Ductile and Compacted Graphite Iron Casting Skin – Evaluation, Effect on Fatigue Strength and Elimination
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Sarum Boonmee Graduate Program in Materials Science and Engineering
The Ohio State University 2013
Dissertation Committee: Yogeshwar Sahai, Advisor Doru M. Stefanescu, Co-advisor Peter Anderson Ji-Cheng Zhao
Copyright by Sarum Boonmee 2013
Abstract Compacted graphite (CG) iron features a good combination of tensile strength, impact resistance, thermal conductivity and damping capacity. This combination makes CG iron a material of choice for various applications, especially for the automobile industry. The mechanical properties of CG iron listed in the standards (i.e. ASTM) are for machined specimens. However, since most iron castings retain the original casting surface (a.k.a. casting skin), the actual performance of the part could be significantly different from that of the machined specimens. Recent studies have shown the negative effect of the casting skin, but little quantification of its effect on mechanical properties is available. Further, the understanding of its mechanism of formation is at best incomplete.
In this research, the effect of the casting skin on mechanical properties in CG and ductile irons (DI) is explored. The differences in tensile and fatigue properties between as-cast and machined samples were quantified and correlated to the casting skin features. It was found that the presence of the casting skin was accountable for 9% reduction of tensile strength and up to 32% reduction of fatigue strength (for CG iron with 40% nodularity).
Several mechanisms of the casting skin formation are proposed in this research. The formation of ferritic and pearlitic rims is explained by decarburizing/carburizing reactions at the mold/metal interface. Mg depletion and solidification kinetics effect were identified ii
as the formation mechanisms of the graphite degradation. A 2-D thermal diffusion model was formulated based on Mg depletion theory. The model can be used to predict the casting skin thickness when Mg depletion is the dominant mechanism. Furthermore, using the asymmetric Fe-Gr phase diagram, some instances of casting skin formation were explained based on solidification kinetics theory.
The experimental microstructural evidence and the theoretical progress were conducive to the development of methods for the minimization of the casting skin formation. A serie of experiments was conducted to evaluate the effect of mold coatings on the casting skin formation. It was found that the thermal conductivity of the inactive coatings played an important role. The results from experiments with FeSi and graphite coatings supported the proposed mechanisms. FeSiMg was found effective in suppressing the casting skin formation. Furthermore, key findings are highlighted including the effect of carbon equivalent (CE) and the effect of section thickness.
Shot blasting is suggested as an effective means for the elimination of the casting skin effect. The removal of the casting skin by shot blasting was observed. In addition, the work hardening created by the shot blasting can be seen through the plastic deformation on the sample surface. Both contributed to the improvement of the tensile and fatigue properties.
iii
Dedication To my family and my teachers
iv
Acknowledgments I would like to first thank my parents who have always been supportive of my education. Without their support, I could have never reached this far. They are essentially the reason why I considered a career in education that changes my life forever.
I would also like to acknowledge Professor Manus Sathirajinda who introduced me the great field of Metalcasting. The inspiration I received will last forever.
I wish to express my sincere gratitude to my advisor, Professor Stefanescu who offered me invaluable guidance throughout this research. Without him, I could have never completed this work at this level of quality. His guidance also constituted an excellent example of a good advisor.
I would like to thank members of the Vision Casting laboratory at Ohio State for their help especially Evan Standish, Bobby Gyesi, Molly Moran, Aitor Loigaza and Elham Momeni. The friendships that they offered to me will never been forgotten. This list of people will never been completed without Ken Kushner and Ross Baldwin, MSE OSU staffs, who sweated with me for my work. Beside the work, they are my best and closest friends at OSU. v
I would like to acknowledge the American Foundry Association, Caterpillar Inc., Cummins Engine Co., and Rio Tinto Iron & Titanium Inc. for supporting this research financially. I am indebted to the members of the 5R Committees and of the Steering Committee (Adrian Catalina chair) for their support and many useful suggestions and comments.
I am grateful to the Government of Thailand and to AFS for the financial support during my Doctor of Philosophy studies at OSU
vi
Vita 1998................................................................B.Eng. Metallurgical Engineering, Suranaree University of Technology 2004................................................................M.Eng. Metallurgical Engineering, Chulalongkorn University 2004 to present ...............................................Lecturer, School of Metallurgical Engineering, Suranaree University of Technology 2011................................................................M.S. Materials Science and Engineering, The Ohio State University
Publications
S. Boonmee and D.M. Stefanescu, "Casting Skin Management in Compacted Graphite Iron Part I: Effect of Mold Coating and Section Thickness", Trans. AFS 121 (2013) paper 1391 S. Boonmee and D.M. Stefanescu, "Casting Skin Management in Compacted Graphite Iron Part II: Mechanism of Casting Skin Formation", Trans. AFS 121 (2013) paper 1392
vii
S. Boonmee and D.M. Stefanescu, “Effect of Casting Skin on the Fatigue Properties of CG Iron”, International Journal of Metalcasting, Spring 2013 D.M. Stefanescu, M.K. Moran, S. Boonmee and W.L. Guesser, "The Use of Combined Liquid Displacement and Cooling Curve Analysis in Understanding the Solidification of Cast Iron", Trans. AFS 120 (2012) pp. 365-374 S. Boonmee and D.M. Stefanescu, "The Effect of Nodularity and Surface Condition on Fatigue Properties of CG Iron", Trans. AFS 120 (2012) pp. 355-364 S. Boonmee and D.M. Stefanescu, "The Mechanism of Formation of Casting Skin and Its Effect on Tensile Properties", Foundry Trade Journal International 186 (2012) pp. 225-228 S. Boonmee, M.K. Moran and D.M. Stefanescu, “On the Effect of the Casting Skin on the Fatigue Properties of CG Iron”, Trans. AFS 119 (2011) pp. 421-430 S. Boonmee and D.M. Stefanescu, "The Mechanism of Formation of Casting Skin and Its Effect on Tensile Properties", Key Engineering Materials 457 (2011) pp. 11-16 S. Boonmee, B. Gyesi, D.M. Stefanescu, “Casting Skin of Compacted Graphite Iron Part I: Evaluation and Mechanism for Formation”, Trans AFS 118 (2010) pp. 205-216 S. Boonmee and D.M. Stefanescu, “Casting Skin of Compacted Graphite Iron Part II: Influence on Tensile Mechanical Properties, Trans. AFS 118 (2010) pp.217-224 S. Boonmee and D.M. Stefanescu, “On the Mechanism of Casting Skin Formation in Compacted Graphite Cast Iron”, International Journal of Metalcasting, Fall (2009) pp. 19-24
viii
Fields of Study
Major Field: Materials Science and Engineering
ix
Table of Contents Abstract ............................................................................................................................... ii Dedication .......................................................................................................................... iv Acknowledgments............................................................................................................... v Vita.................................................................................................................................... vii Publications ................................................................................................................... vii Fields of Study ............................................................................................................... ix Table of Contents ................................................................................................................ x List of Tables .................................................................................................................... xv List of Figures ................................................................................................................. xvii Chapter 1: Background ...................................................................................................... 1 1.1 Introduction ............................................................................................................... 1 1.2 Motivation and objectives ......................................................................................... 4 1.3 The Casting skin ........................................................................................................ 5 1.4 Effect of casting skin on mechanical properties...................................................... 10 1.5 Variables affecting the casting skin formation ........................................................ 16 1.5.1 Section thickness .................................................................................................. 16 x
1.5.2 Molding aggregate................................................................................................ 17 1.5.3 Metallostatic height .............................................................................................. 18 1.5.4 Sand binding system ............................................................................................ 19 1.5.5 Mold coating ........................................................................................................ 21 Chapter 2: Quantification of the casting skin .................................................................. 23 2.1 Introduction ............................................................................................................. 23 2.2 Surface feature......................................................................................................... 23 2.2 Subsurface features ................................................................................................. 25 Chapter 3: Effect of Casting Skin on Mechanical Properties .......................................... 31 3.1 Experimental approach ............................................................................................ 31 3.1.1 Skin factors ........................................................................................................... 32 3.1.2 Casting design and simulation.............................................................................. 33 3.1.3 Production of test casting and samples ................................................................ 34 3.1.4 Thermal analysis .................................................................................................. 38 3.1.5 Chemical analysis ................................................................................................. 39 3.1.6 Tensile and fatigue testing.................................................................................... 39 3.1.7 Metallography ...................................................................................................... 41 3.2 Chemical composition and microstructure.............................................................. 42 3.3 General observations of the casting skin features ................................................... 46 xi
3.3.1 Tensile samples .................................................................................................... 46 3.3.2 Fatigue samples .................................................................................................... 50 3.4 Effect of casting skin on tensile properties ............................................................. 51 3.5 Effect of casting skin on fatigue properties ............................................................. 57 3.6 Conclusions ............................................................................................................. 65 Chapter 4: Mechanism of Casting Skin Formation.......................................................... 66 4.1 Surface roughness ................................................................................................... 66 4.2 Ferritic rim and Graphite depletion ......................................................................... 67 4.3 Pearlitic rim ............................................................................................................. 71 4.4 Graphite degradation ............................................................................................... 71 4.4.1 Mg depletion by sulfur reaction ........................................................................... 71 4.4.2 Mg depletion by Mg oxidation ............................................................................. 74 4.4.3 Diffusion models for Mg depletion zone prediction ............................................ 81 4.5 Solidification kinetics.............................................................................................. 92 4.5.1 Effect of carbon equivalent .................................................................................. 94 4.5.2 Effect of inoculation ............................................................................................. 96 4.5.3 Effect of thermal conductivity.............................................................................. 96 4.6 Conclusion............................................................................................................... 99 Chapter 5: Minimization of the Casting Skin Formation............................................... 101 xii
5.1 Introduction ........................................................................................................... 101 5.2 Experimental approach .......................................................................................... 101 5.2.1 Test casting design ............................................................................................. 102 5.2.2 Mold coating ...................................................................................................... 103 5.2.3 Production of test castings.................................................................................. 107 5.2.4 Microstructural examination and skin quantification......................................... 108 5.2.5 Chemical analysis and microstructure................................................................ 108 5.3 Effect of mold coating on the casting skin formation ........................................... 110 5.3.1 Active coating .................................................................................................... 110 5.3.2 Reactive coating ................................................................................................. 114 5.3.3 Inactive coating .................................................................................................. 117 5.4 Effect of section thickness on the casting skin with mold coatings ...................... 123 5.5 Thermal analysis ................................................................................................... 126 5.6 Conclusion............................................................................................................. 135 Chapter 6: Elimination of the Casting Skin Effect ........................................................ 138 6.1 Introduction ........................................................................................................... 138 6.2 Experimental approach .......................................................................................... 138 6.2.1 Casting design and simulation............................................................................ 138 6.2.2 Production of test casting ................................................................................... 139 xiii
6.2.3. Quantification of casting skin ........................................................................... 139 6.2.4 Tensile and fatigue testing.................................................................................. 140 6.2.5 Shot blasting ....................................................................................................... 140 6.3 Effect of shot blasting on the casting skin effect .................................................. 142 6.3.1 Tensile properties ............................................................................................... 142 6.3.2 Fatigue properties ............................................................................................... 145 6.4 Conclusion............................................................................................................. 150 Chapter 7: Conclusions .................................................................................................. 152 References ....................................................................................................................... 156 Appendix A: Tensile testing data .................................................................................... 160 Appendix B: Fatigue testing data .................................................................................... 163
xiv
List of Tables Table 1. Physical and mechanical properties of cast irons ................................................. 4 Table 2 Casting skin features [11] ...................................................................................... 8 Table 3. Casting skin thickness as a function of casting diameter and mold aggregate [17] ........................................................................................................................................... 18 Table 4 Mg and sulfur content in the graphite degradation zone and the bulk iron [18] .. 21 Table 5. Definitions of measured quantities ..................................................................... 25 Table 6. Measured quantities and their symbols ............................................................... 30 Table 7. Basic dimensions of the reaction chamber ......................................................... 37 Table 8. Dimension of as-cast and machined tensile samples. ......................................... 37 Table 9. Chemical composition of compacted graphite iron for tensile testing ............... 42 Table 10. Chemical composition of compacted graphite iron for fatigue testing ............. 44 Table 11. Correlations between quantities defining the skin thickness ............................ 47 Table 12. Correlations between visual skin thickness and other parameters .................... 48 Table 13. Correlation Analysis of all data ........................................................................ 52 Table 14. Regression analysis for tensile strength (as-cast data)...................................... 55 Table 15. Regression analysis for tensile strength (normalized as-cast data)................... 57 Table 16. Summary of the fatigue skin factor................................................................... 60 Table 17. Heat of formation of some compounds at 298 and 1500 K .............................. 74 Table 18. Chemical composition of the sample investigated with EPMA ....................... 76 xv
Table 19. Gibbs free energy of formation of Mg oxidation reactions at 1600 K .............. 80 Table 20. Diffusivity of various elements in liquid iron at 1600 C [27] ......................... 83 Table 21. Summary of parameters used for simulation .................................................... 87 Table 22. Coating materials and their thermal conductivities ........................................ 104 Table 23. Chemistry of CG iron heats in this study ........................................................ 108 Table 24. Corresponding cooling rates of the virtual thermocouples ............................. 125 Table 25. Visual nodularity of samples from cooling cups at the center of the cups ..... 132 Table 26. Summary of shot blasting conditions used ..................................................... 141 Table 27. Influence of shot blasting on the thickness (visual) of the casting skin ......... 145
xvi
List of Figures Figure 1. Deep-etched microstructure showing graphite morphologies in a) gray iron; b) compacted graphite iron; c) ductile iron [1] ....................................................................... 2 Figure 2. Decarburized layer on the casting surface in gray iron after Reisener [5] .......... 6 Figure 3. Micrograph of the as-cast skin in ductile iron [9] ............................................... 7 Figure 4. Common features of the casting skin .................................................................. 8 Figure 5. Yield strength and elongation of as-cast and machined samples [9]................. 11 Figure 6. Pearlitic rim in ductile iron in Goodrich and Lobenhofer study [9] .................. 11 Figure 7. Comparison of the ultimate tensile strength and elongation of as-cast and machined ductile iron samples [4] .................................................................................... 12 Figure 8. Relationship between ultimate tensile strength and surface roughness in ductile iron; the circles indicate low pouring temperature [4] ...................................................... 13 Figure 9. Relationship between the amount of type-D graphite and ultimate tensile strength along the casting radius [10] ............................................................................... 13 Figure 10. Tensile strength as a function of casting skin thickness [11] .......................... 14 Figure 11. Comparison of endurance limits of as-cast and machined fatigue samples [15] ........................................................................................................................................... 15 Figure 12. S-N curve for as-cast ductile iron sample [15] ................................................ 16 Figure 13. Casting skin thickness variation on section thickness [11] ............................. 17
xvii
Figure 14. Microstructure of the casting skin with various molding aggregate; a) and b) chromite; c) and d) silica; e) and f) Low Density Aluminosilicate (LDASC) [17] .......... 18 Figure 15. Surface roughness as a function of metallostatic height [11] .......................... 19 Figure 16. Relationship between surface roughness and metallostatic pressure [4]......... 20 Figure 17. Sulfur profiles in ductile iron [19] ................................................................... 22 Figure 18. Influence of mold coating types on the casting skin thickness; I – 0.030%Mg; II – 0.036%Mg [20]........................................................................................................... 22 Figure 19. Definition of roughness average [21] .............................................................. 24 Figure 20. Locations and direction of the surface roughness measurements ................... 24 Figure 21. Optical micrographs of compacted graphite iron near the casting skin .......... 27 Figure 22. Graphite shape factors, amount (percent area) and total area versus distance from surface. ..................................................................................................................... 27 Figure 23. Skin thickness measurement based on sphericity............................................ 28 Figure 24. Percent pearlite as a function of distance from casting surface ....................... 29 Figure 25. Overview of the approach ............................................................................... 32 Figure 26. Test casting design for tensile test samples; a) isometric view of the design; b) simulated cooling rate; c) simulated porosity ................................................................... 35 Figure 27. Test casting design for fatigue test samples; a) isometric view of the design; b) simulated cooling rate; c) simulated porosity ................................................................... 36 Figure 28. Reaction chamber; a) schematic drawing; b) during the operation ................. 36 Figure 29. Thermal Analysis (TA) cup ............................................................................. 38 Figure 30. The setup for 4-point bending fatigue testing.................................................. 40 xviii
Figure 31. Representative microstructures from the center of the tensile samples .......... 43 Figure 32. Optical micrographs of compacted graphite iron near the casting skin with various casting section thickness ...................................................................................... 44 Figure 33. Representative microstructures from the center of the plates for different heats ........................................................................................................................................... 45 Figure 34. Representative microstructures underneath the casting surface ...................... 46 Figure 35. Influence of metallostatic height and type of binder on roughness ................. 49 Figure 36. Effect of type of binder on skin thickness (visual) and pearlitic rim thickness ........................................................................................................................................... 49 Figure 37. Effect of type of binder on the pearlitic rim; 4% Nital etched ........................ 50 Figure 38. Effect of the thickness of the casting skin on the tensile properties of CG iron ........................................................................................................................................... 53 Figure 39. Influence of roughness on the tensile properties of as-cast and machined test plates ................................................................................................................................. 54 Figure 40. Plot for regression data in Table 14 ................................................................. 55 Figure 41. S-N curves of as-cast (AC) and machined (M) sample; a) 15% nodularity; b) 30% nodularity; c) 40% nodularity; m = 250 MPa.......................................................... 59 Figure 42. Fatigue skin factor as a function of nodularity ................................................ 60 Figure 43. Comparison of S-N curves tested at different mean stress; a) m = 250 MPa; b) m = 300 MPa ................................................................................................................... 61 Figure 44. Relationship between the cyclic stress and the mean stress of as-cast and machined samples ............................................................................................................. 62 xix
Figure 45. The representative fracture surfaces of as-cast samples .................................. 62 Figure 46. A pin hole close the casting surface ................................................................ 63 Figure 47. Secondary electron images of an as-cast sample; a) at the casting skin; b) cleavage fracture in crack propagation zone; c) ductile-dimple fracture in sudden rupture zone; d) striations in crack propagation zone ................................................................... 64 Figure 48. Molding sand and liquid iron interaction; mild interaction (left) severe interaction (right) .............................................................................................................. 67 Figure 49. Interfacial mold/metal reactions creating decarburized layer on iron casting surface [5] ......................................................................................................................... 69 Figure 50. Relationship between the depth of the graphite-free layer and the time in the temperature range 1150 – 1050 C for grey iron cast in green sand mold [7] .................. 69 Figure 51. Occurrence of graphite depletion in DI cast in permanent mold [24] ............. 70 Figure 52. Measured skin thickness of thermal analysis cups with various types of coatings (CE = 4.23) ......................................................................................................... 73 Figure 53. Effect of section thickness, cooling rate and mold coatings on the skin thickness (CE = 4.23) ........................................................................................................ 74 Figure 54. EPMA maps show solutal distribution at the casting skin .............................. 77 Figure 55. Line profile of Mg from casting surface to 500 m inward ............................ 79 Figure 56. Line profile of Mg from casting surface to 500 m inward ............................ 79 Figure 57. Casting skin in CG iron at the quartz/iron interface ........................................ 81 Figure 58. Relationship between impurity diffusivity in liquid iron and solid solubility at 1350 and 1600 C .............................................................................................................. 83 xx
Figure 59. Comparison of experimental data (squares) and diffusion model prediction (solid line) in CGI casting ................................................................................................. 84 Figure 60. Basic geometry of the model based on the standard cooling cup.................... 86 Figure 61. Comparison of the simulated and experimental cooling curves ...................... 88 Figure 62. Comparison between (a) simulated Mg distribution (b) corresponding microstructure of graphite degraded layer ........................................................................ 89 Figure 63. Effect of x or y on simulation results ......................................................... 90 Figure 64. a) Effect of pouring temperature on the simulated skin thickness; b) corresponding cooling curves ........................................................................................... 91 Figure 65. Effect of initial Mg on the simulated skin thickness ....................................... 92 Figure 66. Schematic asymmetric phase diagram of the iron-graphite system and the effect on casting skin formation ........................................................................................ 93 Figure 67. Mg redistribution at the surface of the casting because of rejection of Mg by the solidifying iron ............................................................................................................ 94 Figure 68. The skin thickness as a function of thermal conductivity (reactive coatings are excluded) ........................................................................................................................... 97 Figure 69. Schematic diagram of the combined effect between two skin formation mechanisms ....................................................................................................................... 99 Figure 70. Summary of casting skin formation mechanism ........................................... 100 Figure 71. Overview of the approach ............................................................................. 102 Figure 72. Test casting design for the mold coating experiment .................................... 103 Figure 73. Core dipping .................................................................................................. 105 xxi
Figure 74. Top layer and penetration depth .................................................................... 106 Figure 75. Cross sections of coated cores ....................................................................... 106 Figure 76. Top layer thickness and penetration depth as a function of the dipping time for various mold coatings ..................................................................................................... 107 Figure 77. Representative microstructure at the center of the test plates ....................... 109 Figure 78. Cooling curves from standard Thermal Analysis cups .................................. 110 Figure 79. Unetched microstructures of the casting skin without coating (top) with ferrosilicon coating (middle) and with graphite coating (bottom) .................................. 112 Figure 80. Color-etched microstructures of the casting skin without coating (top) with ferrosilicon coating (middle) and with graphite coating (bottom) .................................. 113 Figure 81. The influence of carbon equivalent and of active coatings on skin thickness114 Figure 82. Unetched microstructures of samples from Heat 504 (CG iron, CE = 4.23) a) at the skin b) at the middle of the plate ........................................................................... 116 Figure 83. Color micrographs of CG irons with zircon based coating ........................... 118 Figure 84. Average skin thickness as a function of CE (left) and dipping time (right) .. 119 Figure 85. Color micrographs at the casting skin with boron nitride coating................. 119 Figure 86. Color micrographs at the casting skin with mica based coating ................... 120 Figure 87. Unetched microstructure of CG iron (CE = 4.23) with MgO coating ........... 121 Figure 88. The skin thickness from various types of mold coating ................................ 122 Figure 89. Roughness average as a function of mold coating types (Zr1 – zircon coat 1 min. dipping; Zr2 - zircon coat 2 min. dipping; Zr3 – zircon coat 3 min. dipping) ....... 123 Figure 90. Test casting design for step casting [31] ....................................................... 124 xxii
Figure 91. Simulation results a) cooling rate distribution b) nodularity distribution [31] ......................................................................................................................................... 125 Figure 92. Effect of section thickness and mold coatings on microstructure at the casting skin for CG iron with CE = 4.33 ..................................................................................... 127 Figure 93. Effect of section thickness on the casting skin thickness .............................. 128 Figure 94. Example of carbide structure in a sample with graphite coating (CE = 4.06)128 Figure 95. Effect of section thickness and mold coating (CaO, MgO and FeSiMg) on the microstructure at the casting skin for CG iron with CE = 4.23 ...................................... 129 Figure 96. Effect of section thickness, cooling rate and mold coatings (CaO, MgO and FeSiMg) on the skin thickness; CE = 4.23...................................................................... 130 Figure 97. Cooling curves and cooling rates for no-coat and zircon coated cups .......... 130 Figure 98. Results from Thermal Analysis cups with no coating, CaO coating, FeSiMg coating and MgO coating a) Cooling curves and b) derivatives .................................... 131 Figure 99. Cross section of the cooling cup with MgO coating showing a shrinkage cavity around the thermocouple tip ........................................................................................... 133 Figure 100. Unetched microstructure of cooling cups without coating, with CaO, MgO, FeSiMg and Zr based coating at the skin ........................................................................ 134 Figure 101. Unetched microstructure of cooling cups without coating, with CaO, MgO, FeSiMg and Zr based coating at the center of the cups .................................................. 134 Figure 102. Measured skin thickness of cooling cups with various types of coatings (CE = 4.23) ............................................................................................................................. 135 Figure 103. Overview of the approach ........................................................................... 139 xxiii
Figure 104. The shot blasting set up for the fatigue test samples ................................... 141 Figure 105. Effect of shot blasting on visual skin thickness (SB1 = 1 minute, SB2 = 5 minutes)........................................................................................................................... 143 Figure 106. Comparison of casting skin of as cast (left) 1 minute shot blasted (middle) and 5 minute shot blasted sample (right). Unetched. ...................................................... 143 Figure 107. Effect of surface condition on the tensile properties; AC - as-cast; M machined; SB1 - shot blasted 1 min.; SB5 - shot blasted 5 min; a) tensile strength; b) elongation ........................................................................................................................ 144 Figure 108. S-N curves of the as-cast (AC), machined (M), as-cast-shot-blasted (AC-SB) and machined-shot-blasted (M-SB) for 15% nodularity CG iron ................................... 146 Figure 109. Color micrographs of the casting skin before and after shot blasting; a) AC; b) AC-SB; c) M; d) M-SB .............................................................................................. 147 Figure 110. Microhardness profiles of a) AC and M samples; b) AC and AC-SB samples; c) M and M-SB samples; d) AC-SB and M-SB samples ................................................ 149 Figure 111. Microhardness measurement near the casting skin of an as-cast sample; a) before etching; b) after etching with 4%Nital................................................................. 150
xxiv
Chapter 1: Background 1.1 Introduction Cast irons are categorized by their microstructures. White cast iron is the only type of cast iron that has Fe3C as a main carbon rich constituent where carbon is in its metastable state. Other types of cast iron may be referred as “Graphitic iron” because the carbon is in the form of graphite. Even though the graphite phase is the stable form of carbon in cast irons; the kinetic of graphite formation is sluggish. The presence of graphite is practically made possible by adding Si. Because the mechanical properties of irons are governed by the graphite morphology; the graphite morphology is used for classifying types of cast irons. Malleable iron is not discussed here because of its difference in nature of processing and applications.
In regular cast iron with no additions to the melt, carbon appears as lamellar graphite (LG). Cast iron with this form of graphite is known as “gray iron” because of the appearance of its fractured surface. Adding Mg or Ce changes the graphite morphology from lamellar to spheroidal. This type of cast iron is also known as “ductile iron” (DI or SG for spheriodal graphtie) derived from its remarkable ductility compared to gray iron. Compacted graphite (CG) iron has an intermediate graphite morphology between LG and SG. This is obtained by treating the liquid iron with an intermediate level of Mg and/or Ce. The typical Mg content used in CG iron production is in the range of 0.010 – 0.020 1
depending on the section thickness of the castings. Figure 1 demonstrates the change in graphite morphology from lamellar graphite in gray iron to compacted graphite in CG iron to spheroidal graphite in ductile iron.
250×
a)
200×
b)
1700×
c)
Figure 1. Deep-etched microstructure showing graphite morphologies in a) gray iron; b) compacted graphite iron; c) ductile iron [1]
Mechanical properties of cast irons are mainly governed by graphite morphology. Lamellar graphite in gray iron gives limited ductility and promotes cleavage fracture. This results from stress concentration at the tips of the graphite flakes. Spheroidal graphite in ductile iron reduces the stress concentration across the microstructure. Consequently, higher tensile strength, ductility and impact resistance are obtained. CG iron is positioned in between gray and ductile iron because of its intermediate graphite morphology. Therefore, CG iron has become an alternative material to replace gray iron for high performance applications such as engine blocks, transmission gear boxes and exhaust manifolds etc. 2
Thermal conductivity is also an important property for some applications that require good heat dissipation rate. For example, heat is generated during combustion cycles in an automobile engine block. The material used for the engine block should allow the heat to dissipate from the cylinders to prevent overheating; therefore, the materials with high thermal conductivity are preferred. Graphite has higher thermal conductivity than the iron matrix; as a result, the thermal conductivity is governed by graphite morphology. Gray iron is known for its high thermal conductivity. This is because the interconnected structure of lamellar graphite provides a high thermal conductive path. On the other hand, ductile iron has poorer thermal conductivity because the spheroidal graphite particles are isolated. Compacted graphite provides intermediate thermal conductivity between gray and ductile irons.
The damping capacity is another important property of cast irons. High damping capacity suppresses noise created by moving parts. It is known that lamellar and compacted graphite provide higher damping capacity than spheroidal graphite.
Table 1 summarizes the physical and mechanical properties available in the literature. As seen, gray and ductile irons have trade-off properties. For example, gray iron has better thermal conductivity and damping capacity but poorer tensile strength and elongation as compared to ductile iron. CG iron provides a good combination of thermal conductivity,
3
tensile strength, impact resistance and damping capacity. This makes CG iron a material of choice for high performance applications when good thermal conductivity is required.
Table 1. Physical and mechanical properties of cast irons Tensile strength ,MPa [2] Elongation, % [2] Impact energy†, J [3] Thermal conductivity††, W/mK [2] †
Gray iron 150 – 350 0 – 0.2 0 > 0 – 0.2
10 mm 0 0 > 0.4 – 0.6
> 0.4 – 0.6
> 0.4 – 0.6
40 mm > 0 – 0.2 > 0 – 0.2 > 0.4 – 0.6
Figure 14. Microstructure of the casting skin with various molding aggregate; a) and b) chromite; c) and d) silica; e) and f) Low Density Aluminosilicate (LDASC) [17]
1.5.3 Metallostatic height Generally, metallostatic height is known to affect the surface roughness. It is seen in Figure 15 that the surface roughness increased with metallostatic height [11]. Similar
18
relationship was also reported by Dix et al. [4] (Figure 16). Note that the metallostatic height (h) and metallostatic pressure (Pst) are directly related by.
Where is density of liquid metal and g is the acceleration of gravity.
These results were expected, as the greater metallostatic height (or pressure) pushes liquid metal further into gaps between sand grains creating rougher surface finish.
all plates
Roughness, microns
30 25 20 15 10 0
36
72
108 144 180
Metall. height, mm
Figure 15. Surface roughness as a function of metallostatic height [11]
1.5.4 Sand binding system Sand binding systems can have significant impact on the nature of casting skin formation. The effect of using green sand has been reported by several investigators [5] [6] [7] [8]. They suggested that the moisture in green sand promoted the formation of the ferritic rim.
19
Reisener [5] proposed that mold coating with high refractoriness and low permeability such as pyrophyllite was effective in suppressing the decarburizing process.
Figure 16. Relationship between surface roughness and metallostatic pressure [4]
Resin sand binding systems can be divided into non-sulfur-bearing systems (i.e. sodium silicate) and sulfur-bearing systems (i.e. furfuryl alcohol and urea-formaldehyde resin also known as “furan resin”). These binding system are attractive to foundry industry because of the higher productivity and strength. For the casting skin, the sulfur-bearing system gained more attention because of the well-known deleterious effect of sulfur on the spheriodal graphite formation. The effect of sulfur on the casting skin was documented as early as 1979 by Martin and Karsay. [18] The graphite degradation layers with thickness of 500 – 1200 µm were identified in ductile iron poured in various types of no-bake resin molds. The key fact from the study was the difference in sulfur and Mg level between the graphite degradation area and the bulk (shown in Table 4). It is seen 20
that not only the sulfur was increased, but Mg was also depleted as well. As a result, Mg/S was increased by more than 5 times that of the bulk. It was concluded that sulfur combined with Mg to form MgS resulting in less Mg for graphite spheriodization. They also reported that the graphite degradation thickness increased with the section thickness.
Table 4 Mg and sulfur content in the graphite degradation zone and the bulk iron [18] S% 0.018 0.034
Bulk iron Graphite degradation zone
Mg% 0.036 0.013
Mg/S ratio 2.00 0.38
Later work by Xiaogan et al. [19] suggested that the Para Toluene Sulfonic Acid (PTSA) that used as a catalyst for furan resin was the main causes of graphite degradation in ductile iron. It was proposed that SO2 is created when the resin come in contact with molten iron. SO2 then dissociated into sulfur atoms and diffused into molten iron. Mg is then combined with sulfur causing Mg depletion and graphite degradation. Figure 17 shows chemical analyses from the study. It is seen that the sulfur content was significant higher at the surface than at the center. This confirms the result by Martin and Karsay [18] discussed earlier.
1.5.5 Mold coating Mold coating has been thought as a possible solution for the graphite degradation. The effects of mold coating are i) minimizing mold-metal interaction by being a protective layer; ii) providing desulfurizing effect. Ivan et al. [20] conducted an experiment on the effect of the mold coating focusing on the graphite degradation. They reported a negative 21
effect of FeS2 coating. This is expected because FeS2 as a sulfur-bearing coating should promote the deleterious effect of sulfur. On the contrary, all other types of Mg-bearing coatings were effective in reducing the casting skin thickness.
Figure 17. Sulfur profiles in ductile iron [19]
Figure 18. Influence of mold coating types on the casting skin thickness; I – 0.030%Mg; II – 0.036%Mg [20] 22
Chapter 2: Quantification of the casting skin 2.1 Introduction In order to explore the relevant casting skin effect, a methodology for quantification of the casting skin must be established. The brief descriptions of the casting skin features have been discussed in Chapter 1. By recalling Table 2, the casting skin features have been categorized into two groups; surface features and subsurface features. This chapter deals with the details of the quantification methods of all casting skin features. The general observation and example on the measured quantities will also be discussed.
2.2 Surface feature The only casting skin feature in this category is the surface roughness. Surface roughness can be measured directly by a surface roughness tester. The most commonly used surface roughness parameter is the roughness average (Ra) also known as the arithmetic mean roughness. The roughness average was used throughout this research because of its wide acceptance. The definition of Ra is [21]: Ra
1 N
N
Y i 1
i
where N is the number of peaks detected over the evaluation length (2.5 mm) and Yi is the height of peak i over the centerline or the mean surface (Figure 19).
23
Figure 19. Definition of roughness average [21]
In a measurement, there were five evaluation lengths. The five values were averaged to return a reading. In this research, the representative values were obtained from five readings around the sample (typically within the gage area). In a special case, the measurements were done at different metallostatic heights and directions. The objective of the measurement was to verify the effect of the metallostatic height on surface roughness. Figure 20 demonstrates the locations and directions of surface roughness measurements on a tensile sample.
a
5mm
1
b
c
I
2
II 30mm
III
3
Gravity
Figure 20. Locations and direction of the surface roughness measurements 24
2.2 Subsurface features As humans perceive shapes differently from person to person, an image analysis technique was used to characterize the subsurface casting skin features to avoid human bias. The casting skin can be described by several characteristics based on both graphite morphology and matrix structure, such as graphite depletion, graphite degradation, ferritic rim and pearlitic rim. Percent area of graphite reflects graphite depletion, while graphite degradation can be described through graphite shape factors (sphericity, roundness, aspect ratio, compactness). The graphite shape factors are unitless number ranging from 0 to 1 (except for the aspect ratio). The higher numbers indicate higher nodularity (vice versa for the aspect ratio). Finally, the percent pearlite describes the ferritic and pearlitic rim. Note that roundness is the same as ASTM nodularity. Table 5 summarizes the definitions of the measured values.
Table 5. Definitions of measured quantities Feature
Quantity
Equation Agraphite %Gr A field
Graphite depletion
Percent graphite
Ferritic/Pearlitic rim
Percent pearlite
% Pe
Sphericity
S 4
Graphite degradation
A pearlite A field
A P2 4A R 2 l m
Definition : total area of graphite Agraphite A field : field area
Apearlite : total area of pearlite
P: perimeter of graphite particle
Aspect ratio
A = area of graphite particle lm = maxim axis length of particle Aratio= length of longest feret/shortest feret
Compactness
C 4
Roundness
25
A 2 Pconvex
Pconvex : convex perimeter
There were a number of steps of image processing during the image analysis. Firstly, the image was delineated to achieve clearer boundary between graphite particles and matrix. Secondly, the gray level threshold was applied to the image to separate the graphite particles and matrix. Next, a trapping step was added to filter out polishing artifacts that may influence the measurements. This step was done by excluding particles that are smaller than 10 × 10 μm, which correspond to abrasive alumina grain size used for rough polishing (6 μm in diameter). Finally, the graphite particles were measured for quantities. It should be noted that the graphite particles that lied on the frame border were excluded to avoid the bias. At least ten fields of 646 × 150 μm were scanned and averaged for each data point at 200× magnification. All measurements were done in the unetched condition except pearlite measurement that was done after 4% Nital etching.
In order to quantify the skin thickness based on the quantities discussed above, an example is introduced here. Figure 21 shows an example of the microstructure of compacted graphite iron near the casting skin. The casting surface is on the left. In the figure, three regions can be identified starting from the left: -
Region I: graphite depletion/degradation region;
-
Region II: high nodularity region;
-
Region III: bulk microstructure
26
Figure 21. Optical micrographs of compacted graphite iron near the casting skin
Figure 22 presents the quantitative measurements of graphite morphologies of the sample in Figure 21. It is seen that the graphite shape factors vary as a function of distance from surface. Again, three regions can be defined. In Region I, the graphite shape factors and the percent graphite are lower than in the other two regions. As expected because this region is responsible for lowering the mechanical properties, we will define it as the thickness of the skin.
2.4
0.4
1.6
0.2
0.8
0.0
0.0 2000
0
500
1000
1500
Average area (microns^2)
3.2
0.6
Area
300
Aspect ratio
Shape factors
0.8
4.0
% Area
250
10.0 8.0
200
6.0
150 4.0
100 50
2.0
0
0.0 1500 2000
0
500
1000
%Area
Sphericity Roundness Compactness Aspect ratio
1.0
Distance from surface (microns)
Distance from surface (microns)
Figure 22. Graphite shape factors, amount (percent area) and total area versus distance from surface. 27
In Region II, a clear increase in graphite shape factors (nodularity) and in the amount of graphite as compared to the bulk of the sample (Region III), is seen. The same trend occurs in every sample.
To quantify skin thickness based on these measurements, the measured values near the casting surface shall be compared with measured values in Region III. The location of values that demonstrate lower sphericity than the bulk was taken as the skin thickness. Figure 23 demonstrates the skin thickness measurement based on sphericity. Likewise, skin thicknesses based on other quantities were measured in the same fashion.
Figure 23. Skin thickness measurement based on sphericity
28
An example of the pearlitic rim thickness is shown in Figure 24. The same method to evaluate the skin thickness was used. The measured value is regarded as the skin thickness by percent pearlite (tSK-%Pe). Other measured quantities are shown in Table 6. These quantities are to be used for further analyses. The evaluation of the quantification method is discussed in the Section 3.3.
60 %Pearlite
50
%Pearlite
40 30 20 10 0 0
500
1000
1500
2000
Distance from surface (microns)
Figure 24. Percent pearlite as a function of distance from casting surface
29
Table 6. Measured quantities and their symbols Quantity Skin thickness by %graphite Skin thickness by compactness Skin thickness by %pearlite Skin thickness by visual inspection Distance of maximum nodularity by shape factor Distance of maximum nodularity by shape factor visual inspection Graphite sphericity Graphite compactness Nodularity by sphericity by visual inspection Percent of pearlite Thickness of pearlitic rim
30
Symbol tSK-%Gr tSK-cpct tSK-%Pe tSK-vis max Nshpfact max Nvis Nsph Ncpct Nvis %Pe tPe-rim
Chapter 3: Effect of Casting Skin on Mechanical Properties The first goal of this research is to quantify the effect of casting skin on mechanical properties focusing on tensile and fatigue properties. It has been shown in Chapter 1 that the casting skin features have adverse effect on mechanical properties. In this chapter, the detailed approach to the problem is explained. Some correlation analyses were used to explore the relationship between casting skin features and mechanical properties.
3.1 Experimental approach An overview of the approach used in this section is presented in Figure 25. The process started by the casting design. Computer simulation with a casting simulation software package was conducted to confirm that the designing criteria are met. The following step consisted in the production of test castings. Next, the samples were tested for tensile and fatigue properties in as-cast and machined conditions. Also, the skin quantification was performed on all samples to explore the structure – property relationships. All these steps are to be discussed in greater details in following sections.
31
Casting Design Designing and Processing Simulation
Production of
Nodularity
test casting Skin Testing
Deliverable
Tensile Testing
Fatigue Testing
Influence of
Tensile Skin
Fatigue Skin
Skin Features
Factor
Factor
Quantification Effect of %Nod. on Fatigue Skin Factor
Figure 25. Overview of the approach
3.1.1 Skin factors Before discussing the details of the experimental work, the concept of skin factor must be introduced. To quantify the effect of casting skin on tensile and fatigue properties, the samples were tested in the as-cast (AC) and machined (M) conditions. The as-cast condition refers to the testing of samples that retain the casting skin. The machined condition refers to the testing of samples on which the casting skin has been machined off. The skin factor is the ratio between the properties in the as-cast condition to the properties in the machined condition. For tensile strength, the tensile skin factor ( defined as
32
) is
where
and
are the tensile strength in the as-cast and machined conditions
respectively.
Similarly, the fatigue skin factor (
When
and
) is defined as,
are fatigue limit in as-cast and machined condition respectively.
These skin factors are unitless numbers. The skin factors approaches unity when the casting skin effect is minimal.
3.1.2 Casting design and simulation The following designing constraints of the test castings were identified: -
To evaluate mechanical properties, the test samples should be free of shrinkage and porosity.
-
The test castings should be suitable for the evaluation of static mechanical and fatigue properties in as-cast as well as in machined conditions.
-
Various thicknesses should be included in the design to obtain various cooling rates in the same casting. This allows observing the effect of cooling rate on the skin thickness and on the corresponding tensile strength. The section thickness is fixed in the design for fatigue samples.
-
Filling of the mold must be quiescent. 33
-
To avoid the mottled structure and to have comparable cooling rates to industrial applications, the characteristic cooling rate of the sample should be maintained under 10 C/s.
A number of designs were produced under these constraints and then evaluated through the casting simulation software until all the requirements were met. Figure 26 and Figure 27 show the final test casting designs for tensile and fatigue test samples used in this study.
3.1.3 Production of test casting and samples All iron castings in this study were melted in coreless induction furnaces. Two 300-lb heats were produced for tensile test castings. An additional eight 120-lb heats were melted for fatigue test castings. To produce compacted graphite and ductile iron graphite shape modification was required. A reaction chamber was designed for this purpose to provide better consistency in Mg control. Figure 28 shows the cross section of the reaction chamber. The basic dimensions of the chamber are summarized in Table 7. A predetermined amount of a FeSiMg alloy was place inside the chamber (shown in Figure 28). Liquid iron was then poured into the chamber and allowed to react with the magnesium alloy at 1450 C. The treated liquid iron flowed into the pouring ladle. The sand used for the molds was silica sand with 55 – 60 grain fineness number ( 250 - 275 µm in diameter). Sodium silicate and phenolic urethane were used as binders for the
34
molds for the tensile test samples. For fatigue test samples, only sodium silicate was used. The pouring temperature was 1400 C.
a)
b)
c)
Figure 26. Test casting design for tensile test samples; a) isometric view of the design; b) simulated cooling rate; c) simulated porosity
As seen in Figure 26, one test casting produces three dog-bone samples with various thicknesses allowing the observation of the effect of cooling rate. Since the samples are in dog-bone shape, they are ready for testing in as-cast condition. For machined samples, the casting skin was removed all around the samples by machining. The dimensions of the tensile samples are summarized in Table 8. 35
a)
b)
c)
Figure 27. Test casting design for fatigue test samples; a) isometric view of the design; b) simulated cooling rate; c) simulated porosity
Di
Lid
L
H
FeSiMg
Filter
a)
Do
b)
Figure 28. Reaction chamber; a) schematic drawing; b) during the operation 36
Table 7. Basic dimensions of the reaction chamber Feature Cavity height Inlet diameter Outlet diameter Cavity length Cavity width
Symbol H Di Do L W
Dimension 7.25” 3.00” 2.00” 9.45” 3.25”
Table 8. Dimension of as-cast and machined tensile samples. Width, mm Thickness, mm Gage Length, mm
As-cast (AC) Machined (M) 19.1 21.6 31.8 11.4 12.9 19.0 7.6 10.2 15.2 4.5 6.0 8.9 50.8 50.8 50.8 30.0 30.0 30.0
The sample preparation procedures for fatigue testing were as follows. Firstly, the plates were split longitudinally in halves and the thickness was reduced from 13 mm to 8 mm. Note that the machined sample (M) were machined on both top and bottom surface (i.e. 2.5 mm on each side); the as-cast samples (AC) were machine on only the top surface (i.e. 5 mm on one side). As seen in Figure 27, one casting produces four plates that are machined to produce eight fatigue samples. All machined surface were finished by grinding up to 1200 grit paper followed by fine polishing with 6 and 1 m alumina paste, respectively. This was done to minimize the effect of surface roughness. The bottom face of the as-cast samples was left unmodified to retain the casting skin.
37
3.1.4 Thermal analysis During the production of test casting, Thermal Analysis (TA) cups were used for controlling carbon and silicon content before the FeSiMg treatment. Figure 29 shows the geometry of the TA cup. The dimension of the cup is 33 × 33 × 37 mm. The cup is made of furan resin bonded sand by hot box process. Thermocouple tip (type K) is at the geometric center of the cavity and wired to the bottom of the cup. Notice that the thermocouple is protected by a quartz tube. During operation, the cup is set on a stand that connected to the data acquisition system. Data are collected and plotted on a computer screen allowing the operator to make necessary adjustment to the melt. The TA cups can also be used for retrieving cooling curves from the treated iron for interpretation.
Figure 29. Thermal Analysis (TA) cup
38
3.1.5 Chemical analysis During the melting and pouring stages, chilled samples were taken for chemical analysis. The chemical analysis was done by optical emission spectroscopy (OES) which is the standard method for chemical control in the metal casting industry.
Since the major alloying elements in iron are carbon and silicon, iron is a ternary alloy. For the sake of convenience, Carbon Equivalent (CE) has been used in the field as a parameter that combines the effect of both carbon and silicon. The empirical equation for CE is,
where %C, %Si and %P are percentage carbon, silicon and phosphorus in iron respectively. The value can be used directly on the binary Fe-Gr phase diagram.
3.1.6 Tensile and fatigue testing All tensile samples were tested with a universal testing machine with a strain rate of 0.5 cm/min. The yield and ultimate tensile strength were recorded for both as-cast and machined samples. An extensometer was attached to the samples to obtain strain data for percent elongation.
The 4-point bending fatigue testing method was used to assess the fatigue strength of the samples. This setup delivers more consistency than the 3-point bending (by Labreque et al. [15]) that has stress concentrated on a contact point. The distance between the lower 39
supports was 80 mm and the distance between the upper supports was 26.7 mm. The dimensions of the samples were 100 25 8 mm. The set up is shown in Figure 30. In this setup, the upper supports were stationary and connected to a load cell while the lower supports were moving. The force-controlled mode was used throughout this study.
8 mm 26.7 mm
26.7 mm
26.7 mm
80 mm
Figure 30. The setup for 4-point bending fatigue testing
During the test, the load was applied from the lower supports providing the bending moment between them. This action creates tensile stress field along the bottom and compressive stress field along the top of the sample. For the as-cast samples, the casting surface was facing down and was subjected to tensile stress. The applied load was only fluctuated in compression and never in the tensile regime.
40
With this setting, the bending stress () can be calculated from the bending moment (M), the distance from the neutral axis (c) and the moment of inertia of the cross section (I) as
Mc I
For a rectangular cross section, the moment of inertia is I
1 3 wt 12
and the bending moment (M) for this particular set up is
The relationship becomes:
PL wt 2
where P is the applied load, L is the distance between the lower supports (80 mm), w is the sample width (25 mm) and t is the sample thickness (8 mm). The frequency was 50 Hz. The mean stress (m) was kept constant at 250 MPa. The cyclic stress (cyclic) refers to the difference in maximum and minimum compressive stress (max – min). The cyclic stress was varied around the mean stress to find the fatigue limits (i.e. increasing Pmax and decreasing Pmin). Samples that lasted longer than 5 106 cycles were marked as the runouts.
3.1.7 Metallography Common metallographic sample preparation was conducted for general observation and the quantification of the casting skin. The procedure consists of grinding from 240 to 41
1200 grit abrasive paper followed by fine polishing with 6, 1 and 0.3 µm alumina pastes. In some cases, a polishing step by 0.05 µm alumina paste was added for better polishing quality. Quantification of the graphite shape factors was done in the unetched condition. Etching by 4% Nital was used to reveal the matrix structures for ferritic and pearlitic rim quantification.
The color metallography technique was used to reveal microsegregation in selected samples. During the etching, the samples were boiled in a solution consisting of 80 g NaOH, 20 g KOH, and 20 g picric acid in 200 ml distilled water at 120 C for 120 seconds. This technique reveals the dendritic structure in the matrix.
3.2 Chemical composition and microstructure Table 9 shows the chemical composition of the compacted graphite iron produced for the tensile testing experiments. It is seen that the Mg level was in range of 0.007 to 0.009% which produces CG iron with 15% nodularity. Figure 31 shows the representative microstructure at the center of the tensile samples for both heats.
Table 9. Chemical composition of compacted graphite iron for tensile testing Heat 90130 90220
C 3.72 3.73
Si 2.57 2.60
Mn 0.11 0.17
P 0.033 0.035
S 0.013 0.012
42
Mg 0.009 0.007
Ce 0.013 0.008
Cu 0.69 0.64
CE 4.50 4.52
90130
90220
Figure 31. Representative microstructures from the center of the tensile samples
As shown in Figure 26, there were three thicknesses for the tensile test samples. This allows observing the effect of the cooling rate. The microstructure near the casting surface with various sample thicknesses is shown in Figure 32. It is seen that the larger thickness tends to have thicker skin thickness. Note that the nodularity of the microstructure near the casting skin is clearly higher than at the center of the sample (Figure 31). The effect of the cooling rate will be discussed in the next section.
The chemical compositions of all heats for the fatigue experiments are shown in Table 10. It is seen that the Mg level was varied to provide a range of graphite nodularity from 15 – 40%. This allows observing the fatigue skin factor as a function of nodularity. Figure 33 shows the representative microstructure from the center of the plates for each heat. As expected, the nodularity increased with increasing Mg level.
43
a) 7.6 mm
b) 10.2 mm
c) 15.2 mm Figure 32. Optical micrographs of compacted graphite iron near the casting skin with various casting section thickness
Table 10. Chemical composition of compacted graphite iron for fatigue testing Heat 100701.1 100701.2 100707.1 100707.2 101015 101101 101108.1 101108.2
C 3.21 3.10 3.37 3.33 3.25 3.38 3.56 3.54
Si 2.62 2.51 2.46 2.48 2.19 2.18 2.23 2.21
Mn 0.046 0.046 0.061 0.063 0.080 0.033 0.033 0.023
P 0.008 0.012 0.024 0.024 0.030 0.039 0.033 0.033
S 0.009 0.010 0.010 0.009 0.007 0.006 0.006 0.006
Mg 0.020 0.017 0.019 0.020 0.013 0.010 0.008 0.008
Cu 0.034 0.035 0.024 0.024 0.022 0.216 0.098 0.112
CE 4.08 3.94 4.19 4.16 3.98 4.11 4.30 4.28
%N 40 40 40 40 30 15 15 15
%N is percentage nodularity
44
100701.1
100701.2
100707.1
100707.2
101015
101101
101108.1
101108.2
Figure 33. Representative microstructures from the center of the plates for different heats 45
Figure 34 shows the microstructures at and adjacent to the casting surface for the 15, 30 and 40% nodularity CG irons.
a) 15% nodularity
b) 30% nodularity
c) 40% nodularity Figure 34. Representative microstructures underneath the casting surface
3.3 General observations of the casting skin features 3.3.1 Tensile samples To evaluate the skin quantification procedure, a correlation analysis was conducted. Results of correlation analysis for the skin thickness obtained from image analysis and visual measurements are presented in Table 11. The higher numbers indicate better correlation. A perfect correlation will return the number 1. A poor correlation will return 46
values less than 0.5. As seen in the table, all values show very good correlations (higher than 0.75) between the different values of skin thickness. This supports the validity of measurements. Notice that the good correlation between the visual skin thickness and the skin thickness measured by image analysis.
Table 11. Correlations between quantities defining the skin thickness
%Graphite Sphericity Roundness Compactness Aspect ratio %Pearlite Visual skin thickness
Graphite Area 0.96 0.85 0.83 0.75 0.83 0.78 0.84
%Graphite 1.00 0.82 0.89 0.79 0.80 0.81 0.86
Sphericity
Roundness
Compact- Aspect ness ratio
1.00 0.90 0.81 0.90 0.78 0.83
1.00 0.91 0.90 0.87 0.90
1.00 0.81 0.85 0.85
1.00 0.80 0.84
%Pearlite
1.00 0.85
Table 12 shows the correlation between the visual skin thickness and some parameters. Cooling rates were taken from simulations (Figure 26b and Figure 27b). As seen, the skin thickness increased with increasing cooling rate. This supports the observation discussed earlier. The effect of section thickness on skin thickness can be explained by the Mg depletion mechanism which will be discussed in Chapter 4. Furthermore, the distance of maximum graphite shape factors and visual nodularity (i.e. location of Zone I mentioned in Chapter 2) correlates well with the visual skin thickness. This was expected because the larger casting skin should produce larger affected area.
47
The roughness is the only surface feature of casting skin and should be considered separately from subsurface features. When all samples were included in the analysis, roughness did not correlate well with any of the variables. However, roughness was found to be a function of metallostatic height and binder type. As shown in Figure 35, the roughness increases with the metallostatic height, a trend consistently observed in other research. This is because a higher metallostatic pressure increasingly overcomes the ability of the capillary pressure to prevent mechanical penetration. It is also noticed that the plates cast in sodium silicate bonded sand exhibit higher roughness (20.6±1.0 µm) than those produce in the phenolic urethane bonded sand mold (12.3±3.0 µm).
Table 12. Correlations between visual skin thickness and other parameters
Cooling rate Distance of Maximum Shape Factors Distance of Maximum Visual Nodularity
Skin thickness (visual) 0.45 0.55 0.37
Cooling rate 1.00 0.77 0.77
The type of binder also affects the skin thickness and the pearlitic rim (Figure 36). The values shown in the figure were the average on the as-cast samples. A slight difference in the skin thickness was observed between binder types. As for the pearlitic rim, only few samples cast in phenolic urethane bonded sand molds exhibited pearlitic rim. On the other hand, pearlitic rim was readily found on the plates produced in sodium silicate bonded sand molds. In order to have a clear effect of the type of binder, a composite mold was produced with one half with sodium silicate sand and the other with phenolic urethane. Figure 37 shows a comparison of the microstructures of the 10.2 mm plate from 48
the composite mold. It is seen that about 0.08 mm of pearlitic rim occurred on the sodium silicate side, but no pearlitic rim was found on the phenolic urethane side.
Ra, mm
Phenolic Urethane
0.03
Sodium Silicate
0.02
Linear (Phenolic Urethane) Linear (Sodium Silicate)
0.01
0.00 70.0
90.0 110.0 130.0 150.0 Metallostatic Height, mm
Figure 35. Influence of metallostatic height and type of binder on roughness
0.20
Skin Thickness Pearlitic rim
Thickness, mm
0.16 0.12 0.08 0.04 0.00 Phenolic Urethane
Sodium Silicate
Binder Type
Figure 36. Effect of type of binder on skin thickness (visual) and pearlitic rim thickness 49
Sodium silicate
Phenolic-urethane
Figure 37. Effect of type of binder on the pearlitic rim; 4% Nital etched
3.3.2 Fatigue samples The roughness average (Ra) of the as-cast samples ranged from 16 to 26 m, while the Ra of the machined samples was less than 0.09 m. The averages of Ra for as-cast and machined samples are 20.11.95 and 0.080.01 m, respectively. The latter numbers are the standard deviations. Note that there is only a slight difference in Ra between heats.
The averages of the skin thickness were 165.6, 113.9 and 84.1 m for 15, 30 and 40% nodularity CG iron, respectively. Apparently, the higher residual Mg resulted in thinner casting skin. This is because more Mg available in the melt provides a smaller Mg depletion depth. This is to be discussed in greater detail in Chapter 4.
50
In present study, a pearlitic rim was found in all samples. The averages thicknesses of the pearlitic rim were 139.027.3, 86.118.1 and 55.910.9 m for 15, 30 and 40% nodularity CG iron, respectively where the latter numbers are the standard deviations.
3.4 Effect of casting skin on tensile properties To assess the influence of casting skin on the tensile properties, the correlation analysis was run on all experimental data (as-cast and machined). The results are presented in Table 13.
It is seen that the tensile strength (TS) correlates well with roughness average (decreases with Ra) the thickness of the casting skin (TS decreases with tSK-vis) and the thickness of the pearlitic rim (TS decreases with pearlitic rim thickness). The trends are similar for elongation (EL) but the data scatter is larger and the correlations are lower. The section thickness of the test plates did not influence significantly any of the analyzed quantities. A poor correlation is noticed between tensile strength and nodularity (graphite compactness, Ncpct, or graphite sphericity, Nsph). As will be shown later, this is because two distinct groups of samples are included, machined and as-cast, which have different levels of strength. Also, the position of the maximum nodularity across the plate does not affect the properties.
51
Table 13. Correlation Analysis of all data
Elongation Section thickness Cooling rate Roughness average (Ra) Skin thickness by %graphite (tSK-%Gr) Skin thickness by compactness (tSK-cpct) Skin thickness by %pearlite (tSK-%Pe) Skin thickness by visual inspection (tSK-vis) Distance of maximum nodularity by shape factor (max Nshpfact) Distance of maximum nodularity by shape factor visual inspection (max Nvis) Graphite sphericity (Nsph) Graphite compactness (Ncpct) Nodularity by sphericity by visual inspection (Nvis) Percent of pearlite (%Pe) Thickness of pearlitic rim (tPe-rim)
Tensile Strength (TS) 0.77 -0.18 0.23 -0.73 -0.89 -0.82 -0.75 -0.90 -0.13
Elongation (EL) 1.00 -0.15 0.12 -0.62 -0.63 -0.63 -0.53 -0.71 0.01
-0.02
0.10
0.29 -0.02 0.26 0.12 -0.66
0.26 0.03 0.23 0.03 -0.58
The effect of the casting skin thickness on the tensile properties of CG iron is demonstrated through Figure 38. It is seen that both tensile strength and elongation decrease as the skin thickness increases. Again, these data confirm the correlations presented earlier in Table 13.
The influence of surface roughness on the tensile properties can be seen in Figure 39. It is seen that the higher surface roughness decreased the tensile strength. This can be explained by the surface roughness providing the notched effect for stress concentration. Also, the surface roughness creates tensile instability causing localized yielding and necking. Nevertheless, while the trend is clear, a large data scatter persists, as indicated
52
by the regression coefficients. The main reason for this is believed to be the narrow range of roughness in this research.
Figure 38. Effect of the thickness of the casting skin on the tensile properties of CG iron
Regression analysis is a statistical method that establishes mathematic relationships between an independent variable and one or more dependent variables. It was used to derive an equation for the tensile skin factor (SFTS). In this research the following variables were used: -
independent variables: tensile strength (TS), elongation (EL);
-
dependent variables: cooling rate, average roughness (Ra), thickness of the casting skin by visual inspection (tSK-vis), nodularity by visual inspection (Nvis), percent of pearlite (%Pe), and thickness of the pearlitic rim (tPe-rim).
53
Tensile strength, MPa
350
15 12
R2 = 0.5296 9 6
300
Elongation, %
TS Elong Linear (Elong) Linear (TS)
400
3 R2 = 0.3817 250 0 0.000 0.005 0.010 0.015 0.020 0.025 Roughness average, mm
Figure 39. Influence of roughness on the tensile properties of as-cast and machined test plates
Regression analysis for tensile strength using as-cast data for both heats and the visually estimated skin thickness yielded the results presented in Table 14. Using the coefficients in the table the following equation for tensile strength in as-cast condition (TSAC) can be written. TSAC = 324.4 + 1.26Cool rate - 1624Ra - 85.1tSK-vis + 1.58·Nvis + 0.29%Pe+108.8tPe-rim
Note the high R2 of 0.92. This equation can be used to predict the as-cast tensile strength from metallographic information (the thickness of the casting skin, the nodularity, the pearlitic rim and the amount of pearlite) and roughness testing. For a casting without casting skin (e.g. machined) the equation predicts an average tensile strength of 352MPa, while the average tensile strength of the as-cast plates with skin is 338MPa. 54
Table 14. Regression analysis for tensile strength (as-cast data) Equation Intercept Cool rate Ra tSK-vis Nvis %Pe tPe-rim Multiple R 0.96
Coefficient Standard Error 324.41 7.24 1.26 0.59 -1624 495.58 -85.14 23.33 1.58 0.62 0.29 0.39 108.77 92.15 Regression Statistics R2 Standard Error 0.92 6.4
P-value 0.00 0.05 0.00 0.00 0.02 0.48 0.26 Observation 23
The plot for tensile strength versus visual skin thickness (Figure 40) shows good correlation between the measured and calculated data.
Measured TS
Tensile strength, MPa
390
Predicted TS
370 350 330 310 290 270 250 0.0
0.1
0.2
0.3
0.4
0.5
Skin thickness (visual), mm
Figure 40. Plot for regression data in Table 14
55
The average tensile strength on machined plates is 355MPa; while on the as-cast plates, it is 325MPa. This yields the average skin factor SFTS is 0.91. The value indicates that on the average the casting skin, the tensile properties decreases by 9% as compared with machined samples.
The previous regression analysis was done based on unit of each parameter (e.g. millimeter for skin thickness, MPa for tensile strength). This does not allow a clear understanding of the relative weight of the variables. Thus, it is preferable to use normalized data. A normalized value is the value of the variable divided by a reference value chosen such that the range of the normalized values is between 0 and 1. For example, tSK-vis can be normalized by dividing its current value by the maximum value of tSK-vis in the data range of interest.
The results of regression analysis conducted with normalized values for the visually evaluated casting skin are presented in Table 15. The reference normalization factors (the denominators) were as follows: cooling rate = 12 K/s, Ra = 0.025 mm, tSK-vis = 0.39 mm and Nvis = 25%. The analysis was repeated several times, eliminating the variables that generated high P-values. It is now seen that the highest influence on tensile strength is that of nodularity (the coefficient for Ncpct is 38.61) followed by that of the skin thickness (the coefficient of -35.24).
56
Table 15. Regression analysis for tensile strength (normalized as-cast data) Equation Intercept Cool rate Ra tSK-vis Nvis Multiple R 0.96
Coefficient Standard Error P-value 329.28 5.82 0.00 12.01 5.81 0.05 -30.25 8.47 0.00 -35.24 8.67 0.00 38.61 11.48 0.00 Regression Statistics R2 Standard Error Observation 0.92 6.29 23
3.5 Effect of casting skin on fatigue properties Cyclic stress-Number of cycles to failure plots (also known as S-N curves) for 15, 30 and 40% nodularity CG iron are shown in Figure 41. The testing was done at the mean stress of 250 MPa. Although, the surface roughness and pearlitic rim were intentionally kept at the same levels throughout the test, some variations of these features are expected to contribute to the data scattering. It is also seen that the machined samples exhibit higher fatigue limit than the as-cast samples. This is expected because the as-cast samples have higher Ra and undesired casting skin features. The roughness average of as-cast samples was considerably varied in small range and should not have significant impact on the fatigue limits.
By recalling the definition of skin factor, the fatigue skin factor (SFfat) is then a dimensionless number that ranges from 0 to 1. The lower fatigue skin factor indicates more pronounced casting skin effect. Table 16 summarizes the fatigue skin factors from the experiments. It should be noted that the values for gray iron (0% nodularity) and the 57
values for ductile iron (80-100% nodularity) in the table were obtained from Kuwamoto et al. [22] and Labrecque et al. [15] respectively. The sample geometries and testing methods were different in those studies. However, the values are included here to demonstrate the effect of nodularity on the fatigue skin factor (Figure 42). It is seen that lower nodularity causes less damage to fatigue properties. To explain the effect, one may think of the casting skin as a layer of gray iron which has little effect for low nodularity CG iron and more effect for high nodularity CG iron.
However, Labrecque et al. [15] reported fatigue skin factors as high as 0.83 for ductile iron (one-side machined samples versus machined samples). The reason for the higher fatigue skin factors could be rationalized as follows. Firstly, the casting skin in ductile iron is normally thinner than in CG iron. This can be explained through the Mg oxidation model. Since ductile iron has more magnesium available in the melt; the magnesium depletion layer is thinner. Secondly, large undercooling at the mold/metal interface promotes dendritic growth and results in higher fraction of austenite (lower fraction of graphite). As a result, the casting skin of ductile iron can be nearly free of graphite flakes and thinner than for CG iron. Such casting skin would cause less damage to fatigue properties. The presence of surface roughness may be the main contributor to the reduction of the fatigue limit. This hypothesis could be confirmed by metallographic observation. Unfortunately, characterization of the casting skin was not done in the studies.
58
500
500
AC M
M
400
400
Cyclic Stress, MPa
Cyclic Stress, MPa
AC
300
200
100 1.0E+04
300
200
100
1.0E+05
1.0E+06
1.0E+07
1.0E+04
1.0E+05
Nf, cycles
1.0E+06
1.0E+07
Nf, cycles
a)
b) 500 AC M Cyclic stress, MPa
400
300
200
100 1.0E+04
1.0E+05
1.0E+06
1.0E+07
Nf, cycles
c) Figure 41. S-N curves of as-cast (AC) and machined (M) sample; a) 15% nodularity; b) 30% nodularity; c) 40% nodularity; m = 250 MPa
59
Table 16. Summary of the fatigue skin factor *
Fatigue limit, AC Fatigue limit, M Fatigue skin factor
0% 73.5* 78.4* 0.94
Nodularity 15% 30% 40% 185.9 193.6 215.0 218.5 266.1 313.9 0.85 0.73 0.68
80%** 384** 459** 0.83
*
Values for gray iron from Kuwamoto et al.
**
Values for ductile iron from Labreque et al.
1.0
Present study Kuwamoto et al
Fatigue Skin Factor
0.9
Labreque et al
0.8 0.7 0.6 0.5 0% 20% 40% 60% 80% 100% Nodularity
Figure 42. Fatigue skin factor as a function of nodularity
The fatigue strength depends on the mean stress used. Thus, fatigue testing was also performed using a higher mean stress of 300 MPa. The results are shown in Figure 43. As expected, increasing the mean stress lowers the fatigue strength. It is seen that the fatigue 60
strength of as-cast and machined samples decrease by 11.3 and 10%, respectively. The SFfat was calculated to be 0.65.
500
500 AC M
AC
400 Cyclic stress, MPa
400 Cyclic stress, MPa
M
300
200
100
300
200
100
1.0E+04
1.0E+05
1.0E+06
1.0E+07
1.0E+04
Nf, cycles
1.0E+05
1.0E+06
1.0E+07
Nf, cycles
a)
b)
Figure 43. Comparison of S-N curves tested at different mean stress; a) m = 250 MPa; b) m = 300 MPa
The relationship between the mean stress and fatigue strength was then plotted in Figure 44. The normal bending test was conducted for both as-cast and machined samples and used as data points on the extremes (Nf = 1/2 cycle, a = 0, m = bending strength). The average bending strength of as-cast and machined samples are 647.3 and 765.7 MPa, respectively. As linear relationships are seen, the empirical equations for fatigue strength prediction at a given mean stress are,
fat,AC = -0.5356m + 346.51
for the as-cast samples and,
fat,M = -0.6059m + 463.89
for the machined samples. 61
Figure 44. Relationship between the cyclic stress and the mean stress of as-cast and machined samples
The fracture surfaces are free of surface defects except for one sample (marked as in Figure 41c). Note that cracks always originate from the corner of the samples. Figure 45 shows a representative fracture surface for this study. The casting skin is at the top. The fracture surfaces show two distinct zones which are crack propagation zone (bright) and sudden rupture zone (dark).
Figure 45. The representative fracture surfaces of as-cast samples 62
The fracture surface of the sample that exhibits an unusual short fatigue life is shown in Figure 46. As seen, there is a pinhole close to the casting surface which initiated the crack. This sample exhibited 67% shorter fatigue life than its replacement.
Figure 46. A pin hole close the casting surface
Figure 47 shows secondary electron images at the casting skin, crack propagation zone, striations and sudden rupture zone. A cleavage fracture is seen in the casting skin and crack propagation zone, while a ductile-dimple fracture is seen in the sudden rupture zone. This demonstrates the different fracture mechanisms.
At high magnification, striations, which are a signature of fatigue failure, can be seen in the crack propagation zone as well. At the casting skin Figure 47a), a layer graphite depletion/degradation is seen at the top surface followed by the high nodularity zone. The thickness of the layer (0.075 mm) also agrees well with the optical microscopy observation. It should be noted that the secondary electron images shown in Figure 47 are 63
taken from the proximity of the casting surface, therefore, the apparent nodularity seems to be higher than 40%.
a)
b)
c)
d)
Figure 47. Secondary electron images of an as-cast sample; a) at the casting skin; b) cleavage fracture in crack propagation zone; c) ductile-dimple fracture in sudden rupture zone; d) striations in crack propagation zone
64
3.6 Conclusions Good correlations between the skin thicknesses evaluated by different parameters were shown in this study. The measurements appeared to correlate well with the visually evaluated skin thickness. This indicates that the image analysis through graphite shape factors is indeed a viable method for skin quantification.
The average tensile skin factor was found to be 0.91, indicating an average reduction of the tensile strength of 9%. The maximum skin thickness observed in this research was of 0.4 mm. The average tensile strength decreased from 355 MPa on the machined specimens to 300MPa on plates with 0.4 mm skin. This gives a skin factor of 0.845 (or 15.5% decrease).
Regression analysis was used to establish an equation for tensile strength prediction as a function of several parameters such as skin thickness, surface roughness, amount of pearlite and nodularity. The equation indicates that the tensile strength significantly decreases with increasing skin thickness and roughness. Higher nodularity and pearlite content result in higher strength.
The presence of the casting skin was proven to reduce the fatigue strength of CG iron. The fatigue skin factors were 0.85, 0.73 and 0.68 for 15, 30 and 40% nodularity CG iron, respectively. It was seen that the lower nodularity CG iron is less damaged by the presence of the casting skin. 65
Chapter 4: Mechanism of Casting Skin Formation The second goal of this research is to understand the mechanism of casting skin formation. This would lead to the minimization of the occurrence of the casting skin. Proposed mechanisms available in literatures are to be reviewed in this chapter along with new mechanisms hypothesized during this research are included.
4.1 Surface roughness Surface roughness is the result of liquid metal penetration into the sand mold. It can be understood in terms of a pressure balance between the metallostatic pressure, Pst, that pushes the liquid between the sand grains, and the capillary pressure, Pγ, that opposes it [23]. If Pγ > Pst, a mild interaction results, with the metal being prevented from penetrating beyond the first layer of sand grains (Figure 48). If Pγ < Pst, the result is severe interaction and the metal penetrates deeper in the mold resulting in rougher casting surface.
The pressure terms can be calculated as: Pst = gh where is density; g is the gravity acceleration; h is the metallostatic height. Pγ=-2LV cos / dp 66
where LV is surface energy between liquid and vapor phase; is the wetting angle; dp is interparticle spacing.
These equations suggest that the surface roughness will increase with the metallostatic height and grain coarseness, a statement has been confirmed experimentally (Figure 35).
Pγ > Pst
sand
Pγ < Pst
metal
sand
metal
Figure 48. Molding sand and liquid iron interaction; mild interaction (left) severe interaction (right)
Surface roughness can also be decreased by mold coating. The coating materials are typically made by non-wetting materials (large ) with finer grain size (smaller dp). The resulting condition creates higher capillary force (P) in which suppressing the metal penetration.
4.2 Ferritic rim and Graphite depletion Both features share the same formation mechanism. The occurrence of a decarburized skin on gray iron castings was documented as early as 1962. Reisener [5] suggested that 67
the phenomenon is caused by the chemical reaction of the carbon in the iron with the moisture and oxygen in the mold (Figure 49). The direct reaction of C with H with formation of CxHy is less probable than the reaction of C with water vapors. Thus the probable chemical reactions are: C + ½ O2 → CO C + H2O → CO + H2
These chemical reactions could lead to the depletion of carbon in the casting skin. Obviously, this effect is more pronounced in green sand molds because of the inherent moisture. The decarburized layer can appear as a ferritic rim or in more severe case, a graphite depleted region. Matijasevic et al. [6] reported that using seacoal as an additive in green sand molding reduced the ferritic rim thickness in gray iron. The adverse influence of increasing the Si content on the ferritic rim thickness was reported by Narasimha Swamy et al. [8]. This is expected as Si is a strong ferrite stabilizer.
It has been reported that the graphite depletion thickness in gray iron increased with the section size. Rickards [7] found that the graphite depletion thickness also depended on the square root of time in the temperature range 1150 – 1050 C during solidification.
68
Figure 49. Interfacial mold/metal reactions creating decarburized layer on iron casting surface [5]
Figure 50. Relationship between the depth of the graphite-free layer and the time in the temperature range 1150 – 1050 C for grey iron cast in green sand mold [7]
69
The other mechanism of the graphite depletion is the austenite layer formation because of solidification kinetics. As the austenite formation can be induced by large undercooling, the result is the greater percentage of the austenite (i.e. lower percentage of graphite); hence the graphite depletion. A clear evident that supports this theory can be seen through a recent finding by Suarez [24]. The significant formation of the graphite depletion was observed in DI poured in a steel mold. The steel mold created large undercooling in the metal layer adjacent to mold because of its high thermal diffusivity. The resulting microstructure was a well define graphite depletion layer (austenite layer) at the casting skin (Figure 51). Note that as the steel mold has no permeability or moisture, decarburization is improbable. Therefore, the result cannot be explained by decarburization. A comprehensive explanation through the asymmetric phase diagram on this aspect will be provided in the Section 4.5.
a) Before etching
b) After etching
Figure 51. Occurrence of graphite depletion in DI cast in permanent mold [24] 70
4.3 Pearlitic rim Contrary to the formation of the ferritic rim, the pearlitic rim occurs because of carburization at the casting skin. The suggested mechanisms [11] for pearlitic rim formation are: -
Oxidation of carbon in the liquid metal by the oxygen or water vapors in the mold atmosphere, resulting in lower number of nuclei for ferrite nucleation and growth at the A1 transformation temperature.
-
Carburization of the subsurface layer with carbon from the molding material. Diffusion calculation for carbon from the mold [11] predicted reasonable pearlitic rim thickness when compared to microscopic observations.
4.4 Graphite degradation Several mechanisms were proposed to explain graphite degradation at the surface of iron castings. They are all based on the premise of Mg depletion in the subsurface layer of the casting. This causes graphite degradation through the transition from spheroidal graphite– to-compacted graphite–to-lamellar graphite. Two causes for Mg depletion are cited: MgS reaction and Mg – O reaction.
4.4.1 Mg depletion by sulfur reaction It is known that sulfur combines with Mg to form MgS. Consequently, less Mg is available in the melt for graphite spheriodization. This results in lower nodularity when 71
Mg/S is high. The source of sulfur is typically from raw materials used during melting operation. For the casting skin formation, additional source of sulfur can be from binder and catalyst used. It has been discussed in Section 1.5.4 that the combustion of the Para Toluene Sulfonic Acid (PTSA) used as a catalyst in furan and phenolic resin systems generates SO2 which is responsible for the Mg – S reaction in the iron
To minimize the effect of sulfur from the molding materials, mold coating with desulfurizer such as CaO or MgO may be beneficial. The reactions for desulfurization of liquid iron by CaO suggested by Voronova [25] are as follows: CaO + [S] → CaS + [O] CaO + [S] + C→ CaS + CO CaO + [S] + ½[Si]→ CaS + ½SiO2 2CaO + [S] + ½[Si]→ CaS + ½(2CaO·SiO2)
Therefore, a CaO coating should decrease the thickness of the casting skin when sulfur is present in the molding material. In fact, the favorable effect of CaO for furan bonded molds was confirmed through the experiments (more details in Chapter 5), where the skin thickness of CG iron thermal analysis cups (furan binder) was reduced from about 145 µm to about 110 µm when a CaO coating was applied to the inside of the cup (Figure 52). However, in the sodium silicate bonded sand, the CaO coating produced thicker skin than the no-coating condition because of the higher thermal conductivity of the CaO coating. 72
250
Skin Thickness, mm
200
150
100
50
0 NC
FeSiMg
MgO
CaO
Zr based
Figure 52. Measured skin thickness of thermal analysis cups with various types of coatings (CE = 4.23)
The effect of the MgO coating was different than that of the CaO. This is because the thermodynamics of desulfurization by MgO is not favorable. Table 17 shows the heat of formation of MgO, MgS, FeS and MnS at 298 and 1500 K. It is seen that MgS is favored over MgO and FeS at room temperature. At 1500 K, however, MgO formation is favored. This indicates that MgO is an ineffective desulfurizing agent. In addition, MgO has very high thermal conductivity which causes an adverse effect on the skin formation. Results in Figure 52 and Figure 53 confirm this analysis. Note that in both cases, addition of Mg in the skin through a FeSiMg coating significantly decreased skin thickness.
73
Table 17. Heat of formation of some compounds at 298 and 1500 K (adapted from Voronova [25]) Compound
Hf,298K (kJ/mol)
Hf,1500K (kJ/mol)
MgO
-604.0
-738.8
MgS
-4198.3
-556.5
FeS
-158.8
-114.7
MnS
-269.6
-275.5
400 NC
Skin thickness, m
350
CaO
300
MgO
250
FeSiMg
200 150 100 50 0 0
10
20
30
Plate thickness, mm
Figure 53. Effect of section thickness, cooling rate and mold coatings on the skin thickness (CE = 4.23)
4.4.2 Mg depletion by Mg oxidation For sodium silicate bonded sands, there is no sulfur used in either resin or catalyst. Therefore, Mg oxidation should be the main contributor to the Mg loss. 74
Tinebar and Wilson [26] reported that water-bearing binder system produced vermicular or flake graphite on the casting surface of ductile iron castings. They concluded that the moisture in the mold created oxidizing conditions which induced a loss of Mg.
Mg depletion was documented by Goodrich and Lobenhofer [9]. Electron Probe Micro Analysis (EPMA) revealed the absence or reduced level of Mg in the surface region. The Mg oxidation was suggested as the cause of Mg depletion. However, there was no microstructure-measurement-related representation within the work.
In order to confirm the theory, an experiment was set up. The objective of the experiment was to prove the existence of Mg depletion at the casting skin and possibly to explore the nature of Mg segregation. A CG iron sample was evaluated for microsegregation of various elements using EPMA technique. Table 18 shows the chemical composition of the sample analyzed. The sample was cast in a mold that had sodium silicate bonded sand on one side and phenolic urethane bonded sand on the other side. This allowed observing the difference between the binders with identical iron composition. The parameters used for element mapping are as followed, -
Accelerating voltage: 15 kV
-
Beam current:
150 nA
-
Dwell time:
200 msec
-
Map dimension:
512 173 pixels 75
Table 18. Chemical composition of the sample investigated with EPMA Sample before treatment after treatment after correction
C 3.81 3.68 3.70
Si 2.42 2.64 2.65
Mn 0.17 0.17 0.17
P 0.037 0.035 0.035
S 0.012 0.012 0.013
Mg 0.001 0.013 0.012
Ce 0 0.024 0.021
Cu 0.71 0.71 0.71
Cr 0.027 0.027 0.028
Figure 54 shows a backscattered electron image (BEI) along with solutal distribution maps for C, Fe, Si and Mg at the casting skin. The brighter areas seen in the maps indicate the richer content of the elements and vise versa. As expected, the C map shows C rich area where the graphite particles are while the Fe map shows the opposite. The Si map shows Si rich area in the dendritic area around graphite particles and Si depleted area in the interdendritic areas. Unfortunately, it was not possible to see the contrast in the Mg map, because a longer dwell time is required. This would make the total mapping time impractically too long.
However, measuring of the Mg content can be done with line profiling instead of mapping. The line profile was obtained by the same parameter settings as in mapping but with the longer dwell time (5 seconds instead of 200 milliseconds) for better signal-tonoise ratio. Each line profiling has 100 data points. Figure 55 shows the Mg profile from the casting surface to 500 m inward. The profile is compared to the Si map; the redbroken line indicates where the profiling was done. The red squares represent the readings and the green line shows the running average of five readings. As seen, Mg seems to have reverse distribution to Si which means Mg tends to be rich in the interdendritic areas and depleted in the dendritic area. 76
BE image
100 microns
C
100 microns
Fe
100 microns
Si
100 microns
Mg
100 microns
Figure 54. EPMA maps show solutal distribution at the casting skin 77
In addition, two line profiles of Mg were compared side by side to the color-etched microstructure of the measured area (Figure 56). As seen, the Mg level in the interdendritic area (marked by letters) is higher. The Mg depletion in the casting skin was again observed. No significant difference between the binders was seen. The experiment proves the coexistence of graphite degradation and Mg depletion.
Mg oxidation mainly occurs in liquid and liquid-solid (mushy) states when the diffusion rate is significant. The most likely reactions are as follows: Mg + CO2 → MgO + CO Mg + H2O → MgO + H2 Mg +1/2O2 → MgO
The Gibbs free energies of formation of these reactions are given in Table 19. It is seen that the second reaction has the lowest Gibbs free energy of formation which makes it the most favorable reaction. However, this reaction can only occur when moisture is available. In dry sand with little moisture, such as sodium silicate bonded sand, the third reaction must be considered as the predominated reaction.
78
Si
100 m
m ap
0.040 0.035
Mg, %wt
0.030 0.025 0.020 0.015 0.010 0.005 0.000 0
100
200
300
400
500
Distance from surface (m)
Figure 55. Line profile of Mg from casting surface to 500 m inward
A
C
B
0.040
0.040
0.030
0.030
Mg, %wt
Mg, %wt
D
0.020 0.010 0.000
E
F
0.020 0.010 0.000
0
100
200
300
400
500
0
Distance from surface (m)
100
200
300
400
Distance from surface (m)
a) Phenolic urethane
b) Sodium silicate
Figure 56. Line profile of Mg from casting surface to 500 m inward 79
500
Table 19. Gibbs free energy of formation of Mg oxidation reactions at 1600 K Reaction Mg + CO2 → MgO + CO Mg + H2O → MgO + H2 Mg +1/2O2 → MgO
Gf, kJ/mol -285.6 -794.3 -554.0
Once Mg is removed from the surface layer of molten iron through the formation of MgO, additional Mg will be supplied from the bulk iron by diffusion. Thus, this is a diffusion controlled process.
Higher pouring temperature is expected to produce thicker casting skin because of the longer time and higher diffusion rate of oxygen. For the same reason, castings with larger section thickness that have longer solidification times are expected to exhibit thicker skin. This correlation has been confirmed experimentally by Duncan and Kroker [17] and also by our work, as further discussed in Chapter 5. Inactive coatings should reduce the diffusion flux. However, the thermal conductivity of the coating materials plays an important role as well.
While the Mg depletion theory explains successfully many of the experimental findings, it fails to explain all the observations. For example, the occurrence of casting skin at the iron/quartz interface in a thermal analysis cup shown in Figure 57 cannot be explained through depletion of Mg, as there is no evidence of any quartz-iron reaction on the picture, and no oxygen or sulfur that can diffuse from the quartz to the iron. 80
Figure 57. Casting skin in CG iron at the quartz/iron interface
The Mg oxidation theory also fails to explain the effect of carbon equivalent and of the inactive or active coatings that do not affect the oxygen or sulfur level at the interface.
4.4.3 Diffusion models for Mg depletion zone prediction Since Mg oxidation has been confirmed as the main mechanism for the graphite degradation. It is possible to create a diffusion model that predicts the Mg depletion zone which then translated to the graphite degradation. Therefore a 1-D diffusion model was created with following assumptions. -
Mg sink at the mold/metal interface by reaction with oxygen and sulfur
-
no liquid convection in the calculation domain
-
0.019%Mg initial Mg content
-
Constant Mg diffusivity in liquid iron
81
-
Diffusion time was given by solidification time simulated separately using a commercial software package
The governing equation was the Fick’s first laws of diffusion: DL(2C/x2)= C/t where DL is Mg diffusivity in liquid Fe, C is composition, x is distance and t is time for diffusion.
An important element of the database required by the Mg diffusion model is the diffusivity of Mg in liquid iron. Table 20 shows experimental data of diffusivity of impurities in liquid Fe at 1600 C [27]. These data suggest that the diffusivity of alloying elements in liquid iron is related to their solubility in solid iron ( austenite). As seen, the higher the solubility in solid iron, the lower is the diffusivity in liquid iron.
The liquid diffusivity in liquid iron for the elements at a given temperature in Table 20 can be estimated with an Arrhenius equation: DL = D0 exp(Q/RT) where Q is the activation energy, R is the gas constant and T is temperature. These data are plotted in Figure 58 along with calculated diffusivity at 1350 and 1600 C and the power regression equations. For the temperature of 1350 C the predictive equation for diffusivity as a function of solubility is: DL 4 10 9 solubility
0.7416
82
Table 20. Diffusivity of various elements in liquid iron at 1600 C [27] Element Co Mn Ni Cr C Si Mo V Nb W Zr Ti Ru
DL (m2/s) 4.00·10-10 3.00·10-10 4.00·10-10 9.00·10-10 3.70·10-9 2.40·10-9 3.20·10-9 4.10·10-9 4.60·10-9 5.90·10-9 8.10·10-9 1.38·10-8 8.00·10-10
Solid solubility in Fe (at%) 50 50 50 12 8.6 4.2 1.6 1.6 1.4 1 0.5 0.7 30
Do (m2/s) 5.83·10-8 3.12·10-8 5.71·10-8 1.17·10-7 1.57·10-7 7.6·10-8 1.52·10-7 1.11·10-7 2.55·10-7 2.76·10-7 2.25·10-7 3.1·10-7 8.53·10-8
Q (J/mol) 7.99·104 7.32·104 7.91·104 6.74·104 5.77·104 5.4·104 6.07·104 5.15·104 6.36·104 6.02·104 5.19·104 4.81·104 7.11·104
Taking the solubility of Mg in solid iron as 0.17 at% at 1350 C as per Massalski et al. [28] and using the above equation, the estimated value for the diffusivity of Mg in liquid iron is 1.49·10-8 m2/s.
D exp 1600 D cal 1600 D cal 1350 Pow er (D exp 1600) Pow er (D cal 1350)
2
Diffusivity, m /s
1.E-07
1.E-08
-0.6916
y = 6E-09x
1.E-09 y = 4E-09x-0.7416 1.E-10 0.1
1
10
Solubility, at%
100
Figure 58. Relationship between impurity diffusivity in liquid iron and solid solubility at 1350 and 1600 C 83
Figure 59 shows the comparison of measured and calculated Mg level. Both predict a degenerated graphite skin thickness of about 2.3 mm, which agrees well with the observation by Stefanescu et al. [11].
0.025 CG
Mg, %
0.020 0.015 CG+flake
0.010 0.005 0.000 0
2 4 6 8 Distance from surface, mm
10
Figure 59. Comparison of experimental data (squares) and diffusion model prediction (solid line) in CGI casting
It has been shown that the 1-D diffusion model based on Mg oxidation could predict the graphite degradation thickness. However, the model had a number of limitations, as follows: -
The 1-D model could not include the effect of convergent and divergent heat and mass flow at the corners.
-
The diffusivity of Mg was taken as a constant (1.49·10-8 m2/s at 1350 C)
84
-
The solidification time was predicted separately through commercial solidification software and was taken as the diffusion time.
A new 2-D thermal-diffusion model was formulated with the following settings: -
To replicate the standard size of cooling cups for thermal analysis (33 × 33 mm), the cross section of the model was 17 × 17 mm, surrounded by 8 mm of sand. Symmetry planes were used to reduce the computation load (Figure 60).
-
The thermal field was simulated concomitantly with the diffusion field providing the temperature distribution at all time-steps. The solidification time was evaluated directly from the thermal field.
-
Diffusivity as a function of temperature and of the liquid/solid state of the iron was used.
-
The initial Mg level was taken as 0.02%Mg.
-
A Mg sink at the mold/metal interface was assumed.
-
Liquid convection in the casting was ignored.
-
The critical Mg level for graphite degradation was assumed to be 0.015%Mg.
-
The iron was of eutectic composition with a solidification range of 1135 – 1144 C.
85
Sym.
x
25 mm
y
18 mm
25 mm
18 mm
Sym.
TC
Figure 60. Basic geometry of the model based on the standard cooling cup
The computational method used was the explicit Finite Difference Method (FDM). Assuming that the diffusivity and thermal conductivity are not a function of location, the governing equation for mass transfer is: 2C 2 C C D t y x
The governing equation for the heat transfer is: 2T 2T T k t y x
The heat flux at the mold/metal interface was prescribed as: hA(T Tmold ) VH f V
86
T T kA t x
When the temperature is in the solidification range (TS < T < TL); otherwise; hA(T Tmold ) V
T T kA t x
Table 21. Summary of parameters used for simulation Parameter Physical dimensions Mold wall thickness Number of nodes Size of discretized node (x, y) Thermal conductivity - sand (km) - iron(ks) Density - sand (m) - iron (s) Heat capacity - sand (Cp,m) - iron (Cp,s) Heat transfer coefficient (air – sand) Initial Mg level
Value 25 25 mm 8 mm 200 200 1.25 10-4 m 0.135 W/mK 37.5 W/mK 1600 kg/m3 6982 kg/m3 1130 J/kgK 2.7 105 J/kgK 1000 W/m2 0.02 %
By taking 0.17 at% as the solid solubility of Mg in iron, the diffusivity of Mg in liquid iron was estimated as 1.49 10-8 m2/s at 1350 C. Furthermore, the constant (D0) and activation energy (Q) in the Arrhenius equation were estimated as 4.35 10-7 m2/s and 44.9 103 J/mol. Therefore, the diffusivity of Mg in liquid iron can be calculated as a function of temperature.
The diffusivity of Mg in solid was taken as a constant at DS = 1.0 10-11 m2/s. The diffusivity of Mg between liquidus and solidus was estimated by: 87
T TS T T L S
DS / L DS
T TS D 1 L T T L S
where TS and TL are solidus and liquidus temperature respectively.
Note that this model is specifically for Mg depletion and does not take solidification kinetics effects into account. Therefore, the model cannot be used to explain the effects of mold coatings.
A virtual thermocouple was located at the inside corner of the grid to replicate the location of the actual thermocouple (Figure 60). A comparison between the calculated and measured cooling curves is presented in Figure 61.
1250 M odeling Experiment
Temperature, °C
1200
1150
1100
1050 0
50
100
150
200
250
time, sec
Figure 61. Comparison of the simulated and experimental cooling curves
88
The calculated Mg distribution at the corner of the Thermal Analysis (TA) cup is shown in Figure 62a. It is seen that Mg depletion occurred on the exterior edge of the model grid. Taking 0.015%Mg as the critical Mg depletion level (the dotted line), the predicted graphite degradation thickness was in the range of 250 – 300 m depending on location. This is in the same range with the measured skin thickness shown in Figure 62b. This confirms the Mg depletion theory for the graphite degradation in the skin.
2500
0.020
Distance from corner, m
2000
0.017 0.015
0.020
1500
0.012 0.010
1000
0.007
0.005 500 0.002
0
0.007 0.002 0.005
0
500
1000
0.010 0.012 1500
0.015
2000
0.017
2500
Distance from corner, m
(a)
(b)
Figure 62. Comparison between (a) simulated Mg distribution (b) corresponding microstructure of graphite degraded layer
The predicted Mg distribution profile exhibits slightly smaller depletion zone toward the corner. This is because of the divergent heat flow resulting in the shorter solidification time. The shorter solidification time allows less time for Mg diffusion. Unfortunately, the
89
sample has a large radius at the corner; therefore, the same effect could not be seen in the actual sample.
To obtain reliable results, it is important to test the model dependency on the grid fineness. Results are shown in Figure 63. It is seen that the skin thickness decreased with the grid size (x or y). When x approached 0.2 mm, the skin thickness leveled out indicating the independency of the grid size on the skin thickness. Notice that the simulation shown earlier was done with x = 0.125 mm.
Skin thickness, m
300
280
260
240
220
200 0.000
0.200
0.400
0.600
x or y , mm
Figure 63. Effect of x or y on simulation results
The effect of the pouring temperature on the skin thickness is shown in Figure 64. As discussed earlier, the higher pouring temperature allows more time for Mg diffusion; therefore, a larger Mg depletion area is expected. Consequently, the skin thickness (taken 90
by the 0.015%Mg critical value) increased with the pouring temperature. Figure 64b) shows the corresponding cooling curves from the simulations. Of course, longer solidification times are calculated for higher pouring temperature.
1400
350
1200 C 1250 C 1300 C 1350 C 1400 C
300
Temperature, C
Skin thickness, m
1350
250
200
1300 1250 1200 1150
150
1100 100 1200
1250
1300
1350
1050
1400
0
Pouring temperature, C
50
100
150
200
250
Time, s
a)
b)
Figure 64. a) Effect of pouring temperature on the simulated skin thickness; b) corresponding cooling curves
Figure 65 demonstrates the effect of the initial Mg on the skin thickness. Reducing the initial Mg increased the skin thickness exponentially. This is expected because lower initial Mg should produce larger Mg depletion zone. The result implies that ductile iron (typical Mg level of 0.04%) should exhibit smaller casting skin than CG iron (typical Mg level of 0.02).
91
350
Skin thickness, mm
1250 C 1350 C
300
250
200
150
100 0.01
0.02
0.03
0.04
0.05
Initial Mg, %
Figure 65. Effect of initial Mg on the simulated skin thickness
4.5 Solidification kinetics Besides the Mg depletion effects, the occurrence of the casting skin is also governed by the effect of solidification kinetics. Indeed, Mampaey et al. [10] reported the existence of type-D graphite at the casting skin in gray iron with type-A graphite in the bulk. As there is no Mg in gray iron, skin formation must be attributed to some effect of solidification kinetics, such as higher undercooling at the interface. The role of oxygen on the type-A to type–D graphite transition could also be considered.
In the standard binary iron-graphite system, the primary phase is austenite (γ) for hypoeutectic iron and graphite (Gr) for hypereutectic iron. However, for CG iron and DI, solidification starts with the solidification of austenite. Even though for hypereutectic irons some primary graphite may form, there is no significant growth until graphite 92
encapsulation by austenite occurs. This could be understood in terms of the asymmetric phase diagram presented in Figure 66.
Temperature
L L + Gr.
L +
T
Coupled zone FeSi Gr CE
+ eutectic
Gr. + eutectic
High k
Skin thickness, m
A
C
B
250 150 50 3.8
4.0
4.2
4.4
4.6
4.8
CE, % Carbon equivalent
Figure 66. Schematic asymmetric phase diagram of the iron-graphite system and the effect on casting skin formation
The asymmetric phase diagram describes solidification behavior when undercooling and differences in the growth rates of the phases are significant. From the figure it is seen that as the undercooling increases, the coupled (eutectic) region shifts to the right. As the undercooling increases and/or the carbon equivalent decreases, the amount of primary austenite increases. 93
The solidification of primary austenite at the mold/metal interface can be a major cause of skin formation. As shown in Figure 67 for the case of a DI, austenite solidifying at the surface of the casting rejects Mg at the liquid/solid interface, which results in a layer depleted of Mg, and consequently in graphite degradation.
Figure 67. Mg redistribution at the surface of the casting because of rejection of Mg by the solidifying iron
The parameters that affect the amount of austenite at the interface are: i) carbon equivalent ii) inoculation and iii) thermal conductivity.
4.5.1 Effect of carbon equivalent Our recent results (see mold coating experiment in Chapter 5) suggest that carbon equivalent plays an important role on the thickness of graphite depletion. We shall begin this discussion for the no-coating condition. At a given undercooling (the dotted 94
horizontal line in Figure 66), the hypoeutectic iron with low carbon equivalent (CE) (point A) has more dendritic austenite than the medium CE (point B). The hypereutectic iron (point C) starts its solidification with the solidification of austenite and not of primary graphite as predicted by the standard binary Fe-Gr diagram. In this study, the CE values were 4.06, 4.23, 4.33 and 4.53. Skin thickness measurements on plate castings (no-coating) are shown at the bottom of Figure 66 to demonstrate the relationship between CE, the amount of primary austenite and the skin thickness. It is seen that the skin thickness is smaller as the CE increases, approaching the coupled zone.
When a graphite coating is applied, carbon is locally transferred to the iron from the coating by diffusion. The local increase in CE shifts the composition in Figure 66 to the right (indicated by green dots). For low and medium CE (point A and B), the positions are shifted closer to the coupled zone and smaller casting skin is obtained. For high CE (point C), the position may be shifted over and away from the coupled zone producing larger casting skin. The experimental data that confirm this mechanism will be shown in the next chapter.
Ferrosilicon coating should slightly increase the CE because silicon has only one third of the effect of carbon on CE. However, ferrosilicon has a stronger inoculation effect than graphite. This will be discussed in the next section.
95
4.5.2 Effect of inoculation The ferrosilicon coating decreases the undercooling because of its inoculation effect. Positions in the asymmetric phase diagram shown by yellow squares in Figure 66 are shifted upward. The corresponding amount of primary austenite becomes smaller and results in thinner casting skin. However, the improvement is expected to be more effective for the case of medium to high CE (point B and C), while a small improvement is expected for low CE (point A).
4.5.3 Effect of thermal conductivity Examination of the asymmetric Fe-Gr diagram in Figure 66 predicts that higher undercooling will result in more austenite at the beginning of solidification, and therefore thicker skin. Thus, it is reasonable to assume that non-active coatings with high thermal conductivity that increase undercooling at the mold/metal interface will increase the skin thickness.
Figure 68 shows the effect of thermal conductivity over the skin thickness for no-coating and inactive coatings. Increased skin thickness with higher thermal conductivity is seen at all CE levels with only one exception for the mica coating at CE of 4.53 (the green dot).
The effect of the dipping time of zircon coating also supports this line of thought. (See Chapter 5 for the details) that longer dipping time for the zircon coating increased the skin thickness because of the thicker top layer and deeper penetration depth that increase 96
the thermal conductivity. Again, the asymmetric phase diagram can be used to explain this effect. Increasing the undercooling shifts the positions on the asymmetric phase diagram (Figure 66) downward (shown by blue triangles) and results in larger casting skin in all cases because of more primary phase.
300
Skin Thickness, m
250 200 150 CE = 4.06 100
CE = 4.23 CE = 4.33
50
CE = 4.53
0 0
10
20
30
40
Thermal conductivity, W/mK
Figure 68. The skin thickness as a function of thermal conductivity (reactive coatings are excluded)
Similar phenomenon in DI can be seen in Figure 51. The figure was obtained from a DI sample cast in a steel mold. Obviously, steel has much higher thermal conductivity (typically 30 – 40 W/mK) than silica sand; therefore, the effect of undercooling was clearly seen.
Based on this analysis a coating-sand system that decreases the cooling rate, and therefore decreases the undercooling should promote a thinner skin. However, the work 97
by Duncan and Kroker [17] that demonstrates that lower cooling rate resulting from larger section thickness increases the skin thickness. This apparent contradiction can be explained by the contribution of Mg oxidation to the mechanism of skin formation. As the cooling rate is decreased, the solidification time is increased, which also increases the exposure time for Mg depletion.
In summary, two competing mechanisms for the casting skin formation were identified: increased austenite layer (because of solidification kinetics effects) and Mg depletion (through oxidation or sulfur reaction). Solidification kinetics affects skin formation through the change in undercooling at the mold/metal interface, which affects the amount of primary austenite. Indeed, mold coatings with high thermal conductivity favor thicker skin. On the other hand, Mg depletion is dominant at longer solidification times (e.g. large castings). Indeed, thicker casting skin was found for heavy section thickness. In general, high undercooling favors a thicker layer of primary austenite, but discourages magnesium oxidation and vice versa. Figure 69 presents a schematic diagram explaining the combined effect of the two mechanisms. It is seen that skin formation is controlled by Mg depletion at low cooling rate, and by austenite layer formation at the interface at high cooling rate. The 2-D thermal-diffusion model presented before is based on Mg depletion. Therefore it should be valid only on the low cooling rate side of Figure 69.
98
Figure 69. Schematic diagram of the combined effect between two skin formation mechanisms
4.6 Conclusion This chapter reviews the formation mechanisms of the casting skin features based on findings in literatures and advances a more complex theory, that includes not only chemical reactions at the interface, but also the effects of solidification kinetics. The formation mechanisms of casting skin are summarized as follows: -
Surface roughness depends on the fineness of the sand and on the metallostatic height. The mathematics of the phenomenon was described.
-
Decarburization at the casting surface is responsible for the formation of graphite depletion and ferritic rim. Moisture and higher pouring temperature promote decarburization. On the other hand, carburization is the cause for the formation of the pearlitic rim. 99
-
Graphite degradation can be explained either through the depletion of Mg as a result of chemical reactions with the mold and the atmosphere, or by an increase in the amount of the austenite in the surface layer. This last mechanism can be understood through the use of the asymmetric Fe-Gr phase diagram.
-
A 2D thermal-diffusion model was able to predict the skin thickness based on Mg depletion.
Figure 70 summarizes most of the formation mechanisms of casting skin presented in this research.
Mold
Adjacent liquid
Casting skin Graphite
Oxygen Cooling rate
Sulfur
degradation
Mg, Ce
Graphite depletion C Pearlitic
Carbon
rim
Nuclei depletion
C and/or S rich layer
Figure 70. Summary of casting skin formation mechanism
100
Chapter 5: Minimization of the Casting Skin Formation 5.1 Introduction In the previous chapter, Mg depletion and the effect of solidification kinetics were proposed as the main formation mechanisms for the graphite degradation. Mg reaction with oxygen and/or sulfur causes Mg depletion in the iron at the proximity of the mold wall, resulting in graphite degradation. Large undercooling at the mold/metal interface favors dendritic growth of austenite that rejects Mg to the solidification front. This results in low Mg and lamellar graphite structure close to the mold/metal interface, with a high nodularity zone further inside.
Based on these mechanisms, minimizing the casting skin formation may be achieved by minimizing Mg depletion and decreasing undercooling at the mold wall. The main objective of this chapter is to demonstrate these statements through experiments with mold coating.
5.2 Experimental approach The experimental approach for this section is shown in Figure 71.
101
Casting Design & Simulation
Designing and Processing
Mold Coating
Production of
Coating Materials
Carbon Equivalent
Test casting
Step Casting
Plate Casting
Deliverable
Effect of Mold
Effect of CE
Section Thickness
Effect of Cooling rate
Coating
Figure 71. Overview of the approach
5.2.1 Test casting design Similar casting design to the fatigue study (Figure 27a) was used in these experiments. The casting design allows producing four test plates per mold. Cores were added at the bottom of the plates, allowing the use of up to four types of mold coatings in the same mold (Figure 72). The dimensions of the plates were 110 60 13 mm. The cooling rate at the geometric center of the plate obtained from a computer simulation was 1.1 – 1.4 C/s. Risers were located outside the plate to keep cooling rate as uniform as possible. The cores were to be dipped in mold coating slurries, let dry and then glued in place.
102
Parting line Cores
a) isometric view
b) front view showing location of cores
Figure 72. Test casting design for the mold coating experiment
5.2.2 Mold coating The coatings used in this research were classified in three categories: -
Inactive coating – these are coatings that are completely inert with respect to the melt.
-
Active coatings - coatings that alter the local chemistry of the melt. These coatings may have some inoculation effect.
-
Reactive coatings - coatings that have chemical reactions with the melt, such as deoxidation or desulfurization.
A total of eight coatings were used. The mica based, zircon based, boron nitride and MgO coatings are representatives of inactive coatings. Ferrosilicon and graphite are representatives of the active coatings. Finally, the CaO and FeSiMg coatings represent the reactive coatings.
103
Table 22 lists some of the characteristics of the coating materials. The temperature column indicates the range of temperatures within which the thermal conductivity was measured. All types of mold coatings were prepared as slurries in alcohol, with the exception of boron nitride which was water based. The solid fraction was approximately 75 wt%. For ferrosilicon and graphite, the materials were ground and screened down to 325 mesh. FeSiMg, MgO and CaO powders were of 200 mesh. Cores were made out of silica sand having a Grain Fineness Number (GFN) of 55 – 60. The cores dimensions were 33 55 110 mm. A sodium silicate binder was used for both molds and cores.
Table 22. Coating materials and their thermal conductivities Coating No coat (silica sand) Ferrosilicon (75%Si) Graphite (100%C) FeSiMg (6Mg45.4Si) Calcium Oxide (CaO) Magnesium Oxide (MgO) Boron Nitride (BN) Mica based Zircon based
k (W/mK) N/A 0.17 - 0.22 Active N/A Active 30 - 200 Reactive N/A
Temperature (K) 279 – 312 N/A 298 N/A
Mesh size N/A 325 325 -200
Reactive 0.3 – 0.5
390 - 1197
-200
[30] pp. 141 – 143
Types
Reference [29] N/A [30] pp. 79 - 81 N/A
Inactive
27.8 – 36.1 318 – 418
-200
[30] pp. 158 – 167
Inactive Inactive Inactive
25.6 - 36.2 0.50-0.83 4.81-5.86
325 325 325
[30] pp. 656 – 658 [30] pp. 823 – 826 [30] pp. 317 – 319
1047 – 1475 298 298
Figure 73 depicts the dipping method. The face which will be in contact with molten iron was oriented downward and the cores were immersed into the slurry 10 mm under the level of the slurry for predetermined times. The cores were then withdrawn and shaken to
104
smooth the coated layer, and then were left to dry at room temperature for one day before use.
Slurry Figure 73. Core dipping
Dipping tests were conducted to determine the diffusion rate of each mold coating. The cores were cut in half and examined. Two distinct layers were observed, as shown schematically in Figure 74 (since boron nitride was applied as paint it was not included in this analysis). The top layer is a buildup of coating particles on the surface of the core. The penetration depth is a diffused layer that has coating materials that penetrated between the sand grains. Representative photographs of cores after coating are shown in Figure 75. It is seen that graphite, mica and ferrosilicon based coatings exhibit significant penetration depth. The zircon based coating has very limited penetration.
105
Top layer Penetration Depth
Core
Figure 74. Top layer and penetration depth
Zr
Graphite
Mica
FeSi
Figure 75. Cross sections of coated cores
Figure 76 shows the top layer thickness and penetration depths as a function of the dipping time. It is seen that the coatings that develop the top layer faster had smaller penetration depth. A dipping time of 2 minutes was selected for the experiments, as this is the time at which further holding in the slurry had little or no effect on the penetration depth. 106
5.2.3 Production of test castings Four 80 lb CG iron heats were produced in the OSU casting laboratory. The main difference between the heats was the carbon equivalent. The molten irons were treated with FeSiMg in a reaction chamber (see Figure 28). The treatment temperature was approximately 1480 C. The level of Mg was maintained constant at 0.008-0.01% to produce 15 – 20% nodularity CG irons. Silica sand of 55 – 60 GFN was used as molding sand with sodium silicate as binder.
15
1.5
Penetration Depth, mm
Top layer thickness, mm
2.0
1.0 FeSi Gr Mica Zr FeSiMg MgO CaO
0.5
0.0
10
FeSi Gr Mica Zr FeSiMg MgO CaO
5
0
0
50
100
150
200
0
time, s
50
100
150
200
time, s
Figure 76. Top layer thickness and penetration depth as a function of the dipping time for various mold coatings
Thermal analysis (TA) was used to control the carbon and silicon before the FeSiMg treatment. After the treatment, TA cups were poured along with the test castings to retrieve the cooling curves. 107
5.2.4 Microstructural examination and skin quantification Typical metallographic sample preparation was done on all samples. Color metallography was done on selected set of samples. See the section 3.1.6 for complete details.
Quantification of the casting skin was conducted by image analysis. The procedures are as per Chapter 2.
5.2.5 Chemical analysis and microstructure Table 23 shows the chemical composition of the CG irons produced for these experiments. It is seen that CE varied from 4.06 to 4.53%, which covered hypoeutectic, eutectic and hypereutectic compositions. Mg level was in range of 0.008 – 0.01% which delivered 15 – 20% nodularity as intended. Figure 77 shows representative microstructures from the middle of the test plates. Note that within the same sample, the nodularity can appear much higher adjacent to the casting skin.
Table 23. Chemistry of CG iron heats in this study Heat 1114 1222.1 1222.2 504
C 3.47 3.62 3.18 3.32
Si 2.60 2.73 2.63 2.73
Mn 0.145 0.176 0.167 0.103
Mg 0.008 0.008 0.009 0.010
108
S 0.007 0.008 0.009 0.007
P 0.037 0.037 0.039 0.031
Cu 0.023 0.034 0.025 0.036
CE 4.33 4.53 4.06 4.23
The cooling curves from standard thermal analysis cups for all heats are shown in Figure 78. As expected, a high liquidus arrest is seen for Heat 1222.2 (CE = 4.06), while no primary phase arrest or significant undercooling is seen for Heat 1222.1 (CE = 4.53).
CE = 4.06
CE = 4.23
CE = 4.33
CE = 4.53
Figure 77. Representative microstructure at the center of the test plates
109
1250 CE = 4.06 CE = 4.23 CE = 4.33
Temperature, °C
1200
CE = 4.53
1150
1100
1050 0
50
100
150
200
time, sec
Figure 78. Cooling curves from standard Thermal Analysis cups
5.3 Effect of mold coating on the casting skin formation 5.3.1 Active coating The active coatings used in these experiments are ferrosilicon and graphite. For ferrosilicon, the expected effects are: i) to provide local inoculation at the casting skin ii) to slightly increase local carbon equivalent at the casting skin. The effects of FeSi inoculation include decreasing of the undercooling at the mold/metal interface; thus, a lower amount of primary austenite is anticipated. The expected effects of graphite coating are: i) to increase the local carbon equivalent at the casting skin ii) to provide carburizing and reducing atmosphere.
Figure 79 shows unetched microstructures for various types of reactive coatings at different CE. For the non-coated samples, it is seen that higher CE results in thinner casting skin. Note that the high nodularity zone is closer to the casting surface for higher 110
CE as well. For the ferrosilicon coating, large casting skin is seen at CE of 4.06 and 4.23; however, the skin thickness decreases significantly at higher CE (4.33 and 4.53). For the graphite coating, some improvement is seen at low and medium CE; but no improvement is observed at high CE (4.53).
Figure 80 shows color micrographs of the same set of samples as in Figure 79. The color blue is associated with the austenite dendrites. The segregation in the interdendritic regions appears brown. The pearlite at the skin is yellow/orange. It is seen that at the casting skin the predominant structure is austenite dendrites with little segregation, which in some cases transforms in pearlite upon cooling. Away from the skin, the austenite is coarser and allows for significant segregation at the end of solidification. The overall microstructure of ferrosilicon-coated samples appears to be more oriented than that of the graphite-coated samples. This is because the local CE is increased by the graphite coating. Quantitative results of the effects of the carbon equivalent and of the active coatings on the skin thickness are summarized in Figure 81.
The effect of CE for the no-coating condition can be explained by the amount of primary phase formed at the beginning of solidification. Formation of primary austenite dendrites in the hypoeutectic iron produces Mg segregation because the Mg rejected by the solid is pushed by the solidification front, leaving a Mg depleted zone at the casting skin. This creates a highly oriented dendritic structure with fine lamellar graphite between the dendrites. As the CE increases, less austenite is produced, which results in thinner skin. 111
112
CE = 4.06
CE = 4.23
CE = 4.33
CE = 4.53
No Coat
No Coat
No Coat
No Coat
CE = 4.06
CE = 4.23
CE = 4.33
CE = 4.53
FeSi
FeSi
FeSi
FeSi
CE = 4.06
CE = 4.23
CE = 4.33
CE = 4.53
Gr
Gr
Gr
Gr
Figure 79. Unetched microstructures of the casting skin without coating (top) with ferrosilicon coating (middle) and with graphite coating (bottom) 112
113
No Coat
No Coat
No Coat
FeSi
FeSi
FeSi
Gr
Gr
Gr
CE = 4.06
CE = 4.33
CE = 4.53
Figure 80. Color-etched microstructures of the casting skin without coating (top) with ferrosilicon coating (middle) and with graphite coating (bottom) 113
300 NC
Skin thickness, m
250
Gr FeSi
200 150 100 50 0 4.0
4.2
4.4
4.6
CE, %
Figure 81. The influence of carbon equivalent and of active coatings on skin thickness
When graphite coating is used, the local CE is increased. This results in decreased skin thickness for low CE iron as less austenite dendrites are formed. Little effect is expected for high CE.
The FeSi coating decreases undercooling at the mold/metal interface through its inoculation effect. This results in thinner casting skin. The explanation of this effect is based on the asymmetric phase diagram (see Section 4.5.2 for details).
5.3.2 Reactive coating The reactive coatings used in this study are FeSiMg and CaO. These coatings have a desulfurizing and/or deoxidizing reaction with the melt. In addition, FeSiMg has some inoculation effect and provides additional Mg at the mold/metal interface.
114
CaO will react with the sulfur in the molding aggregates such as furan resin and phenolic urethane bonded sands. However, it will behave as an inactive coating when there is no sulfur in the binder, such as when sodium silicate is used as a binder.
Heat 504 (CE = 4.23), was used to evaluate the effect of the FeSiMg and CaO coatings. Unetched microstructures are shown in Figure 82. It is seen that FeSiMg reduced the thickness of the graphite degraded layer as compared to the no-coating condition. In addition, the nodularity at the center of the plate appeared to be higher with the FeSiMg coating. This was expected since there could be some dissolved Mg diffusing into the plate.
CaO was expected to locally desulfurize the melt. However, only small improvement is seen comparing to the no-coating condition. As there is no sulfur associated with the binder used in this experiment, CaO can be thought as inactive coatings.
115
NC
CaO
FeSiMg
a) 116 NC
FeSiMg
CaO
b) Figure 82. Unetched microstructures of samples from Heat 504 (CG iron, CE = 4.23) a) at the skin b) at the middle of the plate 116
5.3.3 Inactive coating Three inactive coatings were used in this study, i.e., mica based, zircon based, boron nitride and MgO. The effects of these coatings are twofold: i) minimizing loss of Mg through oxidation with the atmosphere ii) increasing or decreasing the heat flux at the mold/metal interface depending on the thermal conductivity of the coating materials.
For the experiments with the zircon based coating, the cores were dipped for various times (one, two and three minutes), to demonstrate the effect of the thickness of the top layer. Figure 83 presents microstructures from this experiment. It is clear that the skin thickness decreases with increasing CE, similarly to the no-coating conditions, and increases with increasing dipping time. The average skin thicknesses for all zircon coated samples are shown in Figure 84.
When compared to the no-coating conditions, only the dipping time of 1 minute showed a small improvement. This could be the consequence of reduced Mg oxidation because of the top layer. However, as the dipping time increased the reverse effect is seen. This is because the high thermal conductivity of the coating material increases undercooling, which then favors more austenite dendrites. Indeed, as seen in Figure 83, longer dipping time produces more dendritic structure. It is thus reasonable to conclude that the zircon based coating did not minimize the casting skin. This result agrees with the previous work by Duncan and Kroker [17].
117
CE = 4.06
CE = 4.06
1 min
2 min
3 min
CE = 4.33
CE = 4.33
CE = 4.33
1 min
2 min
3 min
CE = 4.53
CE = 4.53
CE = 4.53
1 min
2 min
3 min
118
CE = 4.06
Figure 83. Color micrographs of CG irons with zircon based coating
118
500
500 Zr -1 min
400
4.33 400
Zr -3 min
Skin thickness, m
Skin thickness, m
4.06
Zr -2 min
300
200
4.53
300
200
100
0
100 4.0
4.1
4.2
4.3
4.4
4.5
4.6
0
CE, %
1
2
3
4
Dipping time, min
Figure 84. Average skin thickness as a function of CE (left) and dipping time (right)
Figure 85 shows color-etched microstructures of CG iron poured in molds with boron nitride coating. The skin thickness decreases with increasing CE. The average skin thickness is higher than for the no-coating conditions. This is expected because of the high thermal conductivity of the boron nitride coating (see Table 22). A pearlitic rim was found in all samples.
CE = 4.06
CE = 4.53
BN
BN
Figure 85. Color micrographs at the casting skin with boron nitride coating
119
Mica has lower thermal conductivity (see Table 22) and therefore it is expected to perform better than the zircon coating. Figure 86 shows color-etched microstructures with different CE. There was no improvement compared to the no-coating condition. This makes sense since mica has higher thermal conductivity than silica sand (0.5 – 0.8 W/mK for mica compared to 0.2 W/mK for silica sand). For the effect of CE, the result did not follow the same trend as shown before. The skin thickness decreases with increasing CE and rebounds back at the hypereutectic composition. This is only observed in mica. The coating does not seem to provide any advantage.
CE = 4.06
CE = 4.33
Mica
Mica
CE = 4.53
Mica
Figure 86. Color micrographs at the casting skin with mica based coating
120
As shown in Table 17, MgO is more stable than MgS at 1500 K. Therefore, MgO does not have desulfurizing effect and can be considered as an inactive coating. Figure 87 shows the unetched microstructure of CG iron with MgO coating. The observed skin thickness was larger than that of the iron without coating. This is because of the high thermal conductivity of MgO that induces dendritic growth of austenite during solidification. The microstructure at the center of the plate also shows higher nodularity than the no-coating condition (Figure 82). The overall effect of the MgO coating is similar to that of the boron nitride coating.
MgO
MgO
a) at the casting skin
b) at the center of the plate
Figure 87. Unetched microstructure of CG iron (CE = 4.23) with MgO coating
A comparison between the effects of different coatings is presented in Figure 88. A clear overall effect of CE is seen. The casting skin thickness decreases with increasing CE. There are two exceptions for mica and graphite coatings. In the case of graphite, the trend is reversed. This could be a consequence of the higher local CE at the mold/metal 121
interface that pushes the CE further to the hypereutectic side and promotes formation of primary graphite.
Figure 88. The skin thickness from various types of mold coating
Figure 89 shows the effect of the various mold coatings on the surface roughness of the test plates. It is seen that all mold coatings with the exception of FeSiMg improved the surface roughness as compared to the no-coating condition. This is mainly because the smaller interparticle distance of coating materials creates greater capillary force opposing the metal penetration by metallostatic pressure. Reaction between FeSiMg and molten iron create gaseous phase that produced higher surface roughness.
122
5.4 Effect of section thickness on the casting skin with mold coatings Duncan and Kroker [17] demonstrated that the casting skin thickness depends on the casting section size. They showed that 40 mm diameter rods had larger casting skin than 10 mm diameter rods without coating. This is also the case for zircon based and mica based coatings. Various types of molding sands were investigated as well. The LDASC (Low Density Aluminosilicate) sand performed the worst as it imposed a very slow cooling rate. The effect of casting section size will be discussed in this section with emphasis on the reactive coatings ferrosilicon and graphite.
30
Roughness average, m
25 20 15 10 5 0 NC
BN
FeSi
Gr
Mica
Zr1
Zr2
Zr3
MgO
CaO
FeSiMg
Mold coating types
Figure 89. Roughness average as a function of mold coating types (Zr1 – zircon coat 1 min. dipping; Zr2 - zircon coat 2 min. dipping; Zr3 – zircon coat 3 min. dipping)
The test casting design is shown in Figure 90. The design includes four thicknesses, i.e. 7, 14, 21 and 28 mm. Each section is 50 mm long and 50 mm wide. The molds were made 123
with sodium silicate bonded sand with a grain fineness number of 55-60. The bottom face of each mold was painted with one of the mold coatings discussed earlier.
T1
T2
T3
28 mm
21 mm
14 mm
50 mm
50 mm
50 mm
T4
7 mm
50 mm
Figure 90. Test casting design for step casting [31]
The variation of the section thickness allows investigating the effect of the cooling rate. However, the results should not be compared directly to the experiments presented in the previous section because of the difference in the application method of the mold coatings.
The solidification of the test casting was simulated to obtain the cooling rates. Four virtual thermocouples were placed at the middle of each step, as shown in Figure 90. The simulation results in the section across the mid plane are presented in Figure 91. It is seen that the cooling rates varied from 1.5 – 15 C/s. The predicted nodularity was in the range of 8 – 18 % which was in line with the results of the microscopic examination. Table 24 summarizes the cooling rates from the virtual thermocouples.
124
a)
b)
Figure 91. Simulation results a) cooling rate distribution b) nodularity distribution [31]
Table 24. Corresponding cooling rates of the virtual thermocouples Thermocouple Section thickness, mm Cooling rate, C/s
T1 28 2.49
T2 21 2.62
T3 14 4.70
T4 7 12.63
Figure 92 demonstrates the effect of section thickness for a CG iron with 4.33% CE. It is obvious that the skin thickness increases with increasing section thickness for all types of mold coatings. A similar trend is seen for higher CEs as well. It is worth noting that the degraded graphite (flake graphite) becomes coarser as the section thickness increases.
The effect of section thickness can be explained by the Mg depletion model. As the section thickness increases, the solidification time increases, allowing more time for Mg to react with the oxygen diffusing from the mold. Therefore, a heavy section is expected to have a larger casting skin. The measured skin thickness as a function of section thickness and cooling rates is shown in Figure 93. 125
All the 7 mm thick plates exhibited some carbides, with one exception. The FeSi coated 7 mm plate was free of carbides, which demonstrates the inoculating effect of the FeSi coating. The inoculation effect of the graphite coating was insufficient, as some carbides persisted in the microstructure. Figure 94 shows examples of etched microstructures of a sample with graphite coating.
The effect of the section thickness on the skin for molds with CaO, MgO and FeSiMg coatings was examined for Heat 504 (CE = 4.23). Unetched microstructures are shown in Figure 95. Similar trends as per the earlier discussion are seen. Apparently, the MgO coating produced the thickest skin compared to other types of coatings. FeSiMg produced the smallest casting skin at all thicknesses. Figure 96 shows a summary of the results from this heat.
5.5 Thermal analysis Thermal Analysis (TA) was conducted on heat 504 (CE = 4.23) to observe the effect of mold coatings on solidification characteristics. Five coating conditions were used as follows: no-coating (NC), CaO, MgO, FeSiMg and zircon coatings. The standard cups (33 33 37 mm) were made of furan resin bonded sand. Two teaspoons of coating slurry were poured into the cups. The cups were then rotated to produce uniform coating and left to dry overnight before pouring. All cups were poured immediately after the melt was treated with FeSiMg treatment. Thus, the pouring temperatures were very close. 126
No coat
Ferrosilicon
Graphite
7 mm
7 mm
7 mm
14 mm
14 mm
14 mm
21 mm
21 mm
21 mm
28 mm
28 mm
28 mm
Figure 92. Effect of section thickness and mold coatings on microstructure at the casting skin for CG iron with CE = 4.33
127
400
CE = 4.06
350
NC
300
FeSi
250
Gr
Skin thickness, m
Skin thickness, m
400
200 150 100
350
NC
300
FeSi
200 150 100 50
0
0
10
20
Gr
250
50 0
30
0
10
Plate thickness, mm
20
30
Plate thickness, mm
400
NC
350
Skin thickness, m
CE = 4.33
CE = 4.53
FeSi
300
Gr
250 200 150 100 50 0 0
10
20
30
Plate thickness, mm
Figure 93. Effect of section thickness on the casting skin thickness
7 mm
7 mm
a) at the casting skin
b) at the middle
Figure 94. Example of carbide structure in a sample with graphite coating (CE = 4.06) 128
CaO
MgO
FeSiMg
7 mm
7 mm
7 mm
14 mm
14 mm
14 mm
21 mm
21 mm
21 mm
28 mm
28 mm
28 mm
Figure 95. Effect of section thickness and mold coating (CaO, MgO and FeSiMg) on the microstructure at the casting skin for CG iron with CE = 4.23
129
400
400 NC
300
MgO
250
FeSiMg
200 150 100
FeSiMg
200 150 100
0
0 20
MgO
250
50
10
CaO
300
50
0
NC
350
CaO
Skin thickness, m
Skin thickness, m
350
0
30
5
10
15
Cooling rate, deg C/s
Plate thickness, mm
Figure 96. Effect of section thickness, cooling rate and mold coatings (CaO, MgO and FeSiMg) on the skin thickness; CE = 4.23
T, NC T, Zr dT/dt, NC dT/dt, Zr
1 0
1200
-1 -2
1150
Cooling rate, deg C/s
Temperature, C
1250
-3 1100
-4 0
50
100
150
200
250
time, s
Figure 97. Cooling curves and cooling rates for no-coat and zircon coated cups
The cooling curves and cooling rates in Figure 97 indicate that the zircon coating produced shorter solidification time than no-coating. This is because of the higher thermal conductivity of the coating and because of a thicker graphite degradation layer 130
produced by the zircon based coating. Note that the Maximum Cooling Rate at the end of solidification (MCR) was slightly changed by the coating, indicating a slightly lower nodularity.
The effect of the reactive coatings CaO, FeSiMg and MgO on the cooling curves is presented in Figure 98. It is seen that the MCR is affected by the type of coating.
1250
1.0 NC CaO
0.0
MgO
1200
dT/dt, C/s
Temperature, C
FeSiMg
1150
-1.0 NC
-2.0
CaO FeSiMg MgO
-3.0 1100
-4.0 0
50
100
150
200
250
0
time, s
50
100
150
200
250
time, s
a)
b)
Figure 98. Results from Thermal Analysis cups with no coating, CaO coating, FeSiMg coating and MgO coating a) Cooling curves and b) derivatives
The change in the MCR is the result of the change in the nodularity of the iron. Indeed, as seen in Figure 98 and Table 25, the nodularity at the geometric center of the samples increases with lower MCR, as thermal conductivity decreases with higher nodule count.
131
The most noticeable effect is that of the FeSiMg coating. This effect could be explained by one of the following hypotheses: -
The coating provides extra Mg that diffuses from the skin to the middle of the sample producing higher nodularity.
-
Local inoculation at the skin decreases the amount of austenite and thus the casting skin.
Table 25. Visual nodularity of samples from cooling cups at the center of the cups Coating No-coating CaO MgO FeSiMg Zr based
MCR, C/s 3.3 2.8 2.4 2.3 3.2
Nodularity, % 15 20 30 40 25
Residual Mg, % 0.017 0.018 0.017 0.018 0.018
Maximum cooling rate at the end of solidification
In order to prove this hypothesis, the cooling cups were cut in half and analyzed for residual Mg by optical emission spectroscopy. Although, the measurements are not accurate in terms of absolute value because of the presence of graphite, differences in the residual Mg between cups should be reliable. As seen from Table 25, the residual Mg level is constant for all samples at 0.017 – 0.018% (higher than results from chilled samples). This indicates that FeSiMg coating had no effect on the residual Mg in the bulk. Thus it is reasonable to assume that the lower MCR and higher nodularity produced by FeSiMg is because of its inoculation effect.
132
Higher nodularity at the middle of the sample was also observed for the MgO coating. MgO has higher thermal conductivity than the silica sand. As a result a shorter solidification time is achieved. Note the shrinkage cavity around the thermocouple observed for the MgO coating (Figure 99). This resulted in a lower MCR than the sound samples. Such shrinkage cavity was not seen in other cups.
Figure 99. Cross section of the cooling cup with MgO coating showing a shrinkage cavity around the thermocouple tip
For the CaO coated cups that produced smaller casting skin than no-coating, longer solidification time was recorded. This is mostly the result of the slightly higher pouring temperature (1251 C for the CaO sample, versus 1245 C for the no-coating sample).
The microstructures of the casting skin and in the middle of the cups are shown in Figure 100 and Figure 101. Figure 102 compares the casting skin thickness of the cooling cups for various coatings.
133
CaO
NC
FeSiMg
MgO
Zr based
Figure 100. Unetched microstructure of cooling cups without coating, with CaO, MgO, FeSiMg and Zr based coating at the skin
NC
MgO
CaO
FeSiMg
Zr based
Figure 101. Unetched microstructure of cooling cups without coating, with CaO, MgO, FeSiMg and Zr based coating at the center of the cups 134
For the step casting (silicate sand), CaO did not improve the casting skin when compared with the no-coating condition (see Figure 96). However, the reverse effect is seen from the thermal analysis cups. This is because the thermal analysis cups are made from furan resin bonded sand which contains sulfur in its composition. Therefore, CaO acts as a desulfurizer decreasing the skin thickness.
250
Skin Thickness, mm
200
150
100
50
0 NC
FeSiMg
MgO
CaO
Zr based
Figure 102. Measured skin thickness of cooling cups with various types of coatings (CE = 4.23)
5.6 Conclusion Three types of coatings were used in this research: a) inactive coatings (mica, zircon, boron nitride and MgO) –coatings that are completely inert with respect to the melt; b) active coatings (ferrosilicon and graphite) - coatings that alter the local chemistry of the
135
melt; c) reactive coatings (CaO and FeSiMg) - coatings that have a chemical reaction with the melt, such as deoxidation or desulfurization
The main effects of the coatings used in this study can be summarized as follows: -
The effect of inactive coatings depends on their respective thermal conductivity. The materials with high thermal conductivities are likely to produce thicker casting skin and vice versa. All inactive coatings in this study were found ineffective in decreasing the casting skin.
-
The ferrosilicon coating decreases the skin thickness. It is more effective for medium and high CE. It is not recommended for low CE.
-
The graphite coating increases CE by adding carbon to the iron. It is effective in decreasing the skin thickness for low CE irons. Graphite coating shall not be used for high CE since it creates more primary graphite and promotes thicker skin.
-
CaO decreases the casting skin when sulfur is present in the molding material. No improvement was found in the absence of sulfur (sodium silicate molds).
-
MgO did not reduce the casting skin because of its high thermal conductivity. The presence of sulfur in the molding material did not have an effect on MgO performance.
-
The FeSiMg was the most effective in reducing the skin thickness of an iron of eutectic composition at all cooling rates used in this work.
136
-
The skin thickness increases with increasing section size. This is because the longer solidification time allows more Mg depletion to occur, which is the dominant mechanism in this case.
137
Chapter 6: Elimination of the Casting Skin Effect 6.1 Introduction This chapter explores the possibility of using shot blasting to eliminate the casting skin effect. Shot blasting is known for improving the fatigue strength by creating compressive stress field on the casting surface. As most foundries own shot blasting equipment no capital investment is required. This makes shot blasting is the best method for surfaces that are accessible to blasting.
6.2 Experimental approach The effectiveness of shot blasting was observed through its effect on skin thickness, tensile and fatigue properties. The results were compared with the as-cast and machined samples discussed in Chapter 3. The other variable was the shot blasting time.
6.2.1 Casting design and simulation The same casting designs as presented earlier for tensile and fatigue testing were used (see the Section 3.1.2 for the complete details). An overview of the experimental approach is shown in Figure 103.
138
Casting Design & Simulation
Designing and Processing
Production of Test casting
Shot Blasting
Testing
Deliverable
Shot blasting time
Skin quantification
Tensile testing
Fatigue testing
Effect on
Effect on
Effect on
skin thickness
tensile strength
fatigue strength
Figure 103. Overview of the approach
6.2.2 Production of test casting The test samples were produced at the same time with the tensile and fatigue samples discussed earlier. Only CG irons with 15% nodularity (Heat 101101, 101108.1 and 101108.2) were used for these experiments because their comparable nodularity to the industrial applications (see the Section 3.1.3 for the complete details).
6.2.3. Quantification of casting skin The quantification procedure of the casting skin was as per Chapter 2.
139
6.2.4 Tensile and fatigue testing The tensile and fatigue testing were as of earlier experiment. See the Section 3.1.6.
6.2.5 Shot blasting Two sets of shot blasting were done at a railroad wheel production facility. The first set was for the tensile test samples and the second set was for the fatigue test samples. Details of both set are as follows.
For the tensile test samples, shot blasting was only applied to the as-cast samples. The samples were shot blasted with bainitic steel shot (grit size S-550) for 1 and 5 minutes (SB1 and SB2) on a rotating table. The test samples were flipped to receive same shot blasting exposure all around. The 1 minute shot blasting represents a moderate level of shot blasting, while 5 minutes shot blasting represents severe level of shot blasting.
For the fatigue test samples, two sets of samples with 15% nodularity were shot blasted with iron shot (S-660). The first set was machined and shot blasted (M-SB); the second set was as-cast and shot blasted (AC-SB). During shot blasting the samples were positioned alternately between as-cast and machined to avoid bias from positioning (see Figure 104). The samples were spot-welded to a dummy wheel. The welded area was on one end of each sample, well outside of the gage area. The wheel was then transferred to the shot blasting chamber where the wheel was rotated automatically to receive uniform shot blasting. The exposure time was 32 seconds. After shot blasting, the samples were 140
polished on the sides again leaving the shot blasted face unmodified. Table 26 summarized the shot blasting conditions used for both sets.
AC
M
AC
M
AC
M
AC
M
Figure 104. The shot blasting set up for the fatigue test samples
Table 26. Summary of shot blasting conditions used Size Type Composition Exposure time Configuration
Tensile samples S-550 (0.06 in dia.) Bainitic steel shot 0.7-1.1%C 0.6-1.2%Mn 0.4-1.2%Si 1 and 5 min all round, on rotating table
141
Fatigue samples S-660 (0.07 in dia.) Iron shot >1.7%C 0.6-1.0%Mn 0.8-1.2%Si 32 sec one face, on steel scrap wheel
6.3 Effect of shot blasting on the casting skin effect 6.3.1 Tensile properties After shot blasting, the roughness average increased from 12.0±3 to 15.1±1.6 and 15.8±1.6 µm for 1 and 5 minutes of shot blasting, respectively. It is seen that the shot blasting time did not have significant influence on the roughness average.
Figure 105 shows the average skin thickness as a function of test conditions. It is seen that one minute of shot blasting decreased the average skin thickness from 0.131mm to 0.071mm. Further increase of the blasting time to five minutes results in additional decrease of the skin to 0.023mm. Figure 106 shows representative microstructures of the casting skin for as-cast and shot blasted samples. The spheroidal graphite particles are closer to the casting surface as shot blasting time increase. Yet, all the figures are from plates having the same thickness (10.2 mm) cast in consecutive phenolic urethane bonded sand molds. Therefore, these samples have comparable cooling rates and pouring temperatures, and therefore should have comparable as-cast skin thickness. It is thus apparent that the casting skin has been decreased by shot blasting. Furthermore, tensile strength was found to be improved by shot blasting even above the level of the machined plates. Therefore, removal of casting skin alone cannot explain the increasing of tensile strength and work hardening generated by shot blasting should contribute to this effect to some extent.
142
Skin thickness (visual), mm
0.15
0.10
0.05
0.00 AC
SB1
SB2
Test condition
Figure 105. Effect of shot blasting on visual skin thickness (SB1 = 1 minute, SB2 = 5 minutes)
As-cast
SB1
SB2
Figure 106. Comparison of casting skin of as cast (left) 1 minute shot blasted (middle) and 5 minute shot blasted sample (right). Unetched.
Tensile testing was performed for test plates having one of the four surface conditions: as-cast (AC), machined (M), shot blasted for 1 minute (SB1), or shot blasted for 5
143
minutes (SB5). A summary of the results is presented in Figure 107, where the average tensile properties for samples having the same surface condition are plotted.
It is seen that shot blasting improves the strength of machined specimens, and that 5 minutes of treatment improves the strength over the one minute treatment. Similar improvements are seen for the elongation except 5 minutes of shot blasting decreased the elongation to 5.1%.
7.0 6.0
370 350
Elongation, %
Tensile strength, MPa
390
330 310 290
5.0 4.0 3.0 2.0
270 250
1.0
AC
M SB1 Surface condition
SB2
AC
a)
M SB1 Surface condition
SB2
b)
Figure 107. Effect of surface condition on the tensile properties; AC - as-cast; M machined; SB1 - shot blasted 1 min.; SB5 - shot blasted 5 min; a) tensile strength; b) elongation
The reason for this increase is twofold. Firstly, it is reasonable to assume that the higher strength is the result of compression stress induced in the skin by shot blasting. Secondly,
144
as shot blasting decreases the thickness of the casting skin (see Table 27), it is expected to increase all tensile properties.
Table 27. Influence of shot blasting on the thickness (visual) of the casting skin Surface condition
As-cast
SB1
SB2
Thickness of casting skin, mm
0.183
0.125
0.0
It can be safely concluded that shot blasting can be used to diminish or even eliminate the effects of the casting skin.
6.3.2 Fatigue properties After shot blasting, the averages Ra for both as-cast and machined samples increased to 22.7 and 23.1 m (from 20.1 and 0.08 m), respectively. The average casting skin thickness decreased from 165.6 to 31.3 m. The average pearlitic rim thickness decreased from 139.0 to 59.4 m. This indicates that the casting skin was partially abraded off by the shot blasting.
Figure 108 shows the S-N curves of the as-cast (AC), machined (M), as-cast-shot-blasted (AC-SB) and machined-shot-blasted (M-SB) for 15% nodularity CG iron. It is seen that shot blasting improved the fatigue limits for both as-cast and machined samples. The percentages of improvement were 43.5 and 21.4% for AC-SB and M-SB samples respectively. No significant difference in fatigue limits between AC-SB and M-SB samples is seen. This implies that the casting skin effect can be eliminated by shot 145
blasting (the fatigue skin factor is close to 1). It should be emphasized that the fatigue limit was improved although the surface roughness was increased. This effect is attributed to the development of a work hardened layer and the removal of the casting skin by shot blasting. This was confirmed by microstructure examination of the shot blasted surfaces.
500 AC M AC-SB M-SB
Cyclic Stress, MPa
400
300
200
100 1.0E+04
1.0E+05
1.0E+06
1.0E+07
Nf, cycles
Figure 108. S-N curves of the as-cast (AC), machined (M), as-cast-shot-blasted (AC-SB) and machined-shot-blasted (M-SB) for 15% nodularity CG iron
Figure 109 shows the comparison of the color micrographs of the structure near the sample surface before and after shot blasting. Color metallography reveals the interdendritic regions (the brown regions in Figure 109) because of microsegregation within the structure. It is seen that graphite particles and the adjacent austenite were severely deformed by shot blasting for both AC-SB and M-SB samples. This indicates the work hardening on the sample surface. It is worth noting that shot blasting did not 146
completely remove the cast skin features. The pearlitic rim and graphite degradation layer can still be seen.
a)
b)
c)
d)
Figure 109. Color micrographs of the casting skin before and after shot blasting; a) AC; b) AC-SB; c) M; d) M-SB
In order to quantify the effect of shot blasting, microhardness measurements were done on samples before and after shot blasting on the 15% nodularity CG iron. Figure 110 shows the hardness profiles of AC, M, AC-SB and M-SB samples. The locations of indentions did not avoid the graphite particles. The raw data are very scattered due to the 147
lack of homogeneity of the microstructure. Therefore, the Gaussian smoothing technique was used to obtain the trends. The Gaussian function, G(x) is [32], 1
G ( x)
2
e
x2 2 2
where is the standard deviation of the distribution and x is the raw data. In this case was set at 5. The results after Gaussian smoothing are shown as solid or dashed lines in Figure 110.
The as-cast sample exhibits higher hardness near the casting surface than the machined sample. This is a result of the presence of the pearlitic rim. The hardness in this region was as high as 350 HV100. Moving inward, the hardness of the as-cast sample decreased to the level to the machined sample. The hardness scatter is the result of pearlite patches. The hardness profile of the machined sample is less scattered because of less pearlite in the structure.
Figure 111 shows an example of the microstructure where the indentations were done before and after etching with 4% Nital. The distances between indentations were approximately 50 m apart from each other.
148
250
Guassian AC M
AC Gaussian AC
350 Microhardness, HV100
225 Microhardness, HV100
400
AC
Guassian M
200 175 150 125
AC-SB Guassian AC-SB
300 250 200 150
100
100 0
200
400
600
800
1000
0
Distance from surface, m
200
600
800
1000
Distance from surface, m
a)
b)
400
400
M Gaussian M M-SB
AC-SB Guassian AC-SB M-SB Guassian M-SB
350 Microhardness, HV100
350 Microhardness, HV100
400
Gaussian M-SB
300 250 200
300 250 200 150
150
100
100 0
200
400
600
800
1000
0
200
400
600
800
1000
Distance from surface, m
Distance from surface, m
c)
d)
Figure 110. Microhardness profiles of a) AC and M samples; b) AC and AC-SB samples; c) M and M-SB samples; d) AC-SB and M-SB samples
149
a)
b)
Figure 111. Microhardness measurement near the casting skin of an as-cast sample; a) before etching; b) after etching with 4%Nital
After shot blasting, the hardness increased for both the AC-SB and the M-SB samples. The affected depth was approximately 750 – 800 m (Figure 110b and Figure 110c). The hardness profile of AC-SB is slightly higher than that of M-SB (Figure 110d). This could be a result of the presence of the pearlitic rim.
6.4 Conclusion For the tensile test plates, one minute of shot blasting was effective in reducing the casting skin, while longer times (five minutes) eliminated the skin. This change in the casting skin was accompanied by a corresponding increase in the tensile strength and elongation, above those recorded on the machined samples. The average machined strength of 355MPa increased to 386MPa after one minute of shot blasting, and to 392MPa after five minutes. This demonstrates that shot blasting can be used successfully to alleviate the negative effects of the casting skin.
150
As for the fatigue properties, it was found that shot blasting improved the fatigue limit by 43.5 and 21.4% for AC-SB and M-SB samples, respectively. The result indicates that shot blasting is an effective method for casting skin elimination. Plastic deformation in microstructures and microhardness measurement demonstrate that the improvement is because of the work hardening created by shot blasting. The hardness on the shot blasted surface was as high as 350 HV100 as compared to 160 HV100 for the unaffected zone. The hardness penetration depth was approximately 750 – 800 m.
151
Chapter 7: Conclusions The occurrence of the casting skin has adverse effect on the mechanical properties of cast iron parts. While the problem was recognized by the cast iron industry, only limited information on the subject is available in the literature, in particular for CG iron. This research addressed many aspects of the casting skin formation and effects in CG and ductile irons. The objectives of the research included: -
To quantify the effect of the casting skin on the static and fatigue properties
-
To understand the mechanism of casting skin formation
-
To minimize the casting skin formation
-
To eliminate the casting skin effect
It was demonstrated that the image analysis technique is a viable approach for the quantification of the skin thickness. Research work was conducted to identify correlations between the casting skin features and the mechanical properties.
In order to quantify the casting skin effect, the skin factor was introduced as the ratio between the properties (i.e. tensile strength, fatigue strength) of as-cast samples and those of the machined samples. For tensile strength the average skin factor was 0.91. This implies that the casting skin decreases the tensile strength by 9%. Surface roughness and graphite degradation are believed to be the major contributors to this negative effect. The 152
maximum skin thickness in the experiment was 0.4 mm which produced 15.5% of tensile strength reduction. The data can be used as an additional safety factor for part designers. The casting skin effect of the casting skin fatigue strength was also quantified. The fatigue skin factors were 0.85, 0.73 and 0.68 for the CG irons with 15, 30 and 40% nodularity respectively. It is worth noting that CG iron is more susceptible to the casting skin effect than gray and ductile iron.
Several mechanisms of formation of the casting skin were identified during this research. Decarburization at the casting surface was suggested as the responsible mechanism of the ferritic rim and graphite depletion. On the other hand, carburization is responsible for the pearlitic rim formation.
The formation mechanisms of the graphite degradation were identified as Mg depletion and the effect of solidification kinetics. Large undercooling at the mold/metal interface induces the formation of primary austenite, which produces thicker skin. The affected area has fine-lamellar graphite with higher amount of austenite comparing to the bulk structure. The concept of solidification kinetics effects on casting skin formation was explained through the use of the Fe-Gr asymmetric phase diagram. Some parameters that promote the formation of austenite were identified such as using mold coating with high thermal conductivity and using low CE iron. On the other hand, using mold coating with inoculation effect inhibited the casting skin formation by decreasing the undercooling. The concept was proven experimentally. 153
Reactions between Mg and oxygen or sulfur create a Mg depletion zone which translates into graphite degradation. It has been proven that the thicker sections in casting produce larger skin thickness. This is due to the longer solidification time that extends the reaction time for Mg depletion. Therefore, Mg depletion is the predominated mechanism for the thicker sections. In general, the skin formation is controlled by Mg depletion at low cooling rate, and by austenite layer formation at the mold/metal interface at high cooling rate. A 2-D thermal diffusion model was developed to demonstrate the Mg depletion mechanism.
Several coating materials were applied to the molds in an attempt to minimize casting skin formation. Three types of mold coatings were used; i) inactive coatings; ii) active coatings; iii) reactive coatings. The inactive coatings (MgO, BN, zircon and mica) did not improve the casting skin because of they had higher thermal conductivities than the silica sand. Active coatings (FeSi and graphite) improved the casting skin by providing inoculation effect (FeSi) or by increasing the carbon equivalent at the mold/metal interface (graphite). CaO as a reactive coating improved the casting skin when sulfur was present in the molding materials. FeSiMg as another reactive coating effectively improved the casting skin by providing extra Mg at the mold/metal interface. Although a high surface roughness was observed for the FeSiMg coating, the composition of the coating can be optimized for better results in future research.
154
An attempt at the elimination of the casting skin effect on the mechanical properties was done by shot blasting. It was found that shot blasting effectively improved both tensile and fatigue strength. The casting skin was abraded off by shot blasting. In addition, the work hardened layer created by shot blasting was the key to property improvement.
155
References
[1] D. M. Stefanescu, Science and Engineering of Casting Solidification, New York: Springer, 2009. [2] ASM Handbook vol. 15, ASM International, 2008. [3] ASM Specialty Handbook "Cast Iron", ASM International, 1996. [4] L. Dix, R. Ruxanda, J. Torrance, M. Fukumoto and D. Stefanescu, "Static Mechanical Properties of Ferritic and Pearlitic Lightweight Ductile Iron Castings," AFS Transactions, pp. paper 03-109, 2003. [5] H. Reisener, "Some aspects of the formation and structure of a skin on iron castings and a metod used to obviate its occurence," The British Foundryman, pp. 362-269, 1962. [6] S. Matijasevic, J. Gomez-Gallardo and J. Wallace, "Ferritic Surface Layers on Gray Iron Castings," AFS Transactions, pp. 571-622, 1974. [7] R. P.J., "Suppression of Ferrite in Shell Molded Gray Iron Castings," AFS Transactions, pp. 101-108, 1976. [8] G. Narasimha and J. Wallace, "Factors Influencing the Ferritic Layer on the Surface
156
of Gray Iron Castings," AFS Transactions, pp. 531-550, 1975. [9] G. Goodrich and R. Lobenhofer, "Effect of Cooling Rate on Ductile Iron Mechanical Properties," AFS Transactions, pp. 1003-1032, 2002. [10] F. Mampaey, P. Li and E. Wettinck, "Variation of Gray Iron Strength Along the Casting Diameter," AFS Transation, pp. paper 03-056, 2003. [11] D. Stefanescu, S. Wills, J. Massone and F. Duncan, "Quantification of Casting Skin in Ductile and Compacted Graphite Irons and Its Effect on Tensile Properties," Internaltional Journal of Metalcasting, no. Fall, pp. 7-26, 2008. [12] I. Riposan, M. Chisamera, S. S. Stan and T. Skaland, "Surface Graphite Degeneration in Ductile Iron Castings for Resin Molds," Tsinghua Science and Technology, vol. 13, no. 2, pp. 157-163, 2008. [13] R. Aufderheide, R. Showman and M. Hysell, "Controlling the "Skin Effect" on ThinWall Ductile Iron Castings," AFS Transactions, pp. 567-579, 2005. [14] M. Starkey and P. Irving, "A Comparison of the Fatigue Strength of Machined and As-cast Surfaces of SG Iron," International Journal of Fatigue, pp. 129-136, 1982. [15] C. Labrecque, M. Gagne, P. Cabanne, C. Francois, C. Becret and F. Hoffmann, "Comparative Study of Fatigue Endurance Limit for 4 and 6 mm Thin Wall ductile iron Castings," International Journal of Metalcasting, pp. 7-17, 2008. [16] J. Torrance and D. Stefanescu, "Investigation on the Effect of Surface Roughness on the Static Mechanical Properties of Thin-Wall Ductile Iron Castings," AFS Transactions, pp. paper 04-013, 2004. 157
[17] D. F.C. and J. Kroker, "A New Test Casting to Evaluate Skin Formation in CGI," AFS Transactions, pp. paper 10-023, 2010. [18] F. Martin and S. Karsay, "Localized Flake Graphite Structure as a Result of a Reaction between Molten Ductile Iron and Some Components of the Mold," AFS Transactions, pp. 221-226, 1979. [19] H. Xiaogan, X. Jin, D. Xuqi and W. Yaoke, "Nodular Iron Surface Deterioration Due to PTSA in Resin," AFS Transactions, pp. 9-15, 1992. [20] N. Ivan, M. Chisamera and I. Riposan, "Mold Coatings to Reduce Graphite Degeneration in The Surface Layer of Ductile Iron Castings," International Journal of Metal Casting, no. Fall, pp. 61-70, 2012. [21] J. Schey, Introduction to Manufacturing Processes, Boston: McGraw Hill, 2000. [22] M. Kawamoto, T. Iwamoto and T. Saeki, "The Effect of Skin Fatigue Resistance of Cast Iron," The Japan Society of Mechanical Engineers, vol. 17, pp. 139-142, 1951. [23] D. Stefanescu, S. Giese, T. Piwonka and A. Lane, "Cast Iron Penetration in Sand Molds Part I: Physics of Penetration Defects and Penetration Model," AFS Transactions, pp. 1233-1248, 1996. [24] R. Suarez, Private communication, 2013. [25] N. Voronova, Desulfurization of Hot Metal by Magnesium, Dayton: The International Magnesium Association, 1983. [26] J. Tinebar and S. Wilson, "Nobake Chemical Binder Systems: Their Effect on Microstructural and Physical Properties of Ductile Iron," AFS Transactions, pp. 169158
174, 1993. [27] F. Woehlbier, Diffusion and Defect Data, Rockport: Trans Tech Publications, 1978. [28] T. Massalski, Binary Alloy Phase Diagram, Metals Park: ASM International, 1986. [29] A. Goldsmith, T. Waterman and H. H.J., Handbook of Thermophysical Properties of Solid Materials, New York: MacMillan, 1961. [30] Y. Touloukian, P. Liley and S. Saxena, Thermal Conductivity: nonmetallic liquids and gases, New York: IFI/Plenum, 1970. [31] A. Catalina, Private communication, 2010. [32] J. Burkill and H. Burkill, A Second Course in Mathematical Analysis, Cambridge: University Press, 1970.
159
Appendix A: Tensile testing data No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Binder
Cond.
PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU
AC AC AC SB2 SB2 SB2 AC AC AC M M M SB1 SB1 SB1 AC AC AC M M M AC AC AC AC AC AC M M M M M M AC
Thickness (mm) 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6
Heat 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90130 90220 90220 90220 90220 90220 90220 90220
160
TS (MPa) 343.4 295.1 329.0 322.7 319.9 295.8 348.2 328.2 334.0 411.0 408.5 362.0 313.0 308.9 291.6 366.7 382.8 338.3 445.9 395.8 367.6 407.7 401.5 376.6 345.4 328.2 274.0 366.4 363.3 347.8 361.2 350.7 347.4 323.4
YS (MPa) 286.8 288.9 279.0 251.5 246.0 243.2 341.3 323.4 281.0 330.0 325.0 305.0 257.7 248.8 228.7 316.5 314.4 176.8 335.0 330.0 310.0 322.0 316.5 285.4 295.0 280.0 255.0 305.0 300.0 290.0 305.0 295.0 295.0 285.0
%EL (%) 4.6 1.6 2.3 1.0 5.0 4.1 2.6 4.8 6.3 3.1 3.5 3.3 0.9 5.5 5.8 4.7 3.7 8.9 4.9 5.0 4.2 0.4 5.2 7.1 6.3 4.6 5.7 5.6 4.2
Ra (m) 16.0 14.5 11.6 16.3 15.7 15.6 15.4 12.7 9.9 0.6 0.6 0.6 17.4 15.5 15.5 13.1 8.6 9.6 0.6 0.6 0.6 20.0 14.2 11.7 18.1 10.7 9.4 0.6 0.6 0.6 0.6 0.6 0.6 13.4
No. 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
Binder
Cond.
PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU PU SS SS SS SS SS SS SS SS SS SS SS SS
AC AC AC AC AC SB1 SB1 SB1 SB2 SB2 SB2 AC AC AC M M M AC AC AC M M M SB1 SB1 SB1 SB2 SB2 SB2 M M M AC AC AC AC AC AC M M M
Thickness (mm) 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2 7.6 10.2 15.2
Heat 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220 90220
161
TS (MPa) 301.3 295.0 323.4 311.6 297.0 391.7 389.7 376.6 400.1 406.3 369.0 336.5 318.6 366.9 366.3 385.7 355.0 351.7 366.9 361.4 378.6 379.5 357.4 358.6 365.3 355.7 357.2 373.8 325.4 360.0 361.1 340.7 324.7 303.4 305.6 398.7 360.7 365.5 427.8 365.4 376.3
YS (MPa) 281.0 275.0 273.2 268.0 264.0 297.8 192.1 299.2 139.6 336.5 181.7 266.0 300.0 315.0 300.0 351.7 197.6 180.4 315.0 310.0 300.0 358.6 167.9 156.2 340.7 140.3 300.0 300.0 295.0 324.0 260.0 261.2 393.9 249.2 245.3 325.0 315.0 310.0
%EL (%) 2.9 0.9 2.1 3.1 1.1 8.7 3.9 4.2 5.5 5.9 7.1 4.7 7.0 4.6 6.4 7.5 4.7 5.5 4.4 6.3 5.2 7.1 5.7 5.7 5.0 3.2 6.3 3.2 3.0 1.4 1.3 8.9 6.5 7.7 7.0 6.3 4.2
Ra (m) 10.6 9.3 16.6 13.8 9.3 16.2 16.4 16.4 15.2 13.7 14.2 12.1 11.6 8.6 0.6 0.6 0.6 10.9 10.4 8.9 0.6 0.6 0.6 13.6 12.5 12.4 17.7 17.4 16.9 0.6 0.6 0.6 21.7 20.9 18.9 21.5 20.6 20.0 0.6 0.6 0.6
No. 76 77 78
Binder
Cond.
SS/PU SS/PU SS/PU
AC AC AC
Thickness (mm) 7.6 10.2 15.2
Heat 90220 90220 90220
TS (MPa) 382.1 372.5 367.6
YS (MPa) 275.7 254.1 194.2
%EL (%) 7.6 6.5
Ra (m) 18.8 21.4 19.5
PU = Phenolic urethane; SS = Sodium silicate; TS = tensile strength; YS = yield strength; %EL = percent elongation; Ra = roughness average
162
Appendix B: Fatigue testing data No.
Heat
Cond.
m
a
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17* 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100707 100701 100701 100701 100701 100701
AC AC AC AC M M M M AC AC AC AC M M M M AC AC AC AC M M M M AC AC AC AC M M M M AC AC AC AC M
MPa 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 300 300 300 300 300 300 300 300 300 300 300 300 300
MPa 206.2 275.5 228.6 215.0 293.7 313.9 333.3 327.0 348.4 303.6 253.2 220.6 348.9 379.9 349.4 305.0 439.0 409.5 439.4 369.5 453.1 400.4 415.3 363.5 219.9 234.6 183.1 197.7 300.0 315.7 265.0 281.5 191.2 170.0 243.1 209.3 289.3
Nf Cycles 1.00E+07 6.20E+05 1.88E+06 1.00E+07 5.50E+06 5.22E+06 8.81E+05 1.28E+06 1.22E+05 2.05E+05 8.46E+05 2.10E+06 6.20E+05 2.09E+05 5.19E+05 5.50E+06 1.15E+04 4.97E+04 3.53E+04 8.87E+04 1.18E+05 1.78E+05 1.51E+05 4.19E+05 7.81E+05 6.86E+05 5.00E+06 1.54E+06 1.43E+06 7.83E+05 5.06E+06 5.05E+06 2.45E+06 5.06E+06 5.03E+05 1.14E+06 1.66E+06
163
Ra µm 20.08 21.30 16.15 18.64 0.11 0.10 0.80 0.13 17.15 17.48 20.11 19.43 0.10 0.11 0.12 0.10 15.29 15.63 18.13 19.42 0.40 0.51 0.38 0.90 18.10 20.11 23.80 21.14 0.15 0.15 0.12 0.16 15.60 16.78 21.34 18.73 0.32
Sk thGr mm 0.075 0.075 0.075 0.075 0.000 0.000 0.000 0.000 0.075 0.075 0.075 0.075 0.000 0.000 0.000 0.000 0.075 0.075 0.075 0.075 0.000 0.000 0.000 0.000 0.075 0.075 0.075 0.075 0.000 0.000 0.000 0.000 0.075 0.075 0.075 0.075 0.000
Nod. % 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40 40
%Pe % 13.5 12.0 15.3 17.7 14.3 12.1 13.4 10.8 13.5 14.5 15.7 14.1 16.2 12.1 10.8 10.7 11.1 12.3 13.9 14.4 16.0 15.2 14.8 13.9 14.3 11.9 12.4 10.6 10.1 9.8 13.9 14.4 12.5 15.3 13.1 14.3 11.9
No.
Heat
Cond.
m
a
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81
100701 100701 100701 100701 100701 100701 100701 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101015 101108.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.2 101108.2 101108.2 101108.2 101108.2 101108.2 101108.2 101108.2 101101.1 101101.1 101101.1 101101.1
M M M AC AC M M AC AC AC AC AC AC AC AC AC M M M M M M M M AC AC AC AC AC AC AC AC AC AC AC AC AC AC AC AC AC-SB AC-SB AC-SB AC-SB
MPa 300 300 300 649.4 645.1 768.8 762.5 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250
MPa 336.9 345.5 365.1 0 0 0 0 324.3 244.3 273.2 224.1 196.7 174.9 186.9 201.3 193.6 338.1 318.2 303.7 280.8 273.4 266.1 250.8 264.2 221.0 256.7 283.2 277.6 301.1 191.0 323.2 238.6 185.9 183.0 227.2 178.7 238.5 288.5 199.4 209.4 355.7 334.4 313.2 304.7
Nf Cycles 4.62E+05 3.00E+05 2.53E+05 0.5 0.5 0.5 0.5 1.45E+05 4.02E+05 2.57E+05 9.95E+05 2.50E+06 1.00E+07 1.00E+07 1.04E+06 5.04E+06 3.56E+05 4.72E+05 8.36E+05 1.22E+06 2.43E+06 5.00E+06 5.10E+06 5.20E+06 5.89E+05 2.70E+05 1.64E+05 1.67E+05 9.11E+04 2.00E+06 6.38E+04 2.75E+05 5.49E+06 9.14E+06 4.68E+05 5.53E+06 3.82E+05 1.36E+05 1.73E+06 1.10E+06 2.97E+05 4.66E+05 5.80E+05 5.32E+05
164
Ra µm 0.41 0.20 0.23 17.88 19.15 0.25 0.24 18.18 17.94 19.83 20.11 21.83 22.06 18.24 19.23 21.84 0.08 0.08 0.08 0.10 0.11 0.09 0.09 0.08 23.65 22.13 22.07 16.97 23.20 20.09 18.45 19.22 20.18 17.83 19.24 19.18 20.18 22.02 19.38 18.37 23.82 21.63 23.67 22.98
Sk thGr mm 0.000 0.000 0.000 0.075 0.075 0.000 0.000 0.125 0.100 0.125 0.075 0.125 0.100 0.125 0.125 0.125 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.175 0.125 0.150 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.125 0.175 0.150 0.175 0.025 0.025 0.025 0.050
Nod. % 40 40 40 40 40 40 40 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 30 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
%Pe % 10.8 13.2 11.6 13.3 12.1 14.2 14.2 11.8 12.3 12.1 10.7 11.4 12.0 10.8 10.3 12.2 11.8 10.5 9.9 12.3 11.6 11.4 10.2 10.7 10.5 10.0 9.7 9.8 9.9 10.3 11.0 10.7 10.4 11.0 12.0 12.5 13.5 15.3 14.3 13.8 14.2 12.1 10.7 16.1
No.
Heat
Cond.
m
a
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 107 108 109 110
101101.1 101101.1 101101.1 101101.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.1 101108.2 101108.2 101108.2 101108.2 101108.2 101108.2 101108.2 101108.2 101101.1 101101.1 101101.1 101101.1 101101.1 101101.1 101101.1 101101.1
AC-SB AC-SB AC-SB AC-SB M M M M M M M M M M M M M M M M M-SB M-SB M-SB M-SB M-SB M-SB M-SB M-SB
MPa 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250 250
MPa 275.3 284.2 266.7 295.3 333.2 282.8 269.5 230.0 221.1 241.0 225.3 232.0 299.6 208.4 317.7 216.0 251.3 282.4 218.5 250.0 316.4 346.2 289.7 364.7 280.0 328.3 276.3 265.3
Nf Cycles 1.46E+06 1.57E+06 6.15E+06 1.04E+06 1.33E+05 2.56E+05 3.74E+05 1.02E+06 2.31E+06 5.56E+05 1.64E+06 8.48E+05 2.16E+05 4.95E+06 1.51E+05 5.00E+06 4.18E+05 2.85E+05 6.00E+06 5.28E+05 4.45E+05 3.52E+05 9.66E+05 1.92E+05 1.49E+06 4.78E+05 2.01E+06 5.00E+06
Ra µm 23.18 22.66 21.65 22.06 0.09 0.08 0.07 0.09 0.08 0.08 0.06 0.06 0.07 0.08 0.06 0.08 0.07 0.06 0.09 0.08 21.56 23.79 20.56 24.28 25.70 20.07 25.97 22.45
Sk thGr mm 0.050 0.025 0.025 0.025 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Nod. % 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
%Pe % 14.8 14.2 14.1 13.2 10.8 12.3 14.1 13.2 12.7 11.1 10.3 13.9 12.1 13.5 9.7 10.0 9.5 12.8 13.2 11.8 13.7 10.2 11.7 11.4 10.2 15.1 14.3 16.7
m = mean stress; a = alternating stress; Nf = number of cycles to failure; Ra = roughness average; Sk thGr = skin thickness by graphite shape factors; Sk thvis = visual skin thickness; Nod = percent nodularity; %Pe. = percentage pearlite; *excluded from analyses and S-N plot
165
