UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigation of erosive-corrosive wear in the low pressure die casting of aluminum A35 Miller, Ainsley Elizabeth 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2005-0274.pdf [ 16.53MB ]
Metadata
JSON: 831-1.0078774.json
JSON-LD: 831-1.0078774-ld.json
RDF/XML (Pretty): 831-1.0078774-rdf.xml
RDF/JSON: 831-1.0078774-rdf.json
Turtle: 831-1.0078774-turtle.txt
N-Triples: 831-1.0078774-rdf-ntriples.txt
Original Record: 831-1.0078774-source.json
Full Text
831-1.0078774-fulltext.txt
Citation
831-1.0078774.ris

Full Text

INVESTIGATION OF EROSIVE-CORROSIVE WEAR IN THE LOW PRESSURE DIE CASTING OF ALUMINUM A356  by  AINSLEY ELIZABETH MILLER  B.ASc, The University of British Columbia, 2003  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE  in  THE FACULTY OF GRADUATE STUDIES  (Metals and Materials Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA April 2005 © Ainsley Elizabeth Miller, 2005  Abstract Low pressure die casting is a process commonly used to produce aluminum wheels. Die surfaces are exposed to a wide range of operating conditions, including high temperatures, high local pressures, and varying melt velocity, which can result in erosive-corrosive wear. The sprue is an integral component of the die and channels liquid aluminum into the wheel cavity. Typically sprues are made from steel and have a protective coating applied to the surface to provide insulation from the effects of liquid aluminum. However, flow-related phenomena can prematurely wear away this coating, resulting in accelerated surface wear.  Erosive-corrosive wear experiments were performed to investigate the effect of geometry on sprue material wear in liquid aluminum alloy A356. Experiments were performed using a laboratory rotating pin apparatus, similar to other tests in literature, examining the effect of pin geometry, velocity, material, temperature, and time on wear behaviour. A profiled crosssection incorporating features from the sprue geometry was used to evaluate the influence of geometry. To ensure that test velocities were realistic, fluid flow models were developed to simulate die cavity filling during low pressure die casting and velocity profiles along the rotating pin surface. The pins were primarily made from 4140 steel, with two profiled pins made from HI3 steel and titanium alloy Ti-6A1-4V. Additional tests were performed with cylindrical pins to validate test set-up and evaluate the effect of additional parameters.  Wear increased with increasing velocity, temperature, and time. The titanium pin showed the least amount of wear, followed by the H13 steel pin. The profiled cross-section produced regions of different pressure and velocity along the surface, promoting accelerated wear at some locations. Most wear was observed along the leading edge of the rotating pin. The flow in the sprue exit region was more closely represented by flow along the pin trailing edge. Along both surfaces, flow simulation predicted a region of flow separation. This region was largest for the 15° pin, which experienced the least wear along the trailing edge and may have implications on accelerated sprue wear. More study needs to be completed in terms of the influence of flow on wear, however, preliminary results suggest that larger pin/sprue draft angles may improve wear performance.  Table of Contents Abstract  ii  Table of Contents  iii  List of Tables  v  List of Figures  vii  Acknowledgements  xi  CHAPTER 1: INTRODUCTION  1  1.1 Die Casting Process  2  1.2 Die Casting Alloys  5  1.3 Sprue Wear  5  CHAPTER 2: LITERATURE REVIEW  9  2.1 Corrosion  9  2.1.1 Dissolution 2.1.2 Intermetallic Compound Formation and Growth 2.1.2.1 Phases in the Fe-Al and Fe-Al-Si Systems 2.1.2.2 Reaction Thermodynamics and Kinetics 2.1.2.3 Corrosion Experiments 2.2 Erosion  :  10 12 12 13 15  ....  16  2.3 Erosion-Corrosion  19  2.3.1 Erosive-Corrosive Wear 2.3.2 Mechanisms for Erosive-Corrosive Wear 2.3.2.1 Phase Boundary Attack 2.3.2.2 Dissolution and Dissociation of Intermetallic Layers 2.4 Work Performed at CAPTIN  20 23 23 24 25  2.5 Summary  26  CHAPTER 3: SCOPE AND OBJECTIVES  28  3.1 Objectives  28  CHAPTER 4: MATHEMATICAL MODEL  29  4.1 Free Surface Model Formulation  29  4.2 Implementation of Filling Model  31  4.2.1 Geometry 4.2.2 Material Properties and Operating Conditions 4.2.3 Boundary Conditions 4.2.3.1 Inlet Pressure Boundary Condition 4.2.3.2 Outlets 4.2.3.3 Screen 4.2.3.4 Symmetry and Wall Boundaries  .,  .......  32 35 35 35 36 38 41  in  4.3 Model Results  41  4.3.1 Effect of Geometry 4.3.2 Effect of Surface Tension 4.3.3 Effect of Screen  41 55 .56  4.4 Mesh Sensitivity  58  4.5 Summary  61  CHAPTER 5: EXPERIMENTAL PROCEDURE  63  5.1 Sprue Examination  63  5.2 Sprue Temperature  67  5.3 Profiled Pin Design  70  5.3.1 Pin Geometry 5.3.2 Pin Design Assessment  70 '.  72  5.4 Experimental Set-up  77  5.5 Test Conditions  80  5.6 Analysis of Pins  82  5.7 Summary  83  CHAPTER 6: RESULTS AND DISCUSSION 6.1 Cylindrical Test Pins 6.1.1 Radius Reduction 6.1.2 Intermetallic Compound Layers 6.2 Profiled Test Pins 6.2.1 Radius Reduction for 4140 Steel Pins 6.2.2 Radius Reduction for Pins of Different Material 6.2.3 Intermetallic Compound Layers  85 85 85 87 92 93 99 102  6.3 Comparison of Profiled Pins with Sprue  106  6.4 Summary  108  CHAPTER 7: SUMMARY AND CONCLUSIONS 7.1 Recommendations for Future Work Bibliography  Ill 112 114  iv  List of Tables Table 1.1: Referenced Die Steel Compositions (in Weight Percent)  2  Table 1.2: Aluminum Alloys Commonly Used in Die Casting (in Weight Percent)  5  Table 1.3: Summary of Sprue Wear Conditions and Geometric Features  7  [6]  Table 2.1: Solubility of Elements in Pure Liquid Aluminum  11  [14]  Table 2.2: Possible Reactions Between Fe-Al Phases and Liquid Aluminum  [19]  13  Table 2.3: Invariant Liquidus Reactions in the Fe-Al-Si System in Die Casting Temperature Range - -  13  1 20 1  Table 2.4: Elemental Composition of Binary Fe-Al Phases  13  Table 2.5: Elemental Composition of Possible Ternary Fe-Al-Si Phases  [21]  13  Table 2.6: Thermodynamic Quantities for the Formation of Intermetallic Phases in the Iron-Aluminum System  14  [21]  Table 2.7: Diffusion Data for Phases in the Iron-Aluminum System  [21]  Table 2.8: Intermetallic Layer Thickness and Radius Reduction  [18]  15 21  Table 2.9: Element Distribution and Hardness of Phases for QRO-90S Pins Rotated in A380 at 1000 rpm for 4 hours at 715°C  [271  22  Table 4.1: Summary of Mesh Geometries  32  Table 4.2: Fluid Properties  35  Table 4.3: Pressure curve data supplied by CAPTIN for the 330N model  36  Table 4.4: Effect of Screen Void Percentage on Pressure Drop for A356 Phase  40  Table 4.5: Mesh Sensitivity  59  Table 5.1: Elemental Compositions of Interface Layers in Weight Percent  65  Table 5.2: Survey of Geometric Features of Sprues in Constriction Region  70  Table 5.3: Composition of Titanium Alloy Ti-6A1-4V .....  80  Table 5.4: Test Matrix For Cylindrical Test Pins  81  Table 5.5: Test Matrix for Profiled Cross-Section Pins  81  Table 6.1: EDS Results from Cylindrical Pin Tests  91  [36]  Table 6.2: Layer Thickness Measurements for Leading Edge of 4140 Steel Profiled Pins ..103 Table 6.3: Layer Thickness Measurements for Trailing Edge of 4140 Steel Profiled Pins...103 Table 6.4: EDS Results from 4140 Steel Profiled Pin Tests  104  Table 6.5: Layer Thickness Measurements for Leading Edge for 10° Profiled Pins of Different Material  105  v  Table 6.6: Layer Thickness Measurements for Trailing Edge for 10° Profiled Pins of Different Material  105  Table 6.7: Composition of Layers Observed Along Interface of H13 Steel Pin (wt%)  105  Table 6.8: Composition of Layers Observed Along Interface of Ti-6A1-4V Pin (wt%)  105  List of Figures Figure 1.1: Substitution of aluminum wheels for steel wheels in passenger car market^  1  Figure 1.2: Illustration of a die casting machine setup at CAPTTN, including holding furnace and die  4  Figure 1.3: Typical pressure curve used in low pressure die casting  4  Figure 1.4: A cross-section of the 330N sprue is shown in (a). This is one sprue at CAPTIN that has consistently experiencing accelerated wear along the exit surface, as shown in the image of the 330N sprue in (b) and the crosssection of the constriction region in (c)  6  Figure 1.5: Comparison of 3 3 ON and 592N LP wheel model sprues  8  Figure 2.1: Intermetallic growth within iron and aluminum, where AB is the initial interface and the arrow length is a measure of the diffusion rate. [23]  14  Figure 2.2: Effect of melt temperature on the H13 pin dissolution in a static A390 melt (solid line is radius reduction, dotted line is intermetallic thickness).  [13]  Figure 2.3: Schematic of the (a) multiple pin test die and (b) pin design^ -  24 1  16 19  Figure 2.4: Corrosion of AISI HI3 steel during exposure in Al-10Si-0.8Fe melt at 735°C. 20 Figure 2.5: Backscattered electron image of the morphology of the interface between H21 steel and A380 after 9 hours at 700°C. 21 [12]  [l8]  Figure 2.6: Microhardness across reaction zone between H21 steel and A380 alloy after 9 hours 22 [,8]  Figure 2.7: A schematic of the mechanism of die soldering, (a) Initial attack of grain boundaries by aluminum, (b) Formation of iron-aluminum phases, (c) Growth of ternary a-(Al,Fe,Si) phases, (d) Growth of intermetallic layers and merging of neighboring pits, (e) Straightening of pits. [19]  23  Figure 2.8: Effect of rotation rate on the (a) radius reduction and (b) intermetallic layer thickness of H13 steel pins in agitated A390 melt at 680°C.  [13]  Figure 4.1: 3D geometry showing exterior surface mesh  25 33  Figure 4.2: 2D axisymmetric mesh of sprue and wheel cavity. The thick spoke section is shown in (a) and the modified spoke section in (b)  34  Figure 4.3: 2D thick spoke section, illustrating outlets at locations C and D  37  Figure 4.4: Velocity magnitude contour plots for the 2D thick spoke model during filling. (m/s)  43  Figure 4.5: Static pressure contour plots for the 2D thick spoke model during filling. (Pa) ..44 Figure 4.6: Velocity magnitude contour plots for the 2D modified spoke model during filling, (m/s)  45  Figure 4.7: Static pressure contour plots for the 2D modified spoke model during filling. (Pa) Figure 4.8: Velocity magnitude contour plots for the 3D model during filling, (m/s)  46 47  Figure 4.9: Static pressure contour plots for the 3D model during filling. (Pa) 48 Figure 4.10: Comparison of the free surface height for the 2D and 3D models. The calculated free surface height by Pascal's theorem is also included for sprue filling 50 Figure 4.11: Comparison of the free surface radial distance for the 2D approximation and 3D models 51 Figure 4.12: Comparison of the A356 velocity magnitude (m/s) for the two symmetry planes during filling  51  Figure 4.13: Exit region of 330N sprue with points indicating locations from which velocity and pressure data were acquired. These points were located 1 mm from the exit surface : 52 Figure 4.14: Comparison of velocity variation for the three model cases at 159 mm height  53  Figure 4.15: Comparison of velocity variation for three model cases at 171 mm height  54  Figure 4.16: Comparison of static pressure for three model cases at 159 mm height  54  Figure 4.17: Comparison of velocity at 159 mm location to examine the effect of surface tension 55 Figure 4.18: Comparison of pressure at 159 mm location to examine the effect of surface tension  56  Figure 4.19: Comparison of velocity for different screen conditions at 159 mm height  57  Figure 4.20: Comparison of pressure for different screen conditions at 159 mm height  58  Figure 4.21: Comparison of velocity variation for the three model cases at 159 mm height  60  Figure 4.22: Comparison of static pressure for three model cases at 159 mm height  60  Figure 5.1: Accelerated wear of (a) 329N and (b) 330N wheel model sprues  64  Figure 5.2: Interface between 4140 steel and A356 from 330N sprue  65  Figure 5.3: SEM image of interface between solidified A356 and 4140 steel. Area indicated in (a) is magnified and shown in an EDS map in (b) Figure 5.4: Illustration of subsurface thermocouple locations in 330N sprue Figure 5.5: Measured temperature at three locations in 33ON sprue during industrial casting run Figure 5.6: Geometry of the (a) 516N sprue in the constriction region and (b) 5° profiled pin Figure 5.7: Profiled pin cross-sections: (a) 5°, (b) 10°, and (c) 15°  66 69 69 71 71  Figure 5.8: Illustration of computation domain  74  Figure 5.9: Velocity for 10° pin at different test rotation rates  74  Figure 5.10: Velocity along pin surface for 5° pin at different test rotation rates  75  Figure 5.11: Velocity along pin surface for 15° pin at different test rotation rates  75  Figure 5.12: Velocity for different pin geometries at 200 rpm from leading edge to trailing edge  76  Figure 5.13: Corresponding velocity and contour plots for 10° pin rotated at 200 rpm:. (a) Velocity magnitude (m/s), (b) Static Pressure (Pa)  77  Figure 5.14: Rotating pin test set-up  79  Figure 5.15: Schematic of experimental set-up  79  Figure 6.1: Effect of temperature on radius reduction of cylindrical pins tested for 4 hours Figure 6.2: Radius reduction for cylindrical pins at 700°C, with data from Yan and Fan  86 86  [18]  Figure 6.3: Aluminum-steel interface along the surface of a cylindrical 4140 steel test pin tested at 700°C and 98 rpm for 4 hours. Arrows indicate three intermetallic layers 88 Figure 6.4: Effect of rotation rate on layer thickness for cylindrical pins  88  Figure 6.5: Effect of temperature on layer thickness for cylindrical pins tested for 4 hours...89 Figure 6.6: Effect of time on intermetallic layer thickness. (Data from Yan and Fan for H21 pins rotated at 300 rpm in A380 at 700°C.)  1 181  89  Figure 6.7: Growth of intermetallic layers for cylindrical pins rotated at 98 rpm and 700°C  90  Figure 6.8: Effect of rotation rate on wear around 5° pin  93  Figure 6.9: Effect of rotation rate on wear around 10° pin  94  Figure 6.10: Effect of rotation rate on wear around 15° pin Figure 6.11: Plot of normalized radius for three pin geometries rotated at 150 rpm. R l and R2 locations are indicated with vertical lines. Leading and trailing edges are indicated with L and T  95  Figure 6.12: Plot of normalized radius for three pin geometries rotated at 200 rpm. R l and R2 locations are indicated with vertical lines. Leading and trailing edges are indicated with L and T  96  96  Figure 6.13: 4140 steel pins tested at 200 rpm and 700°C for 4 hours: (a) 5°, (b) 10°, and (c) 15°. Original pin outline is superimposed on worn pin cross-section. (Scale bars correspond to worn pin cross-sections.) 97 Figure 6.14: Static pressure plots with vectors indicating direction of flow: (a) 5° pin, (b) 10° pin, and (c) 15° pin. (Pa) 98  Figure 6.15: Effect of material on wear around 10° pin at 200 rpm  100  Figure 6.16: Plot of normalized radius for three pin materials rotated at 200 rpm. R l and R2 locations are indicated with vertical lines. Leading and trailing edges are indicated with L and T 100 Figure 6.17: Interface of Ti-6A1-4V pin after rotation at 200 rpm and 700°C for 4 hours....101 Figure 6.18: Interface of HI 3 steel pin after rotation at 200 rpm and 700°C for 4 hours  101  Figure 6.19: Interface of 4140 steel pin after rotation at 200 rpm and 700°C for 4 hours. ...102 Figure 6.20: Representative aluminum-steel interface along the surface of the profiled 4140 steel test pins showing three intermetallic layers  103  Figure 6.21: 5° pin superimposed on 330N sprue with angle indicating region where most wear occurs on trailing edge and arrow indicating where vectors become parallel with sprue again 106 Figure 6.22: Close-up of the exit region of a cross-sectioned 330N sprue. Arrows indicate the constriction point and location of accelerated wear  107  Figure 6.23: Velocity magnitude (m/s) from 330N modified spoke model at 14 seconds.... 108 Figure 6.24: Velocity vector plot for 5° pin geometry rotated at 300 rpm. (m/s)  108  Acknowledgements I would like to thank my research supervisor, Daan Maijer, for his support, guidance, and enthusiasm, which made the completion of this thesis possible.  This research was partially supported by Canadian Autoparts Toyota Inc. and I thank them for their support. I'd like to give particular thanks to Chris Hermesmann and Peter Wilander for their technical expertise and patience in answering my questions.  The experimental work would have taken much longer if it had not been for the support of the machine shop technicians. In particular, I would like to express my gratitude to Ross McLeod for always ensuring that my work was performed in a timely manner and Serge Milaire for helping me through my electronic emergencies. Also, Mary Mager for her technical expertise and assistance with analysis techniques.  Lastly, I would like to extend thanks to my family and friends for their invaluable support and understanding, which helped keep me grounded and my outlook positive through the last two years.  xi  CHAPTER 1: INTRODUCTION  Each year in North America, approximately 16 million cars and light trucks are produced, resulting in the production of about 80 million wheels, not including after-market share . [1]  These wheels are predominantly manufactured from steel (60%) and aluminum (40%). Over the last 25 years, there has been a shift away from steel wheels towards aluminum wheels, as shown in Figure 1.1.  This trend has occurred in spite of the higher material cost of  aluminum almost entirely due to consumer preference.  Substitution Analysis Aluminum vs. Steel Wheels 100% f  • Predicted  • Actual IZ  W  60% f 40% -20% -0%  1975  1980  1985  1390  +  1995  2000  2005  Figure 1.1: Substitution of aluminum wheels for steel wheels in passenger car market . [1]  Aluminum wheels are primarily die cast as the process is well suited to high volume production. Die casting is a high volume production process for non-ferrous metals, producing geometrically complex parts with good surface finish and at low scrap rates . [2]  During die casting, dies are exposed to high thermal and mechanical stresses, placing high demands on the die material. Die wear and failure is a significant issue in the die casting industry due to the high cost of dies. Dies are typically made from alloy tool steels, which offer high strength, wear resistance, and thermal durability^ '. The compositions of several 3  die steels commonly used in industry have been provided in Table 1.1.  1  Table 1.1: Referenced Die Steel Compositions (in Weight Percent) Element Cr Mn C Si Mo W V Ni  AISI4140 0.80-1.10 0.75-1.00 0.38-0.43 0.20-0.35 0.15-0.25  141  -  AISIH13 4.75-5.50 0.20-0.50 0.32-0.45 0.80-1.20 1.10-1.75  AISI H21 3.00-3.75 0.15-0.40 0.26-0.36 0.15-0.50  -  8.50-10.00 0.30-0.60 <0.30  151  0.80-1.20 <0.30  [5]  -  The major wear mechanisms leading to die failure are: (1) erosion or washout, (2) heat [2]  checking, and (3) soldering or corrosion. Erosion of the die surface results from the impingement of liquid metal on the die cavity at high velocity, causing the die material to wash away with the melt. Erosive wear reduces the dimensional tolerances of the die and often requires weld repair to rebuild regions of the die that have experienced excessive wear. Heat checking or thermal cracking is due to thermal fatigue caused by alternate heating and cooling of the die surface during die casting. Soldering is caused by the chemical interaction of the casting alloy with the die material. Intermetallic compounds form along the interface and grow into the die surface, and may result in the sticking of the cast part to the die surface.  Canadian Autoparts Toyota, Inc. (CAPTIN) is a high volume wheel production plant in Delta, BC. While die wear is always an ongoing concern, in the last couple years there have been problems with the accelerated wear of the several sprues at CAPTIN. This chapter discusses the die casting process and the conditions specific to CAPTIN, as well as the problem of accelerated sprue wear.  1.1 Die Casting Process  There are a number of different die casting techniques used in industry, including low pressure (LPDC) and high pressure (HPDC). In LPDC, molten metal is pushed up from a pressurized crucible into the die. LPDC is particularly suited to the production of rotationally symmetric parts, such as wheels, and is this focus of this investigation. Another process referenced at several points in this study and commonly used in industry, although not for wheels, is HPDC. In this process, molten metal is injected at high pressure into a die 2  by a hydraulically powered piston. The pressures in HPDC are significantly higher than in LPDC.  At CAPTIN, two die casting techniques are used: (1) low pressure (LP) and (2) Toyota differential pressure (TDP). Figure 1.2 shows the basic setup for the two die casting techniques. In both techniques, liquid aluminum from a holding furnace moves up through a joint pipe into the die cavity. The die consists of a sprue plate connected to a wheel mould with a screen located in between. One of the main differences between the two techniques is that in the LP process, pressure is applied to the liquid metal surface in the holding furnace, pushing the liquid aluminum up the joint pipe. In the TDP process, a vacuum is applied to the die, pulling liquid aluminum into the die cavity. The pressure differences in TDP are greater than in LP and the fill times shorter. The TDP process also employs water-cooling to extract heat from the die faster, resulting in faster cycle times.  The application of pressure is performed in a series of stages, referred to as a pressure curve. Figure 1.3 provides a sample LPDC pressure curve, showing the variation of applied pressure with time. In thefirsttwo pressure stages, liquid aluminum is pushed up through the joint pipe to the bottom of the sprue. As the pressure increases, liquid aluminum is pushed into the sprue and wheel cavity. Once the wheel cavity isfilled,the pressure is held ensuring that the cavity remains full and preventing losses through vent holes. As the wheel begins to solidify, the pressure is further increased to enhance feeding during solidification. The pressure is held until the wheel has completed solidification. At this time, the applied pressure is released and liquid aluminum in the joint pipe and sprue drops back down into the holding furnace. At this point, the wheel is ejected from the die casting machine. One cycle, commonly referred to as a shot, takes approximately six minutes and a casting run with a single die includes approximately 300 shots with 2 wheels cast in a die set with each shot.  3  t t  Joint Pipe  Liquid Aluminum  olding Furnace  Figure 1.2: Illustration of a die casting machine setup at CAPTIN, including holding furnace and die.  90  0  50  100  150  200  250  300  Time (s) Figure 1.3: Typical pressure curve used in low pressure die casting.  4  1.2 Die Casting Alloys At CAPTIN, all wheels are cast from aluminum alloy A356. Table 1.2 provides the chemical compositions of A3 56, as well as that of two other commonly used aluminum die casting alloys. A356 is heat treatable and offers excellent castability due its resistance to shrinkage, segregation, cracking, and hot tears. A356 is used in the production of machine tool parts, aircraft and automobile wheels, pump parts, and valve bodies. Two other aluminum alloys referenced in this study are A380 and A390, typically used in HPDC. A380 offers a good combination of cost, strength, and corrosion resistance, as well as high fluidity and freedom from hot shortness, and is used to produce housings for lawn mowers and radio transmitters, air brake castings, gear cases and air-cooled cylinder heads. A390 has good wear resistance and strength and is used to make internal combustion engine pistons and blocks and cylinder bodies for compressors, pumps and brakes.  Element Si Mg Fe Ti Cu Mn Zn Ni Sn  A356 6.5-7.5 0.25-0.45 <0.2 <0.2 <0.2 <0.1 <0.1  -  A3 80 7.5-9.5 <0.1 <1.3  3.0-4.0 <0.5 <3 <0.5 <0.35  A390 16.0-18.0 0.45-0.65 <0.5 <0.2 4.0-5.0 <0.1 <0.1  -  -  1.3 Sprue Wear The sprue plate is critical component in the casting operation, since all of the liquid aluminum is channelled into the wheel die through it. Liquid aluminum enters through the bottom of the sprue. From the bottom up, the sprue diameter gradually decreases to the point of minimum diameter, referred to as the constriction point. From mass conservation, it can be deduced that velocity is directly proportional to the area, or in this case diameter, so the melt velocity will be a maximum at the constriction point. Past the constriction point, the diameter gradually increases and flow velocity decreases. It is along this exit surface (illustrated in Figure 1.4) that accelerated sprue wear is observed at CAPTIN. Sprue wear 5  was in some cases uniform, with a gradual thinning of the wall surface. However, in other cases, large amounts of material were removed at selective locations of the sprue. Exit Surface, Location of Accelerated Wear  ^-Exit Diameter  (a)  Accelerated Wear  (c)  Figure 1.4: A cross-section of the 330N sprue is shown in (a). This is one sprue at CAPTIN that has consistently experiencing accelerated wear along the exit surface, as shown in the image of the 3 3 ON sprue in (b) and the cross-section of the constriction region in (c).  To protect the surface of the sprue from both the heat and the liquid aluminum, a ceramic coating is applied to the inside surface prior to each casting run. As long as the coating remains uncompromised, the sprue is protected from interaction with the molten aluminum. Several processes may affect the performance of this coating, including the liquid aluminum flow and the ejection of the casting. Once the coating has been compromised, the sprue surface is exposed to the effects of the liquid aluminum.  In recent years at C A P T I N , there have been problems with some sprue models consistently experiencing accelerated wear. When the project began in May 2003, there were ten sprue plates used in production at CAPTIN (6 LP and 4 TDP). Table 1.3 summarizes the wear conditions of the LP and TDP sprues. Four of these sprues consistently had problems with accelerated wear, two in both LP and TDP. The wheel models associated with these sprues were the 329N LP, 330N LP, 420N TDP, and 517N TDP. The other six sprues had  6  negligible or manageable levels of wear. When a comparison was made of the sprue geometries, it was found that accelerated wear was consistently a problem on sprues which had smaller exit diameters and angles, with the exception of the 516N LP model which was not reported to have problems with accelerated wear. This is illustrated in Figure 1.5 with a superposition of two LP sprue geometries, one of which was reported to consistently experience accelerated wear (330N) and another that did not (592N). Table 1.3: Summary of Sprue Wear Conditions and Geometric Features Wheel Model 329N L P 330N LP 477T L P 516NLP 592N LP 800T L P 420N TDP 517NTDP 800T TDP 800T TDP  Wear Condition X X •-0  • 0 •-0 X X 0  •-o  Exit Diameter (mm) 45 40 58 45 54 62  Exit Angle  52 42 61 61  18 10 20 20  n  5 4 10 5 9 15  0 - - negligible wear • - manageable wear X - high wear, sprue plate life is unpredictable  Each sprue is designed to attach to a specific wheel model, with significant variation in sprue diameter, angle and length. In addition to channelling liquid aluminum into the wheel mould, sprues are supposed to promote even melt flow and minimize melt turbulence in the wheel mould. The sprue design procedure at CAPTIN is based on experience, rather than computational fluid dynamics (CFD), so it is not apparent whether the sprue design actually accomplishes this.  Other than geometry, there are few differences between sprues. The sprue material (4140 steel) is the same for all current production models. All wheels at CAPTIN are cast with aluminum alloy A356 with an inlet temperature of 700±10°C. Sprues experience similar thermal and pressure histories. Considering these facts, it seems as if the accelerated surface wear is strongly related to sprue geometry and therefore a flow-related problem.  7  330N  LP  592N L P  Figure  1.5: Comparison of 330N and 592N LP wheel model sprues.  Sprues are generally expected to last 5-8 casting runs with only minimal repair. However, CAPTIN's experience indicates that 4140 steel sprues only last 2-3 casting runs. At the end of each casting run, sprues are inspected for damage. Sprues that experience accelerated surface wear may require weld repair or replacement after only one casting run. This maintenance issue has become a significant cost for CAPTIN, prompting this study.  The purpose of the research presented here is to improve the understanding of the flow conditions in the wear area of the sprue during low pressure die casting and to develop a laboratory experiment to examine erosive-corrosive wear phenomena. The comprehensive literature review on die wear and related processes is discussed in the next section.  8  CHAPTER 2: LITERATURE REVIEW Die wear occurs when liquid aluminum comes into contact with the die surface during casting. Aluminum is a very reactive and corrosive liquid, dissolving and forming intermetallic compounds with other elements. In addition to chemical attack, the motion of the liquid aluminum can erode or wash away surface layers. These effects individually, or in combination, can result in significant die surface damage and is a major problem in the die casting industry. Understanding the mechanism and interaction of corrosion and erosion processes in the die casting application is critical if die wear is to be minimized. This chapter reviews the available literature on die wear and the interaction between liquid aluminum and die surfaces.  2.1 Corrosion Corrosion is defined as the deterioration of a material due to interaction with its environment. The driving force for the corrosion process is the change in the energy of the system, otherwise known as thermodynamics. This change in energy is a result of the reaction between the metal and its environment forming a more stable and lower energy compound (i.e. an oxide). Traditionally, the term corrosion refers to the chemical reaction that occurs in metals exposed to aqueous solutions, although it has also been used to describe material deterioration occurring at high temperature.  Although not strictly adhering to the above definition, the term corrosion is frequently used in the die casting literature to describe the chemical attack that results from the interaction of liquid aluminum with die materials. The chemical reaction of liquid metals with other materials is similar to corrosion in aqueous solutions, although the interaction that occurs is not electrochemical in nature . Liquid metal corrosion is fundamentally different from 17,8]  aqueous corrosion in that the liquid metal medium, excluding impurities, is in a non-ionized state l Degradation instead results from (1) dissolution and (2) intermetallic compound [9  formation. In the first, iron and other elements from the die material dissolve into the liquid aluminum. In the second, aluminum and other elements from the melt diffuse into the die  9  surface, forming intermetallic compounds . These two processes can occur simultaneously 1  J  and are further discussed in the following sections.  2.1.1 Dissolution  Dissolution is a corrosion reaction that can occur in liquid metal environments. The dissolution reaction is controlled by the elemental solubilities in the liquid metal and the kinetics of the dissolution reaction . The dissolution mechanism is governed by either the [11]  solid/liquid interface reaction or by the transfer of dissolved atoms further into the melt . [12]  Significant surface dissolution occurs upon initial contact of the liquid aluminum and the die surface . After this initial reaction, the dissolution is diffusion-controlled with the [13]  [14]  diffusion rate controlled by a critical element in the die material. Norstrom et a/.  [12]  speculated that this element would be the one with the highest content in the die material in relation to its maximum solubility in the melt, such as iron in the case of a steel die. The diffusion rate is primarily controlled by the difference in the maximum solubility and the nominal content of this critical element in the melt, as well as by its diffusion coefficient in the melt. In fact, aluminum die casters often use casting alloys with iron contents between 0.8 to 1.1 wt% to reduce the chemical potential gradient for dissolution ' . 17 15]  Several factors affecting the dissolution rate are melt velocity, temperature, and alloying elements in the die material. Dissolution has been shown to increase with melt agitation . [14]  The dissolution of different materials in liquids is often investigated with the rotating disk technique l In this technique, cylindrical specimens are pressed into graphite tubes. The [16  specimen and tube are inserted into a liquid melt and rotated, with the tube protecting the lateral surface from liquid metal attack so that only the disk surface is exposed. The dissolution of a solid metal into a liquid metal can be described by the following relation : [14]  In  c-c  = k^  (2.1)  where C is the concentration of the dissolved substance in the bulk liquid at time t, C is the s  saturation concentration at a given temperature, Co is the initial concentration of the  10  dissolving substance, k is the dissolution rate constant, S is the surface area of the solid in contact with the liquid, and Vis the melt volume. The dissolution rate constant, k, can be determined from the Kassner equation : [16]  k =0.554r £> V 'V 1  2 /  1  (2.2)  / 2  where o is the angular rotation speed of the solid metal, v is the kinematic viscosity, D is the diffusion coefficient of the solute across the interface, and / is a correction factor. The Kassner equation is valid for Schmidt numbers less than 1000. The Schmidt number is a dimensionless parameter equal to the kinematic viscosity divided by the diffusion coefficient and is less than 1000 for most liquid metals . With increasing angular rotation speed, the [16]  dissolution rate constant increases, thereby increasing the amount of dissolution. Elements dissolved in the melt at the interface would be carried into the bulk liquid by fluid flow, allowing for more dissolution. As dissolution proceeds in a contained fluid volume, the concentration of dissolved elements in the melt may increase, resulting in a decreasing dissolution rate.  The solubilities of several elements in liquid aluminum are presented in Table 2.1. Both iron and nickel are very soluble in liquid aluminum. Iron, the most important element in die steels, has a solubility limit of 1.87 wt% in liquid aluminum at 655°C . On the other hand, [17]  titanium has a very low solubility in liquid aluminum. If any dissolution occurs, the low solubility of titanium keeps the liquid layer along the surface more readily saturated with titanium solute, inhibiting further dissolution. Table 2.1: Solubility of Elements in Pure Liquid Aluminum Temperature (°C) 700 750 800 850  Ni (wt%) 9.0 13.2 17.3 26.2  Fe (wt %) 3.2 4.9 6.8 7.9  Cr (wt %) 0.7 1.3 2.4 4.2  [14]  Ti (wt %) 0.2 0.3 0.5 0.7  Solubility and consequently dissolution increases with temperature as shown for various elements in pure liquid aluminum in Table 2.1. Considering that in the die casting process,  11  liquid aluminum is injected into the sprue at ~700°C, up to 3.2 wt% iron could dissolve into the aluminum melt.  2.1.2 Intermetallic Compound Formation and Growth  Two processes govern the formation of intermetallic compounds, the diffusion of atoms from the die surface through the intermediate layers (diffusion-controlled) and the continuous reaction of the top layer surface with aluminum to form more intermetallic compounds (interface-controlled) ' l At the same time, the layers are gradually dissolved into the [12  1 5  liquid aluminum.  2.1.2.1  Phases in the Fe-Al and Fe-Al-Si Systems  Of particular interest in this project is the reaction of liquid aluminum with steel. The formation and growth of iron-aluminum intermetallic compounds has been studied extensively in the available literature. When a steel die is in contact with liquid aluminum, a series of iron-aluminum intermetallic layers will form along the die surface and gradually grow into the steel matrix  [14,18]  . Aluminum diffuses into the steel surface forming binary  iron-aluminum intermetallic phases. Successive layers of binary compounds form because of the reaction of each phase with continuously renewed aluminum melt and the diffusion of iron from the steel surface. Ternary iron-aluminum-silicon intermetallic phases form on top of the binary phases. The possible reactions that can occur between iron and liquid aluminum are provided in Table 2.2 and Table 2.3. The fact that two completely different reaction paths have been defined for the same system indicates the uncertainty which remains about the identity of stable phases in the iron-aluminum system and their reactions. The reactions in Table 2.2 seem to have been determined from the phase diagram or equilibrium conditions, while the reactions in Table 2.3 are thought to be experimentally determined. The compositions of confirmed binary and ternary intermetallic phases in the iron-aluminum and iron-aluminum-silicon systems are provided in Table 2.4 and Table 2.5.  12  Table 2.2: Possible Reactions Between Fe-Al Phases and Liquid A l u m i n u m  tl9]  Reaction Fe(s) + Al(l) o cc(Fe)(s) a(Fe)(s) + Al(l) o  pVFeAl(s)  pYFeAl(s) + Al(l)  ^-FeAl (s)  £ - F e A l ( s ) + Al(l) o  ri-Fe Al (s)  2  2  2  n-Fe Al (s) + Al(l) o 2  5  9-FeAl (s)  5  3  9-FeAl (s) + Al-Si(l) o• a(Fe,Al,Si)(T ) + Al(l) 3  5  ,[20]  Invariant Reaction  Temperature (°C)  652 620 615 577 577  L »(Al) + e  L + e o (Ai) + T  5  L + T <=> (Al) + T 5  L o L o  6  (Al) + (Si)  (Al) + (Si) + x  6  Table 2.4: Elemental Composition of Binary Fe-Al Phases , Phase n  . ,. . Stoichiometry  D 4  9  FeAl  il  \  Fe Al FeAl FeAl  p.  Fe Al  3  2  5  2  w  3  Composition, wt % — —— Fe Al c  59.2 54.7 49.1 32.6 13.9  40.8 45.3 50.9 67.4 86.1  Table 2.5: Elemental Composition of Possible Ternary Fe-Al-Si Phases ' 12  Phase  Stoichiometry FeSiAl  t  5  4  Fe Si Ali5 6  5  Composition, wt % Fe  Al  29.1 38.1  56.3 46.0  Si 14.6 16.0  28.9  41.9  29.1  9  42.1  36.6  21.2  *2  Fe Si Ali  41.9  40.5  17.6  Tl  Fe Si Al  55.0  26.6  18.4  T T  4  3  FeSi Al 2  3  Fe Si Al 5  6  5  5  3  2  2  3  1  2.1.2.2 Reaction Thermodynamics and Kinetics  Thermodynamics can be used to predict which phases will form, as well as the driving force necessary for them to form. The thermodynamic values associated with the formation of binary iron-aluminum phases are provided in Table 2.6, including enthalpy, entropy, and free energy of formation values. The more negative the change in free energy, AG, of a phase, 13  the greater its driving force for formation. From Table 2.6, it would seem that the FeAl3 and Fe2Al5 phases would be most likely to form.  Table 2.6: Thermodynamic Quantities for the Formation of Intermetallic Phases in the Iron-  2  (J mol )  (K" mol )  (J mol )  -112560 -194040 -81900 -51240 -57372  95.6 166.7 73.3 51.0 28.0  -22869 -19636 -16999 -11090 -4827  1  1  9-FeAl r|-Fe2Al5 ^-FeAl 3  2  pYFeAl 3,-Fe Al 3  AG973K  AS298K  AH 98K  Phase  1  1  In addition to thermodynamics, kinetics must also be considered in the formation and growth of intermetallic phases. While FeAl3 has the lowest free energy, there have been cases reported where Fe2Ai5 has been the major or only phase to be observed l Table 2.6 [22  presents the diffusion data for iron and aluminum within the various iron-aluminum binary phases.  ft" •  4  Fe  At  Al  m  Fe  At  *>  u  it  (b)  (a)  (c)  Figure 2.1: Intermetallic growth within iron and aluminum, where A B is the initial interface and the arrow length is a measure of the diffusion rate. 1231  To better understand the kinetics of the interfacial reaction, Shahverdi et a/.  [23]  performed a  series of solid iron-liquid aluminum diffusion couple experiments. In these tests, rectangular iron coupons were submerged in liquid aluminum at 700, 800, and 900°C for times ranging from 90 to 3000 seconds. Shahverdi et a/.  [23]  identified the interface layer to be comprised of  two layers, with Fe2Al as the major phase and FeAb as the minor phase, which can perhaps 5  be attributed to the higher activation energy for aluminum diffusion in FeAl3. The growth of the Fe2Al5 layer between iron and aluminum is illustrated in Figure 2.1. The initial growth of the layer into the aluminum in Figure 2.1(b) may be attributed to the ease of diffusion of  14  iron atoms into the liquid aluminum. As the layer is formed, further growth depends on the diffusion of iron and/or aluminum atoms within the phase. The activation energies for diffusion of iron and aluminum through Fe2Als are 131 and 107 kJmol" , respectively, 1  indicating that aluminum would be expected to diffuse more easily and that the phase would grow towards the iron-rich section, as shown in Figure 2.1(c). Table 2.1: Diffusion Data for Phases in the Iron-Aluminum System A l Diffusion (520-630°C) Fe Diffusion (600-1100°C) Phase E (kJmol ) In D In D E (kJmol ) 199.4 101.4 4.018 -12.59 FeAl 131.4 -8.51 107.0 -6.56 Fe Al 145.7 -7.79 FeAl -7.53 111.8 -11.2 93.2 -4.04 166.1 FeAl 3.4 198.2 134.4 Fe Al -6.56 1  1  1  0  0  A  A  3  2  5  2  3  2.1.2.3 Corrosion Experiments  Researchers agree that intermetallic layer formation and growth is primarily governed by the diffusion of atoms through the die material and through any layers that may form. Shankar and Apelian  [15]  showed this through multi-component diffusion couple experiments. In these  experiments, an HI3 tool steel disk was placed into an aluminum alloy A380 melt at 625°C and left submerged for 48, 120 and 168 hours. The disks were sectioned and the thickness of the intermetallic layers measured. Layer thickness increased according to parabolic growth law, indicating that the process was mainly diffusion-controlled.  Diffusion-controlled processes are temperature dependent with high temperatures favouring intermetallic growth. Depending on the temperature, time, die material, and melt alloy, different intermetallic phases will form. Yu et a/.  [13]  performed accelerated corrosion tests to  evaluate the dissolution rate and soldering resistance of HI 3 steel pins in static aluminum A390 melts. In these tests, pins were immersed in liquid aluminum for various exposure times. The tests were referred to as accelerated because the melt temperatures (680 to 750°C) were much higher than those in commercial HPDC processes during the die holding period (500 to 600°C) and the exposure time was as long as 6 hours as opposed to the actual die contact time, which is less than 1 second per cycle. For shorter immersion time tests  15  (0.25 hours), only the t6 phase formed (refer to Table 2.5). For tests longer than 0.75 hours, T5 formed along the T6-steel boundary, and for tests longer than 5 hours, %i formed along the t5-steel boundary. Simultaneously, there was also dissolution of the outer T6 layer into the liquid aluminum. In the static melt tests, Yu et al. attributed the recession of the pin radius to dissolution of the \(, layer and the continuous formation of new its, T5, and T2 intermetallic phases (in order from aluminum to steel).  * •  Corrosion Tim* (hours)  Figure 2.2: Effect of melt temperature on the HI 3 pin dissolution in a static A390 melt (solid line is radius reduction, dotted line is intermetallic thickness). [13]  Yu et a/.  [13]  examined the effect of melt temperature and corrosion time on dissolution and  found that die surface losses could not be accounted for by the removal of intermetallic layers alone. Yu et al. observed that pin radius reduction in a static melt was greater than the intermetallic layer thickness and that remaining die surface losses were probably due to dissolution. Both pin radius reduction (solid line) and intermetallic layer thickness (dotted line) were plotted for 680°C in Figure 2.2. It is clear that the radius reduction was greater than the intermetallic layer thickness, especially for later times.  2.2 Erosion  The basic condition for erosion is liquid motion. The movement of liquid metal can lead to mechanical wear of exposed die surfaces, sometimes referred to as surface washout. Only a  16  small amount o f information was found in literature on the mechanisms o f erosion i n aluminum die casting due to the experimental difficulties involved. However, erosion damage by liquid aluminum shows damage similar to that caused by other well-known liquid systems. Hence, it is reasonable to assume that the knowledge o f erosion mechanisms in these other systems may be applied to aluminum die casting  [12]  .  The erosion o f a solid surface can occur i n either corrosive or non-corrosive liquids, with or without the presence o f abrasive particles. T w o main erosion mechanisms described in literature are: (1) liquid-impingement and (2) cavitation. Liquid-impingement is the mechanical action o f an impinging liquid on a solid surface, resulting i n progressive material loss. Damage may occur from the resulting pressure and velocity or from surface shearing effects . [2]  Cavitation results from the formation and collapse o f bubbles in a liquid against a metal surface, usually occurring on surfaces i n contact with high velocity liquids subject to changes in pressure. If the liquid static pressure drops below the liquid vapour pressure, bubbles can nucleate and grow to a stable size. These bubbles are transported downstream with the flow. In regions o f higher pressure, the bubbles become unstable and collapse, with damage resulting i f the collapse occurs near a solid surface. The pressures produced from the collapse can lead to localized surface deformation and material removal. There has been speculation on the possibility o f cavitation erosion occurring i n aluminum H P D C because o f the high velocities and pressures, however, to date there has been no conclusive evidence indicating that cavitation erosion occurs.  Erosion can result in material loss through the removal o f surface layers (i.e. protective oxide films, coatings, intermetallic compound layers). These layers may separate from the surface due to the high drag forces o f the incoming metal during casting, and can expose the die material to chemical attack.  17  Erosion is controlled mainly by melt velocity, however it can also be said that melt temperature also has an effect. Higher melt temperatures lower the die surface hardness, making die surfaces less wear resistant and therefore more susceptible to erosion.  Venkatesan and Shivpuri  [24]  examined the effect of metal velocity, angle of metal impact, and  melt temperature on the erosive wear of H13 steel pins in a test die shown in Figure 2.3(a). The test die was designed to reflect real production conditions. Liquid aluminum A390 (high silicon content) was injected from the bottom through runners into the cavity. In addition to examining the surface wear of pins at different locations, the effect of attack angle (angle at which the liquid aluminum impinged against the pin surface) was investigated. Figure 2.3(b) shows the design of a pin with an attack angle, a. Venkatesan and Shivpuri's  [24]  experiments revealed that maximum wear occurred at an attack angle of 72°. However, Venkatesan and Shivpuri  [24]  found evidence that the attack angle which produced maximum  wear was a function of the phase impacting the pin. For solid impact, maximum wear occurred at an attack angle of 45°, whereas for liquid impact, the attack angle of maximum wear was 90°. This led Venkatesan and Shivpuri to the conclusion that the metal at the location of maximum wear was partially solidified, as the attack angle was between 45° and 90°, and therefore erosive wear must increase with decreasing temperature. Their conclusion was that the impact of partially solid metal or solid particles, referred to as primary silicon , [25]  at high velocities is the primary driver of erosion, while diffusion effects were not as critical. However, in their review of die erosion in aluminum HPDC, Chen and Jahedi pointed out [26]  two important concerns regarding Venkatesan and Shivpuri's HPDC erosion experiments. Firstly, after testing, their samples were cleaned with an NaOH solution to remove soldering layers, making it uncertain whether the measured material loss was a result of mechanical impact or the soldering reaction. Secondly, Chen and Jahedi considered the hardness of HI 3 steel and silicon particles at temperatures relevant to die casting and found that solidified silicon particles will be softer than HI3 steel during die filling, hence, it is not reasonable to assume that silicon particles can cause erosion damage in the short term.  18  Die Cavity Test pit  .y  Ejector pin  I /  ILL. . Gate  Fixed Die angleV/V/,  Molten aluminum injection direction dii  7777  Tested pin<^ Molten aluminum injection direction  (a)  (b)  Figure 2.3: Schematic of the (a) multiple pin test die and (b) pin design.  [24]  2.3 Erosion-Corrosion Erosion-corrosion is an accelerated form of corrosion caused by the relative motion between a corrosive liquid and a metal surface. In erosion-corrosion, the drag forces imposed by the impinging liquid on the intermetallic layers can cause layer removal, exposing new die material to the liquid aluminum and forming new intermetallic layers along the die surface. This process repeats itself and can result in significant material loss. The erosion-corrosion process has little to do with the erosion of the die material itself, but of the removal of layers along the die surface. At the same time, there is dissolution of the die surface and/or intermetallic layers. Dissolution rates increase with melt agitation because elements from the die surface, dissolved in the melt, are carried into the bulk fluid by melt motion, preventing the liquid layer at the interface from becoming saturated with solute and allowing more dissolution to occur. The mechanisms of erosion-corrosion are complex and not very well understood. The following subsections outline the research and proposed mechanisms for erosive-corrosive wear that are currently available in literature.  19  2.3.1 Erosive-Corrosive Wear  Most erosive-corrosive wear experiments have been performed with a rotating pin set-up, similar to the accelerated corrosion or immersion tests described previously, but including pin rotation. This section discusses the performed experiments and their results.  Norstrom et a/.  [12]  conducted experiments with H13 steel pins immersed (static) and rotated  (stirred) in liquid aluminum alloy Al-10Si-0.8Fe at 735°C for various exposure times. Pin diameter was not reported. After the tests, the pins were sectioned and the radius reduction and intermetallic layer thickness were measured (shown in Figure 2.4). The velocity used in the stirred melt tests was very low, only 0.01 m/s, and consequently the material loss and intermetallic layer thickness measured in both the stagnant and stirred melt tests was very small. In the stagnant melt tests, Norstrom et al. found the material loss and intermetallic layer thickness to increase parabolically, suggesting that the corrosion mechanism was diffusion-controlled. As the diffusion distance (layer thickness) gradually increased, the corrosion rate decreased . In the stirred melt tests, the diffusion distance decreased or [12]  sometimes reached a constant value, leading to faster corrosion. Thickness of intermetallic phases (mm) —  0,02-  material (mm) 0,8  0 0  2 3 Exposure time (h)  4  Figure 2.4: Corrosion of AISI H13 steel during exposure in Al-10Si-0.8Fe melt at 735°C.'  20  With a similar rotating pin test set-up, Yan and Fan  [18]  investigated the chemical interaction  of H21 steel with molten aluminum alloy A380. Cylindrical test pins were rotated at 300 rpm (0.16 m/s) in liquid aluminum at 700°C for different test times. After the tests, the pins were sectioned and the aluminum-steel interfaces were examined. Tests of uncoated H21 steel pins were performed for 1,4, 9, and 16 hours. Figure 2.5 shows a micrograph of the steel surface after immersion for 5 hours. For all test times, Yan and Fan found layers of FeAb and FeaAls along the aluminum-steel interface. For the 9 and 16 hour tests, FeAl also 2  formed adjacent to the steel. The intermetallic layer thickness and pin radius reduction are provided in Table 2.8.  Figure 2.5: Backscattered electron image of the morphology of the interface between H21 steel and A380 after 9 hours at 7 0 0 ° C . [ I 8 ]  Table 2.8: Intermetallic Layer Thickness and Radius Reduction Test time FeAl layer F e A l layer F e A l layer Radius Reduction (hours) (um) (um) (um) (mm) 1 118-123 8-10 0 0.27 4 135-148 8-10 0 0.67 9 194-412 8-10 20-49 0.99 16 458-960 7-10 80-160 1.17 1181  3  2  5  2  The pin radius reduction increased parabolically, similar to the results of Norstrom et al. \ [l2  The Fe2Al5 layer was only 7-10 um thick for all exposure times. The FeAl3 layer was thick and substantially increased with time. Yan and Fan also observed that the FeAb layer was very porous, speculating that this might provide a fast diffusion path for the aluminum to the FeAl3-Fe2Al5 interface and accelerate FeAl3 growth. The thickness of the FeAl layer also 2  increased with test time. Microhardness measurements, shown in Figure 2.6, were performed across the aluminum-steel interface. Both the FeAb and FeAb layers were found  21  to have hardness values greater than that of the steel. The FeaAls layer was too thin for its hardness to be measured.  1200  t H21  innerpart  FciAlj  FeAlj T-ft-  1000  FeAlj  outer part -• '—h-  atuminum  X  w  S c  800  600 •s. in  p w u  s  o - •«  400 200  •o-  -c  0  -100  0  100  200  300  400  500  600  700  DiatancBfrom H21 substrate surface (pm)  Figure 2.6: Microhardness across reaction zone between H21 steel and A380 alloy after 9 hours . [18]  Sundqvist and Hogmark ^ also performed rotating pin tests to investigate the formation of [27  iron-aluminum compounds along the pin surface at different temperatures and exposure times. The pins used in the tests were 10 mm diameter and made from steel alloy QRO-90S, a proprietary version of H13 steel. The melt was aluminum alloy A380. For a pin rotated at 1000 rpm (0.52 m/s) for 4 hours at 715°C, three layers were observed along the interface. The compositions obtained from electron dispersive spectroscopy (EDS) and hardness values are shown in Table 2.9. Similar to Yan and Fan , Sundqvist and Hogmark estimated the [18]  layers to be 9-FeAl3, r|-Fe2Al5, and ^-FeAb, in sequence from the aluminum to the steel.  Table 2.9: Element Distribution and Hardness of Phases for QRO-90S Pins Rotated in A380 at 1000 rpm for 4 hours at 7 1 5 ° C [27]  Phase Aluminum Outer layer Mid layer Inner layer Steel  Al 89 51-56 48-50 39-42 0  Composition, wt% Fe Si 8 6-7 34-36 8-9 41-43 2-5 53-55 1.0 92.8  Cr 1.5-3.0 0 1-2 2.6  HV .5N 0  100 1096 985 239  22  2.3.2 Mechanisms for Erosive-Corrosive Wear  The mechanisms of erosive-corrosive wear are more complex than the individual mechanisms of erosion and corrosion and are much less understood. This section reviews the two prevailing theories available in literature on the erosive-corrosive wear mechanism  2.3.2.1 Phase Boundary Attack  Shankar and Apelian  [15]  performed diffusion couple experiments with molten aluminum and  steel disks (outlined in Section 2.1.2.3) to develop a mechanism for die soldering. Shankar and Apelian postulated that the die soldering mechanism consisted of the following stages, illustrated in Figure 2.7: a) Erosion of phase boundaries on die surface and pitting of die surface b) Formation of iron-aluminum compounds c) Formation of pyramid-shaped intermetallic structures d) Adherence of aluminum to pyramid structures e) Straightening of erosion pits and intermetallic phases  Figure 2.7: A schematic of the mechanism of die soldering, (a) Initial attack of grain boundaries by aluminum, (b) Formation of iron-aluminum phases, (c) Growth of ternary oc-(Al,Fe,Si) phases, (d) Growth of intermetallic layers and merging of neighboring pits, (e) Straightening of pits. [19]  23  Shankar and Apelian  [15,19]  theorized that molten aluminum attacks the weaker intergranular  regions in the steel microstructure where hard carbides are absent. Steel grains are loosened and eventually separate from the surface with the flow of incoming metal and erosion pits form as shown in Figure 2.7(a). Iron diffuses out from the erosion pits and the loosened grains resulting in the formation of binary intermetallic compounds between the iron and aluminum. The intermetallic phases grow radially out of the erosion pits, forming pyramidshaped intermetallic compounds. These intermetallic phases consolidate and prevent further aluminum-steel contact, resulting in sticking or soldering of the aluminum to the steel.  2.3.2.2 Dissolution and Dissociation ofIntermetallic Layers  Contrary to Shankar and Apelian, Yu et a/.  [13]  believed that the destruction of the die surfaces  was due to the gradual dissolution and dissociation of intermetallic layers due to melt agitation. In addition to static melt tests described in Section 2.1.2, Yu et al. rotated H13 pins in aluminum A390 melt. Both radius reduction and intermetallic layer thickness were measured (shown in Figure 2.8). Radius reduction increased and intermetallic layer thickness decreased with increasing rotation rate. Yu et al. speculated that melt velocity promoted the rapid dissolution of intermetallic layers into the turbulent melt, as only the and T phases were observed. The intermetallic phase T2 did not form because the outer x 5  6  layer was continuously removed by the melt motion.  Several researchers observed what are referred to as pyramid or cone-shaped structures along the outer boundaries of the interface following erosive-corrosive attack ' ' ' [14 15  Fan ' [14  18]  I8  27]  . Yan and  observed these structures during stagnant melt or low rotation rate tests and thought  their shape was due to the dendritic growth of the iron aluminates. Sundqvist and Hogmark showed that these structures were made up of layers of intermetallic compounds, [27]  of the same composition as the continuous layers, and speculated that the formation of cones is related to surface condition. Shankar and Apelian  [15]  observed these structures growing out  of pits on the die surface, expanding along the steel surface and merging to form one  24  continuous layer. It was generally agreed that these structures increased the area available for diffusion and could break off at their bases.  Figure 2.8: Effect of rotation rate on the (a) radius reduction and (b) intermetallic layer thickness of HI3 steel pins in agitated A390 melt at 680°C. I13]  2.4 Work Performed at CAPTIN  In an effort to assess sprue wear, CAPTIN performed erosion-corrosion tests to examine p8]  the effect of sprue material and surface treatment. Similar to the experimental set-up described by Y u  [13]  , pins were placed in a sample holder, and lowered into an aluminum melt  and rotated for various test times. The aluminum A356 melt was held in a graphite crucible heated by a gas burner. A pneumatic air motor was used to rotate the samples.  Cylindrical pins, 20 mm in diameter and 85 mm in length, were fabricated from 4140 steel, HI 3 steel, stainless steel, and medium to high silicon content grey cast iron. Some of the 4140 steel pins were plasma-sprayed with coatings while others were nitrided. Ten accelerated corrosion tests were performed testing twelve samples at 60 rpm (0.03 m/s) and 700°C for 4 hours. In each test, a standard 4140 steel pin was simultaneously tested.  Inspection of the pins after testing revealed the presence of an intermetallic layer between the aluminum and the pin surface. In the case of the steel pins, the layer was made up of iron-  25  aluminum intermetallics and had a jagged appearance. Scanning electron microscopy (SEM) and EDS examination revealed that the diffusion layer was made up of two intermetallic compounds, one aluminum-rich and the other iron-rich. The aluminum-rich layer was about 20 um thick and the iron-rich layer was about 30 u,m thick.  Also observed on the pin surface was the development of a series of voids beneath the intermetallic layers. It was speculated that perhaps the iron diffused out of this region without enough aluminum present to form intermetallic compounds. Once the voids had coalesced into cracks, the pin surface tore off.  To compare wear performance of pin materials and surface treatments, pin mass loss measurements were performed. Of the various materials tested, the stainless steel pin performed best. Liquid nitriding of the 4140 steel pin was determined to be the best surface treatment. Recommendations were made for improvement to experimental setup including an alternate heat source to minimize melting time and a more consistent and controllable rotational system.  2.5 Summary  The problem of die wear is complicated, combining the effects of both corrosion and erosion. Die surfaces are coated to prevent contact with the liquid aluminum. However, if this coating becomes compromised, corrosion reactions will likely result. It has been demonstrated in literature that in addition to dissolution of iron and alloying elements from the die surface, a series of intermetallic compounds will likely form along the surface. The drag forces produced by the moving melt accelerate dissolution and remove intermetallic layers, in effect increasing die surface wear. While it has been agreed that die wear is an erosive-corrosive wear problem, however, the mechanism has not been conclusively decided. Some researchers suggest that the liquid aluminum attacks along the steel grain boundaries, while others think the wear is a result of continual layer removal.  26  Researchers have evaluated the erosion-corrosion behaviour of die materials in liquid aluminum in several ways. The most common test methods were the static dip test, dynamic or rotating pin test, and multiple pin test die. In immersion or static dip tests ' ' [12  13  15,23,29]  ,  pins are inserted into liquid aluminum and held for specific time durations. Researchers have used this test to investigate the effect of pin material, melt alloy, melt temperature, and submersion time on corrosion behaviour. The rotating pin test ' ' ' 113  18  21  30]  goes a step beyond  the static dip test, incorporating the effect of melt motion. The experimental set-up is nearly the same, although some means of rotating the pin must be included. The two tests can both be evaluated with weight loss, radius reduction, and intermetallic layer thickness measurements. Both tests offer good control and the ability to examine a wide range of factors. Additionally, in both the static and rotating pin tests, multiple pins can be tested at one time if a pin holder is used.  Lastly, there is the multiple pin test die ' [24  26,3I]  . In this test, test pins are inserted into a die  casting test die. The pins can be of different geometry, location, orientation, and material. Casting conditions, such as temperature, velocity, pressure, and time, can be varied. Weight loss and intermetallic layer thickness can be measured, as well as wear profile. This test die would provide the best reflection of real production conditions and is the only test technique reviewed that included the effect of pin geometry. However, with the more realistic conditions some of the controllability is lost.  In summary, erosive-corrosive wear is a significant problem in the aluminum die casting industry. The problems of erosion and corrosion are accelerated by the combination of their effects. Different test techniques have been used in literature to evaluate erosion and corrosion behaviour, each with their own advantages and disadvantages.  27  CHAPTER 3: SCOPE AND OBJECTIVES The goal of this research programme was to develop a test to assess the effect of flow on erosive-corrosive wear. The rotating pin test was selected for this study as it has the benefits of control, potential variability, and repeatability. A varied cross-section pin was developed, modifying the pattern of flow along the surface and producing regions of different velocity and pressure. It was anticipated that this would produce areas of increased wear.  To ensure that the test velocities were in the range of those in LPDC, a fluid flow model was developed in the commercial CFD program, FLUENT. This model was used to predict the velocity and pressure in the sprue region where accelerated wear is a concern. Another fluid flow model was developed to simulate the flow of liquid aluminum and provide information of the velocity and pressure along the rotating pin surface.  3.1 Objectives The primary objective of the present study is: •  To investigate how the varied pin cross-section changes the flow along the pin surface and how the resulting flow influences erosive-corrosive wear behaviour.  In accomplishing the primary objective, the following sub-objectives were formulated: •  To develop a fluid flow model of the filling process during low pressure die casting which predicts the flow conditions in regions of accelerated wear in the sprue.  •  To develop a rotating pin test set-up similar to those reviewed in literature.  •  To design a test pin using features based on sprue geometry in the region of accelerated wear.  •  To use this test set-up to evaluate the wear behaviour of die materials in liquid aluminum alloy A356 for different temperatures, times, rotation rates, and especially, pin geometry.  •  To relate the observations of wear behaviour back to the accelerated sprue wear problem. 28  CHAPTER 4: MATHEMATICAL MODEL The premise of this project was that accelerated sprue wear was a geometry-related problem, making it necessary to understand the flow conditions during die filling. It is not feasible to physically measure the flow during casting due to the high temperatures, the reactivity of liquid aluminum, and the opaque enclosed die cavity. Instead of attempting to measure pressure and velocity at discrete locations, a model was developed to simulate the filling of the die cavity and provide an indication of these critical parameters throughout the die cavity. This simulation was developed for the 3 3 ON sprue model using the commercial CFD program, FLUENT™. The 33ON wheel model was chosen as it was one of the sprues experiencing accelerated wear. This section details the formulation of the model, including mesh generation, material properties, operating and boundary conditions (BC), and solution techniques, and discusses the results of the model.  4.1 Free Surface Model Formulation Fluid flow can be described using a set of partial differential equations called the NavierStokes equations. The Navier-Stokes equations satisfy the conservation of mass, momentum, and energy. The effects of heat transfer are not being considered at this time, so the energy equation will not be discussed. The continuity and momentum equations can be described by the following equations: f^ + pV.v = 0 dt  (4.1)  —(pv) + V»(pvv) = -Vp + /N v + pg 2  (4.2)  where v is velocity (m/s),p is static pressure (Pa), ji is molecular viscosity (Pa-s), p is density (kg/m ), g is gravitational acceleration (m/s ), and t is time (s). These fluid flow 3  2  equations are highly non-linear, making computations difficult. However, the result of solving these equations is a description of component velocities and pressure throughout the domain of interest.  29  Filling models track the evolution of the free surface. A free surface is defined as the interaction of two or more distinctly different fluids separated by sharply defined interfaces, such as the boundary between liquid aluminum and air in a die cavity. A mathematical model is needed to calculate the position of the free surface, while satisfying the flow equations and boundary conditions. The volume of fluid (VOF) method is commonly used to track the location of transient free surfaces of two fluids. In the VOF formulation, the two fluids are assumed to be immiscible or not mixing. The flow in the entire computation domain (containing both fluids) is treated as a single phase with time and space dependent properties with no boundary conditions being applied to thefreesurface. The fields for all variables and properties are shared by the two phases and represent volume-averaged values. The variables and properties in any given cell are either representative of one phase or of a mixture of the two phases depending on the volume fraction.  A scalar, pseudo-concentration function is used to locate thefronton a fixed mesh and the continuously moving front is tracked by solving a hyperbolic transport equation. This pseudo-concentration or fluidfractionfunction, a , is defined to be 0 in the primary fluid and q  1 in the secondary fluid. The function is averaged over the computational control volume and has a fractional value in elements containing a free surface. A geometric reconstruction scheme is used to preserve the sharpness of the interface and to minimize the amount of free surface distortion. This scheme assumes that the interface between the two fluids is linearly sloped in each cell.  The tracking of the interface between phases is accomplished by solving the continuity equation for the volumefractionof one phase in each computational cell. For the secondary phase, the continuity equation is given by the following equation: ^  ^  +vA7«_^=0  (4.3)  In FLUENT, the volumefractionequation is only solved for the secondary phase. The primary phase volume fraction is calculated with the following constraint:  30  cx —\—<x primary secondary  (4 4) V ' /  The viscosity, ft, and density, p, quantities in the transport equations are determined from an average of the volume fraction of the two fluid phases. For example, in each computational cell the density would be determined from the following equation: P  ^secondaryPsecondary 0  ^secondary) Pprimary  (4.5)  A single momentum equation is solved throughout the domain with the resulting velocity field shared between the two fluids. The momentum equation is dependent on the volume fractions of the two phases through the variables ju and p in Equation (4.2).  To solve the resulting flow equations, FLUENT uses a segregated solver with implicit linearity. The body-force-weighted scheme was used for pressure interpolation and the second-order upwind scheme was used for discretization of the momentum equation. Pressure-velocity coupling was performed with the Pressure-Implicit with Splitting of Operators (PISO) algorithm. These solver options are variables in FLUENT and will not be discussed further. More information can be found in the FLUENT documentation l [32  Convergence tolerances were set to 0.001 for the x-velocity, y-velocity, and z-velocity (3D) residuals. The convergence criterion for the continuity residual was 0.001 and 0.005 for the 2D and 3D models, respectively.  4.2 Implementation of Filling Model Employing the general VOF model formulation for tracking the location of free surface discussed in the last section, the implementation of a fluid flow model to simulate the filling of the sprue and wheel cavity with liquid aluminum during LPDC can now be discussed.  The intent of the model was to simulate melt flow during wheel cavity filling to determine the velocity and pressure conditions in the sprue that may influence accelerated wear. The model was initially developed assuming 2D-axisymmetry, even though the wheel is not 31  actually axisymmetric due to spokes and bolt holes. A 3D model was developed following the 2D-axisymmetric model to evaluate the effect of geometry.  The effects of heat transfer and turbulence were'not included in the analysis. While heat transfer is undoubtedly important in the problem of die wear, the accelerated wear of select sprue models experiencing similar temperature conditions suggested that the problem was geometry dependent. Turbulence was also not included since it was believed that the low applied pressure would not induce sufficiently high velocities. The following sub-sections discuss the development of model geometry and model parameters.  4.2.1 Geometry  Both the sprue and wheel geometries were included in the filling model. The wheel geometry is not fully axisymmetric due to spokes and bolt holes, so it was expected that a 30° section 3D model would provide the most accurate filling model. However, given that the sprue and wheel geometries were nearly axisymmetric and the additional computation time required for a 3D geometry, initial filling models were developed for a 2Daxisymmetric geometry taken through the thick section of a spoke. In an attempt to account for the volume difference between the original 2D-axisymmetric and 3D model geometries, a modified 2D-axisymmetric geometry was also created. This section discusses the three model geometry cases: a 30° section of the 3D wheel geometry, 2D axisymmetric thick spoke and modified spoke. Table 4.1 summarizes the mesh for the three cases. Table 4.1: Summary of Mesh Geometries Model 2D thick spoke  Element ^odes Elements Type  V° Sprue  l U m e  Quad  3394  3132  8.91xl0"  2D modified spoke Quad  3289  3010  8.91X10"  23561  99792  8.92xl0'  3D geometry  Tet  ( ) 30° Section Wheel Total m3  Time step Run time size (s) (days)  m  4.77x10^  5.66X10"  4  5  4.09X10"  4.98X10"  4  5  3.89X1Q-  5  4  4  lf/MO"  5  10' -10" 3  4.78x10-" 10 -10' 3  5  5  1.5 1 -17  The model using the 30° section of the 3D wheel geometry (shown in Figure 4.1) was expected to provide the most realistic flow conditions. However, several difficulties were encountered with this geometry. While it would have been preferable to have the elements  32  a l i g n e d w i t h the d i r e c t i o n o f f l o w , as this w o u l d substantially i m p r o v e c o n v e r g e n c e , creating a h e x a h e d r a l m e s h for s u c h a c o m p l e x geometry p r o v e d too d i f f i c u l t .  Consequently,  tetrahedral elements a p p r o x i m a t e l y 3 m m i n size w e r e u s e d a n d it w a s not p o s s i b l e to a l i g n these elements w i t h the d i r e c t i o n o f f l o w . T h e large n u m b e r o f elements u s e d i n the m o d e l and s l o w c o n v e r g e n c e resulted i n a l o n g r u n time ( o n the order o f s e v e r a l w e e k s ) , so the 3 D m o d e l w a s o n l y r u n o n c e for v e r i f i c a t i o n o f the 2 D a p p r o x i m a t i o n s .  Figure 4.1: 3D geometry showing exterior surface mesh.  Figure 4.2(a)  s h o w s the o r i g i n a l 2 D - a x i s y m m e t r i c a p p r o x i m a t i o n o f the w h e e l c a v i t y , w h i c h  w a s d e v e l o p e d based o n the thickest s e c t i o n o f the spoke. T h e m e s h w a s generated w i t h the c o m p u t e r - a i d e d e n g i n e e r i n g ( C A E ) p r o g r a m I - D E A S . I n this m o d e l , 3 1 3 2 quadrilateral elements a p p r o x i m a t e l y 4 m m i n size w e r e used. Q u a d r i l a t e r a l elements w e r e u s e d because they c o u l d be a l i g n e d i n the d i r e c t i o n o f f l o w , w h i c h is e s p e c i a l l y i m p o r t a n t i n the h u b a n d spoke regions. T h i s results i n better convergence a n d a s u b s t a n t i a l l y l o w e r c o m p u t a t i o n t i m e than w i t h triangular elements.  33  Using the thick spoke section, the corresponding wheel cavity volume is larger than the actual cavity, as the missing volume of the window and bolt holes are not accounted for. This increase in volume was thought to significantly influence the resulting pressure and velocity fields in the sprue, because of the larger surface area through these sections and the increased flow necessary to fill the extra volume.  u  0.1 Radius (m)  0.2  0.1  "0  0.2  Radius (m)  (a) (b) Figure 4.2: 2D axisymmetric mesh of sprue and wheel cavity. The thick spoke section is shown in (a) and the modified spoke section in (b).  A second 2D-axisymmetric geometry was developed by modifying the hub and spoke sections to account for the reduction in volume due to the window and bolt holes. Using the 30° section of the 3D geometry shown in Figure 4.1, circular arcs at given radial distances were extruded through the hub and spoke regions of the wheel and the intersecting area was determined. When multiplied by 12, this area represents the surface area of a cylinder, minus the end caps. The length corresponding to the height of the cylinder was used to define the hub/spoke height in a new 2D geometry at that radial distance. This resulted in a significant area reduction in the hub and spoke regions, shown in Figure 4.2(b), with an equivalent volume that is much closer to the 3D geometry than the thick spoke section. This model used 3010 quadrilateral elements that were approximately 4 mm in size.  34  4.2.2 Material Properties and Operating Conditions  Air and aluminum alloy A3 5 6 were specified as the primary and secondary fluids, respectively. During die casting, liquid aluminum is injected into the sprue at 700±10°C. For this reason, material properties for 700°C were used in the simulation. These properties are presented in Table 4.2. The effect of surface tension between the two fluids was also included. However, a surface tension value specific to air-A356 interaction at 700°C could not be found, so the surface tension for 630°C was used. The effect of this assumption is considered in Section 4.3.2. Table 4.2: Fluid Properties Density (kg/m ) 0.3482 2410  Fluid Air A356  3  [34]  [35]  Viscosity (kg/m-s) 4.42E-05 1.20E-03  Surface Tension (N/m) [33]  0.889  Operating conditions specified in the model were operating pressure and gravitational acceleration. The operating pressure was set to 101325 Pa. The gravitational acceleration was specified as 9.81 m /s in the appropriate direction. 2  4.2.3 Boundary Conditions  Several different types of boundary conditions were required in the formulation of the filling model. The boundary conditions were derived from process information, in an attempt to simulate the flow conditions during the wheel filling process as closely as possible. This process information included the application of the 330N wheel model pressure curve at the inlet, the inclusion of a screen between the sprue and wheel, as well as other boundary conditions.  4.2.3.1 Inlet Pressure Boundary Condition  A pressure inlet boundary condition (BC) was applied across the entrance to the sprue. The transient pressure profile for the filling simulation, shown in Table 4.3, was developed using the pressure curve supplied by CAPTIN for the 330N sprue design. The pressure data refers  to the pressure applied to the metal surface in the holding furnace and was offset to the sprue entrance for use at the pressure inlet. The joint pipe which transports metal from the furnace to the sprue entrance (0.58 m) fills in approximately 8 seconds (step 3 in T a b l e 4.3). Accordingly, the pressure was offset by 13.734 kPa to provide the pressure corresponding to the entrance of the sprue (relative pressure in T a b l e 4.3). The relative pressure, starting at 0 kPa, was then incorporated as a transient pressure profile and applied to the inlet. Table 4.3: Pressure curve data supplied by CAPTIN for the 330N model Step Time  Absolute Time  Relative Time  (s) 4  (s) 0  (s)  1  -  0  -13.734  2  4  4  -  9.4176  -4.3164  10  8 18  0  13.734  0  10 30  19.62 39.24  5.886 25.506  60  39.24  25.506  65  83.385  69.651  Step  3 4 5  20 30  38  5 127  68 73  •  6 7  Pressure (kPa) Relative Absolute  The pressure applied at the inlet is the total pressure, p , of the fluid. For incompressible 0  flow, the total pressure and static pressure,^, are related to velocity through Bernoulli's equation as given below: 1  9  Po=Ps+^P  v  ( -6) 4  At the inlet the static pressure is zero. The velocity components are computed from the resulting velocity magnitude. The volume fraction of the A356 phase is set to 1 at the inlet, signifying that only A356 enters through the inlet.  4.2.3.2 Outlets  Outlets, necessary to compensate for the incoming fluid at the inlet and satisfy continuity, were specified as pressure outlet boundaries with zero gauge pressure. Based on feedback from CAPTIN, two outlets were included in this model and were shown in F i g u r e 4.3. The first outlet was located at the air vent on the flange of the inboard rim where the top and side dies connect. This vent is 1 mm in length and goes all the way around the wheel.  36  B  ^ A  Figure 4.3: 2D thick spoke section, illustrating outlets at locations C and D. The second outlet was included as a gap around the ejector pins. In die casting, ejector pins are used to eject the cast wheel from the die after the wheel has solidified and the pressure in the die has been released. In the 330N design, the ejector pins (12 mm in diameter) actuate through 12.1 mm diameter holes in the die. A small amount of air escapes from the gap around the ejector pins. To resolve issues with entrapped air in the hub section during preliminary filling simulations, an outlet was placed where the ejector pins were located. In the 330N wheel die, there arefiveejector pins in the hub section. The area of the air gap around the five ejector pins was calculated and included in the 2D and 3D models using appropriate outlet BC's.  The centre points of the ejector pins in the hub region are located 50 mm from the central axis. The gap area for an ejector pin is 1.9 mm and forfiveejector pins the equivalent area 2  is 9.5 mm . In the 2D models, the equivalent length was calculated to be 0.03 mm. This 37  length is referred to as equivalent in that when revolved it gave the same surface area as five ejector pin gaps. However, this length is very small and a very fine mesh would be necessary in the hub region to describe it, resulting in increased computation time. Consequently, it was increased to 0.1 mm. In the 3D model, an element surface of approximately 2 mm was specified as the outlet in this region.  4.2.3.3 Screen  In the wheel die casting process at CAPTIN, there is a screen placed at the top of the sprue, held down by the centre core pin of the top die. The screen was included in the model because it was located at the sprue exit and was thought to affect the velocity and pressure in the exit region in the sprue. In the models, the screen was located 1 mm from the top of the sprue for better mesh quality in the sprue exit region. Information provided by CAPTIN indicates that the screen was about 1 mm in thickness with about 50% of the area voids open to flow and holes 2 mm in diameter. Since the screen was located in an area where the flow was essentially uni-directional, a porous jump was used to account for the pressure difference induced by the screen. Since the screen impedes the flow of fluid, there is a drop in pressure across the screen. This effect will increase the pressure and decrease the velocity in the exit region of the sprue.  Porous jump conditions are used to model thin membranes with known velocity or pressuredrop characteristics. In FLUENT, the porous jump is calculated from a combination of Darcy's Law and an additional inertial loss term shown in the following equation: (4.7) where Ap is the change in pressure (Pa), JJ. is the laminar fluid viscosity (Pa-s), a is the medium permeability (m ), C2 is the pressure-jump coefficient (m'), v is the velocity normal to the screen (m/s), and L is the medium thickness (m). In order to use this relationship, the medium permeability and the pressure-jump coefficient need to be calculated. In FLUENT, porous cells are considered to be 100% of the area open to flow, which must be accounted for in the calculation of the permeability and pressure-jump coefficient. 38  The first part of Equation (4.7) represents Darcy's Law. Darcy's Law states that for laminar flows through porous media the pressure drop is proportional to the velocity and the constant C2 can be considered to be zero. Ignoring convective acceleration and diffusion, the porous media model reduces to Darcy's Law: V/> = - - v  1 0 0 % o p e  „  (4.8)  where Vp = Ap IL. Hagen-Poiseuille's Law may be used to calculate the medium permeability, a. Hagen-Poiseuille's Law describes the flow through a pore and is given by the following equation:  \2SpL where Q is the flow rate and d is the pore diameter. The screen is essentially a perforated plate with 50% of the area open to flow. Knowing that for the same flow rate, v  5o%  o p e n  =  2 v  ioo%  o p e  „>  t h e  flow  r a t e  through a pore could be calculated: A  nd  d  n ^  2  Q = Av = —  v  50%open  =—  v  (4.10)  l00%open  Combining this with Equation (4.9) results in the following relation: -Mv  — =— Ap~ 64  m%open  (4.H)  with the negative sign resulting from the decrease in pressure. Combining Equation (4.8) and (4.11) gives the screen permeability:  * 764 7  <- >  =  4  8  12  2  For a screen with 2 mm diameter holes, the permeability was calculated to be 6.25x10" m .  39  The pressure-jump coefficient also needed to be determined while accounting for the void fraction of the screen. The pressure drop through a perforated plate is 0.5 times the dynamic head pressure. The loss factor, KL, is defined as: 1  Ap = K  (4.13)  Voopen  L  An adjusted loss factor, K , based on 100% of the area open to flow is determined from the L  following relation: 50%open  v  = 0.5  2  =2  (4.14)  \ l00%open ) v  The adjusted loss factor is then converted to a loss coefficient per unit thickness of the perforated plate: (4.15)  C = < L 2  For a 1 mm thick screen, the pressure-jump coefficient or inertial loss factor is 2000 m' For a screen of 1 mm thickness and 50% open area, the pressure drop for A356 moving at 0.5 m/s is 612 Pa and 2429 Pa for 1 m/s, summarized in Table 4.4. To examine the assumption of 50% open area, the pressure jump was also determined for 40 and 60% open area, which will be further discussed in Section 4.3.3.  Open Percentage 40 50 60  c  Ratio Of V% pen  a  tO Vion%open  (m )  2.5 2 1.7  5.00xl0" 6.25xl0" 7.50 xlO'  0  2  (m") 1  2  8  8 8  3125 2000 1389  Ap (Pa) at v=0.5 m/s at v=l m/s -953 -3790 -612 -2429 -426 -1690  40  4.2.3.4 Symmetry and Wall Boundaries  The centerline of the 2D-axisymmetric geometries is an axis boundary, where symmetry conditions must be applied (no flow normal to the line). In the 3D model, the mirror planes were specified as symmetry boundaries with zero normal velocity and normal gradients.  The no slip condition was applied to all wall boundaries. With the no slip condition the fluid adheres to the wall with zero velocity, creating a boundary layer along the wall.  4.3 Model Results  The filling model has been used to generate velocity and pressure profiles for the different geometries of the sprue and wheel. Since it was not feasible to experimentally measure the velocity and pressure, the information provided by this simulation can only be considered qualitatively, but nonetheless was important in understanding the mechanism of accelerated sprue wear.  4.3.1 Effect of Geometry  Contour plots illustrating the velocity magnitude and the static pressure in the A356 phase duringfillingare shown in Figure 4.4 through Figure 4.9 for the 2D thick spoke, 2D modified spoke, and 3D models, respectively. In each plot, the flow field has been blanked for volume fractions less than 0.5 to show the flow field for only the liquid A356 phase. As the liquid metal enters the sprue, the sprue cross-section gradually decreases until the constriction point (the point of minimum sprue diameter). As the free surface approaches the constriction point, the velocity of the bulk flow decreases. Beyond the constriction point, the sprue diameter increases allowing the liquid to expand out, leading to an increase in velocity.  In general for all three geometries, the velocity of the A356 phase is fairly low until about 9 seconds, when the liquid A356 startsfillingthe sprue, then velocity increases in the sprue constriction region. In the sprue, velocity is generally higher towards the centerline at a  41  given height. However, near the sprue entrance in the 2D models, the velocity along the centerline is lower than velocity in the bulk flow. This is most likely an entry effect that could be eliminated by adding pipe geometry to ensure that the liquid flow is fully developed when entering the sprue.  For the three models, pressure is applied at the sprue entrance resulting in high pressure originating from there. With increasing time, the pressure increases according to the pressure profile described in Section 4.2.3.1. While the sprue and hub are filling, the pressure contours are horizontally aligned with the sprue entrance and evenly spaced. As the liquid starts filling the spoke, the pressure contours in the sprue exit region are no longer horizontally aligned. Instead, the pressure remains higher along the centerline due to the presence of the centre core pin, and decreasing towards the exit surface at the same vertical height. Additionally, the pressure contours in the sprue exit region become more closely spaced.  In terms of free surface location, all three models show the same location of the liquid/gas front during the filling of the sprue until reaching the hub at approximately 9 seconds. The hub in each model is shaped slightly different. In the hub section of the 2D thick spoke model, the geometric features are straight and near vertical, while the features in the 2D modified spoke model are rounded and more gradual. In the 3D model, the hub section is varied due to the presence of bolt holes and lightner blanks. These features in the 3D model allow the A3 56 to flow along the low side of the hub and into the spoke earlier than in the 2D models.  42  (f)9s  (g)lls (h)12s (i) 14 s Figure 4.5: Static pressure contour plots for the 2 D thick spoke model during filling. (Pa)  (j)17s  10000  9000  8000 7000  000 0  5000 4000  (f>  9 s  (g)Hs (h)12s (i) 14 s Figure 4.7: Static pressure contour plots for the 2 D modified spoke model during filling. (Pa)  (j) 15.5 s  (§) (h)12s (i) 13 s Figure 4.9: Static pressure contour plots for the 3 D model during filling. (Pa) l  l  s  (j)14s  The location of the free surface was plotted for each of the different geometries for the purpose of comparison in Figure 4.10 and Figure 4.11. Figure 4.10 shows the location of the free surface in the vertical direction. When the flow is vertical, it obeys Pascal's Law, given by: p = pgh  (4.16)  where p is the pressure (Pa), g is the gravitational acceleration (m/s ), and h is the liquid 2  metal height (m). This holds for the filling of the sprue, as shown in the agreement of the three models.  The three models agree well up until 9 seconds, at which point the A3 56 reaches the hub and starts spilling into the spoke and the flow direction is no longer primarily vertical. Figure 4.11 plots the free surface location in terms of radial distance from the centerline. The free surface location represents the front of the bulk flow, rather than small disconnected droplets that may have spilled into the spoke.  The difference in the shape of the hub and spoke sections results in different flow patterns in the spoke. The hub section in the 3D model contains bolt hole and lightner blanks. The resulting topography causes the liquid A3 5 6 to flow along the lower side into the spoke. Once the hub is sufficiently filled, the liquid A356 also flows down along the other side, over the lightner blank. Contour plots at two different orientations in Figure 4.12 illustrate the different free surface locations. In order to compare the location of free surface in the 3D model with the 2D models, two free surfaces were tracked: one along the low side (3D spoke region) and the other on the high side over the lightner blank (3D hub region). These two locations appear to start from different radius because liquid enters the spoke early along the low side, while the hub is still filling vertically. Only one free surface was tracked through the hub and spoke sections for each of the 2D models.  In the 3D model, liquid A356 starts filling the spoke after about 9 seconds along the low hub side. The flow along the other side of the hub exhibits slower progress in the radial direction, more in line with that of the 2D models, until about 9.8 seconds, when the free surface makes  49  it over the lightner blank. The timing of this event agrees very well with that of the 2D modified spoke model. At 10.2 seconds in the 2D modified spoke model, the free surface enters the spoke and catches up with the 3D spoke region free surface. The 2D modified spoke model continues to show good agreement with the 3D free surface until about 10.7 seconds when the free surface direction changes back to vertical.  The movement of the bulk flow through the spoke was not as straightforward in the 2D thick spoke model. Due to the steep shape of the hub, only small droplets were able to flow into the spoke. These droplets accumulated in the outboard rim flange. The bulk liquid does not start to spill into the spoke until after nearly 11 seconds, as shown in Figure 4.4(g), explaining the 1 second difference from the other two models. The sharp increase in radial distance at 11.5 seconds represents the bulk flow connecting with the A356 droplets that had collected in the outboard flange rim. 1000 900 800  2D Thick Spoke Model 2D Modified Spoke Model 3D Model Pascal's Theorem  700 /  400 H 300 200 100 0 8 10 Time (s)  12  14  16  18  Figure 4.10: Comparison of the free surface height for the 2D and 3D models. The calculated free surface height by Pascal's theorem is also included for sprue filling.  Given the different volumes of the hub and spoke regions, A3 5 6 starts filling the rim at different times for the three models. However, Figure 4.10 shows that the vertical movement of the free surface is very similar for the 3D and 2D modified spoke models. 50  215 - 3D Hub Region - 3D Spoke Region 2D Thick Spoke 2D Modified Spoke  195 175 155 £ 135  1 115 /  95 75  H  55 35 9.5  10  10.5 Time (s)  11  11.5  12  Figure 4.11: Comparison o f the free surface radial distance for the 2 D approximation and 3 D models.  Figure 4.12: Comparison o f the A356 velocity magnitude (m/s) for the two symmetry planes during filling.  A l l three models had varying amounts of air trapped in the hub at the end of filling. This was most pronounced in the 2D thick spoke model because of the sharp feature into the hub section. There are a couple of reasons why air pockets remained trapped in the hub section at the end of wheel filling. Firstly, the pressure at the inlet is increased after at 10 seconds. The locations of the free surface and the bulk of the A3 56 at that time will affect the flow pattern. 51  Second, the models do not include the effect of temperature, so the temperature dependency of the material properties was not taken into account, which may affect the filling pattern.  Returning to the problem of accelerated sprue wear, the velocity and static pressure along the surface immediately after the constriction point where wear occurs were compared for the three models. In each of the three models, the velocity and pressure were obtained at the two locations indicated in Figure 4.13.  These points are located 1 mm from the sprue surface  and have been referred to by their vertical sprue height, 159 and 171 mm. Center Core Pin  Screen  U  D.15 h  I  171 mm i  d! Sprue •.16  159 mm J , Constristion Point  D.15,  •HI  0.02 Radjif (pi)  Figure 4.13: Exit region of 330N sprue with points indicating locations from which velocity and pressure data were acquired. These points were located 1 mm from the exit surface.  Figure 4.14 and Figure 4.15 compare the velocity at the two locations for each of the three models. The first point is located nearest the constriction point at a height of 159 mm vertical distance. The free surface reaches this height at around 6.4 seconds in each of the three models. Velocity in the 2D models peaks around 12.3 seconds, later than in the 3D model and closer to 0.8 m/s. In the 3D model, the velocity gradually increases until reaching a peak of 0.7 m/s at 12 seconds, when liquid begins filling the rim. The earlier increase in the 3D model can probably be attributed to the shape of the hub. The velocities at a height of 171 mm are very similar to those at the constriction point, although slightly lower.  52  0.9 2D thick spoke 2D modified spoke  0 6  8  10  12 Time (s)  14  16  18  Figure 4.14: Comparison of velocity variation for the three model cases at 159 mm height.  Similarly, the static pressure was plotted in Figure 4.16 at a height of 159 mm for the three models. The 3D model shows a linear pressure increase, until about 9 seconds. The pressure remains nearly constant until after ~12 seconds when pressure exhibits a sharp increase. This corresponds to the increase in velocity observed in Figure 4.14, which peaks at ~12 seconds.  In the case of the 2D models, the rate of pressure increase is nearly linear in both models, showing three distinct regions corresponding to filling in the sprue, spoke, and rim. In the final seconds of filling, the pressure in the 2D modified model is closer to the 3D model than the 2D thick spoke model. The reason for the pressure difference between the 2D and 3D models may be that at 10 seconds, the pressure at the sprue entrance is ramped up. At 10 seconds in the 3D model, the free surface has filled along half the spoke, so the pressure along the sprue exit surface is lower. At this time, the liquid is only just starting to fill the spoke in the 2D models.  53  Time (s)  Figure 4.15: Comparison of velocity variation for three model cases at 171 mm height. 8000  6  8  10  12 Time (s)  14  16  18  Figure 4.16: Comparison of static pressure for three model cases at 159 mm height.  4.3.2 Effect of Surface Tension  To examine the effect of surface tension duringfilling,the surface tension described in Section 4.2.2 was varied by ±10% and compared to the baseline condition. Since the run time of the 3D model was so long, a 2D model was used. The 2D modified spoke model was used, as it was closer to the 3D model in volume and fill time. The effects of surface tension on velocity and pressure in the sprue are shown in Figure 4.17 and Figure 4.18.for the 159 mm location.  Increasing the surface tension changed the velocity and pressure along the exit surface of the sprue only a small amount. The peak velocity occurred a little later than for the baseline surface tension. Almost no change in pressure was observed. Lowering the surface tension significantly changed the velocity along the sprue surface. Velocity did not start to increase until after 11 seconds and it kept increasing until filling was complete. The pressure along the sprue exit surface was also lower than other the two cases during wheel filling.  Figure 4.17: Comparison of velocity at 159 mm location to examine the effect of surface tension.  55  7000  6  7  8  9  10  11 Time (s)  12  13  14  15  16  Figure 4.18: Comparison of pressure at 159 mm location to examine the effect of surface tension.  The surface tension value obtained from literature was for A3 56 and air at 630°C. Surface tension is not critical during the filling of the sprue, given that velocity and pressure did not significantly change until the liquid started filling the spoke. Aluminum is injected into the sprue at 700°C, so it is likely that the temperature in the wheel cavity is lower. If the temperature in the spoke was lower than 630°C, surface tension increases with decreasing temperature and only a small change in velocity and pressure was observed along the exit surface with 10% lower surface tension. This would suggest that the assumed surface tension value is not likely to significantly affect the velocity and pressure along the exit surface of the sprue, although it may influence the filling pattern in the wheel.  4.3.3 Effect of Screen  Another important aspect of the filling model is the effect of the screen on the velocity and the pressure around the exit of the sprue, as it may influence the occurrence of accelerated sprue wear. The screen was implemented through a pressure jump BC, described in Section 4.2.3.3. The base model was formulated assuming that the screen was 50% area open to flow. To examine the effect of the screen on flow in the sprue during filling, the base model  56  was compared with a model with no screen, and models with screens of 40 and 60% area open to flow, respectively. Figure 4.19 and Figure 4.20 show the resulting velocity and pressure curves in the sprue at the 159 mm location. Table 4.4 also provides information on the effect open area percentage has on the pressure jump.  Removal of the screen significantly changed the pressure at the 159 mm height sprue location. There was a sharp drop in pressure around 10 seconds as liquid A356 started filling into the spoke. The pressure increased at approximately 11 seconds when the liquid began blocking the thin spoke section to a pressure higher than observed for the three screen cases. The velocity also peaked earlier and with a sharper drop than in the screen cases.  As for the three screen cases, there was virtually no difference in pressure before 10 seconds. At approximately 10 seconds, liquid starts filling the spoke, until about 13 seconds, when the liquid started filling the rim, the pressure at the 159 mm height sprue location was slightly lower with increasing open area percentage. There were also some small differences in velocity during this time. 1  i  6  7  8  9  10  11 Time (s)  12  13  14  15  16  Figure 4.19: Comparison of velocity for different screen conditions at 159 mm height.  57  The screen properties were approximated based on information provided by C A P T I N . While changing the open area percentage does slightly change the velocity and pressure along the sprue exit surface, the differences were not significant and so small differences i n screen properties are not likely to appreciably affect the results. O n the other hand, ignoring the effect o f the screen would greatly affect the pressure along the sprue exit surface.  7000 n  6  7  8  9  10  11  12  13  14  15  16  Time (s)  Figure 4.20: Comparison of pressure for different screen conditions at 159 mm height.  4.4 Mesh Sensitivity The effect o f mesh (both element size and type) was examined. M a n y locations in the sprue and wheel were locally meshed to obtain flow-oriented quadrilateral elements. A l o n g the surfaces where the mesh had been locally modified, the number o f elements was increased or decreased by 10%. The global element size was also changed i n each case. In addition to element size, the effect o f element type was also considered. Using the 2D modified spoke geometry and similar local element sizes, triangular elements were used instead o f quadrilateral elements. Table 4.5 summarizes the mesh information for these models.  58  Table 4.5: Mesh Sensitivity Model 2D modified spoke -10% no. elements +10% no. elements  Element Type  Nodes  Elements  Quadrilateral Triangular Quadrilateral Quadrilateral  3289 3459 2466 6069  3010 6333 2230 5692  Time step size Computation Time (days) Cs) 1 5xl0" -lxl0lxlOMxlO" 5 1.5 lxlO'MxlO' 1.5 Ixl0" -lxl05  3  4  3  5  3  A large increase in computation time was observed with the triangular element mesh. Meshing such a complex geometry is easier with triangular elements, however, the elements cannot be align with the direction of flow. This resulted in convergence difficulties, with the time step having to be reduced to 10" seconds at times, especially in the spoke region. The 6  change to quadrilateral elements reduced the simulation run time by several days.  Figure 4.21 and Figure 4.22 compare the velocity and pressure at the 159 mm vertical height for the different meshes. No clear trends were observed by changing the mesh. The velocity with the triangular mesh was higher than for the quadrilateral meshes and the pressure lower. The velocity along the exit surface was higher for the finer quadrilateral mesh than the two other quadrilateral meshes. With smaller elements, the boundary layer should be better resolved, so the higher velocity can perhaps be explained. However, the velocity for the coarser quadrilateral mesh was higher than that of the baseline quadrilateral mesh from about 12 to 14 seconds.  Similarly, no clear trends were observed with pressure. The pressure at the 159 mm location was lower with the triangular mesh than with the quadrilateral meshes, corresponding to the higher velocity. There were also some smaller differences for the three quadrilateral mesh cases.  From this analysis, it is clear that the model is sensitive to mesh. Quadrilateral elements were superior to triangular when it came to convergence and computation time. Computation time was substantially minimized when these elements were aligned in the direction of flow. In thin sections, such as in the spoke, a sufficient number of elements must be used to ensure good flow resolution.  59  10  12 Time (s)  Figure 4.21: Comparison of velocity variation for the three model cases at 159 mm height. 7000  6000  Quadrilateral mesh Coarser quadrilateral mesh Finer quadrilateral mesh Triangular mesh  A /•/  //  5000  a  f  £ 4000 3  II)  / /i'  in  2.  a. 3000  2000  1000  10  12  14  16  Time (s)  Figure 4.22: Comparison of static pressure for three model cases at 159 mm heig];ht.  4.5 Summary  A fluid flow model was developed to simulate the flow of liquid A356 during the filling of a 330N sprue and wheel. The filling model uses the VOF technique to track the free surface during die filling. A pressure inlet BC with a transient pressure profile was applied at the sprue entrance. Two outlet locations were included to satisfy continuity, based on their location in the operational wheel die. Additionally, a pressure jump was included at the exit of the sprue, similar to that in the actual process, to simulate the effect of the resulting pressure drop.  Filling was initially modeled with a 2D axisymmetric geometry based on a section taken through the thick portion of the spoke. However, the resulting volume of this model was much greater than the actual 3D geometry, which included spokes and blanks for bolt holes and lightners. A second 2D axisymmetric approximation was made by modifying the hub and spoke regions to give a volume that was equivalent to that of the 3D model. A 3D model was developed to verify the results of the 2D approximations, however, the long computation time made it unfeasible as a tool for investigating the flow during die filling.  The 2D modified spoke model gave a close approximation to the 3D model in terms of free surface location, although it was a little offset in time due the shape of the hub in the model. Both 2D models showed similar velocity and pressure trends, although neither model agreed entirely with the 3D, mostly because of the different hub shapes which influenced filling pattern. The inlet pressure was ramped at the same time for the three models, but the free surface was at different locations during the filling of the spoke and rim, resulting in different velocity and pressure along the sprue exit surface. While not the same, the velocities and pressures in the sprue were of the same magnitude in all three models.  The effects of surface tension, screen open area percentage, and mesh on the velocity and pressure along the exit surface of the sprue were evaluated. The surface tension value used in the baseline model was for A356 and air at 630°C. Surface tension effects are most important during filling of the spoke, and given that aluminum is injected into the sprue at  61  700°C, the temperature there is likely to be considerably lower. Changing the open area percentage of the screen only slightly affected the velocity and pressure along the exit surface of the sprue. Changing the element size and type significantly changed the flow solution, although the trend was not clear. More analysis would have to be performed to optimize the mesh to obtain the best combination of solution and computation time.  The model was developed to provide information on the velocity and pressure in the sprue during filling. The information provided by the model, namely the velocity in the sprue region, was used as a starting point for the erosion-corrosion tests, which will be discussed in the next section. With additional optimization and the addition of heat transfer, this model could be used to optimize the casting process and minimize the occurrence of flow-related defects.  62  CHAPTER 5: EXPERIMENTAL PROCEDURE Processing parameters clearly have a strong influence on reactions at the liquid metal/die interface and die wear. While this interaction has been much studied, the mechanism behind the problem is still not well understood. One aspect lacking in previous experiments was a sufficient link of geometry to wear behaviour. Sprue geometry influences velocity and pressure conditions and appears to be a critical factor in determining the occurrence of accelerated sprue wear. The intent of this project was to investigate the erosive-corrosive wear behaviour of sprue materials exposed to liquid aluminum to shed new light on the mechanism and provide a link to geometry. The rotating pin techniques used by other researched ' " 13  28  30]  were found to be well suited for this purpose, with the inclusion of pin  geometry as a parameter.  This chapter begins with an examination of a sprue that had experienced accelerated wear and the characterization of the thermal conditions to which sprues are subjected to during a casting cycle. Considering the velocity conditions in the sprue region from the filling simulation, developed in the previous chapter, the intent of this initial characterization was to design a pin that would create similar flow conditions to the sprue and determine test conditions which would mimic operation. The methodology used in the design of the profiled pin cross-section is discussed, along with the experimental set-up used and the two sets of tests performed.  5.1 Sprue Examination Two sprues were received from CAPTIN, both showing signs of accelerated wear on the surface after the constriction point. The two sprues examined (shown in Figure 5.1) were taken from dies for the 329N wheel model (which had previously been weld repaired) and the 330N model, both having been replaced after a single casting run. Both sprues were made from 4140 steel. In the case of the 329N sprue shown in Figure 5.1(a), large amounts of material were removed along the top surface. The wear of the 330N sprue was more uniform in that material removal occurred more in the form of wall thinning, as indicated by  63  the arrows in Figure 5.1(b). Sprues made from 4140 steel are expected to last 2-3 casting runs, so experiencing this much wear after one casting run is significant.  (a)  (b)  Figure 5.1: Accelerated wear of (a) 329N and (b) 330N wheel model sprues.  The 3 3 ON sprue was sectioned and polished and the aluminum-steel interface was examined with the secondary electron (SE) mode of a Hitachi S-3000N scanning electron microscope (SEM). It was observed that the ceramic coating, applied as a thermal/erosion barrier, was not present in areas of accelerated wear. Figure 5.2 shows an SEM micrograph from the interface of the wear region of the 330N sprue. Elemental compositions determined using energy dispersive spectroscopy (EDS) are shown in Table 5.1 and correspond to locations indicated in Figure 5.2.  Starting at the top of the micrograph at location 1, a spalled steel layer was observed between the solidified aluminum and the intermetallic layers. EDS confirmed the composition of the steel layer to be mainly iron. Below this, two distinctive intermetallic layers were observed between the aluminum and steel. The layers were approximately 100 nm in thickness and found to be made up of binary and ternary intermetallic compounds, confirmed by EDS to be primarily made up of aluminum, iron, and silicon. The first intermetallic layer had a substantial silicon concentration and may possibly be an iron-aluminum-silicon ternary phase. The second intermetallic layer for the most part had a lower silicon concentration, 0.5 to 2.0 wt% as opposed to 4 to 10 wt% for the first layer, indicating the layer was more likely  64  to be a binary phase. However, there was a large amount of silicon precipitating along the boundary between these two layers, near location 5. It was also speculated that there might be another phase between the second layer and the steel at location 8, although this was not conclusively shown with the compositional analysis.  x600  0000  20kV  50JJM  Figure 5.2: Interface between 4140 steel and A356 from 330N sprue.  Table 5.1: Elemental Compositions of Interface Layers in Weight Percent Location 1 2 3 4 5 6 7 8 9  Spalled layer Layer 1  Layer 2  Steel  Al  Fe  "si 31  0.8 40.4 46.8 39.0 17.8 40.8 43.8 40.2 0.0  93.9 47.2 46.3 48.0 60.1 54.4 53.4 56.5 97.0  1.0 10.2 4.0 10.4 18.6 2.1 0.5 0.8 0.4  Layer Thickness Average Std Dev  42.9  7.5  46.8  5.9  6.3  1.1  Depending on the area of observation, intermetallic layers were made up of one, two or three distinctive phases and consisting mainly of aluminum, iron, and silicon. The layers were severely cracked in places. Layer thickness varied from 5 to 100 um, once again depending on location. At another location of the wear region of the 3 3 ON sprue, the Back Scattered Electron (BSE) mode was used to capture the image shown in Figure 5.3(a). Elemental  m a p p i n g w a s used i n  Figure 5.3(b) to p r o v i d e a qualitative plot o f the a l u m i n u m , i r o n , a n d  s i l i c o n contents a l o n g the a l u m i n u m - s t e e l interface. T h e m a p corresponds to the b o x indicated i n  Figure 5.3(a). R e g i o n s o f h i g h a l u m i n u m c o n c e n t r a t i o n are i n d i c a t e d i n y e l l o w ,  i r o n i n p u r p l e , a n d s i l i c o n i n green, i n d i c a t i n g that the interface layer consists o f a l l three elements a n d that there are regions i n the s o l i d i f i e d a l u m i n u m A 3 5 6 w h e r e a l u m i n u m - s i l i c o n compounds form.  (b) Figure 5.3: S E M image of interface between solidified A356 and 4140 steel. Area indicated in (a) i magnified and shown in an E D S map in (b).  66  5.2 Sprue Temperature Liquid aluminum is injected into the bottom of the sprue at 700±10°C. During a casting cycle, the temperature along the sprue surface will be high and there will be a temperature gradient along the length of the sprue. As temperature is a variable that affects both erosion and corrosion, it was proposed that the temperature in the sprue be measured at several locations during an industrial casting run at CAPTIN. The 330N sprue was chosen to be tested, since it consistently showed problems with accelerated wear and had been used for the wheel filling model.  Thermocouples (Type-K sheathed in stainless steel) were embedded in a sprue at the approximate thermocouple locations indicated in Figure 5.4. Holes of a diameter slightly larger than the thermocouples were drilled into the sprue into which the thermocouples were inserted.  The drilling of the thermocouple holes presented several problems. The interference fit of the sprue with the die along the top portion of the sprue meant that no thermocouples could protrude from the side, which influenced the location of TCI. Accelerated sprue wear occurs on the surface after the constriction point, so it was preferred to have thermocouples located near this surface. Instead, the thermocouple hole had to be drilled at an angle through the support flange, restricting how close thermocouples could be to the surface. Another problem was that the roughness of the outer surface of the sprue made it difficult to know with any great amount of certainty the distance from the outer to the inner surface. To ensure that the holes did not penetrate the inner surface, all holes were drilled to a depth from the inner surface of approximately 5 ± 1 mm. A data acquisition system was used to record the thermocouple data at a frequency of 10 Hz.  The temperature variation for locations TCI, TC2, and TC3 for one casting cycle and a portion of the next are shown in Figure 5.5. Thermocouple TC4 did not respond during the test. As the aluminum fills the wheel and starts solidification, the temperature along the top  67  of the sprue decreases. When solidification is complete and the pressure is released, liquid aluminum in the sprue is released from the sprue and the temperature of the sprue drops.  The thermocouple TCI was located closest to the top of the sprue and showed the greatest temperature variation. Temperature ranged from approximately 485 to 585°C. Considering that the thermocouple is subsurface and sheathed in stainless steel, the temperature is anticipated to be higher. Temperature at TCI increases as liquid aluminum is injected into the sprue. Thermocouple locations TC2 and TC3 showed similar thermal cycles although without as much temperature variation. Thermocouple TC2 was located below the top support flange and showed the lowest maximum temperature. Thermocouple TC3 was located at the mid point of the sprue and showed the least temperature variation, as it was nearest to the sprue entrance. It was speculated that the temperature at TC2 was lower than TCI because it was in close proximity to the support flange, therefore it heated and cooled more slowly. Thermocouple TCI was located in a thin section at the top of the sprue and consequently heated very quickly. It also cooled very quickly due to its close proximity to the die.  Although these temperature measurements do not give the temperature of the sprue at the surface, they do provide information on the thermal variation along the length of the sprue. The temperature of the sprue is varying along the length from approximately 500 to 600°C. It is not unreasonable to expect the temperature at TCI to be in the range of 630 to 640°C.  68  ^ 4 Liquid Aluminum at 700°C Figure 5.4: Illustration of subsurface thermocouple locations in 330N sprue. T  600  Figure 5.5: Measured temperature at three locations in 330N sprue during industrial casting run.  69  5.3 Profiled Pin Design As mentioned before, one problem with using cylindrical test pins in an erosion test is that this provides no link to geometry. It was decided that in order to examine the effect of geometry on wear, a pin with a profiled cross-section would be developed. The profiled cross-section was intended to provide variations in velocity and pressure along the pin surface.  5.3.1 Pin Geometry  Since the purpose of this project was to investigate the problem of accelerated sprue wear, features from the sprue geometry were incorporated into the design of a profiled pin crosssection. Accelerated sprue wear occurs on the surface just beyond the constriction point so the relevant geometric features are the curvature at the constriction point and the exit angle, illustrated in Figure 5.6(a). The exit angles and curvature radii for the ten sprue models were evaluated using the manufacturing drawings and are summarized in Table 5.2.  Table 5.2: Survey of Geometric Features of Sprues in Constriction Region Exit Angle Radius of Wheel Model Curvature (mm) (°) 10 5 329N LLP 10 4 330N LP 477T LLP 10 10 5 516N LLP 10 9 592N LLP 15 800T LP 15 18 420N TDP 15 517NTDP 10 15 20 800T LTDP 20 15 800T TDP All six LP sprue models had a curvature radius of 10 mm at the constriction point and exit angles ranging from 4 to 15°. Profiled pin cross-sections were developed with curvature radii of 10 mm and draft angles of 5°, 10°, and 15°, respectively. In Figure 5.6, the 5° pin design is compared to the constriction region of the 516N sprue from which the geometric features are similar. The radial dimensions Rl and R2 in Figure 5.6(b) represent the 70  minimum and maximum radii, respectively. Three pin geometries, shown in Figure 5.7, were developed keeping the R 2 dimension constant for the three pin geometries.  (a)  (b)  Figure 5.6: Geometry of the (a) 516N sprue in the constriction region and (b) 5° profiled pin.  (a)  (b)  (c)  Figure 5.7: Profiled pin cross-sections: (a) 5°, (b) 10°, and (c) 15°.  The intent of these tests was to produce flow conditions similar to the actual casting process. As discussed in the literature review, velocity is one of the most important parameters influencing die wear, so it is important to use realistic velocities in the laboratory tests. During the tests, the pins are rotated in the liquid aluminum. While it is possible to calculate an approximate velocity at the Rl and R 2 locations from the 'circumference' and rotation rate, similar to how the velocity was calculated for cylindrical pins, this would not provide information on the velocity and pressure along the pin surface and the pattern of flow. (To provide this information, the following section describes a fluid flow simulation that was developed.)  71  5.3.2 Pin Design Assessment  Once the profiled cross-section was designed, a 2D model was developed to simulate the flow of the liquid aluminum around the rotating pin. The model was developed in FLUENT 6.1 using multiple reference frames. Rather than rotating the boundary representing the pin surface relative to the crucible boundary, which would require a moving mesh, rotating reference frames were used with a stationary mesh (shown in Figure 5.8). When solving the equations of motion in a rotating reference frame, the coordinate system rotates with the rotating boundary and, therefore, experiences a constant acceleration in the radial direction. Additional terms are included in the momentum equation to account for the acceleration of the fluid. The momentum conservation equation for an inertial frame (non-accelerating coordinate system) is given by: Q  —(pv) + V*(pvv) = -Vp + pV v +pg dt  (5.1)  2  where v is velocity (m/s), p is static pressure (Pa), p is molecular viscosity (Pa-s), p is density (kg/m ), g is gravitational acceleration (m/s ), and t is time (s). In this problem, the 3  2  flow relative to the rotating (non-inertial) frame is steady, so thefirstterm in Equation (5.1) is ignored. Theflowcan be solved using either absolute velocity, v , relative velocity, v , r  which are related by the following equation: v =v-{Qxr)  (5.2)  r  where Q i s the angular velocity of the rotating frame and f is the position vector in the rotating frame. For a rotating reference frame, the left-hand side of Equation (5.1) can be rewritten in terms of absolute velocity as, V.(pv v) +p(£lxv) r  (5.3)  or in terms of relative velocity, V.(pv v ) + /?(2Qxv +QxQxr) r  r  r  (5.4)  72  where the last term represents the Coriolis and centrifugal forces. For incompressible flow in rotating domains, the continuity equation for steady flow can be written as follows for both absolute and relative velocity formulations: V.(v ) = 0 r  (5.5)  For the rotating pin model formulation, the crucible boundary is specified as a moving wall with an absolute angular velocity of 0 rpm. The pin surface boundary is specified as a moving wall with a relative angular velocity equal to the rotation rate to be used in the rotating pin tests. A no slip condition is applied at both wall boundaries, so that the fluid adheres to the wall and moves with the same velocity. This formulation results in the liquid rotating at the same velocity as the pin along the pin surface and with zero velocity along the crucible wall. This model formulation was verified with a rotating circular cross-section and a manual calculation.  The mesh for the model was created using the computer-aided engineering program I-DEAS. The outer boundary has a diameter of 88.9 mm, the same as the crucible used in the tests described in the next section. The fluid material is aluminum alloy A356, with the properties as provided in Table 4.2.  The rotating pin simulation was helpful in determining rotation rates for the profiled pin tests to ensure that the velocity along the pin surface was similar to velocities in the LPDC process. Velocity along the rotating pin surface is plotted in Figure 5.9 for the 10° pin at 100, 150, and 200 rpm. Velocity increases along the surface until R2, the maximum radius, then decreases down the trailing edge. The flow along the leading and trailing edges is not symmetric. The velocity deviation from symmetry is indicated along the trailing edge Figure 5.9. The deviation becomes more pronounced with increasing rotation rate. The velocities for the cylindrical pin tests rotated at 98 and 196 rpm have been indicated in Figure 5.9 and will be discussed in the next section. Additional velocity plots have been included in Figure 5.10 and Figure 5.11 for the 5° and 15° pins.  73  Crucible boundary (stationary)  Pin surface boundary (rotating)  Figure 5.8: Illustration of computation domain.  Figure 5.12 compares the velocity along the surface of the three pin geometries at a rotation rate of 200 rpm. Maximum velocity occurs at R2, the point of maximum radius, which was held constant for the three pins. The Rl dimension decreased with increasing draft angle and thus shows a different velocity for each pin. 0.6  100 rpm 150 rpm • 200 rpm  Flow Direction >  Indicates deviation om symmetry  0) 13  Trailing Edge  Leading Edge  c D)  io  0.3  o  196 rpm Velocity used In cylindrical pin tests  > 0.2  H  \ 98 rpm  R1  R1 -54  -36  -18  0 Angle (°)  18  36  54  Figure 5.9: Velocity for 10° pin at different test rotation rates.  74  0.75 150 rpm 200 rpm • 300 rpm  Flow Direction  0.65 in  \ v  0.55  73  1 C  Trailing Edge  Leading Edge  O)  0.45  o  > 0.35  *2  R1  R1  -0r25-54  -36  -18  0 Angle (°)  18  36  54  Figure 5.10: Velocity along pin surface for 5° pin at different test rotation rates.  0.6 Flow Direction  150 rpm - 200 rpm  Trailing Edge  Leading Edge •o c  O)  &  0.3  o o >  0.2  R1  R2  R1 -8T4-  -54  -36  -18  0  18  36  54  Angle (")  Figure 5.11: Velocity along pin surface for 15° pin at different test rotation rates.  5° pin - • - 1 0 ° pin 15° pin  Flow Direction  — — >  Trailing Edge  , -54  1 -36  1 -18  0.4-! 0  ,  18'  R1 ,  i  36  54  Angle (")  Figure 5.12: Velocity for different pin geometries at 200 rpm from leading edge to trailing edge.  Figure 5.13(a) shows the predicted velocity vectors for a 10° pin rotated at 200 rpm. The velocity vectors along the pin surface are coloured and indicate that maximum velocity will occur at R2 and minimum velocity at Rl. Near R l , there is a region where the velocity vectors are pointing into the pin. As the vectors move along the leading edge, they straighten out and become parallel with the pin surface. Figure 5.13(b) is a contour plot of static pressure corresponding to Figure 5.13(a). Static pressure is the pressure exerted by the fluid and is highest when velocity is lowest, i.e. R l .  76  (b) Figure 5.13: Corresponding velocity and contour plots for 10° pin rotated at 200 rpm: (a) Velocity magnitude (m/s), (b) Static Pressure (Pa).  5.4 Experimental Set-up  The purpose of these tests was to investigate the erosive-corrosive wear behaviour of sprue materials in liquid aluminum. Through two sets of experiments, the effects of temperature, time, velocity, geometry, and material on pin wear were examined. In the first set of experiments, cylindrical test pins were tested to examine the effects of velocity, temperature,  77  and time. In the second set of tests, pins of profiled cross-section were used to examine the effects of geometry, angular velocity, and material on wear.  All tests were performed using the experimental set-up shown in Figure 5.14, also shown as a labelled schematic in Figure 5.15. Three kilograms of aluminum alloy A356 were melted in a graphite crucible in a resistance furnace. New aluminum was used for each test. The crucible was placed in a top loading resistance furnace equipped with a temperature controller. Pin specimens were lowered in to the liquid metal where they were immersed for static tests or rotated about their axis for dynamic tests. For tests under dynamic conditions, a variable speed DC motor was used to rotate pins. The motor was attached via a chain/gear and bearing assembly to the shaft supporting the pin. The rotation rate was manually controlled with an accuracy of ±5 rpm, while monitoring with a tachometer.  Prior to insertion into the melt, pins were preheated to about 300°C to minimize the initial freezing of aluminum on the steel pins. Once the liquid aluminum reached the target temperature, the pin was lowered into the melt to a depth of approximately 10 mm from the top of the pin. The motor was started immediately after inserting the pin into the melt. To ensure that the melt temperature remained uniform, the temperature was periodically measured with a sheathed Type K thermocouple with an accuracy of ±2°C. At the end of a test, the pin was raised from the melt and air-cooled. The remaining aluminum melt was poured from the crucible.  78  Figure 5.14: Rotating pin test set-up.  Legend: A - Resistance Furnace B - Aluminum Melt C - Graphite Crucible D-Pin E - Sheathed Thermocouple F - Fibrefrax Insulation G - Steel Shaft H - Water-Cooled Bearing Housing I - Gear and Chain Assembly J - Rotation Assembly K-Motor L - Support Column M - Control Arm  Figure 5.15: Schematic of experimental set-up.  5.5 Test Conditions Two sets of experiments were performed to investigate the effect of different parameters on wear. In the first set of experiments, cylindrical test pins were used to determine the effect of rotation rate, temperature, and time on pin wear. The profiled cross-section pins were used in a second set of tests using the same experimental set-up where the effects of rotation rate and material on pin wear were examined.  The cylindrical test pins were made from 4140 steel. The test pins were 100 mm in length and 30.5 mm in diameter. The majority of the profiled test pins were made from 4140 steel, although for comparison one test pin each of HI 3 tool steel and titanium alloy Ti-6A1-4V (composition provided in Table 5.3) were used. These two materials were chosen as they had been considered by CAPTIN for use as replacement materials to 4140 steel. The profiled pins were also 100 mm in length. Test conditions for these two test series are provided in Table 5.4 and Table 5.5, respectively.  Weight Percent 6 0.13 4 0.18  Element Al Fe V O  The conditions for the cylindrical pin tests are shown in Table 5.4. The rotation rate for the cylindrical pin tests was set based on the tests performed by Yan and Fan . These tests [18]  were chosen for comparison because the test set-up and conditions were similar. Although the pin material and melt alloy were different, the intent was to provide some means to determine whether the tests performed were inline with those of other researchers. As described before in Section 2.3, Yan and Fan  tl8J  rotated an H21 steel pin in an A380 melt at  700°C. The pin was 10 mm in diameter and the rotation rate used was 300 rpm (0.16 m/s). In the planned profiled tests, the pins were primarily 4140 steel, so it was desirable to use the same steel grade in the cylindrical pin tests. Given the existing test set-up and material on hand, 4140 steel pins of 30.5 mm diameter were used. In order to be able to make comparisons with available research, it was important that the velocity used in the cylindrical 80  pin tests was the same. The work of Yan and Fan  1  J  was selected for comparison because,  although they used H21 steel and aluminum A380, the experimental set-up and test conditions were similar. However, the diameter of their test pins was 10 mm, so the rotation rate was adjusted to account for the diameter difference. The rotation rate was determined to be 98 rpm and was set as the baseline rotation rate. Rotation rates of 0 rpm (static) and 196 rpm (0.31 m/s) were also used. Other tests examined the effect of temperature and rotation rate.  The conditions for the profiled pin tests are shown in Table 5.5. The baseline test was originally planned to be 4140 steel, 150 rpm, with a draft angle of 10°. However, early tests with the profiled 4140 steel pins did not show significant wear so the baseline rotation rate was changed to 200 rpm. Table 5.4: Test Matrix For Cylindrical Test Pins Temperature (°C) 640 670 670 680 690 700 700 700 700 700 700  Time (hr) 4 4 4 4 4 2 4 4 4 8 8  Rotation Rate (rpm) 196 98 196 98 98 98 0 98 196 0 98  Table 5.5: Test Matrix for Profiled Cross-Section Pins Geometry (°) 10 10 10 10 5 5 5 15 15 10 10  Rotation Rate (rpm) 0 100 150 200 150 200 300 150 200* 200 200  Pin Material 4140 steel 4140 steel 4140 steel 4140 steel 4140 steel 4140 steel 4140 steel 4140 steel 4140 steel H13 steel Ti-6A1-4V  81  5.6 Analysis of Pins  After testing, pins were sectioned normal to the axis about 10 mm from pin bottom. The cross-section was mounted in epoxy resin and polished for profile measurement and microscopic analysis. SEM and EDS were used to characterize any intermetallic layers.  Once polished, pins were scanned with an EPSON Perfection 1240U flatbed scanner with an optical resolution of 720 dpi, with a couple of pins re-checked at 1200 dpi. The image analysis software ImageJ™ was used to analyze the scans. For the cylindrical test pins, the locations of five points were recorded along the surface of the pin. Using three points and the intersection of perpendicular bisectors, the centre point of the circle was calculated. To account for skew, the centre point was an average for the five points. The centre of the profiled pins was determined similarly. Knowing the location of the pin centre, the radius was calculated.  For the profiled pins, the radius was plotted all along the perimeter of the pin. In order to make comparisons between pins of different geometry, the radius was normalized with the initial radius. The initial pin radius was calculated from xy data from the pin model.  Intermetallic layer thickness was measured from SEM micrographs. For each cylindrical pin, intermetallic layer thickness was measured at three locations. Given the variation in layer thickness, three measurements were taken at each location. An average was calculated from the nine measurements. Layer thickness was calculated similarly for the profiled pins, although measurements were taken at three locations for both the leading and trailing edges. At one location for the cylindrical pins and at one location on each leading and trailing edges for the profiled pins, elemental composition of the various layers was determined with quantitative EDS.  For both the radius measurement and intermetallic layer thickness, the error measurement was the sum of the standard of deviation and the distance of 2 pixels in both directions.  82  5.7 Summary  A test was needed to examine the erosion-corrosion behaviour of sprue materials in liquid aluminum alloy A356. Other researchers have used the rotating pin test, which incorporates erosion and corrosion effects in a controlled laboratory test. The usual test parameters are temperature, time, rotation rate, melt alloy, pin material (usually steel), and various surface treatment techniques. While these factors will all have an effect on wear behaviour, at this time geometry has not been linked to wear behaviour. Geometry is important as it influences velocity and pressure, which can accelerate die wear. Given that this project was based on the specific problem of accelerated wear of select sprue models, the effect of geometry was necessary to the analysis.  Examination of the location of accelerated sprue wear indicated multi-phase intermetallic compound layers, mainly consisting of aluminum, iron, and silicon. It was difficult to determine the specific binary and ternary phases given the confusion in literature on phase stoichiometry and the fact that these phases seem to coexist.  Subsurface thermocouples, installed in a sprue, were monitored during a casting run to determine the temperature variation within an operational sprue. Temperature measurements indicated that the top of the sprue experienced the greatest variation in temperature, including the highest and lowest temperatures. This can be explained by the pressure release that occurs at the end of the casting cycle, at which time liquid aluminum in the sprue drops down to the holding furnace and the die is opened. Given that the thermocouples were subsurface and that the maximum temperature recorded was around 585°C, it is not unreasonable to expect the sprue surface to reach 630-640°C.  A profiled pin cross-section was developed with a draft angle and radius of curvature based on geometric features from the sprue. Three pin geometries were designed, each with a different draft angle, representing the variation in exit angles among the LP sprues at CAPTIN. A fluid flow simulation was developed to simulate the flow of liquid aluminum  83  along the pin surface during a rotating pin test. This simulation was used to determine the rotation rate needed to have velocities similar to those in the sprue in LPDC.  Two sets of experiments were performed using the rotating pin test. In the first, the pins were cylindrical and all made of 4140 steel, where the effect of temperature, time, and rotation rate on pin wear was examined. The profiled pins were used in the second set of tests, where geometry, rotation rate, and pin material were considered for their effect on wear. The results of these tests are discussed in the next chapter.  84  CHAPTER 6: RESULTS AND DISCUSSION Rotating pin tests were performed to evaluate the erosive-corrosive wear behaviour of molten A356 with different sprue materials under different test conditions. The effects of temperature, time, rotation rate, and, most importantly, geometry on wear were considered. After each test, pins were cross-sectioned, the centre point was determined and used to calculate the radius, as detailed in the previous chapter. The thickness and composition of the intermetallic layers between the aluminum and the pin surface were also determined. This chapter presents these results and discusses their implication on the accelerated sprue wear problem.  ;  6.1 Cylindrical Test Pins Cylindrical rotating pin tests were completed to investigate the effects of temperature, time, and rotation rate on pin wear. Comparisons were made based on radius reduction, layer thickness, and layer composition.  6.1.1 Radius Reduction  Figure 6.1 and Figure 6.2 show the effect of rotation rate, temperature, and time on radius reduction. A significant increase in radius reduction is evident when comparing the static test to the rotation tests. However, only a small increase in radius reduction is seen between 98 to 196 rpm. This difference may become more significant with higher rotation rates. Wear also increased with increasing temperature, as shown in Figure 6.1 for pins rotated at 98 and 196 rpm for 4 hours. Radius reduction increased with temperature for both 98 and 196 rpm, and was slightly higher for 196 rpm.  Figure 6.2 shows how radius reduction increased with time for the 0 and 98 rpm tests. Included for comparison are the results from the rotation tests of Yan and Fan . Yan and [18]  Fan rotated specimens of H21 steel in molten A380 at 700°C for 4 hours. Their tests used a rotation rate of 300 rpm and a smaller pin diameter (10 mm). Accounting for this diameter  85  difference, the equivalent rotation speed for the samples used in the current investigation 98 rpm. 1.8  •0 98 rpm A 196 rpm  1.5  I 1-2 c  o  u  •i 0.9 a> f£ in  • I  i  i  3  ra 0.6 0.3  630  640  650  660  670  680  690  700  710  Temperature (°C)  Figure 6.1: Effect of temperature on radius reduction of cylindrical pins tested for 4 hours.  1.8 • 0 rpm 1.5  • 98 rpm (0.16 m/s) • Yan and Fan (H21 steel, A380,0.16 m/s)  1.2  •O u.o 4)  ft  It)  3  ra 0.6  0.3 H  —i  0  2  4  6  1 8  1 10  1 12  1 14  1 16  18  Time (hours)  Figure 6.2: Radius reduction for cylindrical pins at 700°C, with data from Yan and Fan  [18]  .  There was only a small increase in radius reduction from 4 to 8 hours for the static test, compared to about 0.8 mm at 98 rpm for the same time period. While the radius reduction of the H21 steel pins was less than that measured in the 4140 steel tests, it appears that the trend of increasing radius reduction with increasing rotation rate for the two test series was similar. H21 steel contains more alloying elements and has a higher hardness than 4140 steel, so it was expected that pins would experience less wear than the 4140 steel.  6.1.2 Intermetallic Compound Layers  Figure 6.3 is an SEM micrograph taken in secondary electron (SE) mode showing a representative location along the aluminum-steel interface. Three intermetallic layers (seen in Figure 6.3) were observed in three of the pins tested. The three pins showing three intermetallic layers were tested at 700°C for 4 hours at 98 rpm and 8 hours at 0 and 98 rpm. Only two layers were observed for all the other pins, which were for 4 hours (except 1 pin for two hours) at various temperatures and rotation rates.  Using measurements taken from the SEM micrographs, the effect of rotation rate, temperature and time on intermetallic layer thickness is shown in Figure 6.4 and Figure 6.5. The layer thickness shown is that of all layers. For the static tests, layer thickness increased by about 10 um from 4 to 8 hours. At 98 rpm, there was an increase in layer thickness from 2 to 4 hours, with virtually no change in layer thickness after 4 hours. Layer thickness decreased slightly with increasing rotation rate. Layer thickness increased with increasing temperature as shown in Figure 6.5. For all tests, the radius reduction was greater than the layer thickness.  87  Spalled intermetallic  Solidified A356  Outer Layer Mid Layer Inner Layer 4140 Steel  A'/ >J@© S3  @3  Figure 6.3: Aluminum-steel interface along the surface of a cylindrical 4140 steel test pin tested at 7 0 0 ° C and 98 rpm for 4 hours. Arrows indicate three intermetallic layers.  100 90 _  E 3  80 70  4  60  I _l  •D O) C  3 E o  o  50 40 30 H 20 10  • • • •  670°C, 4 hours 700°C, 2 hours 700-C, 4 hours 700°C, 8 hours  98  196  Rotation Rate (rpm)  Figure 6.4: Effect of rotation rate on layer thickness for cylindrical pins.  88  630  640  650  660  670  680  700  690  710  Temperature (°C)  Figure 6.5: Effect of temperature on layer thickness for cylindrical pins tested for 4 hours.  1200  • FeAl layer 3  • F^Alj layer • FeAfe layer  1000  •g- 800  a Combined  H  a.  M  s •§ eoo  200  8  10  12  14  18  16  Time (hours)  Figure 6.6: Effect of time on intermetallic layer thickness. (Data from Yan and F a n rotated at 300 rpm in A380 at 700°C.)  [18]  for H21 pins  Figure 6.6 shows a breakdown of the three layers observed by Yan and Fan , identified by [18]  quantitative EDS to be FeAb, Fe2Als, and FeAb from A3 80 to H21 steel (described in  89  Section 2.3). The FeAl3 layer accounted for most of the combined layer thickness and significantly increased with longer test times. The combined layer thickness for the H21 steel was much greater than that observed in the current tests which, combined with the smaller radius reduction, may suggest that the higher alloy content of the H21 steel resulted the growth of a thick protective intermetallic layer.  The thicknesses of the three intermetallic layers observed for the cylindrical test pins in the current investigation are shown in Figure 6.7. Contrary to Yan and Fan , no increase in [18]  combined layer thickness was observed after 4 hours. In fact, the thickness of the outer layer decreased and consequently the thickness of the inner layer increased. The mid layer was not observed until 4 hours and did not show any significant change at 8 hours. The large error associated with the outer and inner layer thickness can be attributed to the variation in layer thickness.  60  50  « Outer layer A Mid layer © Inner layer m Combined layers  —40  E  a. •530  i  L.  J  20  10  Time  (hours)  Figure 6.7: Growth of intermetallic layers for cylindrical pins rotated at 98 rpm and 700°C. The elemental composition of the solidified aluminum, intermetallic compound layers, and steel pins were measured using EDS. The outer and inner layers on pins showing two intermetallic layers had the same composition as the outer and inner layers on pins with three  90  layers with the middle layer being very thin and hard to distinguish. The aluminum, iron, and silicon contents of the different regions are summarized in Table 6.1 for the eleven test pins.  Table 6.1: EDS Results from Cylindrical Pin Tests Layer A356 Outer layer Mid layer Inner layer 4140 steel  Aluminum, wt % Min Max Avg 49.4 90.2 96.8 48.2 52.0 49.9 40.2 48.3 45.1 34.7 46.6 41.8 0.0 0.1 0.0  Min 0.7 35.5 47.8 49.6 96.1  Iron, wt % Max Avg 2.8 1.7 43.0 40.2 51.5 49.3 56.9 52.9 97.8 97.0  Silicon, wt % Min Max Avg 36.1 6.1 1.6 3.9 12.7 8.6 6.4 3.8 1.8 6.7 3.4 1.4 0.0 0.7 0.2  The solidified A356 layer mainly consists of aluminum with some variation, which is primarily due to silicon content. This fluctuation in aluminum and silicon content is probably a result of aluminum-silicon precipitation that occurs during solidification. There is also a small amount of iron in the solidified A3 56 layer, which can be attributed to the separation of intermetallic layers from the pin surface and subsequent dissolution into the melt. Aluminum content gradually decreases with increasing distance from the solidified aluminum layer, through the three intermetallic compounds, with virtually no aluminum diffused into the steel pin. Similarly, iron diffused from the steel pin through the intermetallic layers and into the A3 56 layer. Up to 3 wt% iron was measured in the solidified A356 layer. Like aluminum, the silicon content decreases from the solidified aluminum layer through the intermetallic layers. The silicon content in the steel was very small, a maximum of 0.7 wt%, suggesting that almost no silicon had diffused into the steel. The presence of silicon in the steel may also be explained by the fact that the 4140 steel grade contains 0.2 to 0.35 wt% silicon.  The outer intermetallic layer, closest to the solidified aluminum, has average aluminum, iron, and silicon contents of 50, 40, and 9 wt%, respectively. The silicon content in this layer is higher than the initial silicon content of the liquid A356. The composition of this layer does not exactly match any identified binary or ternary phase, however, it is close in composition to the mid layer identified with EDS by Sundqvist and Hogmark to be Fe Al . The [27]  2  5  aluminum and iron contents are 55 and 45 wt%, respectively, for the Fe2Al compound. 5  91  Considering the aluminum to iron ratio of the outer layer, which is approximately 56 to 44, it does appear possible that the layer is FeaAls, in contrast to the outer layers identified by both Sundqvist and Hogmark and Yan and Fan [27]  [18]  to be FeAl3.  The mid layer had average aluminum, iron, and silicon contents of 45, 50, and 4 wt%, respectively. Once again, this does not match any confirmed iron-aluminum or ironaluminum-silicon phase described in the literature, although it was close in composition to the inner layer identified by Sundqvist and Hogmark to be FeA^. The aluminum to iron t27]  ratio of the mid layer was 47 to 53, close to the FeAl2 composition of 49 and 51 wt%, respectively. Once again, this is in contrast to the mid layers identified by both Sundqvist and Hogmark and Yan and Fan [27]  [18]  to be Fe2Al . 5  The inner layer averaged 42 wt% aluminum, 53 wt% iron, and 3 wt% silicon, which was not close in composition or by aluminum-iron ratio to any confirmed phase. From the composition, perhaps this layer is some intermediate binary phase between FeAl2 and FeAl. EDS is not a very reliable phase identification technique because phases often co-exist and it is not possible to separate the different compositions, which may explain the difficulties in identifying the different phases.  The three layers observed in the present analysis did not come close in composition to any confirmed ternary compound in the aluminum-iron-silicon system, primarily because the silicon content was significantly lower. Of all the literature on erosion-corrosion tests reviewed, Yu et al.  [u]  and Chu et al. were the only researchers to identify intermetallic [1]  layers as ternary compounds. Both researchers performed tests with H13 steel in aluminum A390, which has a higher silicon content than A356 and A380, and none of the layer compositions reported matched compositions in this study.  6.2 Profiled Test Pins  This section presents the results from the rotating pin tests using profiled pins. The effects of pin geometry, rotation rate, and material on erosive-corrosive wear were examined. Similar  92  to the cylindrical pins, comparisons were made using radius reduction, layer thickness, and layer composition.  6.2.1 Radius Reduction for 4140 Steel Pins  Figure 6.8, Figure 6.9, and Figure 6.10 show the variation in radius around the 5, 10, and 15° 4140 steel profiled pins after rotating for 4 hours at 700°C. The dashed line indicates the original pin radius. As expected, radius reduction, or wear, increased with rotation rate for each of the pin geometries.  In the case of the 5° pin, shown in Figure 6.8, most wear occurred along the leading edge, with more wear occurring near R2 than Rl. Ridges developed along several of the leading edges (~25, 100°) of the 5° pin rotated at 200 rpm. These ridges can also be observed along two edges of the pin rotated at 300 rpm (~25, 100°), although they were not as pronounced. The wear at 300 rpm appears to be much less uniform, with greater radius reduction at both the Rl and R2 locations.  I i  1  D  30  1  60  1  90  '  120  T  150  1  180  1  210  i  240  1  270  r  300  *. 300 rpm 1  330  1  360  Angle (•)  Figure 6.8: Effect o f rotation rate on wear around 5° pin.  93  Figure 6.9 s h o w s the radius a l o n g the surface o f the 10° p i n rotated at 100, 150, a n d 2 0 0 r p m , as w e l l as one test w h e r e the p i n was not rotated. T h e p i n that w a s not rotated s h o w e d o n l y a s m a l l amount o f w e a r that o c c u r r e d u n i f o r m l y a l o n g the p i n surface. F o r pins rotated at 100 a n d 150 r p m , most w e a r o c c u r r e d a l o n g the t r a i l i n g edge, near R 2 . I n c r e a s i n g the rotation rate to 2 0 0 r p m , there w a s a substantial increase i n the w e a r at R l .  Several ridges  w e r e also o b s e r v e d at this l o c a t i o n .  — - Original  0  30  60  90  120  150  180  210  240  270  300  •  0 rpm  • *  100 rpm 150 rpm 200 rpm  330  360  Angle (•)  Figure 6.9: Effect of rotation rate on wear around 10° pin. T h e radius a r o u n d the 1 5 ° p i n after rotating at 150 and 2 0 0 r p m is s h o w n i n  Figure  6.10.  T w o p i n s w e r e rotated at 2 0 0 r p m to evaluate the r e p r o d u c i b i l i t y o f the test. T h e 1 5 ° p i n rotated at 150 r p m s h o w e d most w e a r o c c u r r i n g at R 2 . Increasing the r o t a t i o n rate to 2 0 0 r p m increased the a m o u n t o f w e a r o c c u r r i n g a l o n g the l e a d i n g edge. S o m e r i d g e s w e r e o b s e r v e d a l o n g the l e a d i n g edge o f the pins rotated at 2 0 0 r p m . O n l y a s m a l l a m o u n t o f w e a r o c c u r r e d a l o n g the t r a i l i n g edge and at R l for a l l three p i n s , perhaps as a result o f the c h a n g i n g f l o w d i r e c t i o n due to the larger draft angle. T h e t w o 15° p i n s tested at 2 0 0 r p m s h o w e d agreed f a i r l y w e l l , c o n s i d e r i n g the d y n a m i c nature o f e r o s i v e - c o r r o s i v e w e a r . It is interesting to note that for the first p i n tested at 200 r p m a c o u p l e o f r i d g e s c a n be o b s e r v e d m i d w a y a l o n g the t r a i l i n g edge w h i l e the ridges i n s e c o n d p i n o c c u r r e d just before R l . S i n c e  94  the second pin showed more wear than the first pin, perhaps this is indicative of how the wear progressed along the leading edge.  18 H 0  i  i  30  60  1 90  1 120  1 150  1 1 180 210 Angle (")  1  240  1 270  • 200 rpm -1 * 200 rpm-2 1 1 1 300 330 360  Figure 6.10: Effect o f rotation rate on wear around 15° pin.  In order to compare the wear experienced by the different pin geometries, the radius measured along the pin surface was normalized with the initial radius. The normalized radius results are presented in Figure 6.11 and Figure 6.12 for the 5, 10, and 15° pins rotated at 150 and 200 rpm, respectively. Lines have been drawn indicating the angular location of Rl and R2 (minimum and maximum radius). For all pins, maximum radius reduction occurred just to the left of the R2 line on the leading edge. Along the trailing edge, most wear occurred within ~20° of R2.  The 15° pin showed the least amount of radius reduction along the trailing edge. The 15° pin showed the least amount of wear overall except at 200 rpm along the leading edge, where it showed a sharp increase in radius reduction. It is more difficult to comment on the wear of the 5 and 10° pins. The two pins showed similar wear at 150 rpm, although perhaps the 5° pin showed a greater reduction in radius at Rl. At 200 rpm, the 10° pin experienced greater wear at R l , which sometimes carried over to the leading edge.  95  0  30  60  90  120  150  180  210  240  270  300  330  360  Angle (•)  Figure 6.11: Plot o f normalized radius for three pin geometries rotated at 150 rpm R l and R 2 locations are indicated with vertical lines. Leading and trailing edges are indicated with L and T.  R,  V  2  R 2  'A  it  / V I  St  0  30  60  90  120  150  180  210  240  u  270  300  330  360  Angle (")  Figure 6.12: Plot o f normalized radius for three pin geometries rotated at 200 rpm R l and R 2 locations are indicated with vertical lines. Leading and trailing edges are indicated with L and T.  96  (c) Figure 6.13: 4140 steel pins tested at 200 rpm and 700°C for 4 hours: (a) 5°, (b) 10°, and (c) 15°. Original pin outline is superimposed on worn pin cross-section. (Scale bars correspond to worn pin cross-sections.) Figure 6.13 shows the scanned cross-sections of the different pin geometries after rotating at 200 rpm for 4 hours. The original pin outline has been positioned over the worn pin crosssections to indicate locations o f increased wear. (Note that the scale o f the original pin outline is not the same as for the worn pin.) The wear appears to be mostly uniform for the 5° pin, except at indicated ridges along the leading edge, located approximately one third o f the way to R 2 . These ridges are observed along the leading edge o f pin tested at 200 and 300 rpm (shown in Figure 6.10). The 10° pin shows increased wear near R l . In these regions o f increased wear the pin surface appears ridged. The ridges are not as w e l l pronounced in the  97  15° pin. However, upon close inspection, it can be seen that after the R l radius position the whole leading edge has flattened, with wear initiating near Rl. This explains the large drop in radius along the leading edge observed for the 15° pins tested at 200 rpm in Figure 6.10.  (c) Figure 6.14: Static pressure plots with vectors indicating direction o f flow: (a) 5° pin, (b) 10° pin, and (c) 15° pin. (Pa)  Figure 6.14 shows predictions of the static pressure field for the three pin geometries rotated at 200 rpm. Vectors indicate the flow direction and in each case show a region near Rl where the vectors are directed into the pin. This region is more pronounced for the 10° and  98  15° pins and corresponds to the region of higher static pressure where the velocity is lower. The increased wear, in each pin shown in Figure 6.13 seems to correspond to the flat surface immediately after where the flow is directed into the pin surface. Along the trailing edge, just after R2, there is region where the flow is directed away or separates from the surface. In Figure 6.12, there was an increase in wear approximately 20° from R2 along the trailing edge, which seems to correspond to the end of this flow separation region. In this flow separation region, the flow is probably more stagnant, which is supported by the thicker intermetallic layer thickness along the trailing edge.  6.2.2 Radius Reduction for Pins of Different Material  This section considers the wear of the 4140 steel, HI 3 steel, and Ti-6A1-4V pins, rotated at 200 rpm and 700°C for 4 hours. HI3 steel and Ti-6A1-4V were chosen for comparison as they had been considered by CAPTIN for use as replacement materials to 4140 steel. The results for the 4140 steel pin were presented in the previous section and are included again here for comparison.  Figure 6.15 shows the variation in radius along the surface of the 10° pins of different material. The same results are also shown in Figure 6.16, normalized by the original radius. All three pins were rotated at 200 rpm for 4 hours at 700°C. The HI3 steel pin showed a significant amount of wear all along the trailing edge, however, the 4140 steel pin overall showed greater radius reduction. When examining the titanium pin after testing, the surface appeared unaffected by wear. The titanium pin showed a small amount of wear at the R2 location only, where velocity was greatest. Near the R l location in the titanium pin, the radius even appears to have increased at a couple of locations, which can probably be attributed to machining.  99  18  H  0  1  30  1  60  1  90  1  1  1  1  1  1  1  1  1  120 150 180 210 240 270 300 330 360 Angle ( )  Figure 6.15: Effect of material on wear around 10° pin at 200 rpm.  Angle (°)  Figure 6.16: Plot of normalized radius for three pin materials rotated at 200 rpm. R l and R2 locations are indicated with vertical lines. Leading and trailing edges are indicated with L and T.  E x a m i n i n g the aluminum-titanium interface under the microscope, a thick intermetallic c o m p o u n d layer was observed.  This layer is shown i n Figure 6.17 at a location near R2.  100  The layer for the most part appeared intact, with only small areas of local cracking, primarily along the leading edge. Figure 6.18 and Figure 6.19 show the interface of the H13 and 4140 steel pins after rotation at 200 rpm for 4 hours. Extensive cracking was observed along the leading edge of both the 4140 and HI 3 steel pins.  Solidified A356  Intermetallic Layer  Ti-6A1-4V  Figure 6.17: Interface of Ti-6A1-4V pin after rotation at 200 rpm and 7 0 0 ° C for 4 hours.  Figure 6.18: Interface of H13 steel pin after rotation at 200 rpm and 7 0 0 ° C for 4 hours.  101  Solidified A356  Layers  4140 steel  Figure 6.19: Interface of 4140 steel pin after rotation at 200 rpm and 700°C for 4 hours.  6.2.3 Intermetallic Compound Layers  This section presents the results of layer thickness and composition measurements for the leading and trailing edges of the profiled cross-section pins. Table 6.2 and Table 6.3 provide information on the intermetallic layer thickness for the leading and trailing edges of the 10° 4140 steel pins. The intermetallic thickness was measured at three locations along both the leading and trailing edges, with an average of three measurements at each location. Three layers were observed along the interface of all the 4140 steel pins with one exception. Figure 6.20 shows an SEM micrograph that is representative of an interface with three intermetallic layers. Layer thickness was always smaller along the leading edge than the trailing edge, which corresponded to shear flow conditions. The only exception to this was the 10° pin used in the static test.  102  Table 6.2: Layer Thickness Measurements for Leading Edge of 4140 Steel Profiled Pins Pin  5°  10°  15°  Rotation Rate  Outer  Mid  150 rpm  14  16  200 rpm  17  5  300 rpm  14  15  61  0 rpm  12  4  45  100 rpm  17  3  32  150 rpm  7  3  200 rpm  26  150 rpm  19  200 rpm  17  200 rpm  29  Layer Thickness (nm) Inner  Error (um)  Total  Outer  Mid  Inner  0  30  4  2  0  7  22  44  6  1  5  12  91  2  2  8  13  61  4  3  5  12  52  10  1  5  15  32  43  2  1  3  6  6  13  46  4  2  5  11  36  9  63  4  1  1  6  7  14  38  4  2  4  10  4  11  43  5  1  2  8  Total  Table 6.3 : Layer Thickness Measurements for Trailing Edge of 4140 Steel Profiled Pins Pin  Rotation Rate  Outer  Mid  Inner  4  200 rpm  15 21  300 rpm 0 rpm  150 rpm 5°  10°  15°  Layer Thickness (um)  Error (um)  Total  Outer  Mid  Inner  Total  109  128  4  10  135  163  5 4  1  7  2  8  15  30  18  83  132  12  3  19  35  12  4  45  61  4  3  5  12  100 rpm  18  3  74  95  5  1  6  12  150 rpm  22  32  50  104  7  2  6  15  200 rpm  24  7  84  115  4  2  6  13  150 rpm  19  37  72  129  3  1  5  9  200 rpm  21  11  135  167  6  3  7  17  200 rpm  43  15  106  165  10  3  17  30  Figure 6.20: Representative aluminum-steel interface along the surface of the profiled 4140 steel test pins showing three intermetallic layers. 103  For the 10° and 15° pins, layer thickness increased along the trailing edge with increasing rotation rate. No trend was observed for the 5° pins along the trailing edge. Along the leading edge, intermetallic layer thickness increased with increasing rotation rate for the 5° pin. Layer thickness along the leading edge mostly decreased with increasing rotation rate for the 10° and 15° pins. The increase in layer thickness along the trailing edge with rotation rate was probably due to the increased flow separation that occurs at higher rotation rates. Radius reduction was highest along the leading edge where the flow was parallel to the surface. It would be expected that high drag forces would accelerate the removal of the intermetallic layers, which agrees with the measurements along the leading edge of the 10° and 15° pins.  At 150 rpm, layer thickness along the leading edge increased with pin angle. There was no trend to layer thickness among the three pins along the leading edge at 200 rpm or the trailing edge at any rotation rate.  Table 6.4 presents the aluminum, iron, and silicon content determined by EDS for the profiled 4140 steel pins. Similar to the cylindrical test pins, three intermetallic layers were observed on almost all of the pins. The composition of the three layers was nearly identical to the composition of the layers on the cylindrical test pins.  Table 6.4: EDS Results from 4140 Steel Profiled Pin Tests Layer A356 Outer layer Mid layer Inner layer 4140 steel  Aluminum, wt % Min Max Avg 66.8 96.5 90.0 46.3 53.9 50.0 49.2 47.2 45.6 43.2 41.3 45.0 0.2 0.0 0.0  Iron, wt % Min Max Avg 0.7 15.8 2.8 35.2 46.1 40.1 46.7 50.2 47.9 51.8 54.0 53.0 88.3 98.1 97.2  Silicon, wt % Min Max Avg 6.1 31.5 1.2 9.2 4.9 11.4 2.6 6.4 4.0 2.5 3.8 1.4 0.0 0.6 0.2  Table 6.5 and Table 6.6 compare the thickness of the intermetallic layers observed for the 4140 steel, HI3 steel, and Ti-6A1-4V pins along the leading and trailing edges. Two layers were observed along the interface of the H13 steel pin (Figure 6.18) and one layer on the titanium pin (Figure 6.17). The titanium pin had the thickest layer along the leading edge,  104  followed by the combined layers along the HI3 steel. Along the trailing edge, the layers on the HI3 steel pin were the thickest, followed by the 4140 steel.  The elemental compositions of the HI 3 steel and Ti-6A1-4V pins are provided in Table 6 . 7 and Table 6 . 8 . Similar to the 4140 steel pins, aluminum content was highest in the solidified aluminum layer, decreasing through the intermetallic layers. The compositions of the two intermetallic layers along the HI3 steel did not match any confirmed binary or ternary phase, instead the phases seemed to be intermediate in composition. The layer on the Ti-6A1-4V pin has a 48 to 52 ratio of titanium to aluminum in the interface layer, perhaps suggesting the layer is TiAl2. The aluminum content in TiA^ ranges from 51 to 54 wt% . [37]  Table 6.5: Layer Thickness Measurements for Leading Edge for 10° Profiled Pins of Different Material Pin  Layer Thickness (um)  Error (um)  Material  Outer  Mid  Inner  Total  Outer  Mid  Inner  Total  4140 steel  21  6  20  48  4  2  5  11  H13 steel  48  9  0  56  4  2  0  6  Ti-6A1-4V  69  0  0  69  2  0  0  2  Table 6.6: Layer Thickness Measurements for Trailing Edge for 10° Profiled Pins of Different Material Layer Thickness ( u m )  Pin Material  Outer  4140 steel H13 steel Ti-6A1-4V  Error ( u m )  Mid  Inner  Total  Outer  Mid  Inner  Total  24  7  84  115  4  99  36  0  135  15  2  6  13  3  0  18  100  0  0  100  5  0  0  5  Table 6.7: Composition of Layers Observed Along Interface of HI 3 Steel Pin (wt%) Layer A356 Outer layer Inner layer H13 steel  Aluminum 77.0 52.1 43.2 0.0  Iron 10.6 32.9 50.0 90.1  Silicon 4.2 9.1 3.4 0.8  Table 6.8: Composition of Layers Observed Along Interface of TJ-6A1-4V Pin (wt%) Layer Aluminum Titanium Silicon A356 78.7 5.7 14.5 Layer Ti-6A1-4V  46.3 4.2  43.3 91.9  9.1 0.0  105  6.3 Comparison of Profiled Pins with Sprue Profiled pin geometries were designed to reproduce the flow conditions in the exit region of the sprue. Three different pin geometries were developed to encompass the range of exit angles for the 10 different sprue models. All three pin geometries had a curvature of 10 mm, same as in the LP designs (shown in Table 5.2). The intent of the profiled pin geometry was that the flow of liquid aluminum around the R2 location would be similar to the flow along the exit surface of the sprues, and consequently produce similar accelerated wear behaviour. This is illustrated in Figure 6.21 with the superpositioning of 5° pin geometry over the exit surface of the 330N wheel model sprue. (Note that the pin is rotating in the opposite direction to that in the model and experiments.) An angle of 20° has been indicated on the pin, representing the location on the trailing edge where increased wear was observed.  The accelerated wear observed along the sprue exit surface in some locations resulted in significant wall thinning and in others deep localized ridges. Figure 6.22 shows a crosssection from the 330N sprue previously shown in Figure 5.1(b). Arrows indicate the location of the constriction point and the location of the maximum wear. From the appearance of the constriction point, it appears as if accelerated wear started before the constriction point and progressed similarly to the wear observed on profiled test pins. However, this would need to be confirmed by examining more worn sprues to determine where accelerated wear initiates and how it propagates along the exit surface.  Point where vectors becc parallel with sprue surf;  FldU |— Direction I Figure 6.21: 5° pin superimposed on 330N sprue with angle indicating region where most wear occurs on trailing edge and arrow indicating where vectors become parallel with sprue again. 1  106  Accelerated wear  Constriction point  5  mm  Figure 6.22: Close-up of the exit region of a cross-sectioned 330N sprue. Arrows indicate the constriction point and location of accelerated wear. The contour plot in Figure 6.23 shows the velocity in the exit region of the 330N sprue, with a magnified view of the velocity vectors along the constriction point. Vectors are parallel to the sprue surface as they move up the sprue, except around the constriction point in the area indicated in Figure 6.23. The flow is directed away from the sprue surface just before the constriction point until ~3 mm above the constriction point. This point has been indicated with an arrow in Figure 6.21, and corresponds to the section of the 5° pin trailing edge where increased wear was observed. While it is not possible to determine where accelerated wear initiated in Figure 6.22, it is clear that this section of the surface has been heavily worn.  Velocity vectors were plotted for a 5° pin rotated at 300 rpm in Figure 6.24, indicating a region where the vectors are directed away from the pin surface similar to the sprue surface. In this region, it would be expected that the flow would be more stagnant, which is supported by the greater intermetallic layer thickness along the trailing edge. With increasing pin draft angle, this flow separation region increases (Figure 6.14), which may have implications on the accelerated sprue wear problem.  107  Figure 6.23: Velocity magnitude (m/s) from 330N modified spoke model at 14 seconds.  • . * Molten  Figure 6.24: Velocity vector plot for 5° pin geometry rotated at 300 rpm. (m/s) 6.4 Summary  Initial tests involved rotating 4140 steel cylindrical pins in liquid A356 to examine the effects of rotation rate, temperature, and time on pin wear. So as to compare to the results of Yan and Fan , a baseline rotation rate of 98 rpm was used, giving the same velocity along the pin [1]  surface. Wear was measured through radius reduction and intermetallic layer measurements. There was a significant increase in radius reduction and decrease in intermetallic layer thickness between the static immersion tests and the tests where the pins were rotated, 108  were rotated, however, there was little variation between the 98 and 196 rpm tests. Radius reduction and intermetallic layer thickness increased with increasing temperature and time. The radius reduction and intermetallic layer thickness of 4140 steel pins tested at 700°C and 98 rpm were compared to the results of Yan and Fan , who performed tests with H21 steel tl8]  pins in A380 under similar test conditions. The radius reduction experienced by the 4140 steel pins was larger than that of the H21 steel pins, but did show a similar trend. Intermetallic layer thickness was also smaller than that reported for the H21 steel pins.  Profiled pins made from 4140 steel were tested to examine the effect of flow through pin geometry on wear magnitude and location. Significant wear was not observed until 200 rpm. For the 5° and 15° pins, increased wear occurred primarily along the leading edge, coinciding with the location of flow impingement, where simulation indicates that the flow vectors were directed into the pin surface. For the 10° pin, increased wear seemed to be centered around R l , with fairly uniform wear along the leading edge. Tests were also performed with two pins of different material, HI 3 steel and titanium alloy Ti-6A1-4V. These two materials have been considered by CAPTIN for use as replacement materials to the 4140 steel. As expected, the titanium pin outperformed the two steel pins, forming a thick intermetallic layer along the interface. The steel pins also formed intermetallic layers of similar thickness, however, the layers were cracked, particularly along the leading edge.  There were some trends in intermetallic layer thickness along the leading and trailing edge of the profiled pins. Layer thickness increased with increasing rotation rate for the 10° and 15° pins due to the flow separation occurring on the trailing edge. Along the leading edge, layer thickness mostly decreased for the 10° and 15° pins with increasing rotation rate. In contrast, layer thickness increased along the leading edge of the 5° pin with increasing rotation rate.  For both the cylindrical and profiled 4140 steel pins, three layers were present along the aluminum-steel interface. While the mid layer could not be distinguished in some pins, the elemental compositions of the outer and inner layers suggest that the mid layer is present. The elemental compositions of the three layers for the cylindrical and profiled test pins agreed very well. The composition of these phases did not exactly match any confirmed  109  binary or ternary phases, however, two of the layers were close in composition to layers identified by Sundqvist and Hogmark to be FeaAls and FeAl2. Two layers were present at [27]  the interface of the HI 3 steel pin and one layer at the interface of the Ti-6A1-4V pin. The ratio of titanium to aluminum would seem to suggest that the layer is TiAb.  The erosion-corrosion tests with the profiled pins showed how pin angle influenced the location of wear. Although most wear occurred along the leading edge of the rotating pins, the trailing edge more closely resembles the exit region of the sprue. Most of the wear that occurred on the trailing edge occurred within 20° of the R2 location. Using the filling and rotating pin simulations developed in previous chapters, it was shown that in this area along both the pin and sprue geometries, the flow is directed away from the pin/sprue surface. In this region, the flow may be more stagnant, resulting in less intermetallic layer removal. This was supported by the greater intermetallic layer thickness along the trailing edge.  For the 10° and 15° pins, most of the wear was centered around R l . While the sprue exit surface more closely represents the 5° pin, the wear occurring in the R l region may explain how accelerated wear proceeds. The 10° pin at 200 rpm showed significant wear at R l , which resulted in the formation of deep ridges. With the 15° pin, as a result of the change in flow direction from R l to the straight section along the leading edge, the leading edge was flattened. This effect was more pronounced with increasing velocity. In the 330N wheel model received from CAPTIN, areas of localized wear and significant wall thinning were observed. While the sprue exit surface is a flat surface, if it is exposed to liquid aluminum and a cavity forms, the wear observed for the 10° and 15° pins may be indicative of how the wear proceeds.  110  CHAPTER 7: SUMMARY AND CONCLUSIONS  The main contribution of this work to current erosive-corrosive wear research was in the incorporation of geometry to a laboratory erosion-corrosion test. This was important as industrial trials are expensive and given the dynamics of the process, a large number of runs would have to be performed to obtain reliable information. Laboratory experiments provide increased controllability and allow for the evaluation of a wide range of parameters.  The following points summarize the main conclusions of this work: 1. Wear increases with increasing velocity, temperature, and time. 2. For the cylindrical pins, intermetallic layer thickness decreased with increasing velocity and decreasing temperature. Only a small increase in wear was observed from 98 to 196 rpm (0.16 to 0.31 m/s). The difference is probably more pronounced with increased velocity. 3. Using a profiled cross-section pin in the rotating pin test produced variations in wear behaviour around the pin perimeter that were not observed with the cylindrical pins. 4. For the profiled test pins, intermetallic layer thickness was greater along the trailing edge than the leading edge. Along the leading and trailing edges of the 10° and 15° pins, layer thickness decreased and increased, respectively, with increasing velocity. Along the 5° pin leading edge, layer thickness increased with increasing velocity. 5. Three layers were observed for nearly all the cylindrical and profiled test pins, with similar average aluminum, iron, and silicon contents. The compositions of the three layers could not be conclusively identified, however, from the ratio of aluminum to iron and the composition of layers identified by other researchers, it is suggested that the outer and mid layers were Fe2Als and FeA^. 6. The titanium alloy Ti-6A1-4V profiled pin showed better wear resistance than either of the steel alloys. Two intermetallic layers were observed along the interface of the H13 steel pin and one layer along the interface of the Ti-6A1-4V pin. 7. Comparing the liquid aluminum flow predicted around the sprue constriction point and along the trailing edge of the rotating pin, both show a region where the flow is separates from the sprue/pin surface.  111  Overall, the results of this study agree with other literature that wear increases with increasing velocity, temperature, and time. The surface of the profiled pin showed variations in wear behaviour that were not observed with the cylindrical pins. For all three geometries, the greatest amount of wear occurred along the leading edge, which in the fluid flow simulation corresponded to a region where the flow is directed into the pin surface. This area of increased wear shifted with different pin geometry.  The exit surface of the sprue which exhibits increased wear is best represented by the pin trailing edge, however, this edge experienced less wear than along the leading edge. Fluid flow simulation indicated a region along the trailing edge where the flow separatedfromthe pin surface. The greater intermetallic layer thickness along this edge suggests that the flow may be more stagnant in this region, resulting in less layer removal and less wear. Of the three pin geometries, the 15° pin experienced the least amount of wear along the trailing edge and showed the largest flow separation region. This would seem to suggest that a larger exit angle might reduce the amount of wear occurring on the surface immediately after the constriction point. In reviewing the wear conditions and geometries of the different sprues, it was observed that accelerated wear was reported to consistently occur on sprues with smaller exit angle, with only one exception. Given that the 15° pin, the largest draft angle of the three geometries, showed the least amount of wear along the trailing edge, this would suggest that CAPTIN should try to increase the exit angle on problematic sprues.  7.1 Recommendations for Future Work  To fully characterize the accelerated wear problem, more tests with pins of profiled crosssection need to be performed. These additional tests should be performed at higher rotation rates, similar to the peak velocities predicted in the wheel filling models. More tests should also be performed at similar test conditions to verify test reproducibility.  Similarly, more sprues (preferably after different numbers of shots) should be examined to better characterize the location and progression of accelerated wear along the exit surface.  112  While this would be disruptive to the casting process, the information that this would provide may be important in understanding the mechanisms of accelerated wear.  The die filling simulation did not include the effects of heat transfer, solidification, or turbulence. Including these effects in the filling model would aid in understanding the accelerated wear problem, as well as other types of casting defect. The filling simulation included both the sprue and wheel geometries. However, perhaps the mesh resolution in the sprue exit region was not sufficiently small enough to resolve the pattern of flow (such as any regions of flow recirculation) along the sprue exit surface. The information obtained in the sprue-wheel filling model could be used to develop a sub-scale model along the sprue exit surface and provide improved flow resolution.  Lastly, future study could include investigation of the mechanisms of intermetallic layer cracking. The intermetallic layer along the titanium pin surface was substantially less cracked than layers observed along the steel pins. Studying how the cracks in the intermetallic layers develop and why the iron-aluminum layers seem to be more susceptible to cracking could provide insight into the wear mechanism.  113  Bibliography [I]  J.V. Busch, "Wheelmaking Technologies: A Comparison of Their Economics and Aesthetics," presented at AFS 5th International Conference on Permanent Mold Casting of Aluminum, Milwaukee, WI, 2000.  [2]  R. Shivpuri, Y.-L. Chu, K. Venkatesan, J.R. Conrad, K. Sridharan, and M. Shamim, "An evaluation of metallic coatings for erosive wear resistance in die casting applications," Wear, vol. 192, pp. 49-55, 1996.  [3]  The North American Die Casting Association, "A Study in Aluminum Die Casting: Housing for a Thermoelectric Fan", 10 Apr 2005; http://www.diecasting.org/design/case2/TFAN17.htm.  [4]  T.V. Philip, "Medium-Carbon Low-Alloy Steels", ASM Handbooks Online, ASM International, 20 Dec 2004; http://products.asminternational.org/hbk/index.jsp/.  [5]  A.M. Bayer and L.R. Walton, "Wrought Tool Steels", ASM Handbooks Online, ASM International, 25 Jan. 2005; http://products.asminternational.org/hbk/index.jsp/.  [6]  A.L. Kearney, "Properties of Cast Aluminum Alloys", ASM Handbooks Online, ASM International, 5 Dec 2004; http://products. asminternational. org/hbk/index.j sp.  [7]  Y.-L. Chu, P.S. Cheng, and R. Shivpuri, "Soldering Phenomenon in Aluminum Die Casting: Possible Causes and Cures," presented at 17th International Die Casting Congress and Exposition, Cleveland, Ohio, 1993.  [8]  A.G. Guy, Introduction to Materials Science, First ed. New York: McGraw-Hill Book Company, 1972.  [9]  M.I. Marek, "Corrosion in Molten Salts and Liquid Metals", ASM Handbooks Online, ASM International, 17 Jan 2005; http://products.asminternational.org/hbk/index.isp/.  [10]  Q. Han and S. Viswanathan, "Analysis of the Mechanism of Die Soldering in Aluminum Die Casting," Metallurgical and Materials Transactions, vol. 34A, pp. 139-146, 2003.  [II]  P.F. Tortorelli, "Corrosion Reactions in Liquid-Metal Environments", ASM Handbooks Online, ASM International, 5 Dec 2004; http://products.asminternational.org/hbk/index.isp.  114  [12]  L.-A. Norstrom, B. Klarenfjord, and M. Svensson, "General Aspects on "Washout" Mechanisms in Aluminum Die Casting Dies"," Transactions of the 17th International Die Casting Congress, NADCA, pp. T93-075, 1993.  [13]  M. Yu, R. Shivpuri, and R.A. Rapp, "Effects of Molten Aluminum on H13 Dies and Coatings," Journal ofMaterials Engineering and Performance, vol. 4, pp. 175-181, 1995.  [14]  M. Yan and Z. Fan, "Review: Durability of Materials in Molten Aluminum Alloys," Journal ofMaterials Science, vol. 36, pp. 285-295, 2001.  [15]  S. Shankar and D. Apelian, "Die Soldering: Mechanism of the Interface Reaction between Molten Aluminum Alloy and Tool Steel," Metallurgical and Materials Transactions, vol. 33B, pp. 465-476, 2002.  [16]  V.I. Dybkov, Reaction Diffusion and Solid State Chemical Kinetics. Kyiv: IPMS Publications, 2002.  [ 17]  J.R. Davis, "Aluminum and Aluminum Alloys," in ASM Specialty Handbook. Materials Park, OH: ASM International, 1993, pp. 32-33.  [18]  M. Yan and Z. Fan, "The Erosion of H21 Tool Steel in Molten A380 Alloy," Journal of Materials Science, vol. 35, pp. 1661-1667, 2000.  [19]  S. Shankar and D. Apelian, "Mechanism and Preventative Measures for Die Soldering during Al Casting in a Ferrous Mold," JOM, vol. 54, pp. 47-54, 2002.  [20]  V.G. Rivlin and G.V. Raynor, "Phase Equilibria in Iron Ternary Alloys. IV. Critical Evaluation of Constitution of Aluminum-Iron-Silicon System," International Metals Reviews, vol. 3, pp. 133-152,1981.  [21]  R.W. Richards, R.D. Jones, P.D. Clements, and H. Clarke, "Metallurgy of Continuous Hot Dip Aluminising," International Materials Reviews, vol. 39, pp. 191-212, 1994.  [22]  K. Bouche, F. Barbier, and A. Coulet, "Intermetallic compound layer growth between solid iron and molten aluminum," Materials Science and Engineering, vol. A249, pp. 167-175, 1998.  [23]  H.R. Shahverdi, M.R. Ghomashchi, S. Shabestari, and J. Hejazi, "Kinetics of interfacial reaction between solid iron and molten aluminium," Journal of Materials Science, vol. 37, pp. 1061-1066, 2002.  115  [24]  K. Venkatesan and R. Shivpuri, "Experimental and Numerical Investigation of the Effect of Process Parameters on the Erosive Wear of Die Casting Dies," Journal of Materials Engineering and Performance, vol. 4, pp. 166-174, 1995.  [25]  R. Shivpuri, M. Yu, K. Venkatesan, and Y.-L. Chu, "A Study of Erosion in Die Casting Dies by a Multiple Pin Accelerated Erosion Test," Journal of Materials Engineering and Performance, vol. 4, pp. 145-153, 1995.  [26]  Z.W. Chen and M.Z. Jahedi, "Die erosion and its effect on soldering formation in high pressure die casting of aluminum alloys," Materials and Design, vol. 20, pp. 303-309, 1999.  [27]  M. Sundqvist and S. Hogmark, "Effects of liquid aluminium on hot-work tool steel," Tribology International, vol. 26, pp. 129-134, 1993.  [28]  K.-H. Law, "Corrosion protection for die casting sprue plates." Delta, BC: Canadian Autoparts Toyota, Inc., 2002.  [29]  S. Gopal, A. Lakare, and R. Shivpuri, "Evaluation of Thin Coatings for ErosiveCorrosive Wear Prevention in Die Casting Dies," Surface Engineering, vol. 15, pp. 297-300, 1999.  [30]  S. Shankar and D. Apelian, "Die soldering: Effect of process parameters and alloy characteristics on soldering in the pressure die casting process," International Journal of Cast Metals Research, vol. 15, pp. 103-116, 2002.  [31]  K. Venkatesan and R. Shivpuri, "Numerical Simulation of Die Cavity Filling in Die Castings and an Evaluation of Process Parameters on Die Wear," Transactions of the 17th International Die Casting Congress, NADCA, 1993.  [32]  "Fluent 6.1 Documentation." Lebanon, NH: Fluent, Inc., 2003.  [33]  J.P. Anson, R.A.L. Drew, and J.E. Gruzleski, "The Surface Tension of Molten Aluminum and Al-Si-Mg Alloy under Vacuum and Hydrogen Atmospheres," Metallurgical and Materials Transactions, vol. 30B, pp. 1027-1032, 1999.  [34]  D.R. Poirier and G.H. Geiger, Transport Phenomena in Materials Processing. Warrendale: The Minerals, Metals & Materials Society, 1994.  [35]  K.C. Mills, "A1-LM25," in Recommended Values ofThermophysical Properties for Selected Commercial Alloys. Cambridge: Woodhead Publishing Ltd., 2002.  116  D. Eylon, J.R. Newman, and J.K. Thome, "Titanium and Titanium Alloys", ASM Handbooks Online, ASM International, 31 Mar 2005; http://products.asmintemational.org/hbk/index.jsp/. J.L. Murray, "Al-Ti (Aluminum - Titanium)", ASM Handbooks Online, ASM International, 1 Apr 2005; http://products.asmintemational.org/hbk/index.jsp/.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078774/manifest

Comment

Related Items