but the ovalization increases compared to Tube 1. As in the case of
the moment, this happens due to the increase of the yield stress of the steel tube.
Both the ovalization and the moment trendlines follow a linear path. The moment follows a slightly increasing
linear path and the ovalization a constantly decreasing path. The decrease in the ovalization due to the increase
of the friction angle is not as steep as the decrease due the increase of the Young’s modulus.
Finally, it can be seen that the increase of slenderness ratio gives different results for the moment and the
ovalization for changes of different parameters of the slenderness.
116
Chapter 6 - Parametric study of the sand-fill properties with the slice model
5.0E+08
ψ=0.0
4.0E+08 ψ=2.5
3.0E+08 ψ=5.0
ψ=10.0
2.0E+08
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 178: Moment-Curvature diagram of Tube 1 for different dilation angles
ψ=10.0
5.0
4.0
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 179: Ovalization-Curvature diagram of Tube 1 for different dilation angles
5.0E+08
Moment (Nmm)
4.0E+08
ψ=0.0
3.0E+08
ψ=5.0
2.0E+08 ψ=10.0
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 180: Moment-Curvature diagram of Tube 2 for different dilation angles
117
Chapter 6 - Parametric study of the sand-fill properties with the slice model
2.5E+08
ψ=0.0
2.0E+08
ψ=5.0
1.5E+08
ψ=10.0
1.0E+08
5.0E+07
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 182: Moment-Curvature diagram of Tube 3 for different dilation angles
4.0
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 183: Ovalization-Curvature diagram of Tube 3 for different dilation angles
118
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.0E+09
Moment (Nmm)
8.0E+08
6.0E+08
4.0E+08
ψ=0.0
2.0E+08 ψ=5.0
ψ=10.0
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 184: Moment-Curvature diagram of Tube 4 for different dilation angles
6.0
5.0
4.0
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 185: Ovalization-Curvature diagram of Tube 4 for different dilation angles
1.00
Tube 1
Tube 2
Tube 3
0.98 Tube 4
y = 0.0006x + 0.9632
0.96
M / Mpl
y = 0.001x + 0.9151
0.92
0.90
0 2 4 6 8 10
Dilation angle ψ
Figure 186: Effect of dilation angle on the Moment of the Tubes
119
Chapter 6 - Parametric study of the sand-fill properties with the slice model
2.20
Tube 1
Tube 2
Tube 3
1.40
y = -0.234ln(x) + 1.0875
y = -0.31ln(x) + 1.126
1.00
0.60
y = -0.21ln(x) + 0.7938
0.20
0 2 4 6 8 10
Dilation angle ψ
Figure 187: Effect of dilation angle on the Ovalization of the Tubes
In Figure 186 and Figure 187 the normalized moment and normalized ovalization trendlines of each Tube are
shown. It is clear that the moment of each tube is not affected seriously by the dilation angle change (e.g. 0.958
to 0.966 for Tube 2 and 0.963 to 0.968 for Tube 3) but it is affected by the change in the slenderness ratio,
D/(tε2), and even more by the material change. For the tubes 2, 3 as the slenderness ratio increases, the
normalized moment also increases, compared to Tube 1. For the Tube 4 the slenderness ratio also increases but
the normalized moment decreases compared to Tube 1. Therefore, it can be seen that the slenderness change
due to the material properties of the steel has a more significant effect on reaching the plastic moment value.
On the other hand, the ovalization is affected remarkably by the change of the dilation angle, ψ, and of the
slenderness ratio. Especially the transition of the dilation angle from 0 to 5 degrees has a bigger effect on the
ovalization compared to the transition from 5 to 10 degrees. For the tubes 2, 3 as the slenderness ratio increases,
the ovalization decreases, compared to Tube 1. For the Tube 4 the slenderness ratio also increases but the
ovalization increases compared to Tube 1.
The normalized ovalization trendlines follow a logarithmic path. The normalized moment follows a linear,
slightly increasing path (almost no difference, e.g. from 0.958 to 0.964 for Tube 1) and the ovalization a
decreasing path.
Finally, it can be seen that the increase of slenderness ratio gives different results for the moment and the
ovalization for changes of different parameters of the slenderness.
120
Chapter 6 - Parametric study of the sand-fill properties with the slice model
5.0E+08
c=0.0005MPa
4.0E+08 c=0.001MPa
3.0E+08 c=0.005MPa
2.0E+08 c=0.010MPa
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 188: Moment-Curvature diagram of Tube 1 for different cohesion values
c=0.010MPa
C - test
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 189: Ovalization-Curvature diagram of Tube 1 for different cohesion values
5.0E+08
Moment (Nmm)
4.0E+08
c=0.0005Mpa
3.0E+08
c=0.001MPa
2.0E+08 c=0.005MPa
c=0.010MPa
1.0E+08
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 190: Moment-Curvature diagram of Tube 2 for different cohesion values
121
Chapter 6 - Parametric study of the sand-fill properties with the slice model
2.5E+08
c=0.0005MPa
2.0E+08
c=0.001MPa
1.5E+08 c=0.005MPa
1.0E+08 c=0.010MPa
5.0E+07
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 192: Moment-Curvature diagram of Tube 3 for different cohesion values
c=0.010MPa
2.5
2.0
1.5
1.0
0.5
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 193: Ovalization-Curvature diagram of Tube 3 for different cohesion values
122
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.0E+09
Moment (Nmm)
8.0E+08
6.0E+08
4.0E+08
c=0.0005MPa
c=0.001MPa
2.0E+08
c=0.005MPa
c=0.010MPa
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 194: Moment-Curvature diagram of Tube 4 for different cohesion values
4.0
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 195: Ovalization-Curvature diagram of Tube 4 for different cohesion values
1.00
Tube 1
Tube 2
Tube 3
0.98 Tube 4 y = 0.0006x + 0.9651
0.96
M / Mpl
0.92
y = 0.0007x + 0.9193
0.90
0 2 4 6 8 10
Cohesion (kPa)
Figure 196: Effect of cohesion on the Moment of the Tubes
123
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.60
Tube 1
Tube 2
1.40
0.80
0.60 y = 0.832x-0.435
0.40 y = 0.7819x-0.498
y= 0.6156x-0.503
0.20
0 2 4 6 8 10
Cohesion (kPa)
Figure 197: Effect of cohesion on the Ovalization of the Tubes
In Figure 196 and Figure 197 the normalized moment and normalized ovalization trendlines of each Tube are
shown. It is clear that the moment of each tube is not affected significantly by the cohesion change but it is
affected by the change in the slenderness ratio, D/(tε2), and even more by the material change. For the tubes 2, 3
as the slenderness ratio increases, the normalized moment also increases, compared to Tube 1. For the Tube 4
the slenderness ratio also increases but the normalized moment decreases compared to Tube 1. Therefore, it can
be seen that the slenderness change due to the material properties of the steel has a more significant effect on
reaching the plastic moment value.
On the other hand, the ovalization is affected remarkably by the change of the cohesion, c, and of the
slenderness ratio. For the tubes 2, 3 as the slenderness ratio increases, the ovalization decreases, compared to
Tube 1. For the Tube 4 the slenderness ratio also increases but the ovalization increases compared to Tube 1. As
in the case of the moment, this happens due to the increase of the yield stress of the steel tube.
The normalized ovalization values follow a power trendline. The normalized moment follows a linear, slightly
increasing path (almost no difference, e.g. from 0.961 to 0.966 for Tube 1) and the ovalization a decreasing path.
Finally, it can be seen that the increase of slenderness ratio gives different results for the moment and the
ovalization for changes of different parameters of the slenderness.
124
Chapter 6 - Parametric study of the sand-fill properties with the slice model
5.0E+08
4.0E+08
3.0E+08
2.0E+08
ν=0.2
1.0E+08 ν=0.3
ν=0.4
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 198: Moment-Curvature diagram of Tube 1 for different Poisson’s ratio values
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 199: Ovalization-Curvature diagram of Tube 1 for different Poisson’s ratio values
5.0E+08
Moment (Nmm)
4.0E+08
3.0E+08
2.0E+08
ν=0.2
1.0E+08 ν=0.3
0.0E+00 ν=0.4
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 200: Moment-Curvature diagram of Tube 2 for different Poisson’s ratio values
125
Chapter 6 - Parametric study of the sand-fill properties with the slice model
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 201: Ovalization-Curvature diagram of Tube 2 for different Poisson’s ratio values
2.5E+08
2.0E+08
1.5E+08
1.0E+08
ν=0.2
5.0E+07 ν=0.3
ν=0.4
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 202: Moment-Curvature diagram of Tube 3 for different Poisson’s ratio values
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 203: Ovalization-Curvature diagram of Tube 3 for different Poisson’s ratio values
126
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.0E+09
Moment (Nmm)
8.0E+08
6.0E+08
4.0E+08
ν=0.2
2.0E+08 ν=0.3
ν=0.4
0.0E+00
0.00E+00 5.00E-06 1.00E-05 1.50E-05 2.00E-05
Curvature (1/mm)
Figure 204: Moment-Curvature diagram of Tube 4 for different Poisson’s ratio values
5.0
4.0
3.0
2.0
1.0
0.0
0.0E+00 5.0E-06 1.0E-05 1.5E-05 2.0E-05
Curvature (1/mm)
Figure 205: Ovalization-Curvature diagram of Tube 4 for different Poisson’s ratio values
1.00 Tube 1
Tube 2
Tube 3
0.98 Tube 4 y = 0.014x + 0.9612
0.96
M / Mpl
0.92
y = 0.0274x + 0.9119
0.90
0.2 0.3 0.4
Cohesion (kPa)
Figure 206: Effect of Poisson’s ratio on the Moment of the Tubes
127
Chapter 6 - Parametric study of the sand-fill properties with the slice model
1.80
Tube 1
y = -3.4362x + 2.4025 Tube 2
1.60
0.80
0.60
y = -2.8282x + 1.6049
0.40
0.2 0.3 0.4
Cohesion (kPa)
Figure 207: Effect of Poisson’s ratio on the Ovalization of the Tubes
In Figure 206 and Figure 207 the normalized moment and the normalized ovalization trendlines of each Tube
are shown. It is clear that the moment of each tube is not affected by the Poisson’s ratio change (e.g. 0.960 to
0.963 for Tube 1 and 0.917to 0.923 for Tube 4) but it is affected by the change in the slenderness ratio, D/(tε2),
and even more by the material change. For the tubes 2, 3 as the slenderness ratio increases, the normalized
moment also increases, compared to Tube 1. For the Tube 4 the slenderness ratio also increases but the
normalized moment decreases compared to Tube 1. Therefore, it can be seen that the slenderness change due to
the material properties of the steel has a more significant effect on reaching the plastic moment value.
On the other hand, the ovalization is affected notably by the change of the Poisson’s ratio, ν, and of the
slenderness ratio. For the tubes 2, 3 as the slenderness ratio increases, the ovalization decreases, compared to
Tube 1. For the Tube 4 the slenderness ratio also increases but the ovalization increases compared to Tube 1.
The normalized ovalization trendlines follow a path which is linear. The normalized moment follows a linear,
slightly increasing path (almost no difference, e.g. from 0.964 to 0.967 for Tube 3) and the ovalization a
decreasing path.
Finally, it can be seen that the increase of slenderness ratio gives different results for the moment and the
ovalization for changes of different parameters of the slenderness.
128
Chapter 6 - Parametric study of the sand-fill properties with the slice model
6.11 Examining the effect of each parameter on the moment and the
ovalization
The effect of each parameter on the moment and the ovalization is depicted and examined here. In Table 42 the
changes in the moment and the ovalization values in percentages are given for each parametric value and for
every tube type. The values of the moment and the curvature used as comparison basis, are the values derived by
the analyses performed with the lowest value of each parameter.
Table 42: Moment and Ovalization changes in percentage for different parameter values
Tube 1 Tube 2 Tube 3 Tube 4
Moment Ovaliz. Moment Ovaliz. Moment Ovaliz. Moment Ovaliz.
(%) (%) (%) (%) (%) (%) (%) (%)
1.0 0 0 0 0 0 0 0 0
Esand
3.3 0.36 -49.35 0.39 -56.30 0.35 -59.54 0.69 -48.93
/
10.0 0.60 -74.47 0.69 -81.44 0.66 -88.13 1.11 -77.65
30
33.3 0.93 -90.70 0.99 -95.55 1.03 -97.78 1.38 -93.94
30 0 0 0 0 0 0 0 0
φ
35 0.15 -5.59 0.08 -8.79 0.18 -23.20 0.19 -5.34
(0)
40 0.26 -14.12 0.22 -23.88 0.34 -43.16 0.32 -8.86
0 0 0 0 0 0 0 0 0
ψ
5 0.49 -56.10 0.70 -65.87 0.44 -62.85 0.86 -48.94
(0)
10 0.62 -69.41 0.85 -77.64 0.57 -77.02 1.14 -64.67
0.5 0 0 0 0 0 0 0 0
c 1 0.03 -2.96 0.00 -3.93 0.03 -10.09 0.16 -13.37
(kPa) 5 0.36 -59.78 0.41 -67.94 0.45 -70.98 0.30 -24.26
10 0.47 -70.01 0.51 -73.65 0.55 -73.71 0.76 -63.09
0.2 0 0 0 0 0 0 0 0
ν 0.3 0.13 -18.87 0.13 -24.63 0.13 -31.75 0.39 -26.87
0.4 0.26 -41.00 0.33 -49.10 0.29 -53.52 0.60 -39.09
It can be seen that the moment percentage is increasing as the parametric value increases. The increase though, is
almost with no serious effect on the outcome, since the highest percentile increase is around 1% for the highest
parametric value. Therefore, as it has been seen also previously in this chapter, the moment is barely affected by
the change of the parameter values.
On the other hand, the ovalization is also favourably affected by the increase of the parametric values, but in a
higher grade than the moment. The decrease of the ovalization values is significant when increasing the value of
each parameter and from Table 42, it can be seen that the highest impact on the ovalization comes from the
Young’s modulus (90.70% reduction for Tube 1) and the lowest from the friction angle, φ (14.12% reduction for
Tube 1). Taking into consideration though, property values that can be combined for a sand, such as
Esand/30=3.3, φ=350, ψ=50, c=1kPa and ν=0.3, then the reduction percentages for each parameter are 49.35%,
5.59%, 56.10%, 2.96% and 18.87% respectively. In this case the dilation angle and then the Young’s modulus
have the most significant impact on the ovalization reduction.
129
Chapter 7 - Conclusions and Recommendations
7.1 Conclusions
After the completion of this master thesis research study, the conclusions derived are:
130
Chapter 7 - Conclusions and Recommendations
- It can be concluded that the tube types A and C numerical diagrams were closer to the analytical ones
because of their material input in the analysis. Since the analytical formulas account for an elastic-perfectly
plastic material behaviour, these two tubes are closer to the analytical because of the plateau right after
yielding in the stress-strain diagram.
- Comparing the empty tubes to the filled ones, it is clear that the sand-fill provides radial support to the steel
tube which helps for the tube to have a better post-yielding behaviour. In the numerical and analytical
models, it can be seen that the sand-fill contributes to both the post-yielding behaviour and the bending
moment capacity since the filled tubed exhibit higher moment values.
- The comparison between the two numerical models, slice model and full scale model, shows that the local
buckling failure cannot be depicted by the slice model. Apart from that, both the moment-curvature and the
ovalization-curvature diagrams coincide until the failure in the full scale model.
- The comparison of the ovalization results between the numerical model and the measurements shows that
there is a significant difference in almost all tube types. Especially for the sand-filled tubes the difference is
much higher. One of the sources of this difference is the lack of reliable data input for the sand-fill
properties.
- By changing the sand-fill’s properties and selecting values which are in the normal value range of a typical
sand soil, then the ABAQUS ovalization results come closer to the measured ones.
- For the full scale model five Imperfection/Thickness percentages (4%, 8%, 12%, 16%, 20%) were chosen
for the analyses (except for Tube D in which only 4% and 20% were implemented), since there weren’t any
measurements available for the actual imperfections. Imperfection/Thickness percentages from 4%-12%
are typical for spirally welded steel tubes.
- Also for the full scale model the ovalization results are not close to the measured ones. In the case of Tube
B, the failure in the numerical model happens in far too early curvature values compared to the
experimental results, something that shows high uncertainty for the measured critical curvature.
131
Chapter 7 - Conclusions and Recommendations
7.2 Recommendations
Future research studies, which will be related to the local buckling behaviour of the steel tubes used in
combination walls have to take into consideration certain aspects such as:
- Perform experiments with detailed imperfection measurements such as the locations of the welds and
the possible offsets in the welding. This should be done before the experimental procedure.
- Input these imperfections into the numerical model in order to design a model as close to the real
experiment as possible. Designing an accurate and complete numerical model improves the simulation
results prominently.
- Perform extensive material testing in order to derive the properties of both the steel tubes and the sand-
fill. These material properties are critical parameters in order to achieve accurate results in the numerical
analyses, since they have a substantial impact on the equilibrium path, the maximum moment, the
critical curvature and the ovalization. Testing on different locations of the tube and in every direction
should be executed in order to have a clear view of the tube’s material properties. Also the material at
the weld locations has altered properties compared to the base material.
- Record the moment-curvature and ovalization-curvature curves during the experimental procedure, so
that they can be compared to the ones produced from the numerical analyses and to the analytical
curves. This way also the equilibrium path of the numerical and analytical curves can be verified.
- Calculate the through thickness distribution of the residual stresses for tubes with different welding
scheme. The residual stresses distribution used in the current thesis project was based upon tubes with
spiral weld. This scheme might apply also for the longitudinal welded steel tubes but there has been no
available research upon the matter. It would be for the best, that an extensive investigation would be
performed.
- Perform a parametrical study on the full scale model for different sand properties. This will give a clear
view on whether the sand-fill properties have an effect also on the critical curvature, since the
parametric study, which is performed with the slice model, clearly showed that there is a dependency of
the ovalization of the tube to the sand-fill properties. Due to lack of time, in this study has only been
implemented a rather insufficient number of numerical analyses and restricted variety in the properties
of the sand because each numerical analysis of the full scale model with the sand-fill taken into account,
demands at least one day of running -with the current means- in order to reach to an acceptable point.
- Perform a parametrical study on the full scale model for different soil properties (sand, silt, clay).
132
APPENDICES
133
134
Appendix A
ABAQUS contact model
The ABAQUS software has various relationships which can be used to define a contact model. The contact
model is described by inputting the mechanical contact properties in the software. In a mechanical contact
simulation, the interaction between contacting bodies is defined by assigning a contact property model to a
contact interaction. The interaction properties which were chosen to be input in the current research are a
Normal Behaviour model and a Tangential Behaviour model.
For the Normal Behaviour, a “Hard” Contact property is selected, which a pressure-overclosure relationship that
minimizes the penetration of the slave surface into the master surface at the constraint locations and does not
allow the transfer of tensile stress across the interface. When surfaces are in contact, any contact pressure can be
transmitted between them. The surfaces separate if the contact pressure reduces to zero. Separated surfaces
come into contact when the clearance between them reduces to zero. In Figure 208 the pressure-overclosure
relationship is shown.
For the Tangential Behaviour, is implemented with a stiffness (penalty) method which is practically describing
the Coulomb friction model. In ABAQUS, its general form allows the friction coefficient to be defined in terms
of slip rate, contact pressure, average surface temperature at the contact point, and field variables.
The basic concept of the Coulomb friction model is to relate the maximum allowable frictional (shear) stress
across an interface to the contact pressure between the contacting bodies. In the basic form of the Coulomb
friction model, two contacting surfaces can carry shear stresses up to a certain magnitude across their interface
before they start sliding relative to one another; this state is known as sticking. The Coulomb friction model
defines this critical shear stress, τcrit, at which sliding of the surfaces starts as a fraction of the contact pressure, p,
between the surfaces (τcrit = μ*p). The stick/slip calculations determine when a point transitions from sticking to
slipping or from slipping to sticking. The fraction, μ, is known as the coefficient of friction.
The basic friction model assumes that μ is the same in all directions (isotropic friction). For a three-dimensional
simulation there are two orthogonal components of shear stress, τ1 and τ2, along the interface between the two
bodies. These components act in the local tangent directions for the contact surfaces or contact elements. You
can specify an optional equivalent shear stress limit, τmax, so that, regardless of the magnitude of the contact
pressure stress, sliding will occur if the magnitude of the equivalent shear stress reaches this value.
Figure 209: Stick region for the basic Coulomb friction model
135
This shear stress limit is typically introduced in cases when the contact pressure stress may become very large (as
can happen in some manufacturing processes), causing the Coulomb theory to provide a critical shear stress at
the interface that exceeds the yield stress in the material beneath the contact surface. A reasonable upper bound
estimate for τmax is fy/√3, where fy is the Mises yield stress of the material adjacent to the surface; however,
empirical data are the best source for τmax. A value of zero is not allowed.
136
Appendix B
Photographs of failure mode shapes
137
138
139
140
141
142
143
Appendix C
Model implementation in ABAQUS
Initially, the model geometry of the two parts is created:
Module ‘Part’ → Create Part
144
Module ‘Property’ → Create Section
Next, the material orientation and element normal for the steel part are defined:
Module ‘Property’ → Assign Shell/Membrane normal
If the outside surface of the tube is brown, then the shell orientation has been assigned correctly.
In order to define the material orientation, a local cylindrical coordinate system has to be defined, because the
residual stresses are given in terms of hoop stress and axial stress. The origin is at 0,0,0.
Module ‘Property’ → Create Datum CSYS: 3 Points
145
The new coordinate system should be selected, and then the correct axis should be selected as the normal
direction. Direction 1 should correspond to the axial direction of the tube and direction 2 should correspond
with the hoop direction. This is important for defining the residual stresses correctly.
Next, the global seeds have to be set for the mesh. A mesh size of about 2-5% of the steel diameter is followed,
in order to get the form of the waves of the mode shapes with enough elements.
Module ‘Mesh’ → Seed part instance
This is done for the steel part. For the sand part the Seed Edges is used in order to mesh the sand-fill cross-
section and set the mesh size in the longitudinal direction. In the longitudinal direction, a mesh size equal to the
steel mesh size is selected and at the cross-section face a 30-40mm size is selected.
146
The elements to be selected are chosen by:
Module ‘Mesh’ → Seed part instance
The element S4R for the steel tube and the element C3D8 for the sand-fill are selected.
In order to ensure that the mesh is assigned correctly, the mesh controls should be checked. If quadrilateral
elements were selected, a quad element shape is appropriate:
Next, the Assembly can be meshed by clicking the Mesh Part Instance button and selecting both parts.
Next, two reference points RP are created at both ends of the tube. One is located at 0,0,0 and the second at
0,0,11000. In addition, datum planes are created to make it clear where the centre of the tube is. Some sets have
to be assigned now in order to make it easier to export the output and in order to assign residual stresses. An
element set of all steel elements is created to which the residual stresses are assigned and three sets are created at
the points where the output is required. For each RP a set is assigned and lastly a set is assigned to the node
located at one of the middle sides of the steel tube in order to measure the horizontal ovalization.
147
Next, the analysis steps are defined:
Module ‘Step’ → Create step
Two steps are created. The first (Static, General) in order to input the residual stresses and the imperfection
mode and the second (Static, Riks) in order to apply the rotation at the edges and get the local buckling failure.
The Static Riks procedure is capable of tracing the entire equilibrium path but it is not always robust because
sometimes it was not able to achieve convergence past the limiting moment. To surpass these difficulties some
changes to the convergence criteria are used. The activation of ‘Discontinuous analysis’ in the Step module →
Other → General Solution Controls → Manager → Time Incrementation. Also the first two parameters of the
Field Equations tab are doubled in value, since in the ABAQUS manual ([1] , §7.2.2) is stated that these criteria
are rather strict and can be changed when convergence is difficult to be achieved.
It should also be verified that the correct output is being requested under "Field Output Requests" and "History
Output Requests". Three History Output Requests are created for the two RP sets and the node for ovalization
measurement.
148
Next, the interaction and the interaction properties are defined:
Module ‘Interaction’ → Create Interaction Property
The contact is a surface-to-surface type and the master and slave surfaces have to be selected. As master surface,
the steel tube’s internal surface is selected and as slave surface, the sand-fill’s surface which is in contact with the
steel is selected.
Next, two constraints are defined:
Module ‘Interaction’ → Create Constraint
149
Now the boundary conditions can be defined. One end should be constrained in the directions U1, U2, UR2,
and UR3, and the other end should be constrained in U1, U2, U3, UR2, and UR3. These conditions prevent the
tube from rotating or moving, but allow one end of the tube to slide along the tube axis.
The rotations are applied to each RP and they are equal and opposite to each other. Usually a rotation of 0.15-
0.20 radians is more than adequate.
Next, the imperfection and initial conditions (residual stresses) are applied. They are both applied by editing the
input file. This can be done by right-clicking on the model’s name in the GUI and selecting Edit Keywords. This
input text must be entered in a specific way and in a specific place. Before the Step-1 initiation and the
imperfection before the initial conditions keyword.
Before this analysis, a buckling analysis has to be performed. The tube model is simply copied and the following
changes are made:
1. The static steps are suppressed and a Linear perturbation → Buckle step is added. The Subspace eigensolver is
used and the number of eigenvalues requested can be specified (usually 10).
2. A rotation is applied. The calculated eigenvalues will be proportional to this rotation, so a rotation such as 1
radian is a good choice.
3. The following keyword should be added to the input file right before the keyword *End Step:
NODE FILE
u
150
Now the buckling analysis can be submitted. First a job is created. The name of the job must match the name of
the job from which the initial imperfections will be taken. In order to speed up the analysis, CPU parallelization
can be selected by going to the "Parallelization" tab.
After the buckling analysis is complete, the static analysis can be submitted in the same way. The jobs can be
monitored by right clicking on the job and clicking on "Monitor", and the results can be viewed by clicking on
"Results".
One way to export the output is to go to Tools→XY Data→Manager, in the results viewer or just double-click
on XYData in the ‘Visualisation’ Module. XY data can be created based on the ODB field output. In the
dialog box, the position "Unique Nodal" should be selected. Then the required output variables can be selected.
To generate a moment-curvature diagram, the output variables CM1 and UR1 should be selected. The initial
empty static step should also be deselected, so that only output from the Riks analysis is requested.
The node set corresponding to the reference point at the support can be selected under the "Elements/Nodes"
tab. The output can now be viewed by clicking "Plot", and prepared to be exported by clicking "Save".
Finally, an Excel file can be generated from this output by going to the Plug-ins→Tools→Excel Utilities. The
data can now be further manipulated in MS Excel.
One way of running multiple analyses is to use batch files. An input file can be created from within the graphical
interface by right clicking on the job and selecting "Write Input". Next a batch file has to be created to call this
input file. The syntax is as follows:
abaqus job=name_of_input_file interactive
Multiple batch files can be created which correspond to different input files. In order to call them, a master
batch file has to be created, which has the following syntax in order to call the batch files defined above. Multiple
jobs can be called this way:
call name_of_batch_file1.bat
call name_of_batch_file2.bat
…
Then, run the main .bat file through the command prompt.
151
152
Bibliography
153