The Effects of Copper on the Metallurgical, Mechanical, and Fracture Properties of 0.90 Carbon Rail Steels

THE EFFECTS OF COPPER ON THE METALLURGICAL, MECHANICAL, AND FRACTURE PROPERTIES OF 0.90 CARBON RAIL STEELS.

Glenn T. Eavenson

A thesis submitted to the Faculty and Board of Trustees of the Colorado School of Mines in partial fulfillment of the requirements for the degree of Master of Science (Metallurgical and Materials Engineering).

Golden, Colorado Date: ______

Signed:______Glenn T. Eavenson

Signed:______Dr. David K. Matlock Thesis Advisor

Golden, Colorado Date:______

Signed:______Professor Ivar Reimanis Professor and Interim Head Department of Metallurgical and Materials Engineering

ABSTRACT

The effects of variations in copper content on the metallurgical, mechanical, and fracture properties of pearlitic rail steels with a base composition (in wt pct) of 0.9 C, 1.0 Mn, 0.35 Si, and 0.01 Ti were evaluated. Six industrial heats with copper content ranging from (in wt pct) 0.07 to 0.85 were cast, re-heated, rolled and air-quenched with identical industrial processing parameters to produce full rails of the 136RE section. The materials were tested to determine the influence of copper content on austenitic grain size, pearlitic interlamellar spacing, microstructure, hardenability, hardness profile, tensile and yield strength, Charpy U-notch impact toughness, K1c fracture toughness, and fatigue crack growth rate according to standard ASTM testing methodologies. The austenitic grain size, as determined by the McQuaid-Ehn method, suggested that copper does not influence the austenite grain growth characteristics in the temperature range of the test for the steels evaluated. Jominy end-quench hardenability testing showed that copper acts to delay pearlite transformation to a small degree. Pearlite interlamellar spacing measurements, determined via scanning electron microscopy, indicated that increased copper content refines the interlamellar spacing, which agrees with the Jominy hardenability data. The refinement in pearlite interlamellar spacing with increasing copper content increased hardness and strength according to a Hall-Petch type relationship. Both the impact toughness and fatigue crack growth rates were essentially independent of copper content. There was a slight decrease in fracture toughness with increasing copper content most likely due to the increase in yield strength with increasing copper content. Nonetheless, copper does not diminish the strength-toughness balance in pearlitic rail steels, and copper appears to primarily act as another hardenability element during the production of steel rails. The results of this study suggest that the current maximum allowable copper content of 0.4 wt pct specified by the American Railway Engineering and Maintenance-of-Way Association (AREMA) may be too restrictive for modern steel rail production technology.

iii TABLE OF CONTENTS

ABSTRACT ...... iii

LIST OF FIGURES ...... vi

LIST OF TABLES ...... xi

ACKKNOWLEDGEMENTS ...... xii

CHAPTER 1 INTRODUCTION ...... 1

CHAPTER 2 BACKGROUND AND LITERATURE REVIEW ...... 4 2.1 Introduction ...... 4 2.2 Microstructural considerations ...... 5 2.3 Hardenability ...... 8 2.4 Mechanical Properties ...... 10 2.5 Fracture Mechanics...... 11 2.5.1 Charpy Impact Toughness ...... 12

2.5.2 Plane Strain Fracture Toughness (K1c) ...... 12 2.5.3 Fatigue Crack Growth Rate ...... 14

CHAPTER 3 EXPERIMENTAL DESIGN ...... 16 3.1 Experimental Alloys ...... 16 3.2 Microstructural Evolution...... 18 3.2.1 Austenite Grain Size ...... 18 3.2.2 Interlamellar Spacing ...... 19 3.2.3 Digital (Optical) Microscopy ...... 20 3.2.4 Hardenability ...... 20 3.3 Mechanical Properties ...... 21 3.3.1 Tensile Properties ...... 22 3.3.2 Hardness ...... 22 3.4 Fracture Mechanics...... 23 3.4.1 Charpy Impact Test ...... 23

3.4.2 K1c Fracture Toughness ...... 24 3.4.3 Fatigue Crack Growth Rate ...... 25

CHAPTER 4 EXPERIMENTAL RESULTS...... 28 4.1 Microstructural Evolution...... 28

iv 4.1.1 Austenitic Grain Size Comparisons...... 28 4.1.2 Pearlite Interlamellar Spacing ...... 32 4.1.3 Optical / Digital Microscopy ...... 34 4.1.4 Hardenability...... 35 4.2 Mechanical Properties ...... 46 4.2.1 Tensile Results ...... 46 4.2.2 Hardness at Depth ...... 46 4.3 Fracture Mechanics ...... 50 4.3.1 Charpy Impact Toughness ...... 50 4.3.2 Fracture Toughness ...... 52 4.3.3 Fatigue Crack Growth Rate ...... 52

CHAPTER 5 DISCUSSION ...... 59 5.1 Copper effect on pearlitic rail microstructure ...... 59 5.2 Mechanical Properties ...... 61 5.3 Fracture Mechanics ...... 65

CHAPTER 6 CONCLUSIONS ...... 69

CHAPTER 7 FUTURE WORK ...... 70 7.1 Wear rates ...... 70 7.2 Copper strengthening mechanisms ...... 70 7.3 Weld testing ...... 70

REFERENCES CITED ...... 72

APPENDIX A ...... 78

v LIST OF FIGURES

Figure 1.1 Photograph showing hot shortness during hot rolling of Standard Strength rail steel. Black arrows indicate plane of polish for Figure 1.2. Photo courtesy of Evraz Pueblo ...... 2

Figure 1.2 Light optical micrograph of tranverse section from Figure 1.1. Bright white areas (two denoted by black arrows) are copper as verified by energy dispersive spectroscopy. Unetched light optical micrograph courtesy of Evraz Pueblo...... 2

Figure 2.1 Rail wear as a function of number of wheel passes for pearlitic steels of various hardness (HBW). Adapted from Stock et al. [11]...... 5

Figure 2.2 Macro photographs of polished and etched transverse sections of (a) off- line, and (b) in-line head hardened pearlitic rail, 2% nital etch. Photos courtesy of Evraz Pueblo...... 5

Figure 2.3 Pearlite spacing as a function of isothermal transformation temperature for a hypereutectoid steel austenitized at four different temperatures to produce four different austenite grain sizes (γGS). Plot reconstructed from Elwazri et. al [19]...... 6

Figure 2.4 Comparison of interlamellar spacing measurements obtained from laser scaning confocal (LSCM) and field emission scanning electron (FE-SEM) microscopy on a hypereutectoid (V0) and micro-alloyed hypereutectoid (V) steel at two isothermal transformation temperatures. Plot constructed from the work of Elwazri et al. [30]...... 8

Figure 2.5 Calculated hardenability curves for a high (4140) and low (A36) hardenability steel constructed from Li et al. [42]. Calculations are in accordance with ASTM-A255 [40]...... 9

Figure 2.6 Strength dependence on interlamellar spacing for 0.80 wt pct C rail steel. Created from Gomes et al. [14]...... 11

Figure 2.7 Yield strength dependence on interlamellar spacing for 1.5 and 1.8C wt pct ultra-high carbon steels. Re-constructed from Taleff et al. [16]...... 11

Figure 2.8 Effect of temperature and carbon content on notch toughness. Reconstructed from Metals Handbook [54]...... 12

Figure 2.9 Relationship between tensile ductility (reduction of area) and plane strain fracture toughness for five rail steels. Adapted from the work of Ochi et al. [56]. Steel compositions are shown in Table 2.1...... 13

vi Figure 2.10 Macrophotograph of a Detail Fracture showing longitudinal cracking and transverse fatigue in the head of a rail. Photo courtesy of Evraz Pueblo / TTCI [58] ...... 14

Figure 2.11 Fatigue crack growth rate behavior of four rail steels. Reconstructed from the work of Ochi et al. [56]...... 15

Figure 3.1 136 RE Rail section excerpted from AREMA Chapter 4 [69]. Dimensions are in inches...... 17

Figure 3.2 Sectioning schematic for digital (optical) microscopy specimens...... 20

Figure 3.3 Jominy hardenability test schematics: sectioning diagram (a), and specimen dimensions (b)...... 21

Figure 3.4 Tensile specimen blank sectioning schematic...... 22

Figure 3.5 Rockwell C hardness traverse schematics; sectioning diagram (a), and indentation map (b)...... 23

Figure 3.6 Charpy U-notch impact test schematics; sectioning diagram (a), and specimen dimensions (b)...... 24

Figure 3.7 K1C fracture toughness specimen schematics; sectioning diagram (a), and specimen dimensions (b)...... 25

Figure 3.8 Fatigue crack growth rate specimen schematics; sectioning diagram (a) and specimen dimensions (b)...... 27

Figure 4.1 Light optical micrographs showing prior austenite grain size images of (a) 7 Cu, (b) 11 Cu, (c) 22 Cu, (d) 29 Cu, (e) 38 Cu, and (f) 85 Cu alloys. Specimens prepared from un-quenched portion of Jominy specimens, saturated picric acid etch...... 29

Figure 4.2 Light optical micrographs showing prior austenite grain size images of (a) 7 Cu, (b) 11 Cu, (c) 22 Cu, (d) 29 Cu, (e) 38 Cu, and (f) 85 Cu alloys. Specimens prepared from un-quenched portion of Jominy specimens, saturated picric acid etch...... 30

Figure 4.3 Average austenitic ASTM grain size number as a function of copper content using the circular intercept method. Error bars represent one standard deviation from the mean...... 31

vii

Figure 4.4 Representative FE-SEM (10.0 kV) secondary electron micrographs used to measure pearlite inter-lamellar spacing of the specimens. The white layers are cementite and the dark layers are ferrite. White arrows indicate possible copper precipitates. Measurements taken at depths of 6 mm (a, b, c), and 10 mm (d, e, f) from the running surface. Saturated picric acid etch...... 33

Figure 4.5 Inter-lamellar spacing as a function of copper content at depths of 6 mm and 10 mm using the linear intercept method. Error bars represent one standard deviation from the mean for the overall average...... 34

Figure 4.6 Representative transverse light optical micrographs of 7 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeutectoid ferrite. Nital (2%) etch...... 36

Figure 4.7 Representative transverse light optical micrographs of 11 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeuctectoid ferrite. Nital etch (2%)...... 37

Figure 4.8 Representative transverse light optical micrographs of 22 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeuctectoid ferrite. Nital (2%) etch...... 38

Figure 4.9 Representative transverse light optical micrographs of 29 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeutectoid ferrite. Nital (2%) etch...... 39

Figure 4.10 Representative transverse light optical micrographs of 38 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeuctectoid ferrite. Nital (2%) etch...... 40

Figure 4.11 Representative transverse light optical micrographs of 85 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeutectoid ferrite. Nital (2%) etch...... 41

Figure 4.12 Light optical micrographs for each alloy selected to show the “worst case” surface condition for the 7 Cu (a), 11 Cu (b), 22 Cu (c), 29 Cu (d), 38 Cu (e), and 85 Cu (f) specimens. Nital (2%) etch...... 42

Figure 4.13 Hardenability curves for the six alloys over the first 12 hardness readings...... 43

viii

Figure 4.14 Relationship between multiplying factor used to determine DI and the wt pct carbon based on ASTM-A255 [40]...... 43

Figure 4.15 Relationship between the multiplying factor used to determine DI and the wt pct copper based on ASTM-A255 [40]...... 44

Figure 4.16 Comparison of experimental vs. predicted hardenability curves for the 7 Cu (a), 11 Cu (b), 22 Cu (c), 29 Cu (d), 38 Cu (e), and 85 Cu (f) alloys...... 47

Figure 4.17 Tensile and 0.2% offset yield strength as a function of copper content. Error bars are one standard deviation from the arithmetic mean ...... 48

Figure 4.18 Hardness as a function of depth from running surface for the 7 Cu, 38 Cu and 85 Cu alloys. Error bars represent one standard deviation from the arithmetic mean...... 49

Figure 4.19 Macro photograph images of representative broken Charpy U-notch (2 mm) specimens for the (a) 7 Cu, (b) 11 Cu, (c) 22 Cu, (d) 29 Cu, (e) 38 Cu, and (f) 85 Cu alloys...... 51

Figure 4.20 Average Charpy U-notch absorbed impact energy as a function of copper content. Error bars represent one standard deviation from the mean...... 52

Figure 4.21 Average KIc fracture toughness as a function of copper content. Error bars represent one standard deviation from the mean...... 53

Figure 4.22 Average K1c fracture toughness as a function of austenitic grain size. Error bars represent one standard deviation from the mean...... 54

Figure 4.23 da/dN vs. ∆K plots for the 7 Cu (a), 38 Cu (b), and 85 Cu (c) alloys. The data from all three alloys are combined in (d)...... 56

Figure 4.24 Representative S.E.M. images of the 7 Cu, 38 Cu, and 85 Cu alloys at the fatigue crack initiation (a, b, c) and at stable crack growth (d, e, f) locations respectively. White arrows indicate direction of crack propagation...... 57

Figure 4.25 Representative S.E.M. images of 7 Cu, 38 Cu, and 85 Cu alloys at the unstable crack growth (a, b, c) and final fracture (d, e, f) locations respectively. White arrows indicate direction of crack propagation...... 58

Figure 5.1 Comparison of calculated hardenability curves for various steel alloys including the 7 Cu, 38 Cu, and 85 Cu alloys from this study. Curves adopted from other sources include pearlitic rail steels from: Ochi (Cr-Mo) [56], Grigorivich (Cr-Si) [78], Aglan (HH) [60], and Dauebler (Carbon) [77] as well as a SAE 4140 [79] and SAE 1045 [4] for comparitive purposes...... 61

Figure 5.2 Hall-Petch type relationship between hardness and pearlite spacing for the 7 Cu, 38 Cu, and 85 Cu alloys at a depth of 10 mm from the running surface...... 62

Figure 5.3 Hall-Petch type relationship between strength and pearlite spacing for the 7 Cu, 38 Cu, and 85 Cu alloys at a depth of 10 mm from the running surface...... 62

Figure 5.4 Hall-Petch relationship between yield strength and pearlite spacing for the 7 Cu, 38 Cu, and 85 Cu alloys of this study along with a 1080 and 1080 Nb steels from Gomes et al. [14] and two ultra high carbon steels from Taleff et al. [16]...... 63

Figure 5.5 Hall-Petch relationship between yield strength and interlamellar spacing highlighted to illustrate the comparison between this study and the equivalent range of pearlite spacing for the data from Taleff et al. [16]...... 64

Figure 5.6 Relationship between plane strain fracture tougness and tensile ductility. Horizontal and vertical error bars represent one standard deviation from the mean...... 66

Figure 5.7 Relationship between plane strain fracture toughness and tensile ductility for the alloys of this study along with published results from Ochi et al. [56]...... 66

Figure 5.8 Plastic zone size calculated from Equation 5.1 as a function of tensile ductility as measured by percent reduction of area...... 67

Figure 5.9 Comparison of steady state region of fatigue crack growth rate curves plotted from the Paris equation for various pearlitic steels...... 68

x LIST OF TABLES

Table 2.1 Chemical Composition of Rail Steels from Ochi et al.[56]...... 13

Table 3.1 Chemical Composition of Selected Alloys in wt pct...... 17

Table 4.1 Summary Statistics of the Circular Intercept Method for ASTM Grain Size Determination ...... 31

Table 4.2 Summary Statistics for Interlamellar Spacing (nm) ...... 32

Table 4.3 P-values for Statistical Comparison of Lamellar Spacing ...... 34

Table 4.4 Calculated Ideal Diameter from Chemical Content According to ASTM- A255 [40] ...... 45

Table 4.5 Average Ultimate Tensile Strength (UTS), Yield Strength (YS), and % Reduction of Area (% R/A) for the Alloys Tested...... 46

Table 4.6 Average Hardness and Standard Deviation Values for the Hardness Measurements at Depth; These Values are Shown Graphically in Figure 4.18 ...... 49

Table 4.7 Hardness Comparison Rockwell C to Brinell ...... 50

Table 4.8 Room Temperature Charpy U-notch (2 mm) Absorbed Impact Energy ...... 50

Table 4.9 Average K1c Fracture Toughness for the Six Alloys Tested. Data are Shown in Figure 4.21 ...... 53

Table 4.10 Paris Law Regression Results for the Fatigue Crack Growth Rate Analysis ...... 55

Table 5.1 Austenite Grain Size Summary ...... 59

Table 5.2 Results of Linear Regression from Figure 5.4 ...... 63

Table 5.3 Linear regression results from Figure 5.5 ...... 65

xi ACKKNOWLEDGEMENTS

I would like to thank my advisor, Dr. David Matlock for the many years of mentorship, belief in me, and the opportunity to finish what I had started twenty years ago. I would also like to thank my thesis committee members, Dr. Chester Van Tyne, and Dr. Emmanuel De Moor for their selfless support, guidance and assistance. To the gentlemen that originally sparked my interest in metallurgy – Dr. David Matlock, Dr. Chester Van Tyne, Dr. George Krauss, Dr. Glen Edwards, Dr. Gerard Martins, Dr. Robert Frost, Dr. John Hagar, Dr. Stephen Liu, and Dr. Rex Bull, your passion for the metallurgical science is contagious. For my sponsor company and employer of over twenty years, Evraz Pueblo, thank you for all of the challenges and experiences during my career as well as the opportunity and necessary resources to complete this lifetime goal. Many thanks to my career mentor and supervisor, Mark Erspamer, for providing the laboratory resources and time to complete this research. The assistance, guidance, and expertise of my Industrial Mentors and ASPPRC alumni Dr. Mark Richards, Dr. Greg Lenhoff, and Joe Kristan is greatly appreciated and crucial to the success of this project. Gratitude is also owed to my friend and mentor Bob Glodowski of Evraz Stratcor for his expertise and constantly challenging me to become a better metallurgist. Thanks to Dr. Shahrooz Nafisi of Evraz Regina for the Gleeble simulations and SEM work. Thanks are also due to the men and women of the metallurgical laboratory of Evraz Pueblo for their help and support in the seemingly endless sample preparations. Specifically Jim Thompson for teaching me how to use the EDM machine, Jominy quench unit as well as sharing his sample polishing techniques, Tommy Duran for the numerous CNC programs written and perfected, and lab supervisor Mike Breece for providing the time and human resources in the lab. I am indebted to my good friend and confidant, Charlie Bocchietti for his friendship, unique perspective and endless humorous stories. I would also like to thank my wonderful daughters, Jordan and Jessica, for their complete faith, support and patience during this process as well as bringing me food during the many long nights at the lab. Finally, thank you Ma, for the genetic code to have a shot at this, for being my lifetime mentor and friend, and for never doubting me even when I did.

xii CHAPTER 1

INTRODUCTION

The scope of this project is to investigate the effects of increasing residual copper content on the metallurgical, mechanical, and fracture properties of fully pearlitic, hypereutectoid rail steels. The results are intended to scientifically illustrate the possible detriments and/or benefits of this residual chemical element, i.e. copper, commonly found in the scrap used to produce electric arc furnace steel. This project operates under the premise that best practices were performed during rail manufacturing to mitigate manufacturing risks that are associated with copper, such as hot shortness. All test specimens were sectioned from full rails, which were rolled from full industrial heats produced at Evraz Pueblo.

When metallic elements and compounds are intentionally added to the steelmaking process for the purposes of modifying or enhancing the final properties, they are considered as alloying additions. When these elements or compounds are entrained in the feedstock of the steelmaking process, they are considered residual, or tramp, elements. Copper (Cu) is generally considered as a residual element in plain carbon and micro-alloyed steels made from Electric Arc Furnaces (EAF’s). There does not currently exist an economically feasible method to remove Cu from steel once melted, thus the amount of residual Cu in steel will continue to rise over time. The increasing level of residual copper in available scrap has been well-documented [1 – 4]. The primary interest in maintaining lower copper levels is the prevention of hot shortness during steel hot forming. Hot shortness is a condition that is created when a sufficient quantity of liquid copper exists on the steel surface during forging and forming operations, resulting in a lack of ductility due to liquid penetration of austenite grain boundaries. An example of the visual appearance of hot shortness is shown in Figure 1.1 in a sample of Standard Strength commercially produced rail steel. Figure 1.2 is an unetched micrograph and illustrates how the copper penetrates the austenite grain boundaries during reheating and hot rolling. All of the bright white areas in Figure 1.2 are copper with two of the areas identified with black arrows. It should be noted that modern rolling mills, such as the mill in Pueblo, CO where the alloys of this study were manufactured, utilize reheating furnaces with advanced technologies in controlling the reheat furnace atmosphere and temperature. These vast improvements in thermal cycle control have greatly reduced the occurrence of hot shortness from the era of soaking pits. Furthermore, the likelihood of shipping a rail that exhibits surface defects as a result of hot shortness is nonexistent as this issue would be identified in the rail making, finishing, and inspection processes.

Figure 1.1 Photograph showing hot shortness during hot rolling of Standard Strength rail steel. Black arrows indicate plane of polish for Figure 1.2. Photo courtesy of Evraz Pueblo

2 For EAF shops to accommodate customer specifications for maximum copper contents the use of iron substitutes such as Direct Reduced Iron (DRI) and Hot Briquetted Iron (HBI) act to dilute the overall copper concentration within any given heat, and the use of these additions increases production costs. In situations where the copper content is too high for dilution, EAF shops will cast a scrap heat to prevent stopping the continuous caster due to the high cost of lost production and restarting the cast. The billets from such heats are then sequestered and sent back to the scrap preparation step to be spread out over multiple heats of future production. The three domestic producers of railroad rails (Evraz, Arcelor Mittal, and Steel Dynamics) all utilize EAF steel for the production of rolling feedstock. According to the Steel Manufacturers Association, the overall percentage of EAF production in the United States has risen from less than 10% in 1970 to 64% in 2011 [5] and most likely higher at the time of this publication, and thus understanding the effect of higher copper content is important to multiple steel products.

3 CHAPTER 2

BACKGROUND AND LITERATURE REVEIW

This chapter discusses the necessary background information prompting the need for this research. Relevant information on eutectoid steels, particularly rail steels, from the literature is also included.

2.1 Introduction One of the largest operating costs for North American Railroads is the maintenance of the railroad infrastructure, namely the steel rails. Industrial research efforts over the last few decades have focused on improving the life of rails by decreasing the wear rate via optimizing the metallurgical structure. For the preferred fully pearlitic microstructure of North American Railroad rails, research has shown that maximizing the hardness of the steel minimizes the wear rates in both laboratory and field tests [6 – 11]. Stock et al. compared the wear rates of three pearlitic rail grades denoted by their hardness values, namely 260, 350, and 400 HBW, on a full scale wheel test rig [11]. The authors suggest the increase in hardness resulted from an increase in carbon content (cementite fraction) and cooling rate (pearlite spacing). The test rig was comprised of a 1.5 m length of rail on a mobile carriage, sliding back and forth beneath a full size rail wheel. The amount of wear was determined by taking measurements along the top head, side head, and gauge corner via a profilometer at predetermined cycle intervals. Figure 2.1 displays the results of this research and shows the reduction in total wear with increasing rail hardness. The most economical method of increasing the hardness of fully pearlitic steel is by increasing the cementite fraction via increasing the carbon content [12]. With the introduction of accelerated cooling technologies, the hardness of pearlitic steel at a given carbon content can be further increased through reduction of the pearlite interlamellar spacing [6, 13 – 17]. In-line rail hardening technologies in use today capitalize on the latent heat of rolling and provide a fully pearlitic rail steel that possesses high hardness at greater depths from the running surface than the former off-line methods. Off-line rail hardening typically consists of re-heating the finished rail in an induction furnace, followed by accelerated air cooling. Figure 2.2 displays the different rail head etching responses of the off-line (Figure 2.2 a) and in-line (Figure 2.2 b) methods of rail head hardening. Note the homogenous etching response for the in-line specimen which illustrates the microstructural and hardness uniformity. As a consequence of improvements in depth of hardness and subsequently rail wear rates, rail life has been extended and other failure modes, i.e. rolling contact fatigue (RCF), are becoming more predominant as reasons for rail replacement [7]. RCF is caused by repeated wheel/rail contacts during normal railroad operations. The non-proportional multiaxial stress-strain condition causes fatigue damage to the rail

4 which initiates cracks in the surface or subsurface of the rail [18]. This damaged layer can be removed by profile grinding of the rail, which in turn reduces the life of the rail from a material loss perspective.

Figure 2.1 Rail wear as a function of number of wheel passes for pearlitic steels of various hardness (HBW). Adapted from Stock et al. [11].

(a) (b) Figure 2.2 Macro photographs of polished and etched transverse sections of (a) off-line, and (b) in-line head hardened pearlitic rail, 2% nital etch. Photos courtesy of Evraz Pueblo.

2.2 Microstructural considerations The decomposition of austenite to pearlite is a diffusion controlled process and as such, lower transformation temperatures decrease the diffusivity of carbon in austenite, resulting in a finer pearlitic structure [19]. For a given chemical composition, a finer pearlitic structure will improve the hardness and

5 tensile properties. Herman and Leroy suggested that steels with higher residual element contents have a higher flow stress during hot working leading to a delay in the static recrystallization between passes and ultimately leading to finer microstructures on cooling [1]. In a study on 0.95 C, 5 Ni, 7 Cu (wt pct) steels, Fourlaris et al., and Chairuangsri et al. noticed the existence of ε-Cu interphase precipitates within the pearlitic cementite and only within the pearlitic ferrite if transformation happened near the eutectoid temperature (i.e. low undercooling) or if the steel was given a subsequent ageing treatment [20,21]. However, Khalid et al. observed ε-Cu precipitation in both the pearlitic ferrite and the pearlitic cementite of 0.70 wt pct C hot rolled rods. The primary effect of the ε-Cu precipitates was a change in the morphology of the microcracks in the cementite plates during deformation to an irregular or blunted crack, as opposed to the straight microcracks observed in non-copper-bearing steels [22]. There appears to be a disagreement in the literature as to the effect of prior austenite grain size on the final pearlitic inter-lamellar spacing. Hyzak and Bernstein suggested that the pearlitic spacing can be refined to a small degree by increasing the austenitic grain size [23]. Their observation with respect the effect of grain size is in contrast to the viewpoint of Reed-Hill who suggest that the distance between lamellae is independent of the prior austenite grain size [24]. Elwazri et al. reported a proportional relationship between prior austenite grain size and pearlite spacing. They performed a series of transformation experiments on a hypereutectoid steel (0.91C, 0.49Mn, 0.21 Si wt pct) by austenitizing at four different temperatures and isothermally transforming to pearlite at three different temperatures [19]. The results of this work are shown in Figure 2.3.

6 Pickering and Garbarz [25] argued that although prior austenite grain size does not affect the inter-lamellar spacing, it does have an effect on the pearlite morphology, with nodular pearlitic structures formed from coarse austenite, whereas fine austenitic grains decomposed into individually formed pearlite colonies. In a related report, Elwazri et al. observed an increase in the transformation kinetics with an increase in the pearlite start temperature associated with a finer austenite grain size [26]. Although their work focused on ultrafine grain steel, Lian et al. observed that austenitic grain size has a great influence on the pearlitic transformation. Above a 4 µm grain size lamellar pearlite formed, whereas below 4 µm a divorced eutectoid transformation product formed [27, 28]. The transformation from austenite to pearlite has been observed to initiate at prior austenite grain boundaries [25, 26, 29]. As most authors agree that pearlite interlamellar spacing is primarily controlled by the transformation temperature or degree of undercooling, it stands to reason that any mechanism that retards pearlite transformation results in a finer pearlitic structure. Elwazri et al. compared the benefits of the various techniques to measure the pearlite spacing, such as light optical microscopy (LOM), field emission scanning electron microscopy (FE-SEM), transmission electron microscopy (TEM), atomic force microscopy (AFM), and laser scanning confocal microscopy (LSCM) [30]. Of these methods, Elwazri et al. illustrated that measurements using LSCM provide similar results to FE-SEM, but with a greater speed of evaluating larger areas. In their work, they compared the interlamellar spacing measurements derived from FE-SEM and LSCM methods on a hypereutectoid (0.91C, 0.49 Mn, and 0.21Si wt pct denoted “V0”) and a micro-alloyed hypereutectoid (1.1C, 0.63Mn, 0.23Si and 0.17V wt pct denoted “V”) steel, austenitized at 1200 °C for 20 min and isothermally transformed at 550 and 620 °C. The results of these measurements are illustrated in Figure 2.4. To obtain adequate measurements using a FE-SEM, near-perpendicular lamellae must be identified for measurement using either linear or circular intercept methods. With LSCM, a greater variation from perpendicular is acceptable as the method involves rotating the image generated from the surface relief to account for non-perpendicular lamellae. To eliminate the need for isolating only near-perpendicular lamellae, Saltykov showed that the mean true spacing is equal to ½ of the mean random spacing for pearlite that has a constant spacing within each colony [31], a method repeated by other authors [17, 25, 32, 33]. Vander Voort and Roosz expanded on Saltykov’s work by incorporating a distribution of true spacings derived from continuous cooling rather than the constant spacing assumed from isothermal transformation [32]. Fong utilized the results of Vander Voort and Roosz and applied a central weighting theory to the apparent spacing segments to improve the accuracy of the results [34]. A common and simple method to determine the interlamellar spacing is to take direct measurements of pearlite spacing from near perpendicular plates of ferrite and cementite using scanning electron microscope images [16, 35, 36]. The purpose of the present work is to determine if there are any relative differences between

7 interlamellar spacing with varying copper content, thus the linear intercept method on FE-SEM images was employed.

In the absence of micro-alloying, steels are typically described as either coarse grain or fine grain by the method of deoxidation practice used. Silicon deoxidized (killed) steels are considered coarse grain, while aluminum deoxidized steels are considered fine grain. The fine grain structure of aluminum killed steels is derived by the dispersion of AlN particles that “pin” the austenite grain boundaries and prevent coarsening during certain thermal treatments [37]. Other elements, such as Ti, Nb, and V also form nitrides and carbo-nitrides that act in a similar fashion, although their effects are varied due to solubility differences. Herman et al., mentions that tramp elements, such as copper, can increase the deformation resistance during hot rolling and delay the inter-pass static recrystallization [1]. If delayed for sufficient time, a reduction in the austenitic grain size could be observed similar to a thermomechanical rolling condition, provided the rolling mill can handle the increased motor load.

2.3 Hardenability The hardenability of steel is a relative measure of the ease of which martensite can form with respect to the thermal history of the material [38, 39]. For instance, steel with a low hardenability will require a more severe cooling rate to produce martensite at a specific depth within a specimen than steel with a high hardenability. For comparative purposes, the concept of the critical ideal diameter, DI, was

8 developed to facilitate comparisons between different steel grades. The DI has units of distance (either mm or in) and describes the diameter of a round specimen that will contain 50% martensite at the center of the specimen after quenching with infinite severity. ASTM – A255 [40] describes a standardized testing methodology to determine DI. While the carbon content controls the maximum hardness that can be achieved in a quenchant of infinite cooling capacity, other factors such as alloying elements, austenitic grain size, and thermal history, all influence the hardenability of a steel [41]. The elements listed in the standard as having an increasing effect on hardenability are C, Mn, Si, Ni, Cr, Mo, Cu, V, and Zr. The value of DI derived from the chemical content of an alloy is calculated by the product of all the multiplying factors for each of the listed elements. These multiplying factors are valid assuming an austenitic grain size of ASTM 7. The values for the multiplying factors can be found either directly from the supplied tables, or by the equations used to calculate the multiplying factors, both of which are included in the standard. The graphical representation of hardenability encompasses plotting the measured hardness as a function of the distance from the quenched end of the specimen. The curve starts from the maximum hardness dictated by the carbon content for plain carbon steels, and transitions to the air-cooled hardness of the steel, the rate of which is governed by the chemical content. Figure 2.5 shows the predicted hardenability curves for an A36 and 4140 steel taken from the literature to illustrate the difference between low and high hardenability [42].

Figure 2.5 Calculated hardenability curves for a high (4140) and low (A36) hardenability steel constructed from Li et al. [42]. Calculations are in accordance with ASTM-A255 [40].

There exists a vast amount of information in the literature concerning hardenability for numerous alloy systems [4, 41 – 45]. In addition to the chemical contributions to hardenability, increasing austenite grain size has also been shown to increase hardenability most likely by restricting the number of nucleation

9 sites for diffusional austenite decomposition products [46]. According to ASTM-A255, an austenite grain size of ASTM 7 is assumed since experience shows that most steels with hardenability control fall within that grain size. To more accurately predict the hardenability of steels that deviate from an austenitic grain size of 7, a multiplying factor to account for grain size should be used. Pavlina et al. studied the effects of copper on the hardenability of a medium carbon steel using the methods described in ASTM-A255 [4]. In this work, they developed an equation to determine a multiplying factor for austenitic grain size other than ASTM 7 that will be used in this research. For the purposes of this project, the test methods described in ASTM-A255 will be utilized to evaluate the effect of Cu on hardenability of 0.90 wt pct carbon rail steels while keeping the other alloying elements as constant as possible.

2.4 Mechanical Properties Several methods exist to alter the mechanical properties of steel. For instance, raising the carbon content increases the pearlite or cementite fraction, which increases the strength and hardness properties. Grain refinement can also increase the mechanical properties and can be accomplished through micro- alloying additions to retard austenite grain growth (i.e. Al, V, Nb, and Ti). Grain refinement can also be accomplished through colder final rolling temperatures that reduce austenite grain growth after recovery and recrystallization. Accelerated cooling after rolling may also refine the final grain structure. Strengthening can also occur from precipitation in the final structure, as is found with vanadium carbo- nitrides. Solid solution strengthening from alloying additions of substitutional elements (i.e. Mn, Cr, Si, Mo) can also influence the mechanical properties. All strengthening mechanisms share the common property of restricting dislocation movement in the lattice. Similar to grain size, the mechanical properties of a pearlitic steel have been shown to be inversely proportional to the inter-lamellar spacing of the pearlite in a Hall-Petch type relationship [13 – 15, 23, 24, 47]. Gomes et al. studied the effect of various austenitizing and transformation temperatures on an eutectoid (0.76C, 0.87Mn, and 0.14Si wt pct), and a micro-alloyed eutectoid (0.76C, 0.87Mn, 0.15Si, and 0.045Nb wt pct) rail steel and found a Hall-Petch type strength dependence on interlamellar spacing as shown in Figure 2.6 [14]. Taleff et al. also reported a strong strength dependence on interlamellar spacing in two ultra-high carbon (1.5 and 1.8 wt pct. C) steels austenitized and transformed at various temperatures to impart interlamellar spacing variation. The results of this work are illustrated in Figure 2.7.

Figure 2.6 Strength dependence on interlamellar spacing for 0.80 wt pct C rail steel. Adapted from Gomes et al. [14].

Figure 2.7 Yield strength dependence on interlamellar spacing for 1.5 and 1.8C wt pct ultra-high carbon steels. Re-constructed from Taleff et al. [16].

2.5 Fracture Mechanics Pearlitic rail steels are generally considered brittle materials in the sense that cleavage-type fracture is prevalent at typical service temperatures. Impact toughness, plane strain fracture toughness, and fatigue crack growth rate testing can be used to evaluate the fracture properties of rail steels.

11 2.5.1 Charpy Impact Toughness The Charpy impact toughness test is a simple and widely used test to evaluate the behavior of a material during high rates of impact loading in the presence of a notch [48 – 52]. This test is also a very useful tool in determining the temperature at which a normally ductile body centered cubic (BCC) metal will fracture by cleavage instead of ductile rupture. This temperature, known as the ductile to brittle transition temperature or DBTT is typically more pronounced in medium and low strength BCC metals [53]. Most high carbon, high-strength pearlitic steels do not exhibit a large change in the absorbed energy of the impact test with increasing temperature in the same manner as low and medium carbon structural steels. Figure 2.8 illustrates the dynamic notch toughness of steels with varying carbon content [54]. Most rail steels are on the order of 0.80C wt pct or higher and are thus not expected to have a high resistance to impact loading in the presence of a notch. It is because of the ambient environmental conditions of structural steels that selection of the steels must take into consideration the DBTT. Since all pearlitic rail steels are high carbon and high-strength, this test is not normally performed. However, this test is a requirement of both the Canadian and Russian railroads, presumably due to the extremely low temperatures seen in these geographical regions.

Figure 2.8 Effect of temperature and carbon content on notch toughness. Reconstructed from Metals Handbook [54].

2.5.2 Plane Strain Fracture Toughness (K1c) The linear elastic plane strain fracture toughness test outlined in ASTM – E399 provides insight into how a material will respond during loading in the presence of a sharp stress concentrator (i.e. a fatigue crack) [55]. The linear elastic plane strain fracture toughness is denoted as K1c, where K is the

12 stress intensity factor in units of MPa √m (ksi √in), the subscript “1” describes tensile loading in a direction perpendicular to the crack growth, and the subscript “c” refers to the critical value of stress intensity factor above which fracture is imminent under plane strain loading conditions. In a study on fatigue crack growth and fracture toughness of five different pearlitic rail steels, Ochi et al. observed an increase in the fracture toughness with increasing tensile ductility as measured by reduction of area percent from a standard tensile test [56]. The chemical compositions of the steels evaluated in this study are listed in Table 2.1 and the relationship between fracture toughness and tensile ductility is shown in Figure 2.9.

Table 2.1 Chemical Composition of Rail Steels from Ochi et al.[56]. Rail Steel C Mn P S Si Cr Mo V Carbon 0.77 1.15 0.014 0.016 0.25 0.21 - - HT 0.76 0.95 0.018 0.014 0.15 0.08 0.023 - Cr-Mo 0.75 0.66 0.021 0.019 0.33 0.68 0.18 - Cr-V 0.76 1.10 0.022 0.009 0.69 1.01 0.01 0.09 HH 0.79 0.91 0.017 0.007 0.25 - - -

The Carbon, Cr-Mo, and Cr-V grades were ambient cooled, the HT grade was fully heat treated by quenching in oil and stress-relieving, and the HH grade was subjected to a rail head air-quench after rolling, i.e. “Head Hardened”.

13 2.5.3 Fatigue Crack Growth Rate The ability of a rail steel to resist fracture in the presence of a fatigue crack is of paramount importance to railroads as car derailments represent a safety concern and can be quite costly. Most freight and passenger rail lines utilize in-track ultrasonic flaw detection cars to detect the presence of internal cracking from either fatigue or other sources (i.e. hydrogen embrittlement). Knowledge of fatigue crack growth and fracture toughness properties of rail steels can provide the railroads with information needed to modify maintenance schedules to reduce the risks of derailments [57, 58]. Figure 2.10 is an example of sub-surface fatigue crack in a rail that initiated at the interface between the cold-worked running surface and the parent metal known to the railroad industry as a Detail Fracture [58].

Figure 2.10 Macrophotograph of a Detail Fracture showing longitudinal cracking and transverse fatigue in the head of a rail. Photo courtesy of Evraz Pueblo / TTCI [58]

A common method of determining the fatigue crack growth rates as a function of imposed loading cycle is provided in ASTM – E647 [59]. This test method encompasses the determination of fatigue crack growth rates from the near threshold to the unstable crack growth regimes. A variety of specimen sizes and geometries can be used for this test provided the dimensions of the specimen are of sufficient planar size to remain predominantly elastic during testing. Similar to the test method for fracture toughness testing, this method uses a measure of the stress intensity factor in the calculations and allows comparisons to specimens of different geometry and/or chemical composition. This comparison is made fairly simple by a log-log plot of the incremental crack growth, typically denoted in units of meters per cycle (da/dN), as a function of the stress intensity factor range (ΔK). Stress intensity factor range is calculated by subtracting the stress intensity factor at the minimum load from the stress intensity factor of

14 the maximum load within a given fatigue loading cycle. Gray et al., Algan et al., and Ochi et al., have performed fatigue crack growth rate evaluations on rail steels [56, 60, 61]. Typically, ΔK threshold is determined at growth rates on the order of 10-10 m/cycle. In most studies of pearlitic steel, the steady- state crack extension, or Stage II, region of the da/dN versus ΔK plot occurs between 10-8 m/cycle and 10- 5 m/cycle. Within this linear region of the log-log plot, the behavior can be described by the Paris

m equation (da/dN = C0ΔK ) which has been referenced in many studies [60 – 68]. With continued crack extension in a constant load test, the crack growth becomes unstable and grows at an exponential rate in Stage III, where the maximum stress intensity approaches the fracture toughness of the material. Figure 2.11 displays the results of Ochi et al for four of the alloys tested in their study [56]. Of note is the close proximity of the curves for all alloys, especially in the slope of the linear portion of the curve, i.e. that portion described by the Paris law.

Figure 2.11 Fatigue crack growth rate behavior of four rail steels. Reconstructed from the work of Ochi et al. [56].

15 CHAPTER 3

EXPERIMENTAL DESIGN

This chapter outlines the methodology by which the selected alloys were tested. A description of the selected alloys is discussed first, followed by microstructural, mechanical property, and fracture mechanics evaluation methods.

3.1 Experimental Alloys Chapter 4 of the American Railway Engineering and Maintenance of Way Association (AREMA) manual for railway engineering is the principal industry specification for the design, manufacture, joining, and maintenance of railroad rails [69]. Similar to other steel products (i.e. ASTM – A510 for wire rod), individual rail customers typically require adherence to the accepted industry standard along with their own specification limits. In many cases customer specific requirements are more stringent than the industry standard. For instance Table 4-2-1-4-2a of AREMA Chapter 4 [69] limits the maximum carbon content for intermediate and high strength rails to 0.86 wt pct, however the actual acceptable maximum carbon content for high strength rail can be much higher provided the microstructure is fully pearlitic. This table also lists the maximum copper content at 0.40 wt pct, which is a value that the railroads are not currently willing to increase. To ensure industrial relevance for this research, production heats were carefully selected to minimize composition variations other than copper. Five alloys with copper levels varying from 0.07 wt pct to 0.38 wt pct were produced to the same rail section following the same processing procedures used for all rail production at Evraz Pueblo. Processing consisted of heating the round cast blooms with dimensions of 311 mm (12.25 in) diameter by 5.89 m (19.33 ft) length in a natural gas walking beam re-heat furnace, a total rolling reduction of 88.7% via 13 passes of hot reduction to the final nominal dimensions shown in Figure 3.1, a final rolling temperature above 870 °C (1600 °F), and forced cooling through pearlitic transformation in a production head hardening unit. The exact cooling conditions are proprietary, and are therefore not reported in this document. Additionally, billets of a heat with high copper content (0.85 wt pct) were also rolled for the purpose of this study, resulting in the evaluation of a total of six rail steel chemistries. For each heat, one 12.2 m (40 ft) long section was cut from a rail that represented the middle of the heat with respect to casting sequence to eliminate inherent transitional chemical variation of a continuous casting process. Table 3.1 summarizes the chemical compositions of the six alloys selected. Note that the copper content varied from 0.07 to 0.85 wt pct. For reference each alloy is identified by its corresponding copper content and these designations are also summarized in Table 3.1.

16 Table 3.1 Chemical Composition of Selected Alloys in wt pct. Alloy C Mn P S Si Cu Ni Cr Mo Ti N H(ppm)

7 Cu 0.93 1.02 0.007 0.012 0.38 0.07 0.04 0.23 0.007 0.007 0.0060 0.8 11 Cu 0.94 1.02 0.007 0.007 0.37 0.11 0.05 0.23 0.012 0.007 0.0071 0.9 22 Cu 0.91 1.02 0010 0.011 0.36 0.22 0.07 0.24 0.017 0.008 0.0108 1.5 29 Cu 0.92 1.03 0.009 0.009 0.34 0.29 0.09 0.24 0.018 0.009 0.0082 1.4 38 Cu 0.93 1.01 0.012 0.010 0.33 0.38 0.08 0.22 0.018 0.010 0.0109 0.9 85 Cu 0.93 1.05 0.013 0.009 0.32 0.85 0.07 0.21 0.019 0.010 0.0081 1.0

Figure 3.1 136 RE Rail section excerpted from AREMA Chapter 4 [69]. Dimensions are in inches.

17 3.2 Microstructural Evolution This section describes the methods used to evaluate the pertinent concepts of the microstructural evolution. The method used to determine prior austenite grain size is followed by the method used to determine the interlamellar spacing followed by standard optical microscopy. The method to determine the hardenability of the alloys is also given in this section.

3.2.1 Austenite Grain Size The austenitic grain size can influence the transformation characteristics of steel and therefore, must be examined for the experimental alloys. To ensure complete solution of all cementite and to eliminate all thermal history of the samples, the McQuaid – Ehn method in ASTM – E112 [70] was employed. To reduce specimen preparation time, a 25.4 mm (1 in) specimen was cut from the top, un- quenched portion of each of the 18 Jominy bars (see Section 3.2.4 below) and used for this experiment. Cast-iron sample holders were lined with approximately 25.4 mm (1 inch) of Wilcarbo® Pack Hardening Compound. The one-inch specimens were then placed in the holders and completely covered with the Wilcarbo. A layer of cast-iron chips were then placed on top to minimize the rapid oxidation of the pack hardening compound. A Lindberg/Blue® sample furnace was preheated to 927°C, the cast-iron holders containing the specimens were placed into the furnace, and an eight hour timer was started once the furnace recovered back to the target temperature. After the eight hours expired, the furnace was turned off and the samples were allowed to furnace cool for 16 hours. The thick black crust from the hardening compound was removed with a glass bead shot blaster. The specimens were ground on a Struers® AbraPlan-20 to remove 1.5 mm of material, then polished using a Struers AbraPol-10 machine using 1 µm diamond solution for the final polishing step. The specimens were etched for eight seconds in ambient saturated picric acid, which consisted of 6 g picric acid and 300 mL distilled water. Grain size measurements were performed with Olympus PME3 optical microscope equipped with an automatic stage and Clemex® CL-vision32 software with built-in ASTM – E112 three concentric circles test method [70]. Analyses were all performed at 100 X magnification. The Clemex® software automatically calculates the ASTM grain size number for each field evaluated according to the procedure outlined in ASTM-E112 three circles method, and is summarized as follows. A series of three concentric circles with a total perimeter of 207 mm was superimposed at random locations on micrographs taken at 100X magnification (i.e. those shown in Figure 4.1) The total number of grain boundary intersections with the test pattern is used to calculate the number of intercepts per unit length of test line according to Equation 3.1

푁 푁 = 푖 퐿 퐿 (3.1) ⁄푀 where 푁퐿 = number of intercepts per unit length of test line 푁푖 = number of intercepts 퐿 = total test line length in mm 푀 = magnification used

This process is repeated on all fields and the average NL is used to calculate the ASTM Grain Size number according to Table 6 of ASTM-E112, summarized in Equation 3.2.

퐺 = (6.643856 × 퐿표푔10푁̅퐿) − 3.288 (3.2) where 퐺 = ASTM Grain Size Number

푁̅퐿 = Average number of intercepts per unit length of test line

To ensure robustness of the evaluation, 35 fields per area in three areas per specimen were analyzed. This methodology resulted in a total grain count of over 10000 grains per alloy which easily exceeds the recommended 500 grain count [70].

3.2.2 Interlamellar Spacing After hardness testing, described in section 3.3.2, a 12.7 mm (0.50 in) square specimen was cut from the middle of the head to include the running surface on the 7 Cu, 38 Cu, and 85 Cu alloys. The specimens were ground and polished following the same procedures used for the austenite grain size specimens. The specimens were etched for eight seconds in ambient saturated picric acid which consisted of 6 g picric acid and 300 mL distilled water. The specimens were packaged and sent to Evraz’s research laboratory in Regina Saskatchewan for scanning electron microscope (SEM) imaging using the field emission SEM located at the research lab. Each specimen was scanned at depths of 6 mm (0.236 in) and 10 mm (0.394 in) from the running surface of the rail head at 10 kV and various magnifications to enable direct measurement of the lamellar spacing. The interlamellar spacing analysis was performed by measuring the distance between cementite lamellae in pearlite colonies with lamellae that appeared perpendicular to the plane of polish. For each alloy, 35 images of perpendicular lamellae per depth were acquired and submitted for analysis. Direct measurement of the inter-lamellar spacing was performed using Image J® software.

19 3.2.3 Digital (Optical) Microscopy Full head specimens of 22.2 mm (0.875 in) thickness were sectioned from the 7 Cu, 38 Cu, and 85 Cu alloys with an Advanced Machine & Engineering® MSaw R300 and an H.F. Wells® Model F-16- 2 water-cooled horizontal band saw according to the schematic in Figure 3.2. The specimens were ground and polished following the same procedures described earlier. The specimens were etched for eight seconds in ambient saturated picric acid which consisted of 6 g picric acid and 300 mL distilled water. The specimens were evaluated at various magnifications from 100X to 1000X with the Keyence® Digital Microscope to verify a fully pearlitic microstructure and to identify any deleterious microstructural constituents, including any evidence of near surface inter-granular attack.

Figure 3.2 Sectioning schematic for digital (optical) microscopy specimens.

3.2.4 Hardenability Full cross sections of rail with a length of 178 mm (7 in) were cut with an Advanced Machine & Engineering® MSaw R300. Individual specimens for standard Jominy End-Quench hardenability tests were sectioned from the gauge corner of the head (Figure 3.3 (a)) using H. F. Wells® Model F-16-2 water-cooled horizontal band saws. Three Jominy bars from each alloy were normalized, machined, austenitized and quenched according to the requirements of ASTM – A255 [40]. The procedure consisted of normalizing the pre-machined blanks at 871°C (1600°F) for one hour and air cooling. The cooled blanks were machined on a HAAS® Model SL-20TB Lathe to the preferred test specimen dimensions (Figure 3.3 (b)) with a diameter of 25.4 mm (1 in), length of 98.4 mm (3.875 in), with a 31.8 mm (1.25 in) diameter by 3.2 mm (0.125 in) thick lip for hanging in the quenching unit. The machined

20 specimens were then austenitized at 843°C (1550°F) for 30 minutes and immediately water quenched using a Bueller® Metaserve End Quench Unit; the water spray was maintained for 10 minutes. Water was gravity fed from a 38 L (10 gal.) tank at a height of approximately 1 m (3.3 ft.) above the quenching unit to ensure laminar water flow. The tank was supplied with city water via a float valve to maintain constant pressure. Parallel flats were prepared on a Swisher Automatic ® Mark VI oscillating grinding machine by removing material to a depth of 0.1 mm (0.020 in) to eliminate any effects of decarburization during the austenitizing step. A Wilson® Rockwell series 2000 equipped with an automatic stage, Jominy holder, and pattern software performed the hardness indentations at the prescribed 1.6 mm (0.063 in) intervals. Several “dummy” flats were prepared on unused Jominy bars to “calibrate” the first hardness punch at a distance of 1.6 mm ± 0.1mm from the quenched end. The indentation distances were confirmed using a Starrett Model 797 digital caliper and an Olympus® SZX16 stereoscope at 25X magnification. To ensure sufficient data, three Jominy specimens were prepared from each of the six alloys and two sets of flats were ground and tested instead of the normal one set of parallel flats. Thus, each Jominy bar received four sets of Rockwell hardness indents.

(a) (b) Figure 3.3 Jominy hardenability test schematics: sectioning diagram (a), and specimen dimensions (b).

3.3 Mechanical Properties This section describes the methods used to determine the mechanical properties of the selected alloys. The method used to determine the tensile properties is followed by the method used to determine the hardness profile.

21 3.3.1 Tensile Properties Tensile properties were measured on samples machined from 254 mm (10 in) lengths of full rail cross-section using an Advanced Machine and Engineering® MSaw R300. Longitudinal tensile specimens with a gage diameter of 8.75 mm (0.350 in) were sectioned from the gauge corner of the head (Figure 3.4) using H. F. Wells® Model F-16-2 water cooled horizontal band saws. The centerline of the tensile specimens corresponded to a depth of 6 mm (0.236 in) from the rail head surface. Five tensile specimens were machined from each alloy, stress relieved at 93.3°C (200°F) for two hours, and tested in accordance with ASTM – A370 [71] and ASTM – E8 [72] on a 489 kN (110 kip) tensile frame. Prior to testing each specimen was marked with a 35 mm (1.4 in) gauge punch and ground with 180 grit sandpaper to ensure a consistent surface roughness. The tensile test was performed at a stressing rate of 345 MPa (50 ksi) per minute until the 0.2% offset yield strength had been determined. The program then switched to cross-head speed control at a rate of 12.7 mm (0.5 inch) per minute through specimen fracture. A 35 mm (1.4 in) Satec Model B2M extensometer was used to determine the 0.2% offset yield strength. The software program, Partner® performs the machine controlling functions as well as the calculations.

Figure 3.4 Tensile specimen blank sectioning schematic.

3.3.2 Hardness Full head specimens with a thickness of 22.2 mm (0.875 in) were sectioned from the 7 Cu, 38 Cu, and 85 Cu alloys with an Advanced Machine & Engineering® MSaw R300 and an H.F. Wells® Model F- 16-2 water-cooled horizontal band saw according to the schematic in Figure 3.5(a). Both cross-sectional surfaces were machined flat to remove the rough surface from the cold saw cutting operation and to

22 ensure parallel surfaces. The surface to be tested was given an additional 120 grit sandpaper grinding step. Rockwell C scale hardness testing was performed using a Wilson® Rockwell Series B2000 equipped with an automatic stage and Instron® ATA software. A 4 mm x 2 mm grid was superimposed on the specimen face and programmed into the software, providing hardness readings at 2 mm depth intervals. After data collection, each specimen was re-machined to remove 2 mm (0.08 in) and retested for hardness. This process was repeated five times per specimen, providing a minimum of 35 hardness measurements per depth increment. Figure 3.5(b) illustrates the testing locations.

(a) (b)

Figure 3.5 Rockwell C hardness traverse schematics; sectioning diagram (a), and indentation map (b).

3.4 Fracture Mechanics This section describes the methods used to characterize the fracture mechanics of the selected alloys. The Charpy Impact test is described first, followed by the plane strain fracture toughness and fatigue crack growth rate test methods.

3.4.1 Charpy Impact Test Charpy impact specimens were tested at room temperature to assess the influence of copper on the dynamic fracture behavior of the steels. Eight specimens per alloy were prepared for testing at room temperature. Due to the inherently high strength and fully pearlitic microstructure of the alloys evaluated, along with the author’s personal previous experience, a 2 mm U – notch was employed rather than the typical 2 mm V – notch or 5 mm U – notch as referenced in ASTM – E23 [73]. The Charpy blanks were

23 sectioned out of the gauge corner (Figure 3.6 a) and machined to 10 mm x 10 mm x 55 mm with the long dimension parallel to the rolling direction. A 1 mm radius U-notch, parallel to the transverse direction, on the surface closest to the running surface, was introduced by broaching with the geometry illustrated in Figure 3.6 (b). The specimens were tested on a 406.7 J (300.0 ft-lbf) capacity Satec Machine at an ambient temperature of 22.7°C (72.9°F) and 80% relative humidity. Energy absorbed during the test was recorded and fracture surfaces were evaluated visually for percent shear.

8.0 mm 10.0 mm

55.0 mm

(a) (b) Figure 3.6 Charpy U-notch impact test schematics; sectioning diagram (a), and specimen dimensions (b).

3.4.2 K1c Fracture Toughness Five compact tensile specimens were machined from all six alloys from 330.2 mm (13 in) long full cross-section rail specimens using a HAAS® VF3 milling machine and GF Agie Charmilles® wire electric discharge machine (WEDM) according to the sectioning schematic (Figure 3.7 a). The nominal dimensions of the specimen were; W = 38.1 mm (1.500 in), B = 19.05 mm (0.750 in), h = 2.56 mm

(0.10 in), and a0 = 16.26 mm (0.640 in) as illustrated in Figure 3.7 (b). The entire test was performed on a 250 kN (55 kip) MTS® Model 370 Servohydraulic Load Frame according to the prescribed method in ASTM – E399 [55]. Pre-cracking was performed under load control at 15 Hz frequency and a load ratio of 0.10. Fracture toughness tests were performed at a ramp rate of 222.4 N s-1 (50 lbf s-1). The software program MTS® TestSuite MP Elite® was used to control the parameters of the pre-cracking and fracture testing, and the software program Fracture Analyzer® was used to perform the validation tests and

24 fracture toughness calculations. Crack lengths were measured at five locations on 50 X magnification images taken on a Keyence® Digital Microscope.

(a) (b)

Figure 3.7 K1C fracture toughness specimen schematics; sectioning diagram (a), and specimen dimensions (b).

3.4.3 Fatigue Crack Growth Rate Ten compact tensile (CT) specimens were machined from 330.2 mm (13 in) long full cross-section rail specimens of the 7 Cu, 38 Cu, and 85 Cu alloys using a HAAS® VF3 milling machine and GF Agie Charmilles® wire electric discharge machine (WEDM) according to Figure 3.8. Prior to final machining, the blanks were stress relieved at 204°C (400°F) for one hour to remove any residual stresses. The nominal dimensions of the fatigue specimens were: W = 38.1 mm (1.500 in), B = 8.0 mm

(0.315 in), h = 1.0 mm (0.04 in), and a0 = 10 mm (0.40 in) as illustrated in Figure 3.8 (b). Both faces of the CT specimens were ground using 220 grit and 600 grit cameo discs, followed by 6 µm and 3 µm diamond solution polishing to provide a surface conducive to observing the fatigue crack on the Keyence® Digital Microscope. Each test was performed on a 250 kN (55 kip) MTS® Model 370 Servohydraulic Load Frame retrofitted with a 55 kN (12.5 kip) load cell according to the prescribed method in ASTM – E647 [73]. Pre-cracking was performed at 15 Hz frequency and a load ratio of 0.10.

Test parameters were selected that resulted in a final Kmax of 9.9 MPa √m (9.0 ksi √in). Verification of crack extension uniformity was performed by measuring the pre-crack length on both faces of the CT specimens at 50 X using the Keyence® Digital Microscope. Fatigue testing was performed under ΔK control at 15 Hz frequency, a normalized K gradient of 4.0 and an initial ΔK of 8.9 MPa √m (8.1 ksi √in).

25 The software program MTS® TestSuite MP Elite® was used to control the parameters of the pre-cracking and fatigue cracking. A Crack Opening Displacement (COD) gauge was used to measure crack opening displacements, which determine crack propagation from compliance calculations. Data for constructing the crack propagation as a function of the stress intensity range (i.e. da/dN vs ΔK) curves were automatically captured by the software at a rate of one data point every 2000 cycles. All fatigue cracks were grown to specimen fracture to facilitate crack uniformity measurements, which were performed at 30 X magnification on the Keyence® Digital Microscope. The fatigue crack growth rate data are plotted on a log-log scale with the incremental crack extension per cycle, denoted da/dN, as a function of the stress intensity factor range, denoted ΔK. The calculation for ΔK is given in Equations 3.3 – 3.5.

∆퐾 = 퐾푚푎푥 − 퐾푚푖푛 (3.3)

Where 퐾푚푎푥 = = the maximum stress intensity factor in a given cycle

퐾푚푖푛 = the minimum stress intensity factor in a given cycle

( ) ∆푃 2 + 훼 2 3 4 ∆퐾 = × 3 × (0.886 + 4.64훼 − 13.32훼 + 14.72훼 − 5.6훼 ) (3.4) 퐵√푊 (1 − 훼) ⁄2

푎 훼 = (3.5) 푊 where ∆푃 = loading range (maximum load minus minimum load) 푎 = crack length 퐵 = specimen thickness 푊 = specimen width

The full plot of fatigue crack growth rate (da/dN vs ΔK), has three distinct stages. Stage I exhibits the growth behavior of an incipient crack and it is generally agreed to occur between 10-10 to 10-8 m/cycle. The value of ΔK at these crack extension rates are considered the threshold value of ΔK, below which fatigue cracks do not initiate, and is denoted as ΔKth. Stage II is considered the stable crack extension range typically found above 10-7 m/cycle, encompasses the linear portion of the da/dN vs ΔK log-log plot, and can be expressed by the Paris equation (Equation 3.6). Stage III is the unstable crack extension range where ΔKmax approaches the material fracture toughness in an asymptotic fashion.

26 푑푎 Paris Law: = 퐶 ∆퐾푚 (3.6) 푑푁 0 where da/dN = incremental crack extension per cycle C0 = crack growth constant determined experimentally ΔK = stress intensity factor range from equation 4.29 m = slope of the linear portion of the da/dN vs ΔK log-log plot.

(a) (b) Figure 3.8 Fatigue crack growth rate specimen schematics; sectioning diagram (a) and specimen dimensions (b).

27 CHAPTER 4

EXPERIMENTAL RESULTS

This chapter provides the results of the experiments described in Chapter 3. Test results for the microstructural evaluation are listed first, followed by the mechanical properties and fracture mechanics.

4.1 Microstructural Evolution The microstructural evolution begins with the results on the prior austenite grain size evaluation, followed by the measurements of the interlamellar spacing and optical microscopy.

4.1.1 Austenitic Grain Size Comparisons. Figure 4.1 shows representative light optical micrographs of the six alloys examined. These micrographs were taken at the central portion of each specimen and clearly show the prior austenite grain boundaries from the proeutectoid cementite formation during slow furnace cooling. To aid with visual clarity, Figure 4.2 shows the prior austenite grains at a higher magnification. The percent relative accuracy (%RA) of the grain size measurements is calculated by dividing the 95% confidence interval (95% CI) by the average mean lineal intercept (ℓ̅) and expressing as a percentage according to Equations 4.1 through 4.4. According to ASTM E-112, a relative accuracy of less than 10% is an acceptable level of precision [70]. Table 4.1 summarizes the average grain size, standard error, and percent relative accuracy for the austenitic grain size evaluation.

1 ℓ̅ = (4.1) 푁̅퐿

1⁄ ∑(푋 − 푋̅)2 2 푠 = [ 푖 ] (4.2) (푛 − 1) 푡 × 푠 95% 퐶퐼 = (4.3) √푛 95% 퐶퐼 %푅퐴 = × 100 (4.4) ℓ̅ where ℓ̅ = mean lineal intercept s = sample standard deviation Xi = individual measurement n = number of observations t = Statistical multiplier interpolated from the t-distribution [74] = 1.960

(a) (b) (c)

(d) (e) (f) Figure 4.1 Light optical micrographs showing prior austenite grain size images of (a) 7 Cu, (b) 11 Cu, (c) 22 Cu, (d) 29 Cu, (e) 38 Cu, and (f) 85 Cu alloys. Specimens prepared from un-quenched portion of Jominy specimens, saturated picric acid etch.

(a) (b) (e)

Table 4.1 Summary Statistics of the Circular Intercept Method for ASTM Grain Size Determination ASTM Grain Standard % Relative Alloy Size No. Error Accuracy 7 Cu 5.0 0.4 1.4 11 Cu 5.2 0.5 1.8 22 Cu 5.3 0.5 1.8 29 Cu 5.1 0.4 1.6 38 Cu 5.4 0.4 1.4 85 Cu 4.9 0.4 1.6

Figure 4.3 shows the average austenitic grain size as a function of Cu content for the circular intercept method. The data show that grain size is essentially independent of copper content as all data are within one standard deviation as depicted in the plot. Furthermore, the data in Table 4.1 indicate that all samples exhibited an ASTM grain size number of five which is consistent with fine grain practice. This indicates that the titanium micro-alloying referenced in Table 3.1 is effective in restricting austenitic grain growth in these hypereutectoid silicon killed steels.

Figure 4.3 Average austenitic ASTM grain size number as a function of copper content using the circular intercept method. Error bars represent one standard deviation from the mean.

31 4.1.2 Pearlite Interlamellar Spacing Pearlite interlamellar spacing was measured on three alloys, the 7 Cu, 38 Cu, and 85 Cu alloys, which were selected to span the copper contents evaluated in this study. Data were obtained using a line intercept method on scanning electron micrographs obtained with the Field Emission Scanning Electron Microscope (FESEM) located at Evraz in Regina, Saskatchewan. Figure 4.4 shows representative micrographs for the three tested alloys at the listed depths. Single transverse specimens of each alloy were analyzed at depths of 6 mm and 10 mm from the running surface until 35 images of perpendicular lamellae were captured for each alloy and depth. Of particular interest are the particles observed between the lamellae on the 38 Cu and 85 Cu alloys in Figure 4.4 (e) and (f). The FE-SEM lacked the resolution to identify these particles, however, the fact that they are only present on the higher copper alloys suggests that they might be copper precipitates. The summary statistics for interlamellar spacing analysis are listed in Table 4.2. These data are also displayed graphically in Figure 4.5 which shows the relationship between inter-lamellar spacing and copper content for data obtained at the two measurement depths. The measurements at 6 mm and 10 mm were combined to calculate an overall average for each alloy.

Table 4.2 Summary Statistics for Interlamellar Spacing (nm) 6 mm 10 mm Overall Alloy Avg. Std. Error Avg. Std. Error Avg. Std. Error 7 Cu 98.9 16.4 106.7 19.1 102.8 18.2 38 Cu 117.9 16.3 99.8 15.6 108.9 16.8 85 Cu 88.0 12.6 97.5 21.9 92.8 18.6

The trend in overall average inter-lamellar spacing shows a slight reduction in pearlite spacing with increasing copper content. Although the averages are within one standard deviation of each other suggesting that copper does not have a strong influence on pearlite transformation kinetics, a statistical analysis provided mixed results. The data acquired for this analysis did not follow a normal distribution as a result of the bias towards near perpendicular lamellae. The Mann-Whitney test for median comparison was selected to analyze the data. This statistical test is similar to the two-tailed t test used for normal distributions in that the null hypothesis states that the medians are equal. A 95% confidence level (α = 0.05) was selected for this test. Table 4.3 displays the results of the statistical analysis in terms of the P-value.

(a) (b) (c)

(d) (e) (f) Figure 4.4 Representative FE-SEM (10.0 kV) secondary electron micrographs used to measure pearlite inter-lamellar spacing of the specimens. The white layers are cementite and the dark layers are ferrite. White arrows indicate possible copper precipitates. Measurements taken at depths of 6 mm (a, b, c), and 10 mm (d, e, f) from the running surface. Saturated picric acid etch.

Similar to the two-tailed t test, the P-value ranges from 0.0 to 1.0 and can be loosely defined as the probability of being wrong in rejecting the null hypothesis. A P-value less than the confidence level suggests the differences between the data sets are statistically significant. From a statistical analysis standpoint, the difference between interlamellar spacing at 6 mm is statistically significant for the three alloys. The analysis of the interlamellar spacing measurements at 10 mm indicates that there is not a statistically significant difference between the 7 Cu and 38 Cu alloys, nor the 38 Cu and 85 Cu alloys, however, there is a significant difference between the 7 Cu and 85 Cu alloys at 10 mm. From a practical standpoint, there appears to be a downward trend in interlamellar spacing with increasing copper content but not a significant difference between the alloys as the error bars in Figure 4.5 overlap considerably.

Table 4.3 P-values for Statistical Comparison of Lamellar Spacing Data set P Data set P 7 Cu 6 mm vs 38 Cu 6 mm 0.004 7 Cu 10 mm vs 38 Cu 10 mm 0.109 38 Cu 6 mm vs 85 Cu 6mm 0.000 38 Cu 10 mm vs 85 Cu 10 mm 0.231 7 Cu 6 mm vs 85 Cu 6 mm 0.001 7 Cu 10 mm vs 85 Cu 10 mm 0.025

4.1.3 Optical / Digital Microscopy Figure 4.6 through 4.11 illustrate the light optical micrographs of transverse sections taken from the head of the finished rails of the six alloys at various depths. There is a small amount of pro-eutectoid

34 ferrite (white etching areas indicated by arrows) near the surface of all six specimens, most likely due to a small amount of decarburization from the re-heating operation. The balance of the microstructure is fully pearlitic with no evidence of deleterious microstructural constituents such as bainite, martensite or retained austenite. The variations in color in the micrographs are simply an etching response of the different orientations of the pearlite colonies. The pearlite lamellae could not be resolved with the digital (optical) microscope, reinforcing the very fine pearlite results of the previous section. Examination of the photomicrographs also suggests an increasing pearlite colony size with increasing depth (distance away from quenched surface), although colony size was not quantified in this research. To ascertain the impact of Cu level on surface quality, full cross-sectional head specimens were surveyed and the worst-case surface conditions are illustrated in Figure 4.12 for all six alloys. The light gray etching areas are high temperature mill-scale (mixture of FeO and Fe3O4). Although the 85 Cu alloy did have the worst surface condition in the samples evaluated, it could not be determined if the copper content was directly responsible for the condition as no evidence of hot shortness was found microscopically near rolled-in scale shown in Figure 4.12 (f).

4.1.4 Hardenability Three Jominy specimens per alloy were prepared and tested in accordance with Section 3.2.4 for all six alloys. The hardenability curves for all six alloys are displayed on the same plot in Figure 4.13. To assist visual clarity, only the first 12 hardness indentations are shown in the plot. All six alloys exhibit the same maximum hardness of about 62 HRc, which is consistent with the literature suggesting that the maximum hardness of martensite is controlled by the Carbon content. The 7 Cu and 11 Cu alloys transitioned from a fully martensitic microstructure at a distance of 8 mm from the quenched end as indicated by a reduction in hardness below 60 HRc. The 22 Cu, 29 Cu, and 38 Cu alloys transitioned at a distance of 9.6 mm while the 85 Cu alloy transitioned at a distance of 11.2 mm. These data suggest that hardenability increases with increasing copper content. The plot also indicates a slight increase in pearlitic hardness associated with the lower plateau at a greater distance from the quenched end with increasing copper content, i.e. at distances ≥ 14.4 mm from the quenched end. The calculation of the Ideal Critical Diameter (DI) from chemical composition is the product of the multiplying factors as outlined in Tables 6 and 11 of ASTM-A255 for non-Boron steels [40]. An extrapolation of the multiplying factors for carbon had to be employed since Table 6 of ASTM-A255 ends at a carbon level of 0.90 wt pct, and all six of the experimental alloys are above this level. A simple plot of multiplying factor versus carbon content is shown in Figure 4.14 and shows a linear relationship with a distinct change in slope at a carbon level of 0.40 wt pct.

(a) (b) (c)

(d) (e) (f) Figure 4.6 Representative transverse light optical micrographs of 7 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeutectoid ferrite. Nital (2%) etch.

(a) (b) (c)

(d) (e) (f) Figure 4.7 Representative transverse light optical micrographs of 11 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeuctectoid ferrite. Nital etch (2%).

(a) (b) (c)

(d) (e) (f) Figure 4.8 Representative transverse light optical micrographs of 22 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeuctectoid ferrite. Nital (2%) etch.

(a) (b) (c)

(d) (e) (f) Figure 4.9 Representative transverse light optical micrographs of 29 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeutectoid ferrite. Nital (2%) etch.

(a) (b) (c)

(d) (e) (f) Figure 4.10 Representative transverse light optical micrographs of 38 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeuctectoid ferrite. Nital (2%) etch.

(a) (b) (c)

(d) (e) (f) Figure 4.11 Representative transverse light optical micrographs of 85 Cu alloy at the surface (a), and at depths of 6 mm (b), 10 mm (c), 18 mm (d), 26 mm (e), and 36 mm (f) measured from the running surface. White arrow indicates proeutectoid ferrite. Nital (2%) etch.

(a) (b) (c)

(d) (e) (f) Figure 4.12 Light optical micrographs for each alloy selected to show the “worst case” surface condition for the 7 Cu (a), 11 Cu (b), 22 Cu (c), 29 Cu (d), 38 Cu (e), and 85 Cu (f) specimens. Nital (2%) etch.

Figure 4.13 Hardenability curves for the six alloys over the first 12 hardness readings.

Figure 4.14 Relationship between multiplying factor used to determine DI and the wt pct carbon based on ASTM-A255 [40].

Linear regression of the multiplying factor for carbon was performed over the carbon range of 0.0 to 0.39 wt pct (Equation 4.5) and from 0.40 to 0.90 wt pct (Equation 4.6). The R2 values for the linear regressions were 1.0 and 0.9959 respectively, suggesting a strong linear relationship and suitability for extrapolation. Based on the carbon content for the six alloys analyzed, Equation 4.6 was used to calculate the multiplying factor from carbon for the six alloys.

43 푀퐹 (퐶) = 0.5402[퐶]; 0 ≤ [퐶] ≤ 0.39 (4.5)

푀퐹 (퐶) = 0.2152[퐶] + 0.1308; [퐶] ≥ 0.40 (4.6) where [퐶] = wt pct carbon.

In a similar fashion, the multiplying factor for Cu contents > 0.55 wt pct was employed to estimate the multiplying factor for Cu for the 85 Cu alloy. Linear regression of the multiplying factor for copper listed in ASTM-A255 was also performed. Figure 4.15 shows the multiplying factor plot for Cu with the corresponding linear regression equation having a favorable R2 value of 0.9976. Equation 4.7 summarizes the Cu linear regression and was used to calculate the multiplying factor for the 85 Cu alloy.

Figure 4.15 Relationship between the multiplying factor used to determine DI and the wt pct copper based on ASTM-A255 [40].

푀퐹(퐶푢) = 0.3668[퐶푢] + 1.0 (4.7) where [퐶푢] = wt pct copper.

The multiplying factors in ASTM-A255 were derived based on a constant austenitic grain size of ASTM 7. It is generally accepted that hardenability increases with grain size due to the lower number of nucleation sites (i.e. grain boundaries) for the diffusion controlled, pearlitic decomposition of austenite.

44 To account for the effect of grain size on hardenability, the work of Pavlina et al. [4] was adopted with the multiplying factor expressed in Equation 4.8.

MF (DI) = 1.5286 – 0.0755dASTM (4.8) where dASTM is the austenitic grain size number determined in Section 4.1.1 of this research.

Table 4.4 summarizes the calculated DI values for the six alloys based on the methodology outlined in ASTM – A255 along with the grain size factor. These data suggest that the hardenability curves for the 7 Cu and 11 Cu alloys should be very close to each other, as should the 22, 29 and 38 Cu alloys, with the highest hardenability seen in the 85 Cu alloy. The experimental data shown in Figure 4.13 are consistent with the predicted values in terms of order of increasing hardenability.

Table 4.4 Calculated Ideal Diameter from Chemical Content According to ASTM-A255 [40] Alloy DI (in) DI (mm) 7 Cu 3.4 75.4 11 Cu 3.4 75.9 22 Cu 3.6 82.6 29 Cu 3.8 85.1 38 Cu 3.7 82.8 85 Cu 4.4 96.5

From the calculated DI values and initial hardness at the first Jominy position (J = 1.6 mm), hardenability curves from the bulk chemistry can be generated by following the method outlined in Section 10.5 of ASTM – A255 [40]. Namely, the predicted value of hardness at each Jominy position is calculated by dividing the starting or maximum hardness by the corresponding coefficient listed in Table 2 of ASTM-A255. This relationship is expressed in Equation 4.9. Individual plots of calculated and experimental hardenability curves are displayed in Figure 4.16.

퐻푚푎푥 퐻퐽푖 = (4.9) 푓푖

where 퐻퐽푖 = Rockwell C hardness at Jominy position i (in mm or in) 퐻푚푎푥 = Maximum hardness of 100% martensite determined experimentally. 푓푖 = Proportional coefficient at position i from Table 2 of ASTM-A255.

45 The model outlined in ASTM-A255 under-predicts the depth of hardenability, and over-predicts the pearlitic hardness for all six alloys. The variation between predicted and actual hardness is small and most likely due to the linear nature of the predicted model vs. the reality of a change in crystalline structure between martensite and pearlite.

4.2 Mechanical Properties This results from the mechanical property evaluation are listed in this section. The results from the uniaxial tensile test are followed by the results from the hardness profile evaluation.

4.2.1 Tensile Results Five tensile specimens per alloy were prepared and tested in accordance with section 3.3.1 of this study. The data in Table 4.5 summarize the average tensile and yield strengths along with the standard error (standard deviation) for the tests performed. Figure 4.17 illustrates the 0.2% offset yield and ultimate tensile strengths as a function of copper content for the six alloys. The data suggest an increase in strength level with increasing copper content, consistent with the increase in hardness observed at Jominy depths greater than 14.4 mm shown in Figure 4.13.

Table 4.5 Average Ultimate Tensile Strength (UTS), Yield Strength (YS), and % Reduction of Area (% R/A) for the Alloys Tested.

UTS std. error YS std. error Alloy UTS MPa (ksi) YS MPa (ksi) % R/A MPa (ksi) MPa (ksi) 7 Cu 1350.2 (195.8) 9.6 (1.4) 850.5 (123.3) 17.4 (2.5) 24.1 11 Cu 1386.0 (201.0) 13.1 (1.9) 876.9 (127.2) 15.0 (2.2) 21.3 22 Cu 1372.2 (199.0) 10.5 (1.5) 870.4 (126.2) 15.2 (2.2) 20.2 29 Cu 1414.2 (205.1) 11.8 (1.7) 905.0 (131.3) 17.1 (2.5) 19.1 38 Cu 1379.1 (200.0) 15.8 (2.3) 921.6 (133.7) 21.0 (3.0) 18.5 85 Cu 1433.7 (207.9) 10.6 (1.5) 962.0 (139.5) 14.3 (2.1) 18.9

4.2.2 Hardness at Depth Table 4.6 displays the average hardness data as a function of distance from the running surface for the 7 Cu, 38 Cu, and 85 Cu alloys and is graphically represented in Figure 4.18. The average hardness data were obtained from the seven hardness traverse locations shown schematically in Figure 3.5.

(a) (b) (c)

(d) (e) (f) Figure 4.16 Comparison of experimental vs. predicted hardenability curves for the 7 Cu (a), 11 Cu (b), 22 Cu (c), 29 Cu (d), 38 Cu (e), and 85 Cu (f) alloys.

47 A maximum depth of 26 mm was selected as this depth encompasses all of the other tests performed in the study. The results show an expected decrease in hardness with increasing depth due to the decreasing cooling rate from the rail surface to the center of the head. The graph also illustrates an overall increase in hardness value with increasing Cu content, similar to the observations on tensile and hardenability properties.

Figure 4.17 Tensile and 0.2% offset yield strength as a function of copper content. Error bars are one standard deviation from the arithmetic mean

The conversion from Rockwell C to Brinell Hardness Number used in Figure 4.18 was developed by AREMA Committee Four, is referenced in AREMA Chapter 4 [69], and is expressed here in Equation 4.10. 퐻퐵 = 165.77 + 2.3597퐻푅퐶 + 0.0777퐻푅퐶2 (4.10) where: HB = Equivalent Brinell hardness HRC = Rockwell hardness “C” scale

In consideration of the obscurity of the AREMA developed hardness conversion to most readers, Table 4.7 exhibits the conversion from the measured Rockwell C hardness to both the AREMA and ASTM- A370 Brinell Hardness numbers. The AREMA HB numbers were calculated using Equation 4.12 and the A370 BHN were taken from Table 2 of ASTM-A370 [71]. The table indicates that the AREMA calculation for Brinell Hardness to be 13 to 15 points higher than the more familiar ASTM A370 conversion in the applicable hardness range for pearlitic rail steels.

Table 4.6 Average Hardness and Standard Deviation Values for the Hardness Measurements at Depth; These Values are Shown Graphically in Figure 4.18

Depth 7 Cu 7 Cu 38 Cu 38 Cu 85 Cu 85 Cu (mm) Avg HRc Std error Avg HRc Std error Avg HRc Std error 2 40.8 0.40 41.6 0.92 43.1 0.30 4 41.5 0.48 42.0 0.62 42.8 0.31 6 41.7 0.43 42.4 0.42 42.9 0.22 8 41.5 0.40 42.3 0.38 42.8 0.32 10 41.2 0.44 42.0 0.38 42.5 0.27 12 40.7 0.46 41.5 0.40 42.0 0.31 14 40.2 0.45 41.0 0.32 41.8 0.27 16 39.7 0.37 40.5 0.41 41.2 0.39 18 39.3 0.39 40.1 0.38 40.6 0.38 20 38.8 0.35 39.6 0.34 40.3 0.36 22 38.4 0.42 39.1 0.44 39.8 0.36 24 38.0 0.61 38.6 0.63 39.4 0.43 26 37.7 0.82 38.0 0.98 39.0 0.54

Figure 4.18 Hardness as a function of depth from running surface for the 7 Cu, 38 Cu and 85 Cu alloys. Error bars represent one standard deviation from the arithmetic mean.

Table 4.7 Hardness Comparison Rockwell C to Brinell Rockwell C AREMA HB A370 BHN 46 439 432 44 420 409 42 402 390 40 384 371 38 368 353 36 351 336

4.3 Fracture Mechanics This section displays the results from the characterization of the fracture mechanics for the selected alloys. Impact toughness results are followed by plane strain fracture toughness and fatigue crack growth rate.

4.3.1 Charpy Impact Toughness Table 4.8 summarizes the observed average room temperature (22.7 °C) Charpy U-notch (2 mm) impact energies for the six experimental alloys along with the calculated standard deviations based on eight replicates for each condition. All measured impact energies were low, between 9.0 and 10.8 Joule with an average standard deviation of about 1.6 Joule. The low toughness values indicate all specimens fractured in a brittle manner, a conclusion that is consistent with analysis of the fracture surfaces shown in the fractographs summarized in Figure 4.19. The fractographs show samples from the location identified in Figure 3.6. All samples exhibited brittle fracture without the presence of discernable shear lips.

Table 4.8 Room Temperature Charpy U-notch (2 mm) Absorbed Impact Energy

Alloy Impact Energy Standard Deviation (Joule) (Joule) 7 Cu 10.7 1.45 11 Cu 9.5 1.59 22 Cu 9.4 1.67 29 Cu 9.0 1.47 38 Cu 10.8 1.79 85 Cu 10.7 1.82

(a) (b) (c)

(d) (e) (f) Figure 4.19 Macro photograph images of representative broken Charpy U-notch (2 mm) specimens for the (a) 7 Cu, (b) 11 Cu, (c) 22 Cu, (d) 29 Cu, (e) 38 Cu, and (f) 85 Cu alloys.

Figure 4.20 Average Charpy U-notch absorbed impact energy as a function of copper content. Error bars represent one standard deviation from the mean.

4.3.2 Fracture Toughness Table 4.9 and Figure 4.21 summarize the average fracture toughness for the six alloys. All data were from specimens that met the validity requirements in ASTM-E399 [55] as shown in Appendix A of this report. The 7 Cu alloy had the highest average K1c at 38.6 MPa √m while the 38 Cu alloy had the lowest average K1c at 34.2 MPa √m. This difference is most likely the result of normal variation experienced when testing the same grade of steel. The results in Figure 4.21 show an excellent reproducibility as evidenced by small standard deviation represented by the error bars. Furthermore, the average values of K1c fracture toughness in this study are consistent with those found in the literature [47, 56, 57] for data on standard carbon, fully heat treated, Cr-Mo, Cr-V and head hardened rail steels. Figure 4.22 correlates fracture toughness with ASTM grain size number for the steels evaluated in this study and shows that fracture toughness is essentially independent of grain size. However, it can be argued that the austenite grain size in this study remained constant over the range of alloys evaluated, and therefore insufficient variations in grain size are available to establish austenite grain size dependence.

4.3.3 Fatigue Crack Growth Rate Figures 4.23a – 4.23c individually summarize the dependence of fatigue crack growth rate, da/dN, on imposed ∆K values for the 7 Cu, 38 Cu, and 85 Cu alloys. Also included in Figure 4.23d is a superposition of the individual data sets. For each material, data were obtained to assess stable crack growth behavior according to the Paris equation shown in Equation 4.11.

Table 4.9 Average K1c Fracture Toughness for the Six Alloys Tested. Data are Shown in Figure 4.21

Alloy Avg. K1c Std. Dev. (MPa √m) (MPa √m) 7 Cu 38.6 1.25 11 Cu 36.0 0.88 22 Cu 36.4 0.73 29 Cu 36.7 0.46 38 Cu 34.2 0.71 85 Cu 35.8 0.40

Figure 4.21 Average KIc fracture toughness as a function of copper content. Error bars represent one standard deviation from the mean.

Figure 4.22 Average K1c fracture toughness as a function of austenitic grain size. Error bars represent one standard deviation from the mean.

푑푎 Paris Law: = 퐶 ∆퐾푚 (4.11) 푑푁 0 where da/dN = incremental crack extension per cycle C0 = crack growth constant determined experimentally ΔK = stress intensity factor range from Equation 3.4 m = slope of the linear portion of the da/dN vs ΔK log-log plot.

Assessment of threshold crack behavior, i.e. at growth rates ≤ 10-8 m/cycle was beyond the scope of this study. A regression analysis was used to determine the C0 and m terms of the Paris Law equation and are summarized in Table 4.10. Superposition of the crack growth data in Figure 4.23d and the essential equivalence of the Paris Law constants summarized in Table 4.10 indicate that copper level has no discernable influence on fatigue crack growth rate. The R2 values in Table 4.10 is a measure of “goodness of fit” for the regression and has a maximum value of 1.0. The calculated Paris Law constants fit the experimental data very well, since all of the R2 values are greater than 0.99. Figure 4.24 and 4.25 show representative SEM images of the crack initiation, stable crack extension, unstable crack extension, and final fracture morphologies respectively for the 7 Cu, 38 Cu, and 85 Cu alloys. In all micrographs, the white arrow indicates the direction of crack propagation. Each specimen exhibited multiple fatigue crack initiation sites along the WEDM machined notch at either an inclusion or a pearlite colony that had an orientation perpendicular to the applied load, i.e. parallel to the desired direction of the crack propagation. The micrographs of fatigue crack initiation sites were chosen at random, however, the subsequent micrographs representing the stable crack growth, unstable crack

54 growth, and final fracture morphologies were directly in line with the initiation site for the respective material. The stable crack growth morphology is identifiable by the uniform “river bed” structure devoid of evidence of preferential path, i.e. grain boundaries. The unstable crack growth morphology was a mixture of the same features as the stable crack growth, denoted with the number “1” in Figure 4.25a, b and c, as well as quasi-cleavage, denoted with the number “2” and secondary cracking, denoted with the number “3”. The final fracture morphology consisted primarily of quasi-cleavage along with some secondary cracking. No evidence of fatigue could be seen in the final fracture morphology shown in Figure 4.25 d, e and f.

Table 4.10 Paris Law Regression Results for the Fatigue Crack Growth Rate Analysis

2 Alloy C0 m R 7 Cu 1.95 E-12 3.48 0.9926 38 Cu 4.22 E-12 3.31 0.9942 85 Cu 1.42 E-12 3.64 0.9922

(a) (b)

(c) (d) Figure 4.23 da/dN vs. ∆K plots for the 7 Cu (a), 38 Cu (b), and 85 Cu (c) alloys. The data from all three alloys are combined in (d).

(a) (b) (c)

(d) (e) (f) Figure 4.24 Representative S.E.M. images of the 7 Cu, 38 Cu, and 85 Cu alloys at the fatigue crack initiation (a, b, c) and at stable crack growth (d, e, f) locations respectively. White arrows indicate direction of crack propagation.

(a) (b) (c)

(d) (e) (f) Figure 4.25 Representative S.E.M. images of 7 Cu, 38 Cu, and 85 Cu alloys at the unstable crack growth (a, b, c) and final fracture (d, e, f) locations respectively. White arrows indicate direction of crack propagation.

58 CHAPTER 5

DISCUSSION

The results presented in the previous chapter are discused here to assess the potential influence of copper on the material properties of the rail steels investigated. The effect on microstructural evolution is discussed first, followed by the effects on mechanical and fracture properties.

5.1 Copper effect on pearlitic rail microstructure The results of the McQuaid-Ehn test described in Sections 3.2.1 and 4.1.1 suggest that copper does not influence the austenitic grain growth characteristics to a measurable degree. The overall difference between the maximum and minimum grain sizes observed in the six alloys tested was only one half of one ASTM grain size number which is within the uncertainty of the grain size measurement. Table 5.1 summarizes the average grain size number for the six alloys converted to average grain area and average grain diameter per the equations in ASTM-E112 [70].

Table 5.1 Austenite Grain Size Summary

Alloy ASTM Grain ̅A (µm2) ̅d (µm) Size No. 7 Cu 5.0 4032 63.5 11 Cu 5.2 3511 59.3 22 Cu 5.3 3276 57.2 29 Cu 5.1 3763 61.3 38 Cu 5.4 3056 55.3 85 Cu 4.9 4322 65.7 where G = ASTM grain size number ̅A = average grain area (µm2) ̅d = average grain diameter (µm)

It should be noted that the average grain diameter has no physical significance as it is merely the square root of the average grain area and austenitic grains are not typically square. However, it does offer a relative measure of the expected grain dimensions. As was discussed in Section 3.1, all six alloys of the present study were processed identically with respect to melting, casting, re-heating, hot reduction and final cooling. Thus, it can be concluded that copper content, at the levels studied, did not influence the final austenitic grain size prior to air quenching. It should be noted that analyses of the effects of copper

59 content on processing parameters such as rolling loads or microstructural characteristics, such as, static or dynamic recrystallization behavior, were beyond the scope of this thesis. Measurements of the pearlitic interlamellar spacing did show a slight decrease in spacing with an increase in copper content. Since all alloys were given identical thermo-mechanical treatments, i.e. re-heat (austenitizing) temperature, hot reduction ratio, final rolling temperature, start-quench temperature, and quenching parameters, it can be inferred that copper influences the diffusion controlled pearlitic decomposition of austenite. The smaller lamellar spacing with increasing copper content suggests that, on cooling, copper delays slightly the transformation to lower temperatures. It could not be determined in this study if the delay was caused by a solute drag effect with copper being rejected to the transformation front by either the pearlitic ferrite/cementite, or if the delay was due to ε-Cu precipitation as has been observed in other studies [20, 21, 75, 76]. The small differences in interlamellar spacing observed with the scanning electron microscope were unresolvable with the light optical microscope as all alloys displayed a similar fine pearlitic structure at each depth from the running surface. Although not measured, the apparent pearlite grain size (colony size) increased with increasing depth from the running surface as seen in Figures 4.6 through 4.12. This observed variation in colony size with depth was most likely due to a combination of the hot-rolling operations in which the steel surface experienced a higher amount of reduction as compared to the interior of the bar and the quenching operation in which the cooling rate was higher closer to the surface. Increased hot reduction may have resulted in a finer austenite grain size prior to pearlite transformation, thus providing increased nucleation sites. Higher cooling rates during quenching should lead to lower pearlite transformation temperatures, also promoting pearlite nucleation. Hardenability measurements reinforced the concept that copper delays pearlitic transformation. With increasing copper content, the depth of martensite formation from the Jominy test increased, as can be expected for most austenite stabilizing elements. Figure 5.1 shows a comparison of calculated hardenability curves for the 7, 38, and 85 Cu alloys along with the calculated hardenability of other pearlitic rail steels found in the literature [56, 61, 77, 78] using an ASTM 5 grain size to ensure consistency with this research. The curves show that the 7 Cu and 38 Cu alloys are consistent in hardenability with the HH rail steel studied by Aglan [61] and the additional copper in the 85 Cu alloy shared a similar hardenability with the Cr-Si rail steel studied by Grigorivich [78]. All alloys studied in this work possess a hardenability less than that of the Cr-Mo rail steel studied by Ochi [56]. Also included in the comparison is a nominal chemistry for a SAE 4140 and the 1045 steel used by Pavlina et al. for frame of reference [4, 79].

It is important to remember that North American rail steels are designed to be fully pearlitic [80], so whereas a Jominy test would not normally be used to qualify the performance of a particular grade, it does provide a useful comparative measure of the risk of martensite formation. Martensite formation is of interest in rail steels; AREMA recommends that if martensite is formed in electric flash butt rail welds due to rapid cooling, then the welding procedure used must be confirmed to produce welds that pass the full-section slow bend test [80].

5.2 Mechanical Properties As mentioned above, increasing copper content delays the pearlite reaction to a small degree by lowering the transformation temperature, which results in a finer pearlite spacing from the reduced carbon diffusivity at lower temperatures. For the 7, 38 and 85 Cu steels in this study the corresponding measured average pearlite spacings were 107, 100, and 98 nm at 10 mm depth from the running surface respectively. The slight reduction in pearlite spacing lowers the mean-free path for dislocation motion in the pearlitic ferrite. This increases the overall resistance to plastic deformation which is manifested in higher hardness and strength properties via a Hall-Petch type relationship. Figure 5.2 shows the Hall-Petch type relationship for transverse hardness at a depth of 10 mm from the running surface of the rail.

Figure 5.2 Hall-Petch type relationship between hardness and pearlite spacing for the 7 Cu, 38 Cu, and 85 Cu alloys at a depth of 10 mm from the running surface.

The Ultimate Tensile and Yield Strengths also followed the Hall-Petch type relationship with decreasing pearlite spacing although the relationship was not as strong. Figure 5.3 displays the effect of decreasing lamellar spacing on the yield and ultimate tensile strengths for the 7 Cu, 38 Cu, and 85 Cu alloys at a depth of 10 mm from the running surface.

Figure 5.3 Hall-Petch type relationship between strength and pearlite spacing for the 7 Cu, 38 Cu, and 85 Cu alloys at a depth of 10 mm from the running surface.

62 Figure 5.4 compares the relationship between interlamellar spacing and yield strength for the 7 Cu, 38 Cu, and 85 Cu alloys used in this study to four different alloys found in the literature.

The relationship between strength and pearlite spacing can be expressed mathematically by a Hall-Petch type correlation as shown in Equation 5.1. −1/2 휎푦 = 휎0 + 푘휆 (5.1) where 휎푦 = actual yield strength (MPa) 휎0 = nominal yield strength (MPa) 푘 = strengthening coefficient (MPa √nm) 휆 = interlamellar spacing (nm)

Linear regression of the data from Figure 5.4 provides the slope or “k” term of the Hall-Petch relationship and is summarized in Table 5.2.

Table 5.2 Results of Linear Regression from Figure 5.4

Alloy 푘 (MPa √nm) 휎0 (MPa) Taleff UHCS 1.8C 6370 419 Taleff UHCS 1.5C 5810 448 Gomes 1080 5450 133 Gomes 1080 Nb 10600 19 This study 24200 -1500

63 The higher slope observed in the 1080 Nb and the alloys of this study suggest that strengthening mechanisms other than interlamellar pearlite spacing refinement, such as precipitation hardening, may contribute to the increase in strength. The highly negative value for 휎0 obtained for the alloys in this study suggests that there are insufficient data to accurately model the strengthening effect. To help illustrate this point, Figure 5.5 highlights the results of this study in comparison with the data from the UHCS 1.8C steel evaluated by Taleff et al. [16] over the same range of interlamellar spacing and shows that the overall strengths measured here are reasonable and consistent with others. Some data and regression lines have been removed for visual clarity. Table 5.3 displays the results of the linear regression of the data in this study with the results of a linear regression of the Taleff et al. data over the same range of interlamellar spacing. The variation in measured yield strength and observed interlamellar spacing suggests a wider range of strength and spacing measurements are needed for accurate modeling. Additional differences in the comparisons shown in Figures 5.4 and 5.5 may also reflect differences in the specific procedures each study used to measure interlamellar spacing.

64 Table 5.3 Linear regression results from Figure 5.5

Alloy 푘 (MPa √nm) 휎0 (MPa) Taleff UHCS 1.8C 19600 -910 This study 24200 -1500

5.3 Fracture Mechanics The results of the dynamic impact loading test (Charpy) were consistent with literature [54], i.e. the absorbed energies for the high carbon high strength pearlitic alloys studied here were essentially the same for all the alloys studied. High strength pearlitic steels exhibit low impact toughness values at all temperature ranges of practical importance and are thus considered to be in the brittle lower shelf regime in testing and application. It is because of this low impact toughness that AREMA requires a maximum manufacturing defect depth of 0.51 mm (0.020 in) for defects formed during hot rolling [80]. This specification limit is reduced to 0.25 mm (0.010 in) for defects that are formed below 371 °C (700 °F) because of the higher strain fields from plastic deformation at ambient temperatures and the possibility of forming surface martensite. Surface martensite is prone to cracking and the high strength pearlite has limited ability to tolerate the cracks under dynamic loading. As discussed in Section 4.3.2, the plane strain fracture toughness exhibited a limited dependence on copper content. Specifically the 7 Cu alloy did exhibit the highest average toughness value of 38.6 MPa √m and toughness decreased slightly with increasing copper content. The 7 Cu alloy also had the lowest yield strength with an average result of 850 MPa, and as shown in Figure 5.3, the yield strength increases slightly with copper content. To understand how mechanical properties influence fracture properties, a plot of fracture toughness versus yield strength is shown in Figure 5.6. This plot suggests that fracture toughness has a small inverse relationship with yield strength. To expand on this idea, Figure 5.7 displays the fracture toughness as a function of ductility as measured by reduction of area percent from tensile testing along with data from Ochi et al. [56]. As the plot indicates, the results of this study are in very good agreement with published data. From the preceding discussion, it is observed that reduction of area and yield strength obtained from the tensile test along with the plane strain fracture toughness correlate with one another in the experimental alloys. Of these three parameters, yield strength and fracture toughness can be combined into a single parameter using the concept of the plastic zone at the crack tip according to

Equation 5.1 [81, 83] with K = K1c.

Figure 5.6 Relationship between plane strain fracture tougness and tensile ductility. Horizontal and vertical error bars represent one standard deviation from the mean.

Figure 5.7 Relationship between plane strain fracture toughness and tensile ductility for the alloys of this study along with published results from Ochi et al. [56].

1 퐾 2 푟푦 = ( ) 5.1 6휋 휎푦푠 where

푟푦 = size of plastic zone ahead of crack tip.

Figure 5.8 shows the correlation between plastic zone size and reduction of area. In this context, the plastic zone size not only serves as a useful parameter to capture the combined behavior of yield strength and fracture toughness, but it may also provide insight into the observed differences in fracture behavior and the role of yield strength. Namely, with an increase in plastic zone size (a consequence of a decrease in yield strength), the amount of energy that contributes to plastic deformation processes ahead of the crack tip increases, thus reducing the amount of energy available to promote brittle fracture at the crack tip [83]. Similarly, reduction of area in the tensile test is a measure of the deformation required to produce fracture [84]. Thus, the correlation observed in Figure 5.8 has some physical significance, even though the data used in the plot originate from two different types of testing.

Figure 5.8 Plastic zone size calculated from Equation 5.1 as a function of tensile ductility as measured by percent reduction of area.

With the understanding that fatigue in metals can lead to catastrophic failure, knowledge of the fracture toughness and fatigue crack growth rate of a material can help in the scheduling of periodic ultrasonic in-service inspections performed in the railroad industry. The ability to detect a fatigue crack in a rail before it reaches a critical size improves the safety and cost effectiveness of the railroads. The

67 fatigue crack growth rates of the alloys investigated in this study are very consistent with published data [56, 59, 63, 65, 82]. As mentioned in Section 4.3.3, the steady-state crack extension, i.e. Stage II crack growth, can be represented by the Paris equation. Figure 5.9 illustrates the data derived from this study along with the expected plot for ferrite/pearlite steels from Barsom and Rolfe [82], as well as other commercial pearlitic rail steels in use [63].

Figure 5.9 Comparison of steady state region of fatigue crack growth rate curves plotted from the Paris equation for various pearlitic steels.

The slopes of the fatigue crack growth rate curves for the 7 Cu, 38 Cu, and 85 Cu steels are in very good agreement with other pearlitic steels of differing mechanical properties which suggests that the fatigue crack growth rate (da/dN) is not strongly dependent on the microstructural variables that control other mechanical properties of the pearlitic steel. It can be inferred that the rate of fatigue crack growth through pearlite is independent of the increase in hardness and strength with copper content. This conclusion agrees with the observations of Barsom and Rolfe [112] that the fatigue crack propagation in the steady state crack extension regime is not strongly influenced by minor microstructural changes within a particular family of alloys [82]. Correspondingly, they proposed a universal equation for ferrite/pearlite steels given in Equation 5.2. The values for C0 and m in Equation 5.2 are very similar to the observed behavior of the steels in this study as shown in Table 4.10.

푑푎 = 6.8 × 10−12(∆퐾)3.0 (5.2) 푑푁

For da/dN in units of m/cycle and ∆K in units of MPa √m

CHAPTER 6

CONCLUSIONS

This section briefly describes the conclusions that can be drawn from the results of this work.

1. The austenite grain growth characteristics in rail steels are independent of copper contents for steels with copper contents between 0.07 and 0.85 wt pct when billets are re-heated in a suitable atmosphere and hot-rolled with a reduction ratio between 8:1 and 20:1 since all of the alloys have an average McQuaid-Ehn austenitic grain size of approximately ASTM 5. 2. Copper acts as a hardenability element that delays pearlite transformation in rail steel, likely due to its role as an austenite stabilizer and poor solubility in pearlitic cementite. 3. The ability of copper to delay pearlite transformation results in a finer pearlite interlamellar spacing with increasing copper content through head hardening operations during rail manufacturing. 4. The finer pearlite spacing with increased copper content results in higher hardness, tensile strength, and yield strength. There appears to be an additional active strengthening mechanism in the alloys studied (i.e. precipitation or solid solution strengthening) not explained solely through interlamellar spacing refinement.

5. The fracture characteristics, as measured by Charpy U-notch (2 mm) impact tests, standard K1c fracture toughness testing on compact tension specimens, and fatigue crack growth rate measurements, are predominantly independent of copper content for the rail steels evaluated.

69 CHAPTER 7

FUTURE WORK

Whereas the experiments performed in this research were able to answer some of the questions as to the effects of rising copper levels in rail steels produced in EAF shops, there is still benefit in developing additional understanding. This section makes a few suggestions as to the next logical steps in fully characterizing the influence of rising copper levels in rail steels.

7.1 Wear rates As mentioned in Section 2 of this work, the best measure of wear performance in pearlitic rail steels is the surface hardness of the rail. With the hardness gradient observed and discussed in Section 4.2.2, it can be expected that the wear resistance decreases as material is removed via wear over the life of the rail. It was observed in this study that the overall hardness profile curves increased with increasing copper content. It would be interesting to examine if the small increase in hardness observed over the range of the alloys tested manifested in an observed decrease in rail wear.

7.2 Copper strengthening mechanisms With the ability to minimize the variability in the selected alloys with respect to chemistry, total hot reduction, and thermal history, this study was able to isolate small differences in overall strength, hardness, and hardenability as a function of varying copper content. A slight decrease in the interlamellar spacing was observed which at least partially explains the increase in strength and hardness properties. An evaluation of the Hall-Petch relationship, as determined by linear regression, showed a very strong strengthening coefficient (푘) along with a highly negative nominal yield stress (휎0). Developing thermal cycles to create a wider range of interlamellar spacing within each composition could help provide more reasonable values for the 휎0 and 푘 terms of the Hall-Petch equation and isolate the additional strengthening effects of copper, i.e. precipitation and/or solid solution strengthening. Transmission Electron Microscopy (TEM) and Atom Probe Tomography (APT) analysis could elucidate the location of the copper in the final microstructure.

7.3 Weld testing To provide for a smooth running surface, most rails are electric flash-butt welded into quarter- mile long strings in a weld plant and transported to the installation site via specialized rail cars. Field welding is accomplished through both mobile electric flash-butt welding and thermite welding. With only the ends of the rails being locally austenitized in electric flash-butt welding, the remaining mass of rail acts as a heat sink and could potentially rapidly quench the weldment to form martensite. As

70 increasing copper content was shown to increase the hardenability of the steel in Section 4.1.4, an evaluation of the impact on weldability is warranted to ensure that there is not an undue increase in the likelihood of martensite formation at or near the fusion line of electric flash-butt welds at rail copper contents exceeding the currently accepted maximum level of 0.4 wt pct Cu [80]. It is also important to point out that consideration of martensite formation in welds applies to all hardenability elements such as Mn, Cr, and Si that are prevalent in all rail steels, not just copper. Conversely, secondary copper precipitation may provide an increase in the hardness in the weld heat affected zones (HAZs) and provide a negation of the softening effect of spheroidization of the lamellar pearlitic cementite that occurs as a result of welding. Higher hardness in this region would improve the integrity of the running surface in the soft HAZs, thus improving the performance of the rail weld. This aspect of the possible influence of copper on rail welding also warrants study.

71 REFERENCES CITED

[1] J.C. Herman and V. Leroy, “Influence of Residual Elements on Steel Processing and Mechanical Properties,” Presented at “Metal Working and Steel Processing,” Cleveland, October 14-17, 1996.

[2] L. Savov, E. Volkova, and D. Janke, “Copper and Tin in Steel Scrap Recycling,” Materials and Geoenvironment, vol. 50, pp. 627-640, 2003.

[3] K. Shibata, S. Seo, M. Kaga, H Uchino, A. Sasanuma, K. Asakura and C. Nagasaki, “Suppresion of Surface Hot Shortness due to Cu in Recycled Steels,” Materials Transactions, vol. 43, pp. 292-300, 2002.

[4] E.J. Pavlina, S.J. Lee, E.T. Virtanen, L.M. Rothleutner and C.J. Van Tyne, “Effects of Copper on the Hardenability of a Medium-Carbon Steel,” Metallurgical and Materials Transactions A, vol. 42A, pp. 3572 3576, 2011.

[5] Steel Manufacturer’s Association CRU World Steel Conference, London, England, Feb. 2012, power point presentation.

[6] G. Krauss, Steels – Processing, Structure, and Performance, Materials Park, OH: ASM International, 2005, pp. 281-295.

[7] R.O. Olivares, C.I. Garcia, A. DeArdo, S. Kalay and F.C.R. Hernandez, “Advanced Metallurgical Alloy Design and Thermomechanical Processing for Rail Steels for North American Heavy Haul Use,” Wear, vol. 271, pp. 364-373, 2011.

[8] P. Clayton and D. Danks, “Effect of Interlamellar Spacing on the Wear Resistance of Eutectoid Steels under Rolling-Sliding Conditions,” Wear, vol. 135, pp. 369-389, 1990.

[9] P. Pointner, “High Strength Rail Steels – The Importance of Material Properties in Contact Mechanics Problems,” Wear, vol. 265, pp. 1373-1379, 2008.

[10] J. Kalousek, D.M. Fegredo, E.E. Laufer, “The Wear Resistance and Worn Metallography of Pearlite, Bainite and Tempered Martensite Rail Steel Microstructures of High Hardness,” Wear, vol. 100, pp. 199-222, 1985.

[11] R. Stock and R. Pippan, “RCF and Wear in Theory and Practice – The Influence of Rail Grade on Wear and RCF,” Wear, vol. 271, pp. 125-133, 2011.

[12] G. Krauss, Steels – Heat Treatment and Processing Principles, ASM International, Materials Park, OH, 1989, pp. 133-138.

[13] K. Nakase and I.M. Bernstein, “The Effect of Alloying Elements and Microstructure on the Strength and Fracture Resistance of Pearlitic Steel,” Metallurgical Transactions A. vol. 19A, pp. 2819-2829, 1988.

[14] M.G.M.F. Gomes, L.H. de Almeida, L.C.F.C. Gomes and I.L. May, “Effects of Microstructural Parameters on the Mechanical Properties of Eutectoid Rail Steels,” Materials Characterization, vol. 39, pp. 1-14, 1997.

72 [15] A.M. Elwazri, P. Wanjara and S. Yue, “The Effect of Microstructural Characteristics of Pearlite on the Mechanical Properties of Hypereutectoid Steel,” Materials Science and Engineering A, vol. 404, pp. 91-98, 2005.

[16] E.M. Taleff, C.K. Syn, D.R. Lesuer and O.D. Sherby, “Pearlite in Ultrahigh Carbon Steels: Heat Treatments and Mechanical Properties,” Metallurgical and Materials Transactions A, vol. 27A, pp. 111-118, 1996.

[17] K.M. Wu and H.K.D.H. Bhadeshia, “Extremely Fine Pearlite by Continuous Cooling Transformation,” Scripta Materialia, vol. 67, pp. 53-56, 2012.

[18] J.W. Ringsberg, T. Lindback, “Rolling Contact Fatigue Analysis of Rails including Numerical Simulations of the Rail Manufacturing Process and Repeated Wheel-Rail Contact Loads,” International Journal of Fatigue, vol. 25, pp. 547-558, 2003.

[19] A.M. Elwazri, P. Wanjara and S. Yue, “Empirical Modelling of the Isothermal Transformation of Pearlite in Hypereutectoid Steel,” Materials Science and Technology, vol. 22, no. 5, pp. 542-546, 2006.

[20] G. Fourlaris, A.J. Baker and G.D. Papadimitriou, “A Microscopic Examination of the Precipitation Phenomena Occurring During the Isothermal Pearlitic Transformation in High Carbon-Copper Nickel Steels,” Acta Metallurgica et Materialia, vol. 43, no. 12, pp. 4421-4438, 1995.

[21] T. Chairuangsri and D.V. Edmonds, “The Precipitation of Copper in Abnormal Ferrite and Pearlite in Hyper-Eutectoid Steels,” Acta Materialia, vol. 48, pp. 3931-3949, 2000.

[22] F.A. Khalid and D.V. Edmonds, “On the Properties and Structure of Micro-Alloyed and Copper-Bearing Hot-Rolled Steels,” Journal of Materials Processing Technology, vol. 72, pp. 434-436, 1997.

[23] J.M. Hyzak and I.M. Bernstein, “The Role of Microstructure on the Strength and Toughness of Fully Pearlitic Steels,” Metallurgical Transactions A, vol. 7A, pp. 1217-1224, 1976.

[24] R.E. Reed-Hill, Physical Metallurgy Principles, Second Edition, Boston: PWS Engineering, 1973, pp. 670-676.

[25] F.B. Pickering and B. Garbarz, “The Effect of Transformation Temperature and Prior Austenite Grain Size on the Pearlite Colony Size in Vanadium Treated Pearlitic Steels,” Scripta Metallurgica, vol. 21, pp. 249-253, 1987.

[26] A.M. Elwazri, P. Wanjara and S. Yue, “Continuous Cooling Transformation Temperature and Microstructures of Microalloyed Hypereutectoid Steels,” ISIJ International, vol. 46, pp. 1354-1360, 2006.

[27] F.L. Lian, H.J. Liu, J.J. Sun, X.J. Sun, S.W. Guo and Y.N. Liu, “Ultrafine Grain Effect on Pearlitic Transformation in Hyperteutectoid Steel,” Journal of Materials Research, vol. 28, no. 5, pp. 757-765, 2013.

[28] Y. Liu, T. He, G. Peng and F. Lian, “Pearlitic Transformations in an Ultrafine-Grained

73 Hypereutectoid Steel,” Metallurgical and Materials Transactions A, vol. 42A, pp. 2144-2152, 2011.

[29] E.B. Hawbolt, B. Chau and J.K. Brimacombe, “Kinetics of Austenite-Pearlite Transformation in Eutectoid Carbon Steel,” Metallurgical Transactions A, vol. 14A, pp. 1803-1815, 1983.

[30] A.M. Elwazri, P. Wanjara and S. Yue, “Measurement of Pearlite Interlamellar Spacing in Hypereutectoid Steels,” Materials Characterization, vol. 54, pp. 473-478, 2005.

[31] D. Markov, “Laboratory Tests for Wear of Rail and Wheel Steels,” Wear, vols. 181-183, pp. 678-686, 1995.

[32] G.F. Vander Voort and A. Roosz, “Measurement of the Interlamellar Spacing of Pearlite,” Metallography, vol. 17, pp. 1-17, 1984.

[33] F.G. Caballero, C. Garcia de Andres and C. Capdevila, “Characterization and Morphological Analysis of Pearlite in a Eutectoid Steel,” Materials Characterization, vol. 45, pp. 111-116, 2000.

[34] H.S. Fong, “Determining True Pearlite Lamellar Spacings from Observed Apparent Spacings,” Metallography, vol. 23, pp. 173-188, 1989.

[35] J.F. Tilbury, T.D. Mottishaw and G.D.W. Smith, “Short Communication: Measurements of the Spread in Pearlite Spacings,” Metallography, vol. 19, pp. 243-246, 1986.

[36] G. Krauss, “Evaluation of Pearlitic Microstructures in Six Samples of Rail Steel (Krauss Report 151),” unpublished correspondence, 2011.

[37] G. Krauss, Steels – Heat Treatment and Processing Principles, ASM International, Materials Park, OH, 1989, pp. 193-201.

[38] G. Krauss, Steels – Heat Treatment and Processing Principles, Materials Park, OH: ASM International, 1989, pp. 145-178.

[39] C.A. Siebert, D.V. Doane and D.H. Breen, The Hardenability of Steels, ASM International, Materials Park, OH, 1977.

[40] “Standard Test Methods for Determining the Hardenability of Steel,” A-255-10, ASTM International, West Conshocken, PA, 2010.

[41] E.C. Bain, “Factors Affecting the Inherent Hardenability of Steel,” Journal of Heat Treating, vol. 1, no. 2, pp. 57-100, 1932.

[42] M.V. Li, D.V. Niebuhr, L.L Meekisho and D.G. Atteridge, “A Computational Model for the Prediction of Steel Hardenability,” Metallurgical and Materials Transactions B, vol. 29B, pp. 661-672, 1998.

[43] G.T. Brown and B.A. James, “The Accurate Measurement, Calculation, and Control of Steel Hardenability,” Metallurgical Transactions, vol. 4, pp. 2245-2256, 1973.

[44] C.J. McMahon, Jr., “The Role of Solute Segregation in Promoting the Hardenability of Steel,”

74 Metallurgical Transactions A, vol. 11A, pp. 531-535, 1980.

[45] Symposium: Transformation and Hardenability in Steels, Climax Molybdenum Company of Michigan & The University of Michigan, February 1967.

[46] R.V. Fostini and F.J. Schoen, Transformation and Hardenability in Steels, Climax Molybdenum Company of Michigan & The University of Michigan, pp. 195-209, Feb. 1967.

[47] F.P.L. Kavishe and T.J. Baker, “Effect of Prior Austenite Grain Size and Pearlite Interlamellar Spacing on Strength and Fracture Toughness of a Eutectoid Rail Steel,” Materials Science and Technology, vol. 2, pp. 816-822, 1986.

[48] P.R. Sreenivasan, S.K. Ray, K.G. Samuel and S.L. Mannan, “An Empirical Relation between Yield Stress and General Yield Load for a Charpy U-Notch Specimen,” Engineering Fracture Mechanics, vol. 42, no. 6, pp. 1047-1049, 1992.

[49] B. Nageswara Rao and A.R. Acharya, “A Comparative Study on Evaluation of Fracture Toughness from Charpy V-Notch Impact Energy and Reduction-In-Area,” Engineering Fracture Mechanics, vol. 41, no. 1, pp. 85-90, 1992.

[50] B. Strnadel and P. Hausild, “Statistical Scatter in the Fracture Toughness and Charpy Impact Energy of Pearlitic Steel,” Materials Science and Engineering A, vol. 486, pp. 208-214, 2008.

[51] A. Rossoll, C. Berdin, P. Forget, C. Prioul and B. Marini, “Mechanical Aspects of the Chapy Impact Test,” Nuclear Engineering and Design, vol. 188, pp. 217-229, 1999.

[52] R.E. Reed-Hill, Physical Metallurgy Principles, Second Edition, Boston: PWS Engineering, 1973, pp. 780-795.

[53] G.E. Dieter, Mechanical Metallurgy, SI Metric Edition, London, McGraw Hill Book Co., 1988, pp. 471-486.

[54] F. Alwahdi, F.J. Franklin, A. Kapoor, “The Effect of Partial Slip on the Wear Rate of Rails,” Wear, vol. 258, pp. 1031-1037, 2005.

[55] “Standard Test Methods for Linear-Elastic Plane-Strain Fracture Toughness K1C of Metallic Materials,” E399-12ε3, ASTM International, West Conshocken, PA, 2012.

[56] Y. Ochi and A.J. McEvily, “An Evaluation of the Fatigue Crack Growth and Fracture Toughness Characteristics of Rail Steels,” Engineering Fracture Mechanics, vol. 29, no. 2, pp. 159-172, 1988.

[57] R. Ravaee and A. Hassani, “Fracture Mechanics Determinations of Allowable Crack Size in Railroad Rails,” Journal of Failure Analysis and Prevention, vol. 7, pp. 305-310, 2007.

[58] G. Garcia, “Flaw Characterization of Rail Service Failures”, Transportation Technology Center, Inc. Report No. R-963, p. A-4, July 2003.

75 [59] “Standard Test Method for Measurement of Fatigue Crack Growth Rates,” E647-13a, West Conshoshoken, PA: ASTM International, 2013

[60] G.T. Gray III, A.W. Thompson, J.C. Williams and D.H. Stone, “The Effect of Microstructure on Fatigue Crack Propagation in Pearlitic Eutectoid Steels,” Canadian Metallurgical Quarterly, vol. 21, no. 1, pp. 73-78, 1982.

[61] H. Aglan and Y.X. Gan, “Fatigue Crack Growth Analysis of a Premium Rail Steel,” Journal of Materials Science, vol. 36, pp. 389-397, 2001.

[62] J. Toribio, B. Gonzalez and J.C. Matos, “Transient and Steady State Regimes of Fatigue Crack Growth in High Strength Steel,” Key Engineering Materials, vols. 525-526, pp. 553-556, 2013.

[63] B. Lu and X. Zheng, “Prediction of Fatigue Crack Growth Rate in Steels,” International Journal of Fracture, vol. 55, pp. 21-31, 1992.

[64] O. Orringer, J. M. Morris and D.Y. Jeong, “Detail Fracture Growth in Rails: Test Results,” Theoretical and Applied Fracture Mechanics, vol. 5, pp. 63-95, 1986.

[65] R.C. Dimitriu and H.K.D.H. Bhadeshia, “Fatigue Crack Growth Rate Model for Metallic Alloys,” Materials and Design, vol. 31, pp. 2134-2139, 2010.

[66] H.W. Liu and D Liu, “A Quantitative Analysis of Structure Sensitive Fatigue Crack Growth in Steels,” Scripta Metallurgica, vol. 18, pp. 7-12, 1984.

[67] J. Toribio, J.C. Matos and B. Gonzalez, “Micro- and Macro-Approach to the Fatigue Crack Growth in Progressively Drawn Pearlitic Steels at Different R-Ratios,” International Journal of Fatigue, vol. 31, pp. 2014-2021, 2009.

[68] J.V. Cartensen and T. Magnin, “Characterization and Quantification of Multiple Crack Growth during Low Cycle Fatigue,” International Journal of Fatigue, vol. 23, pp. S195-S200, 2001.

[69] “American Railroad and Maintenance-of-Way Association Manual for Railway Engineering” Chapter 4, pp.18 -20, 2013

[70] “Standard Test Methods for Determining Average Grain Size,” E112-13, ASTM International, West Conshocken, PA, 2013.

[71] “Standard Test Methods and Definitions for Mechanical Testing of Steel Products,” A370-14, ASTM International, West Conshocken, PA. 2014

[72] “Standard Test Methods for Tension Testing of Metallic Materials,” E8-13a, ASTM International, West Conshocken, PA. 2013

[73] “Standard Test Methods for Notched Bar Impact Testing of Metallic Materials,” E23-12c, ASTM International, West Conshocken, PA, 2012.

[74] E.L. Crow, F.A. Davis and M.W. Maxfield, Statistics Manual, Mineola, N.Y., 1960, Appendix Table 3 p.231

76 [75] S.K. Ghosh, P.S. Bandyopadhyay, S. Kundu and S. Chatterjee, “Copper Bearing Microalloyed Ultrahigh Strength Steel on a Pilot Scale: Microstructure and Properties,” Materials Science and Engineering A, vol. 528, pp. 7887-7894, 2011.

[76] S.K. Ghosh, A. Haldar and P.P. Chattopadhyay, “Effect of Ageing on the Mechanical Properties of Directly Quenched Copper Bearing Microalloyed Steels,” Materials Chemistry and Physics, vol. 119, pp. 436-441, 2010.

[77] M.A. Daeubler, A.W. Thompson and I.M. Bernstein, “Influence of Microstructure on Fatigue Behavior and Surface Fatigue Crack Growth of Fully Pearlitic Steels,” Metallurgical Transactions A, vol. 21A, pp. 925-933, 1990.

[78] K.V. Grigorovich, A.S. Trushnikova, A.M. Arsenkin, S.S. Shibaev and A.K. Garber, “Structure and Metallurgical Quality of Rail Steels Produced by Various Manufacturers,” Russian Metallurgy (Metally), vol. 2006, no. 5, pp. 427-438, 2006.

[79] “Classification and Designation of Carbon and Low-Alloy Steels” in ASM Metals Handbook Volume 1, 10th edition, Materials Park OH: ASM International, p. 152, 1990.

[80] “American Railroad and Maintenance-of-Way Association Manual for Railway Engineering” Chapter 4, p.28, 2013

[81] J. Barson and S. Rolfe, Fracture and Fatigue Control in Structures – Applications of Fracture Mechanics, second edition, New Jersey, Prentice-Hall Inc., 1987, p. 56

[82] J. Barsom and S. Rolfe, Fracture and Fatigue Control in Structures – Applications of Fracture Mechanics, second edition, New Jersey, Prentice-Hall Inc., 1987, pp. 278-305.

[83] R.J. Sanford, Principles of Fracture Mechanics, Pearson Education, Inc., 2003, pp. 211-234.

[84] G.E. Dieter, Mechanical Metallurgy, SI Metric Edition, London, McGraw Hill Book Co., 1988, p. 295

77 APPENDIX A

K1c VALIDITY REQUIREMENTS

Although the software program, Fracture Analyzer®, performs all of the test validity requirements, the following example illustrates the calculations for the fracture toughness on sample 1 of the 29 Cu alloy. These requirements are excerpted from ASTM – E399 [55].

푊 Validity requirement #1: 2 ≤ ≤ 4 (A.1) 퐵 푎 Validity requirement #2: 0.45 ≤ ≤ 0.55 (A.2) 푊 Validity requirement #3: 푎1 푎푛푑 푎5 ≥ 푙푎푟푔푒푟 표푓 0.025푊 표푟 1.3푚푚 (A.3) Validity requirement #4: |푎2 − 푎3|, |푎3 − 푎4|, |푎4 − 푎2| ≤ 0.1푎 (A.4)

Validity requirement #5: 푃푙푎푛푒 표푓 푎 ∥ 푃푙푎푛푒 표푓 푎푛 ± 10° (A.5) Validity requirement #6: 0.85푎 ≤ 푎1, 푎5 ≤ 1.15푎 (A.6) Validity requirement #7: 0.9푎 ≤ |푎1 − 푎5| ≤ 1.1푎 (A.7) Validity requirement #8: −1 ≤ 푅 ≤ 0.10 (A.8)

Validity requirement #9: 푃푟푒푐푟푎푐푘 퐾푚푎푥 ≤ 0.8퐾푄 (During entire crack growth) (A.9)

푃푟푒푐푟푎푐푘 퐾푚푎푥 ≤ 0.6퐾푄 (During terminal stage of crack Validity requirement #10: (A.10) growth (last 2.5% of 푎))

퐾푚푎푥 Validity requirement #11: ≤ 0.0003√푚 (A.11) 퐸

푃푚푎푥 Validity requirement #12: ≤ 1.10 (A.12) 푃푄 퐾 2 Validity requirement #13: 푊 − 푎 ≥ 2.5 ( 푄 ) (A.13) 휎푦푠 where 푊 = specimen width = 38.2 mm (1.505 in) 퐵 = specimen thickness = 19.1 mm (0.752 in) 푎 = average crack size = 1/3 (푎2 + 푎3 + 푎4) = 20.4 mm (0.803 in) 푎3, 푎2, 푎4 = fatigue crack length at the mid-thickness and quarter-thickness points (20.6, 20.5, 20.2 mm) respectively.

78 푎1 푎푛푑 푎5 = fatigue crack length as measured on faces of the specimen (19.3, 18.7mm)

푎푛 = starter notch = 16.3 mm (0.640 in) 푅 = loading ratio = minimum pre-cracking load / maximum pre-cracking load = 0.10

퐾푄 = conditional stress intensity factor (equation 4.26). If all criteria are met, 퐾푄 = 퐾퐼푐

퐾푚푎푥 = maximum stress intensity factor during last 2.5% of a = 17.79 MPa √m (16.2 ksi√in) E = Young’s modulus = 2 x 105 MPa (29 x 106 psi)

푃푚푎푥 = the maximum force the specimen was able to sustain = 13.4 kN

푃푄 = maximum value of the force crack mouth displacement curve at or below a 5% secant line w/r/t the linear portion of the curve = 12.9 kN.

휎푦푠 = 0.2% offset yield strength = 905 MPa (131.2 ksi)

The calculation of 퐾푄 is found in Annex A4.5 of ASTM – E399 [T 61] and is given in Equation A.14.

푃푄 푎 퐾 = ( ) × 푓 ( ) (A.14) 푄 푊 √(퐵퐵푛) × √푊

where 퐾푄 = conditional stress intensity factor

푃푄 = maximum value of the force crack mouth displacement curve at or below a 5% secant line w/r/t the linear portion of the curve = 12.9 kN 퐵 = specimen thickness = 19.1 mm (0.752 in)

퐵푛 = thickness of the specimen at groove (if used) – for our case 퐵푛 = 퐵 푊 = specimen width = 38.2 mm (1.505 in) 푎 푓 ( ) = a polynomial function depicted in Equation A.15 푊

푎 푎 푎 푎 [0.886 + 4.64 ( ) − 13.32( )2 + 14.72( )3 − 5.6( )4] 푎 푎 푊 푊 푊 푊 푓 ( ) = (2 + ) × 3 (A.15) 푊 푊 푎 (1 − )2 푊

Substituting our values: KQ = 37.1 MPa √m for this specimen. Since all criteria were met for all specimens tested, KQ = K1c.