UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The detection of mould-strand interaction employing load cells in the continuous casting of billets Brendzy, J. Lorraine 1990

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

Item Metadata


831-ubc_1991_0516.pdf [ 7.08MB ]
JSON: 831-1.0078579.json
JSON-LD: 831-1.0078579-ld.json
RDF/XML (Pretty): 831-1.0078579-rdf.xml
RDF/JSON: 831-1.0078579-rdf.json
Turtle: 831-1.0078579-turtle.txt
N-Triples: 831-1.0078579-rdf-ntriples.txt
Original Record: 831-1.0078579-source.json
Full Text

Full Text

THE DETECTION OF MOULD-STRAND INTERACTIONEMPLOYING LOAD CELLS IN THE CONTINUOUS CASTING OFBILLETSBYJ. LORRAINE BRENDZYB.ASc (Metals and Materials Engineering)University of British ColumbiaVancouver, British Columbia1987A THESIS SUBMI1TED IN PARTiAL FULFILLMENT OF THE REQUIREMENTS FOR THEDEGREE OFMASTER OF APPLIED SCIENCEinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF METALS AND MATERIALS ENGINEERINGWe accept this thesis as conformingt, the required standardTHE UNIVERSITY OF BRITISH COLUMBIAJanuary 1990J. Lorraine BrendzyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of aínoL ‘7fl cxb’The University of British ColumbiaVancouver, CanadaDate 1t&(a’iP) /99c)DE-6 (2/88)iiABSTRACTInstrumentation of the mould with load cells and linear variabledisplacement transducers was completed to evaluate mould-strand interaction ina continuous casting machine for steel billets. A plant trial was conducted inwhich several lubricants were tested at four rates. Three carbon grades werecast under similar mould lubricating conditions. In the analysis of theaccumulated data, correlation between the mould loading as determined by theload cell signals and the oscillation cycle as indicated by the linear variabledisplacement transducers was achieved. The load cell data indicated that thereare two modes of strand-mould interaction occurring during each oscillationcycle. During the positive-strip period, lubrication conditions and carbon contentwere observed to affect the loading of the mould but during the negative-stripperiod, the effects of lubrication and carbon are masked due to the mould bearingdown on the strand. During the period which the mould travels downward fasterthan the strand, the load cell response indicated a smooth decrease in mouldloading. It is also shown that the minimum load reached is related with changesin casting speed: as casting speed decreases, the mould attempts to overtakethe strand to a greater extent than when the casting speed is higher. The resultis a greater decompression of the load cell when casting speeds are slower. Thisfinding supports a proposed mechanism for the formation of oscillation marks inbillets. It is shown that different lubricants produce different loading responsesand that a reduction in flow rate causes increases in load on the mould duringupstroke. A theoretical analysis employing a fluid flow model has also beenundertaken to determine the presence of lubricant at the meniscus during anoscillation cycle. The results from this analysis indicates that lubricant reachesthe meniscus generally during the downstroke and that replenishment of lubricantat the meniscus does not occur during the upstroke leaving only residuallubricant to provide lubrication at the mould-strand interlace. This findingsuggests that two modes of lubrication may operate: hydrodynamic lubricationoccurring when lubricant is present at the meniscus and boundary layer or solidlayer lubrication operating when bulk lubricant does not reach the meniscusregion. From the load cell response, binding was evident which was correlatedwith the presence of depressions found on the surface of examined billets.iiiivTABLE OF CONTENTSPAGEABSTRACT IITABLE OF CONTENTS ivLIST OF TABLES ixLIST OF FIGURES xiiLIST OF NOMENCLATURE xxACKNOWLEDGEMENT xxiiCHAPTER 1 - INTRODUCTION 1CHAPTER 2- LITERATURE REVIEW 52.1 Introduction 52.2 Measunng Devices for Mould-Strand Interaction 62.2.1 Accelerometers 62.2.2 Strain Gauges 72.2.3 Load Cells 132.3 Oscillation Mark Formation and Mould-Strand Interaction 212.4 Billet Defects and Mould-Strand Interaction 292.4.1 Transverse Depression 292.4.2 Longitudinal Corner Cracking and Romboidity 302.4.3 Off-Corner Internal Cracks 332.4.4 Surface Roughness 342.4.5 Pinholes 34V2.5 Lubrication Oils 352.5.1 Lubricating Mechanisms 372.5.1.1 Hydrodynamic lubrication 372.5.1.2 Boundary lubrication 392. Fatty acids 402. Oxidation of fatty acids 402. Fatty acids and lubrication 442.5.1.3 Mixed lubrication 462.5.1.4 Extreme-Pressure (EP) lubrication 472.5.1.5 Elastohydrodynamic/Plastohydrodynamiclubrication 472.5.1.6 Solid-film lubrication 47CHAPTER 3- SCOPE AND OBJECTIVES 49CHAPTER 4- EXPERIMENTAL PROCEDURE 514.1 Industrial Operations 514.1.1 Details of casters 514.1.2 Mould assembly 524.1.2.1 Western Canada Steel 524.1.2.2 Ivaco Rolling Mills 524.1.3 Lubrication Distribution System 584.2 Mould Instrumentation 604.2.1 Load Cells 604.2.2 Linear Variable Displacement Transducers 674.2.3 Thermocouples 674.2.4 Casting speed 704.3 Data Acquisition System 704.4 Lubricating Oils 744.5 Billet Samples 74vi4.6 Plant Trials 754.6.1 Trial at Western Canada Steel 754.6.2 Procedural Changes 754.6.3 Trial at lvaco Rolling Mills 76CHAPTER 5- RESULTS OF INDUSTRIAL TRIALS 865.1 Western Canada Steel Results 875.2 lvaco Results 925.2.1 Analysis of Lubricants 925.2.1.1 Flash points 925.2.1.2. Viscosities 965.2.1.3 Fatty acid content 995.2.2 Load Cell Results 1005.2.2.1 Load cell profiles 1005.2.2.2 Load cell profiles and oscillation cycle 1025.2.2.3 Minimum load and casting speed 1065.2.2.4 The influence of lubricant type,f low rate,and carbon content on load cellresponse 1215.2.2.5 Qualitative and quantitative analysisof load cell response 1345.2.2.6 Peak-to-peak change in rear load cells 1575.2.2.7 Frequency analysis 1615.2.2.8 Billet quality analysis 1615.3 Lubrication of Meniscus During the Oscillation Cycle 195viiCHAPTER 6- DISCUSSION 2166.1 Introduction 2166.2 Mould-Strand Interaction 2196.2.1 Mould-Billet Dynamics 2196.2.2 Net Force Between Billet and Mould 2236.2.3 Load Cell Decompression 2236.2.3.1 Point of initiation of the downwardInteraction of the mould on the strand 2276.2.3.2 Time Delay Between Minimum Loadand Maximum Downward Velocity 2306.2.3.3 Severity of load cell decompressionas a function of casting speed 2316.3 Lubrication 2336.3.1 The Effect of Changing Acceleration on Total Load 2346.3.2 Shear Stresses at the Meniscus 2366.3.3 Mould Lubrication 2376.3.3.1 Mould lubricant type 2386.3.3.2 Mould lubricant flow rate 2396.4 Depressions 2436.4.1 The Effect of Carbon Content on LoadCell Response 2436.4.2 Depressions and the Load Cell Signal 244viiiCHAPTER 7- SUMMARY AND CONCLUSIONS 2477.1 Summary 2477.2 Conclusions 2497.3 Future Work 252REFERENCES 253APPENDIX A Load Partitioning of External Forces 257APPENDIX B Minimum Load Versus Casting Speed for MouldLubricants A and C 260APPENDIX C ANOVA: A Statistical Comparison of Averagesand Standard Deviations for Multi-samplesDrawn from Different Populations 274ixLIST OF TABLESTable 2.1: Melting and boiling points for some fatty acids. 42Table 2.2: Relative rates of oxidation of unsaturated fatty acids. 43Table 4.1: Details of the caster and mould parameters duringthe plant trials at Western Canada Steel and IvacoRolling Mills. 53Table 4.2: Parabolic tapers of the curved sides of the mouldtube employed in the Ivaco plant trial. 54Table 4.3: Parabolic tapers of the straight sides of the mouldtube employed in the Ivaco plant trial. 55Table 4.4: Steel compositions of the heats monitored at WCS. 76Table 4.5: Conditions employed for data acquisition during theWestern Canada Steel plant trial. 77Table 4.6: Composition of heats monitored during the Ivacoplant trial. 79Table 4.7: Casting conditions of heats monitored at Ivaco planttrial. 80Table 4.8: Lubricating conditions for runs at Ivaco planttrial. 83Table 4.9: List of billets collected during the Ivaco planttrial. 84Table 5.1: Peak-to-peak averages and standard deviations forthe Western Canada Steel plant trial results. 89Table 5.2: Flash points of the three mould lubricants testedat lvaco. 93Table 5.3: Flash points as determined by the Pensky-Martensmethod. 96Table 5.4: Kinematic viscosities of the three mould oilstested at the Ivaco plant trial. 97Table 5.5: Fatty acid content of the oils used in the Ivacoplant trial. 99Table 5.6: Linear regression results for minimum load andcasting speed. 117xTable 5.7: Qualitative classification of load cell outputdunng the upstroke for Lubricant A mould oil.The carbon content is 0.09%. 140Table 5.8: Qualitative classification of load cell outputduring the upstroke for Lubricant B mould oil.The carbon content is 0.12%. 142Table 5.9: Qualitative classification of load cell outputdunng the upstroke for Lubricant C mould oil.The carbon content is 0.15%. 143Table 5.10: Quantitative analysis of upstroke period forLubricant A. 146Table 5.11: Quantitative analysis of upstroke period forLubricant B. 147Table 5.12: Quantitative analysis of upstroke period forLubricant C. 148Table 5.13: Distribution of delta loads during upstrokefor Lubricant A- 0.09% Carbon. 150Table 5.14: Distribution of delta loads during upstrokefor Lubricant B - 0.12% Carbon. 151Table 5.15: Distribution of delta loads during upstrokefor Lubricant C - 0.15% Carbon. 152Table 5.16: Qualitative classification of load cell outputduring the upstroke for several carbons whenC mould lubricant is used at 54 mI/mm. 153Table 5.17: Quantitative analysis of upstroke period forseveral carbons when mould lubricant C is usedat a flow rate of 54 mI/mm. 154Table 5.18: Variation in maximum load as a function of carboncontent of steel. 155Table 5.19: Distribution of delta loads during upstroke forseveral carbon steels when C mould lubricant is usedat a flow rate of 54 mI/mm. 156Table 5.20: Average peak-to-peak change in rear load for thethree mould lubricants used in the Ivaco plant trial. 159Table 5.21: Average peak-to-peak change in rear load forseveral carbon contents when Lubricant C was usedat54ml/min. 160Table 5.22: Billet evaluation summary by steel grade. 164Table 5.23: Billet profilometer measurements. 166Table 5.24: Average surface roughness along the centrelinefor several steel grades. 177Table 5.25: Oscillation mark depth and standard deviationfor several oils and flow rates. 195Table 6.1: Initiation and length of negative strip time forseveral casting speeds for a mould oscillation cyclehaving a 2 Hz frequency and a 9.5 mm stroke. 220Table 6.2: Relative velocity differences between mould andstrand during the downstroke period for specifictimes. 231xixiiLIST OF FIGURESFigure 2.1: Examples of mould friction signals (MF) for threedifferent mould lubricants: a) commercial oil;b) improved oil; and c) mould powder [9]. 7Figure 2.2: Strain gauges positioned on the coupling rodduring the investigation by Foussal et al. [10]. 8Figure 2.3: Phase differences between displacement andforce signals for a casting speed of 1.3 rn/mm(51 in/min)[10]. 9Figure 2.4: Phase differences between displacement andforce signals for a casting speed of 0.7 rn/mm(28 in/mm) [10]. 9Figure 2.5: Typical observed data of frictional force undersinusoidal mould oscillation; strand speed 1.2 rn/mm(48 in/mm) [5]. 11Figure 2.6: Typical observed data of frictional force undersinusoidal mould oscillation; strand speed 1.5 m/min(60 in/mm) [5]. 11Figure 2.7: Oscillation curves on mould side (1) andon driving side (2) [7]. 12Figure 2.8: Example of mould motion, velocity, and frictionforce during the investigation by Mairy and Wolf [8]. 14Figure 2.9: Casting speed and mould friction of 0.15% C steelin 200x200mm with two different lubricants [6,8]. 15Figure 2.10: Friction force and carbon content [8]. 16Figure 2.11: Friction force and oxygen lancing of pluggednozzles [8]. 16Figure 2.12: Friction force and breakouts [8]. 17Figure 2.13: Relative velocity at the mould-strand interfaceas determined by the mathematical model developedby Schacht[11,12]. 18Figure 2.14: The cyclic cam force as determined by themathematical model developed by Schacht [11,12]. 19Figure 2.15: The field measured cam force from the bloomcaster [11,12]. 19xiiiFigure 2.16: Position of rigid interface after variousintervals t from time of contact of liquid withmould [25]. 24Figure 2.17: Predicted temperature distribution in the mouldflux and fraction solidified at the meniscus aftera time 0.3 s. Molten steel assumed to be stagnant [29]. 25Figure 2.18: Predicted axial pressure profiles in the fluxchannel near the meniscus [29]. 26Figure 2.19: Schematic diagram showing the formation oftransverse depressions and cracks in billets due tobinding in the mould [21]. 30Figure 2.20: Schematic diagram illustrating the formation of asubsurface crack on diagonal at obtuse-angle cornersof rhomboid billet [23]. 32Figure 2.21: Schematic diagram showing generation of aninternal crack due to bulging of the billet shellin the mould and a hinging action in theoff-corner region [19]. 33Figure 2.22: a) Ploughing on a surface, and b) chip formationwith an abrasive [35]. 37Figure 2.23: Newtonian method of fluid flow [34]. 39Figure 2.24: The structural arrangement of a fatty acid,a glyceride, and a triglyceride [37]. 41Figure 4.1: Ivaco parabolic taper for the curved wall. 56Figure 4.2: Ivaco parabolic taper for the straight wall. 57Figure 4.3: Original plate design for lubrication system usedin the Western Canada Steel plant trial. 59Figure 4.4: Schematic illustration of load cell. 61Figure 4.5: Schematic illustration of positioning of load cellbetween mould housing and mould oscillating tableand the bolt-spring arrangement used in the Ivacoplant trial. 62Figure 4.6: Schematic illustration of mould cut out for loadcell placement. 63Figure 4.7: Load cell positioner used to ensure contact ofload cell button on oscillating table. 65Figure 4.8: Step down circuit for load cells. 66xivFigure 4.9: Placement of LVDTs and load cells employed in theIvaco plant trial. 68Figure 4.10: Resistive network for signal conditioners usedin conjunction with LVDTs. 69Figure 4.11: Positioning of thermocouples in a straight wall. 71Figure 4.12: Step down circuit for the casting speed signal. 72Figure 5.1: Typical, periodic response of a rear load cellobtained during both trials. 88Figure 5.2: Dampened waveform of the front load cell from thefirst plant trial. 91Figure 5.3: Flash temperatures of mould lubricants as determinedby laboratory and supplier analyses. 95Figure 5.4: Measured kinematic viscosity versus temperature. 98Figure 5.5: Response of rear load cells from the second planttrial. 101Figure 5.6: Key components of the mould oscillation cycleimposed on the load cell response; 0.09% carbonwith Lubricant A fed at 54 mI/mm. 103Figure 5.7: Key components of the mould oscillation cycleimposed on the load cell response; 0.12% carbonwith Lubricant B fed at 54 mI/mm. 104Figure 5.8: Key components of the mould oscillation cycleimposed on the load cell response; 0.05% carbonwith Lubricant C fed at 54 mI/mm. 105Figure 5.9: Key components of the mould oscillation cycleimposed on the load cell response; 0.15% carbonwith Lubricant C fed at 54 mI/mm. 107Figure 5.10: Key components of the mould oscillation cycleimposed on the load cell response; 0.17% carbonwith Lubricant C fed at 54 mI/mm. 108Figure 5.11: Key components of the mould oscillation cycleimposed on the load cell response; 0.42% carbonwith Lubricant C fed at 54 mI/mm. 109Figure 5.12: Casting speed imposed on the load cell responsefor Lubricant A fed at 54 mI/mm (0.09% C). 110xvFigure 5.13: Casting speed imposed on the load cell responsefor Lubricant A fed at 24 mI/mm (0.09% C). 111Figure 5.14: Dependence of minimum load on casting speedfor Lubricant B fed at 54 mI/mm (0.12% C). 112Figure 5.15: Dependence of minimum load on casting speedfor Lubricant B fed at 44 mI/mm (0.12% C). 113Figure 5.16: Dependence of minimum load on casting speedfor Lubricant Bfed at 34 mI/mm (0.12% C). 114Figure 5.17: Dependence of minimum load on casting speedfor Lubricant B fed at 24 mI/mm (0.12% C). 115Figure 5.18: Slope dependence on casting speed for the threemould lubricants tested. 118Figure 5.19: Maximum load dependence on Casting speed forLubricant B at 34 mI/mm. 119Figure 5.20: Maximum load dependence on Casting speed forLubricant C at 24 mI/mm. 120Figure 5.21: Traces of the rear load cell response forLubricant Cat 44 mI/mm (0.15% C). 122Figure 5.22: Traces of the rear load cell response forLubricant Cat 34 mI/mm (0.15% C). 123Figure 5.23: Traces of the rear load cell response forLubricant Cat 24 mI/mm (0.15% C). 123Figure 5.24: Traces of the rear load cell response forLubricant A at 54 mI/mm (0.09% C). 124Figure 5.25: Traces of the rear load cell response forLubricant A at 44 mI/mm (0.09% C). 126Figure 5.26: Traces of the rear load cell response forLubricant A at 34 mI/mm (0.09% C). 127Figure 5.27: Traces of the rear load cell response forLubricant A at 34 mI/mm (0.09% C). 128Figure 5.28: Traces of the rear load cell response forLubricant A at 24 mI/mm (0.09% C). 129Figure 5.29: Traces of the rear load cell response forLubricant B at 54 mI/mm (0.12% C). 130Figure 5.30: Traces of the rear load cell response forLubricant B at 44 mI/mm (0.12% C). 131xviFigure 5.31: Traces of the rear load cell response forLubricant B at 34 mI/mm (0.12% C). 132Figure 5.32: Traces of the rear load cell response forLubricant B at 24 mI/mm (0.12% C). 133Figure 5.33: Traces of the rear load cell response forLubricant C at 54 mI/mm (0.035% C). 135Figure 5.34: Traces of the rear load cell response forLubricant Cat 54 mI/mm (0.051% C). 136Figure 5.35: Traces of the rear load cell response forLubricant Cat 54 mI/mm (0.18% C). 137Figure 5.36: Traces of the rear load cell response forLubricant C at 54 mI/mm (0.42% C). 138Figure 5.37: Maximum, minimum, and change in load duringthe upstroke cycle. 149Figure 5.38: Average peak-to-peak change in rear load forthe three mould lubricants at different flow rates. 158Figure 5.39: Frequency analysis for the load cell results of a0.17% carbon cast with Lubricant Cat 54 mI/mm. 162Figure 5.40: Orientation of the billet sides and profilometertraces. 165Figure 5.41: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant C at 54 mI/mm. Grade was a 1008. 169Figure 5.42: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant C at 54 mI/mm. Grade was a 1008. 170Figure 5.43: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant A at 44 and 34 mI/mm.Grade was a 1010. 171Figure 5.44: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant A at 34 and 24 mI/mm.Grade was a 1010. 172Figure 5.45: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant B at54 mI/mm. Grade was a 1012. 173xviiFigure 5.46: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant C at 44 and 34 mI/mm.Gradewasal0l5. 174Figure 5.47: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant C at 54 mI/mm. Grade was a 1018. 175Figure 5.48: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant Cat 54 mI/mm. Grade was a 1039. 176Figure 5.49: Photograph of the surface of a 1008 grade billetfrom strand 2 (control) showing depressions and asurface crack. 179Figure 5.50: Longitudinal section of the 1008 grade billetfrom strand 2 (control) showing the presence of acrack at the base of a depression. 180Figure 5.51: Transverse section of a 1008 grade billetfrom strand 3 (test) showing the presence ofcracks beneath a depression. 181Figure 5.52: Longitudinal section of a 1008 grade billetfrom strand 3 (test) showing the presence ofcracks beneath a depression. 182Figure 5.53: Photograph of the surface of a 1008 gradebillet showing the irregular spacing andnonparallelism of the oscillation marks. 183Figure 5.54: Photograph of the surface of a 1010 gradebillet showing the surface roughness typicalof this grade of steel. 184Figure 5.55: Photograph of the surface of a 1039 gradebillet showing the shallower oscillation markscompared to the other two grades (1008, 1010). 185Figure 5.56: Profilometer traces for a 1008 steel grade. 187Figure 5.57: Profilometer traces for a 1018 steel grade. 188Figure 5.58: Profilometer traces for a 1039 steel grade. 189Figure 5.59: Profilometer traces for a 1010 steel grade atwhen Lubricant A is fed at 24 mI/mm. 190Figure 5.60: Profilometer traces for a 1010 steel grade atwhen Lubricant A is fed at 34 mI/mm. 191xvii ±Figure 5.61: Profilometer traces for a 1010 steel grade atwhen Lubricant A is fed at 34 mI/mm. 192Figure 5.62: Profilometer traces for a 1010 steel grade atwhen Lubricant A is fed at 44 mI/mm. 193Figure 5.63: Profilometer traces for a 1010 steel grade atwhen Lubricant A is fed at 54 mI/mm. 194Figure 5.64: Average oil velocity along the mould wallwhen the inlet water temperature is 10 0 C. 197Figure 5.65: Average oil velocity along the mould wallwhen the inlet water temperature is 200 C. 198Figure 5.66: Average oil velocity along the mould wallwhen the inlet water temperature is 300C. 199Figure 5.67: Average oil velocity along the mould wallwhen the inlet water temperature is 40 0 C. 200Figure 5.68: Average oil velocity along the mould wallwhen the inlet water temperature is 50 0 C. 201Figure 5.69: Schematic illustration of the oil lubricationat the meniscus during an oscillation cyclewhen the water temperature is 10 0 C. 203Figure 5.70: Schematic illustration of the oil lubricationat the meniscus during an oscillation cyclewhen the water temperature is 50 0 C. 204Figure 5.71: Duration of time oil is present at the meniscusfor different stroke lengths when the oscillationfrequency is 1.333 Hz (80 cpm). 206Figure 5.72: Duration of time oil is present at the meniscusfor different stroke lengths when the oscillationfrequency is 1.667 Hz (100 cpm). 207Figure 5.73: Duration of time oil is present at the meniscusfor different stroke lengths when the oscillationfrequency is 2.0 Hz (120 cpm). 208Figure 5.74: Duration of time oil is present at the meniscusfor different stroke lengths when the oscillationfrequency is 2.333 Hz (140 cpm). 209Figure 5.75: Lubrication at the meniscus for an oscillationfrequency of 1.333 Hz (80 cpm) when cooling watertemperature is 50 0 C. 211xixFigure 5.76: Lubrication at the meniscus for an oscillationfrequency of 2.0 Hz (120 cpm) when cooling watertemperature is 50 0 C. 212Figure 5.77: Lubrication at the meniscus for an oscillationfrequency of 2.0 Hz (120 cpm) when cooling watertemperature is 100C. 213Figure 6.1: Mould dynamics for the IVACO casting operation. 221Figure 6.2: Influence of casting speed on the length of thenegative strip period. 222Figure 6.3: Relative mould and strand velocities. 224Figure 6.4: The resulting direction of the net forceduring the dynamic oscillation of the mould. 225Figure 6.5: The formation of an oscillation mark due to themechanical interaction between mould and strandduring the negative strip period. 226Figure 6.6: The key oscillation features imposed on a loadcell response when strand velocity is 36.0 mm/s(85 ipm). 228Figure 6.7: The key oscillation features imposed on a loadcell response when strand velocity is 42.3 mm/s(100 ipm). 229Figure 6.8: The fluctuating casting speed imposed on loadcell response to indicate the effect of speedon minimum load. 232Figure 6.9: Empty mould force during an oscillation cycleis shown. 235xxLIST OF NOMENCLATUREA area of the bolt, rn2Ar area of real contact, m2am mould acceleration, rn/s2CL centre linedb extension of the bolt, mdc direct current,Adl extension of the load cell, mE Young’s modulus, MPaF tangential force, NFM mouldforce,Ng gravity, rn/s2icw inside curved walli/p input/output voltage, vkb spring constant of the bolt, MPakj spring constant of the load cell, MPaLVDT linear variable displacement transducert viscosity, kg/rn/sMn/S manganese/sulphur ratioMn/Si manganese/silica ratioMF mould frictionn normal force, Nocw outside curved wallPb load on the bolt, NP1 load on the load cell, NPL change in load on the load cell, Na normal stress, N/rn2SD standard deviationStd. Dev. standard deviationt shear stress, N/rn2tneg negative strip timeV voltage,vVM mould velocity, rn/sVmax maximum velocity of the mould, rn/sVS strand velocity, misxxixxiiACKNOWLEDGEMENTI would like to express my sincere gratitude to my supervisors, ProfessorJ.K. Brimacombe and Doctor LV. Samarasekera, for their assistance during thisresearch.My sincere thanks to Ian Bakshi, Neil Walker, and Nick Hemingway fortheir contributions to the successful plant trials and for their continuous support,encouragement, and discussion and for the use of work completed by TheCentre for Metallurgical Processes. My appreciation for all those fellowgraduates who helped me during the course of this work, particularly DaveMcQuistan and Andrew Shook.I would like to especially thank my husband, Russ, and my children,Aaron, Ty, and Cara, for their continual understanding and love throughout thiswhole project.Financial support for this work was obtained from the Natural Science andEngineering Research Council.1CHAPTER 1. INTRODUCTIONComprehension of the role of mould lubrication in the continuous castingof steel billets is important for successful design and for excellent billet quality.Oils are currently employed as lubricants in the casting of smaller sized billets toprovide lubricity at the interface between the mould and the strand. Until thepresent, little work has been undertaken to examine the lubrication of the mouldin billet casting.The liquid lubricant is pumped, sometimes for great distances, fromstorage through pipes to the caster and the mould. Weather conditions varyingbetween winter and summer may play havoc with the physical properties of theoil; which, depending on temperature, may be viscous or fluid. Othertypes ofproblems relating to the handling of the oil, namely water condensation in storagecontainers and unheated pipes, may or may not be present depending on plantprocedures. Also, in some operations, the bulk oil may not be heated prior tobeing pumped. At the top of the mould, the lubricant is pumped into an oilchannel where it weeps through a narrow slot onto the hot mould face. Thedistribution of the lubricant is assumed to be uniform around the periphery of themould. However, measurements made by Bakshi and co-workers [1] haveshown that uniform oil distribution is a misconception for most plant operations.As a result of these measurements, Bakshi et al. [1] have developed adistribution system which is effective in providing uniform oil flow along all fourmould faces.2Once the oil has arrived at the mould, it begins its downward flow alongthe mould face. The velocity of the oil down the mould wall is dictated by itsviscosity, which is a function of the mould wall temperature. Depending on thecooling water practices of the operation and the current season, the wateremployed to provide mould cooling can vary dramatically in temperature.Because the oil above the meniscus assumes the temperature of the wall, anywall temperature variation can produce a change in the amount of lubricantreaching the meniscus to provide the lubrication required.The mechanism of lubrication at the meniscus in billet casters employingoil has not been addressed as of today. The most likely mode of lubrication ismixed which combines the mechanisms of hydrodynamic and boundary layerlubrication. Hydrodynamic lubrication provides a physical barrier between twosliding surfaces and is dependent upon the viscosity of the lubricant. In boundarylayer lubrication, it is the ability of the molecules in the lubricant to adhere to thesubstrate and develop a system of long molecular tentacles which allow the othermoving surface to glide easily by. Not all molecules present in the oil areconducive to achieving this mechanism of lubrication.When mixed lubrication occurs, sufficient lubricant is present to providehydrodynamic lubrication but for some period of the oscillation cycle, the supplydecreases leaving remnants of the oil to operate under boundary lubricationconditions.3Since the environment at the meniscus is extreme due to the hightemperatures, the oil in this region must be able to survive for some small timeperiod determined by the parameters of the oscillation cycle, namely strokelength and oscillation frequency. It will be shown that oil reaches the meniscusgenerally only during the downstroke and that bulk lubricant is absent during theupstroke. This finding implies that hydrodynamic lubrication is operating duringthe downstroke and that if flows are not sufficient then the boundary lubricationmechanism assumes greater importance in providing lubricity at the mould-strandinterface.The ability of the oil to survive down the mould wall particularly close to themeniscus is dependent on its flash point. The mould wall temperature at themeniscus during casting is a function of such parameters as wall thickness andcopper conductivity, the presence of deposits on the cold face of the mould, andthe carbon content of the steel being cast. If, finally, mould temperatures arehigh, an oil having a high flash point must be employed; conversely, iftemperatures are low, an oil with a lower flash point may be utilized.Surface defects such as pinholes, transverse depressions, rhomboidity,and longitudinal cracking, originate in the mould resulting from interactionsbetween the mould and the strand. Some of these defects occur due toinappropriate cooling conditions, mould taper, or lubrication. Mould oscillationwas introduced as a feature in the continuous-casting process to minimizesurface defects resulting from the mould-strand interaction. Other attempts toimprove billet quality in the mould include tightening of casting machinetolerances, use of thicker and more softening resistant coppers, determiningoptimum cooling water conditions, and reassessing mould oscillation4characteristics and strand velocity. Lubrication of the mould remains the lastfrontier to be explored for the production of high quality billets.The fundamental objective of this project was to elucidate the lubricationphenomena at the meniscus region in the mould of a billet caster by examiningthe loading of the mould during casting. The objective was achieved in part byemploying load cells which were placed in the mould housing. This technique formeasuring the mould-strand interaction is relatively new and novel in billetcasting. Several properties of the lubricants were examined, particularly theirability to reach and remain functioning at the meniscus. Several carbon gradeswere cast under similar mould lubricating conditions in order to assess the effectof carbon content on mould-strand interaction. Also, three oils were tested underfour flow rates to determine the effect of lubricant type and flow rate on theloading of the mould. Without question, this project provided insight into theinteraction between the mould and the strand during the casting of billets. Thiswork is only a beginning in the further understanding of lubrication phenomena inthe continuous casting of billets.5CHAPTER 2. LITERATURE REVIEW2.1 IntroductionInvestigation of mould-strand interaction began in 1970 with the efforts ofthe Russian researchers Efremov et al. [2] who built a laboratory model to studythe apparent friction coefficient between the strand and the mould and its effecton billet surface quality. Several other researchers [3,4,5] have alsosubsequently employed laboratory models to investigate this interaction. Asearly as 1977, electronic sensors were being utilized to elucidate the frictionbetween the mould and the strand in actual industrial operations [6- 12]. Wolf[6], Mairy et al. [7], and Mairy and Wolf [8] have conducted extensive studies onthe mould-strand interaction in slab casters. Corresponding work on billetcasters is limited to a single study [9].Three types of measuring devices have been employed in theinvestigation of friction forces in the mould, namely accelerometers, straingauges, and load cells. Attached to an outside mould face, accelerometers havebeen employed to detect mould-strand friction in several studies [7,8,9]. Straingauges have been positioned on an outside mould face [2] and on the oscillatingarm which transmits the oscillating motion to the mould [5,10]. Load cellsmounted onto the oscillating system have been utilized by several researchers[3,4,8,11,12] to determine the degree of mould-strand interaction in slab casting.Interestingly, in 1983 a patent [13] was issued to Werke Klockner andKlockner-Humboldt-Deutz for a breakout warning device which also detectsmould-strand interaction using a method different to those discussed so far. Thisdevice monitored noise in the region of the mould by acoustical means.6Predetermined frequency ranges were selected to indicate interaction betweenthe mould and the strand and jamming conditions in the mould. No literature wasfound which indicated this device had been utilized in operation.2.2 Measuring Devices for Mould-Strand Interaction2.2.1 AccelerometersShort et al. [9] replaced variable linear displacement transducers (LVDTs)with accelerometers to monitor the interaction between mould and strand in abillet caster because the LVDTs were found to be simpler and more sensitive.From their report, the position of the accelerometers is unclear but possibly mayhave been on the oscillator. In their experiments, they tested different proprietarymould oils and began to blend naturally occurring vegetable oils which containcarboxylic acid esters with synthetics which have greater chemical stability.These researchers blended esters with mineral oils to optimize the wettingbehaviour, mould heat transfer, and mould friction of the lubricants, although nodetails were provided on how this was accomplished. In Figure 2.1 the frictionsignals for a commercial oil, an improved oil, and a mould powder, as recordedby the accelerometers, are compared. Assuming that the abbreviation for mouldfriction is MF, this figure clearly indicates the lower friction associated with theimproved oil as compared to that of the proprietary oil and the similarity in frictionresponse of the former with that of the mould powder.MF % SAE 1144,140mm, 1.9 rn/mm100____________________________8060 %4q j48t 7 %4020—j.16Z%——— —,_..— i.1O S 1%0 10 20 30 40 50 60 70 tIme mmFTgure 2.1 Examples of mould frIction sgnale (NF) to.’ three diffmmould bjbflcants: a) commeroai oil; b) imptod oil; andC) mould powder [9).The brief report by this research group is relevant because they addressthe question of which properties are important in the production of a good mouldlubricant. However, they fail to comment specifically on the additives and onwhich lubricating mechanisms may be operating.2.2.2 Strain GaugesThe research completed by Foussal et al. [101(1985) found that the forcesignal from strain gauges positioned on the coupling rod of a mould shown inFigure 2.2 was periodic. Furthermore, they learned that a phase change existedbetween displacement and the force signal which was dependent upon castingspeed as shown in Figures 2.3 and 2.4. The casting speeds in Figure 2.3 and2.4 are 1.3 rn/mm (51 in/mm) and 0.7 rn/mm (28 in/mm), respectively. At slowercasting speeds, the oscillation frequency, force, and displacement signalsapproach a zero phase difference as the signals indicate in Figure 2.4. The8steel meniscusCoflicityIIIFigure 22: Strain gauges positioned on the coupflrig rodduring the investigation by Foussal et aL[1O1.heat extractionaI 4 — — —— — — —44.I— — — — — — — — — —— torce— — — — — — —. — —a.. •a91.71.41.10tension (v)time (10 s)Figure 2.3: Phase diflerences between disp’acement and force signalsfor a casting speed of 1.3 rn/mm (51 in/mm) [10].tension (v) v=0.7 rn/mmtime (10 s)Figure 2.4 Phase differences between displacement and force signalsfor a casting speed of 0.7 rrvmin (28 in/mEn) (101.50 1001.71.41.10 50 100 15010opposite is true for faster casting speeds as shown in Figure 2.3. Yamanaka etal. [5] in their investigation into the lubrication phenomena in the casting of slabsalso found that a difference in phase exists between mould displacement andmould friction and that this difference is dependent on casting speed. Therelationship between phase and casting speed found by Yamanaka et al. [5], whoemployed a laboratory model of a slab caster, was opposite to that concluded byFoussal et al. [10], who monitored an operational caster. Results fromYamanaka et al. [5] reveal that at slower casting speeds (1.2 rn/mm, 47 in/mm),the frictional force lags 90 degrees behind mould displacement as shown inFigure 2.5. The phase difference is negligible at the higher casting speed of 1.5m/min (59 in/mm) as shown in Figure 2.6. The Japanese [5] monitored friction ata site on the mould system different from the French [10]. In the mould simulatordesigned by the Japanese [5], strain gauges were attached to a center rod whichpulled the solidified shell downward while the French [10] placed strain gaugeson the coupling rod as previously mentioned. Mairy et al. [7] positionedaccelerometers on the mould of an operating caster and on the oscillating armand found a phase shift of 180 degrees as shown in Figure 2.7. This resultsuggests that the contrary findings published by Foussal et al. [10] andYamanaka et al. [5] may be the result of the placement site of the measuringdevices or of the different measurement systems.While Foussal et al. [10] offer no explanation for the phase shift betweenthe frictional force and mould displacement as a function of casting speed,Yamanaka et al. [5] suggest that this phase difference is due to the rheologicalbehaviour of the mould powder under various casting speeds. Without explaininghow casting speed affects the powder, they conclude that mould powders aremore “viscous” at lower casting speeds and exhibit “elastic” behaviour at higher110U0C0UU--I‘1* .,_— I — — — 9_-\:_____I /- /_ — —Figure 2.5 Typical observed data of frIctional foi’ce under sinusoidalmould oscdlation; strand speed 1.2 rn/nn (48 irVnin) (51.0C.)CU-C0EIFigure 2.6 Typical observed data of frictional force under sriusoidalmould oscillation; strand speed 1.5 rn/rrn (60 in/mm) [51.12casting speeds.Time(s)Figure 2.7 OscillatIon curves on mould s4d (1) andon dnvlng sld. (2) (7].Yamanaka et al. [5] found that friction increases as the thickness of themould powder film decreases and as casting speed increases. Comparison ofthe force profiles in Figures 2.5 and 2.6 illustrates the effect ofcasting speed onfriction force. At higher casting speeds, the thickness of themould powderdecreases providing only a thin lubricant layer causing an increase in mould-strand friction.The work completed by Foussal et al. [10] and Mairy et al. [7] clearlyindicates that monitoring the caster could provide useful information regardinginteraction between the mould and the strand. From the work published by Mairyet al. [7] and due to the confusion of the resuLts from Foussal et al. [10] andYamanaka et al. [5], it has become apparent that mould-strand monitoringdevices should be in contact with the mould to adequately detect mould friction.13In the current project, load cells were placed in such a position that theymonitored the exact loading behaviour of the mould.2.2.3 Load CellsOn a small-scale experimental caster, Komatsu et al. [3] placed load cellson the mould oscillating table to measure the friction force which was defined asthe difference between the apparent mould weight and the mould inertial force.This study found that the load decreases as casting of a heat progresses. Theyalso determined that friction forces were greater during positive-strip periods thanduring negative-strip periods and that the steel shell experiences greater tensilestresses in a tapered mould than in a straight mould. Friction forces were alsoobserved to be larger for rectangular moulds than for round moulds.By placing load cells between the oscillating plate and mould housing andaccelerometers fixed to the outside of the mould of a slab caster, Mairy and Wolf[8] were able to relate mould friction to mould displacement. The friction forcevariation with time was found to be periodic and reproducible but not sinusoidal;this finding wa also evident in the current work. Figure 2.8 illustrates the time-dependent variations in mould displacement, velocity, and force. From thisfigure, it is apparent that the force begins to decrease at the beginning ofnegative-strip. At the end of negative-strip the load is comparable with that whichoccurred at the beginning of the negative-strip period. The load during upstrokereaches a maximum and then slightly decreases. Wolf [6], in his discussion ofthese curves, interpreted the force response during positive- and negative-stripperiods as the result of different operating modes of friction at the interfacebetween mould and strand at the meniscus. He proposes that during positive-14E•1C0EC)CC)(00strip, the load is highest due to solid friction. The presence of a molten powderlayer provides liquid lubrication giving rise to a reduction in friction duringnegative-strip. Wolf further suggests that the steep increase in maximum frictionwhich coincides with maximum mould velocity may indicate that lubricant feedinglags behind mould displacement. But as the mould velocity decreases, lubricanttransport improves or catches up providing better lubrication.4-C00>C)00U-Figure 2.8 Example of mould motion, velocity, and fnction forcedunng the investigation Dy Mairy arid Wo [81.15LLPowderOHMairy and Wolf [6,8] also performed mould friction measurements forpowder versus oil lubrication during bloom casting. Figure 2.9 illUstrates thefriction signals during powder and oil lubrication. The change from powder to oilresults in an increase in mould friction which fluctuates more than that forFigure 2.9 Casting speed arid mould friction for 0.15% C steelin 200x200mm with Iwo different lubflcants (6,61.powder. This observation supports that of Short et al. [9] in their investigation oflubricants in billet casting.In their paper published in 1982, Mairy and Wolf [8] explored the effect ofcarbon content, mould powder, casting speed, alumina absorption in slag anddeslagging, oxygen lancing of pouring nozzles, and breakouts on the mouldfriction. They found that a 0.12% carbon yielded a significantly higher frictionlevel than that for lower carbons (0.08%,0.06%) when consecutively cast with thesame mould powder, as illustrated by Figure 2.10, and concluded that differentmould powders should be employed when casting different carbon grades.16>4 .-O.06%C 0.O8%C O.12%CFigure 2.10 Friction force and carton content 81.As with Foussal et al. [10], the analysis of the friction responses by Mairyand Wolf [8] showed that the responses could reveal casting problems.Reoxidation of the steel during pouring led to steep increases in friction as didoxygen lancing of plugged nozzles. Breakouts or excessive sticking and tearingcould be sensed during monitoring as the friction signal increased gradually thenrose dramatically. These increases in friction may reflect a change in mould fluxviscosity due to the absorption of reoxidation products by the flux. Figures 2.11and 2.12 illustrate the friction behaviour for these two conditions. Mairy and Wolf[8] further examined the effects of casting parameters on oscillation markformation in slabs which will be discussed in a later section.02 lancingI LO.06%CTimeFigure 2.11 Frlcbon force and oxygen Lanng of plugged nozzles [8.17breakout—. . IiiJifhWLF01J6%CTineFigur• 2.12 Friction forc and breakouts [8J.Investigation of mould friction by load cell measurements has beendiscontinued by Mairy and Wolf [8] due to the difficulty of handling the equipmentand the maintenance of the system in the severe environment of slab casting.They have selected accelerometers for their subsequent investigations intomould friction.Schacht published two papers [11,12] on his investigation into the mould-strand behaviour. He developed a mathematical model incorporating thesignificant portions of the caster: the displacement profiles of the parabolic [11]and sinusoidal [12] cams, the caster structure supporting the mould, the frictionforce at the mould-strand interface, the gap in the starter-bar linkage, and thestarter-bar stiffness. Figure 2.13 shows the relative velocity between mould andstrand. Casting conditions included mould oscillation at 1 Hz, stroke length of 13mm, and strand withdrawal at 1.32 rn/mm (52 in/mm). The analysis identified twoconditions which occurred at the beginning and the end of negative-strip whenisthe relative velocities are zero and are shown in Figure 2.13. Schacht [11,121refers to thee, conditions as stick conditions and proposes that the friction forcebetween strand and mould causes the stick condition. The friction force, hesuggests, arises due to the development of friction in the lower mould regionwhere ferrostatic pressure causes mould-strand interaction. The calculatedcyclic cam force is presented in Figure 2.14 while the measured cam force from abloom caster is shown in Figure 2.15. Thus, the measured and predicted forceresponses are seen to be similar. Furthermore, as will be discussed in a futuresection, the measured force profile for slabs in the study by Schacht [11,12] willbe seen to be similar to the force profiles of billets obtained in this current study.The main deviation between the slab and billet profiles lies in, the negative-stripperiod. In bloom or slab casting, the shape of the profile during negative-strip isflatter than that seen in billet casting which will be illustrated later.E>1Time (s)Figure 2.13: Relative velocity at the mould-strand interalaceas detemiried by the mathematical model developed by Schacht[11,12].zU)010U.E(UC)CUC19OX oai oio ‘a ‘a ‘i aTime, sFigure 2.14 The cyclic cam force as determined by the mathematical modeldeveloped by Schacht 1 1,12J.za)0L1E(UC)C)0(UC)Time, sFigure 2.15 The field measured cam force from the bloom caster [11.121.In all three investigations - Komatsu et al. [3], Mairy and Wolf [8], andSchacht [11,12] - load cells were positioned between the mould housing and theoscillating table of a slab caster. The load cell response was not shown for thework completed by the Japanese group [3]; but for the other two groups, thepublished responses closely reflect the signals obtained during this study. Theinterpretation of the signals by Mairy and Wolf [8] differ significantly from thisinvestigator as will be shown iii Chapter 6 of this thesis.20212.3 Oscillation Mark Formation and Mould-Strand InteractionSeveral researchers [6,16,17,18,25,28] proposed mechanisms foroscillation mark formation. Of these, only Saucedo et al. [16,25], Samarasekeraand Bnmacombe [21], and Takeuchi and Bnmacombe [28] have providedsupporting evidence and detailed descriptions rendering credence to theirproposed mechanisms for the formation of oscillation marks.Figures 2.1 and 2.9 indicate that friction forces are higher when castingwith oil lubrication than with powder during both billet and slab casting. However,Wolf [6] has found that strands cast with oil lubricant have shallower oscillationmarks than those cast with mould powder. Therefore, he concludes that frictionforces are not important in the formation process of oscillation marks. Wolf [6]also dismisses cooling intensity of the strand as being significant in oscillationmark formation because the local heat flux is greater for oil lubrication than forpowder.Irrespective of the lubrication type, Wolf [6] considers consumption oflubricant to be the controlling factor in the severity of oscillation mark depth.Employing load cells to monitor mould friction and accelerometers to record theconcurrent mould motion, Wolf [6] proposed a mechanism of oscillation markformation based on the relationship shown in Figure 2.8 without distinguishingbetween oil or powder lubrication. Oscillation mark formation initiates at thepoint where maximum friction occurs during the negative-strip period. Theoscillation mark is formed and traps available lubricant until positive-strip begins.Thus, the greater the amount of lubricant trapped within the oscillation mark, thedeeper the mark.22Tomono et al. [17] constructed a refractory mould to elucidate themechanisms which determine surface quality. The mould was water cooled andthe steel was bottom poured. No indication was given as to whether lubricationwas employed in the experimental arrangement. Their experiments dispute thepremise by Wolf [6] that heat flux is not important in oscillation mark formation.They found that the depth and pitch of the oscillation marks could be related tothe cooling intensity of the mould wall and the surface tension of the moltenmetal. By experimenting with different gas atmospheres at the meniscus, theydetermined that the magnitude of heat transfer plays an important role in thecharacteristics of oscillation marks. Large pitch and deep marks formed under areducing atmosphere while smoother surfaces were formed under a normalatmosphere, ie air. Under a reducing atmosphere of hydrogen gas the depth ofthe oscillation mark was 0.2 mm and the pitch was 4.7 mm. The depth and pitchdecrease substantially to 0.08 and 1.5 mm respectively under normalatmosphere conditions. Tomono et al. [17] determined the surface tensionvalues for each atmospheric condition by fitting experimental points to anequation which describes the profile of any liquid cylinder in mechanicalequilibrium. With this technique, the surface tension of the steel was determinedand was found to be three times greater for the hydrogen atmosphere than forthe air atmosphere; this difference partially accounts for the significant differencein oscillation mark characteristics between the two atmospheres examined aspitch and depth of the oscillation marks increase with surface tension. Alsocontributing to the smaller mark characteristics is the lower heat extractioncapability of the air atmosphere as compared to that of the hydrogenatmosphere. Since heat transfer at the mould-steel interface is controlled byconduction through the gap, the higher conductivity of the hydrogen atmosphereenhances heat transfer and promotes solidification which, in turn, provides23stronger mechanical support for the liquid meniscus. Therefore, under anatmosphere of hydrogen, the combination of the higher surface tension of theliquid steel and the higher heat transfer enhances the formation of deeperoscillation marks.Investigations by Saucedo et at. [16] showed that heat transfer isimportant in oscillation mark formation supporting the findings of Tomono et al.[17]. In their experiments with a chill mold and uphill teeming technique,Saucedo et al. [16] found that the width and depth of ripples of the ingot of a0.1% carbon steel were greater when solidification occurs against a water-cooledcopper plate than with a mild steel plate. They attributed the severity of ripples tothe higher heat flux which occurs with the water-cooled copper plate. Contrary toWolf [6], Saucedo et al. [16] emphasize the importance of heat transfer in themechanism of ripples and lap formation. Any mechanism which reduces theextent of solidification at the meniscus decreases the severity of rippling. Highsuperheats and high casting speeds decrease rippling as these parametersretard the solidification process at the meniscus.Saucedo et al. [25] developed a mathematical model based on heattransfer and solidification to determine the solidification behaviour of the interfacein the meniscus region during casting. Utilizing this model, the authorsinvestigated the amount of solid which could form under short time periods whichare comparable to the contact time of a melt and mould in reciprocating systems.This time period can vary between 0.2 and 0.6 seconds, but typically is 0.3seconds. The authors selected a value of 0.2 for the solid fraction as their rigiditycriterion. There is some dispute as to the validity of 0.2 as a rigidity factor but theshell thickness results appear to be reasonable.24Figure 2.16 illustrates the predicted shell thickness for two rigidityfractions: 0.2 and 0.6. Assuming 0.3 seconds as the duration period for mould-strand contact, Saucedo et al. [25] indicate that a solid shell has grown over asignificant distance along the meniscus surface. They stress that solid formationEE(50C00C(5S:45 —::Da::4:xe ::csDistance From Mold Wall (mm)Figure 2.16: Position of rigid interface after vanous intervalsfrom time ol contact of quid with mould [25].00 C20 0181>02—‘N ‘S>crCt1. •5(.10008125along the meniscus occurs only fractions of a second after the melt contacts themould wall due to the steep temperature gradient and despite the fact that thebulk of the steel maintains its initial superheat. For a rigidity criterion of 0.2 usedby Saucedo et al. [251 and a contact duration of 0.3 seconds, the predicted solidwhich has formed along the meniscus is approximately 0.7 mm which Saucedo etal. [25] believe to be in agreement with measurements of oscillation mark depthson the surface of continuously cast steel slabs.Takeuchi and Bnmacombe [28] also undertook an analysis of heat flow atthe meniscus in slab casting. Basing their model on actual heat flux distributionsfrom an instrumented slab mould, their results agree with Saucedo et al. [25] alsopredicting a solidified thickness of 0.7 mm along the meniscus as shown inFigure 2.17. However, the mechanism they propose for oscillation markformation differs significantly from that of Saucedo et al. [25]..1IIIiIFigure 2.17: Predicted terrverature distribution in the mould fluxand fraction solidified at the meniscus after a time 0.3 s.Molten steel assumed to b stagnant [29J.4 6 8Distonce Fron Mo’d *oa (mm)26Takeuchi and Bnmacombe (28] produced the most complete and detailedanalysis of the oscillation mark formation process found in this literature search.These researchers examined slab surfaces finding that the oscillation mark depthranged from approximately 0.2 mm to 1.0 mm but subsurface hooks extendedfrom 1.5 to 2 mm. The hook length suggests that the overflow mechanismproposed by Tomono et al. (17] and others is incomplete, inadequately describingthe meniscus behaviour. In an earlier work, Takeuchi and Brimacombe (29]further investigated the effect of pressure which arises due to the meniscusshape and narrowing of the flux channel. Their resutts shown in Figure 2.18*Figur 2.18: Predicted axial pressure profiles in the fluxChannel nw the meniscus (29J.2 4 6 8 ‘0D,stonce Down Mold Woll (mm)27indicate that positive pressures are generated in the mould flux on thedownstroke and negative pressures on the upstroke. The magnitude of thepressure during upstroke is greater than during the downstroke due to the largervalue of the relative motions between mould and strand.The mechanism for oscillation mark formation proposed by Takeuchi andBrimacombe [28,29] incorporates the effect of these pressures. During negative-strip the positive pressure developed in the mould flux pushes the solidifiedmeniscus away from the mould wall and then during positive-strip, the meniscusis drawn back towards the mould wall. Subsurface hooks form if the shell isstrong and overflow occurs; subsurface hooks will not form if the shell is weakand is pushed back to the mould wall. Takeuchi and Bnmacombe [28,29] alsofound that hooks in the 0.09% carbon slabs formed a smaller angle with thesurface than hooks in the 0.26% carbon slabs. They surmise that the differencein angle is due to the ability of the stronger solidified shell of the lower carbon atthe meniscus to resist the flux pressure.Metallographic examination and the heat flow analysis of Takeuchi andBrimacombe [28,29] led to their proposal of a mechanism for the development oftransverse crack formation specifically at the bottom of an oscillation mark.Examination of oscillation marks on slab sections indicated that larger dendritearm spacing and coarser secondary solidification structures occurred at thebottom of the oscillation mark compared to the top. This difference in structurewas attributed to the difference in cooling which exists in these two regions of theoscillation mark. In the outer regions of the oscillation mark, due to the closerproximity of the shell to the mould, cooling is strong. On the other hand, the gapbetween mould and strand at the base of the oscillation mark increases the28thermal resistance reducing cooling and allowing coarse structures to form uponsolidification. Then tensile stress or strain due to increases in mould-strandfriction in these areas can initiate cracks.Samarasekera et al. [18] have proposed a mechanism for the formation ofoscillation marks in billets in which oil is employed as the lubricant. In theirinvestigation, three significant relationships became evident from which theydeveloped their mechanism. The first was that the faster the velocity of thecooling water, the lower the heat flux of the mould at the meniscus and theshallower the oscillation mark. Secondly, the negative taper resulting from thethermal distortion of the mould decreased with increasing cooling water velocity.Thirdly, the pitch of the oscillation mark correlated to the ratio of casting speedand oscillation frequency. This relationship infers that the oscillation marks formduring an oscillation cycle [18]. Samarasekera and Brimacombe have shownthat mould heat transfer causes the mould tube to becomes thermomechanicallydistorted resulting in a bulged shape with negative taper at the meniscus [18].Samarasekera et al. [18] propose that during negative-strip, when the thermallydistorted mould moves down faster than the strand and squeezes downward onthe newly forming shell, it deforms and buckles creating an oscillation mark. Thedegree of shell deformation is dependent on the degree of distortion, the durationof the negative-strip period, and the mechanical properties of the shell which varywith temperature and composition. Solidification of the meniscus would occur inresponse to the high heat extraction resulting from the excellent contact betweenshell and mould during the jamming down of the mould. During positive-strip themould moves upward disengaging from the strand and a gap is created.Overflowing liquid steel may or may not form a lap or a hook depending on thedegree of solidification which has occurred at the meniscus. As previously29indicated, the load cell results indicate that there is mechanical interactionbetween the mould and strand dunng the negative-strip period when the mould isbearing down on the strand. In this study, casting speed was also discovered toaffect the load cell response during the negative-strip period. Together, thesetwo observations, discussed in Chapter 6, will give support to this theory ofoscillation mark formation.2.4 Billet Defects and Mould-Strand Interaction2.4.1 Transverse DepressionsThe mechanism for transverse depression formation proposed bySamarasekera et at. [21] is based on axial tensile stresses produced duringmould-strand interaction and is illustrated in Figure 2.19. If, as indicated at thetop of the figure, binding occurs, friction resufts and together with the withdrawalforce of the strand, axial tensile stresses result. Plastic deformation occurs andnecking of the steel produces a depression on the surface whilst cracks can format the solidification front beneath the depression owing to reduced ductility attemperatures close to the solidus.Excessive taper will exacerbate binding and interaction between the billetand mould. Samarasekera et at. [21] also found in their study, that binding couldbe deleteriously affected by decreasing casting speed. Then, owing to increaseddwell time, a point is reached in the mould where the resistance of the shellbecomes comparable to the resistance of the air gap. The heat transferdecreases giving rise to lower shrinkage and an increased tendency to bind inthe lower part of the mould.30Binding is enhanced when casting low-carbon billets (0.1 - 0.14% C) [21].Heat transfer is retarded reducing the cooling and shrinkage of the shell.However, billets in the carbon range of 0.17 to 0.24 percent also occasionallyexhibit transverse defects as these steel grades have particularly diminishedductility at elevated temperature [21].Binding(High Friction)Liquid SteelNecking ofDuctile Shell Crackto Form DepressionZone of Low DuctilityWithdrawal ForceFigure 2.19: SchematIc diagram showrng the formation of transi,ersedePressions and cracks in billets due to binding in the mould [ Longitudinal Corner Cracks and RhombolditySamarasekera et al. [23,24], in an investigation into billet qualityemploying heat flow and stress analyses of the mould wall, examined the mould-related problems of rhomboidity, longitudinal corner cracks, and breakouts.31Rhomboidity and corner crackingwill be discussed in this subsection.In these studies, corner cracks intransverse sections were shownto be 1to 2 mm deep but in extreme cases they could extend to cracks 6or 7 mm indepth. Examination indicated that the cracks occurred between primarydendrites and that the crack surfaces are smooth indicative of hottearing nearthe solidification front where residual liquid is present.The dynamics of billet rhomboidity is responsible for the presenceof thesecorner cracks. The mechanismfor rhomboidity proposed by Samarasekera et al.[23] involves nucleate boiling in the cooling water channel. Their mathematicalmodel predicted that under conditions of intermittent boiling, the mould assumesa rhomboid shape dependent on the instantaneous boiling events at the fourfaces. The mould face respondsto the heat flux by altering the air gap around itsperiphery: the colder faces contract more than the hotter face. The affected billetface cools differently from the other faces causing nonuniform shrinkage andrhomboidity.Corner cracking [23,24] may occur in those areas where tensile strains aregenerated due to the dynamics ofrhomboidity. If the strain at the obtuse cornersis sufficient, internal cracks may develop close to the solidification front where theductility is low. Figure 2.20 illustrates the strain at the corners. If theobtusecorner moves away from the mould, a locally wide corner gap is created and theheat flow to the mould is reduced.Tensile strains may develop at thesolidification front if surface reheating occurs. Penetration of the crackto thesurface may result depending on thecrack depth, the extent of reheating,and themagnitude of the generated tensile strains.32CrackstraincrackFigure 2.20: ScernatIc agram ilbstrattnQ the forrndon of a subsurfacecrack on agonaJ at octuse-angis corners of rhorrtoid il.t (23].If rhomboidity of the billet in the mould becomes excessive, binding canoccur at the acute angle corners producing transverse corner cracks. The cornerradius affects the location of the corner cracks. Longitudinal cracks appear at thecorners when large radii are employed while longitudinal cracks occur at off-corner sites [24] when small radii are employed.To control longitudinal corner cracking in high-carbon steels, cooling-waterflows have been reduced providing water quality was adequate [24]. Horizontalserrations have been machined on the cold face of the mould which also havereduced the prevalence of this type of surface defect. Machine alignment andcorrect adjustment of foot rollers also controls rhomboidity and longitudinal cornercracking [24].strain332.4.3 Oft-Cornr Internal CracksIn a study on mould behaviour and billet solidification, Bommaraju et at.[19] found that internal cracks were observed within 15 mm from any corner at adepth of 5 mm from the surface in many of the transverse bWet sectionsexamined. The cracks were random and in some samples associated with asurface depression. These off corner cracks are hot tears which originate at thesolidification front and, as they initiate within 5 mm from the surface, can beassumed to form in the mould region. Bommaraju et al. [19] have postulated thatbulging of a given billet lace due to inadequate taper is responsible for this qualitydefect. Bulging is inhibited at the corner where the billet temperature is lowerand strength is higher than the remainder of the face. Hinging occurs asillustrated in Figure 2.21 about the off-corner region at the point where the cooler,stronger corner resists the outward push by the ferrostatic pressure. Generationof the tensile strains at the solidification front creates internal cracks..Lu’dFigure 2.21: Schematic diagram showing generationof an internal crackdue to bulging of the biflet shell in the mould and ahinging actionin the ottcomer region L191.Mould wallFerrosta pressure000iSo’ed spiell342.4.4 Surface RoughnessSurface roughness has been observed by several researchers[6,16,20,25,26,27] to be carbon dependent. Billet surfaces of 0.1-percent carbonsteels are wrinkled as compared to billet surfaces of higher carbon grades. Grilland Brimacombe [27] have suggested that the peritectic reaction is responsiblefor this difference in surface quality. The 0.1-percent carbon steel transformsfrom delta to gamma in the solid state causing the shell, which initially contractsto pull away from the mould wall, to experience a large reduction in volume ofabout 0.38% [27]. Intermittent air gaps are created reducing the heat transferrate.2.4.5 PinholesPinholes may be detrimental depending on the application of the steel.Donaldson [31] in his examination into the quality of billets found that pinholesoccur more prevalently on the sides of the billets than at the corners and thatpinholes had a tendency to form in zones. The first zone includes those pinholeswhich are at the surface. The second zone occurs about 1.6 mm (0.0625 in)below the surface.Excessive lubrication has sometimes been claimed to be responsible forthe creation of pinholes. The oil pyrolyzes due to the elevated temperaturesproviding excess hydrogen and subsequently creating pinholes. However,Donaldson [31] suggests that it is steelmaking techniques which influencepinhole formation. Pickup of hydrogen due to the presence of excess moisture inthe ladle or tundish and of nitrogen from the atmosphere also contribute to the35formation of pinholes or blowholes in the steel. Care given to hydrogen andnitrogen pickup along with control of oxygen content and casting temperaturescan minimize the occurrence of pinholes. Brown [32] in his investigation intosurface quality of billet surfaces found that by tapping at lower temperatures andsubstituting calcium-silicon for 75% ferrosilicon in the ladle, pinholes becamenegligible. Singh and Blazek [25] noted that the pinhole problem was improvedby using an elevated reservoir and high sulphur contents.2.5 Lubrication OilsOriginally oils for mould-strand lubrication in the continuous casting ofsmaller steel sections such as billets were derived from animal and vegetablesources. However, mineral, and more recently synthetic oils, are being chosenfor utilization in mould lubrication. Synthetic lubricants are fluids that have adefinite molecular structure. Mineral and synthetic oils are being blended withvegetable oils to enhance the overall lubricity and stability of the lubricants [34].Lubricants are employed to reduce the contact or friction between twosurfaces. The most common theory of friction involves adhesion betweenasperities of two contacting bodies in which the load is supported by a relativelyfew asperities producing high normal stresses. Higher loads can plasticallydeform either material and form adhesive bonds referred to as microwelds whichare shown in Figure 2.22. The nature and strength of these microwelds dependson many factors such as the mutual solubility and diffusion of the two surfaces,the temperature, the contact time, the nature and thickness of either the oxidefilms or contaminants at the interface, and the properties of the lubricating film.36The coefficient of friction is defined as the ratio of the tangential forcerequired to cause shearing between the two surfaces and the normal force[34,35]:(2.1)NOAErGwhere t : shear stressa : normal stressAr : real area of contactEquation (2.1) indicates that the friction coefficient can be reduced by loweringthe shear stress at the interface of two surfaces. The magnitude of the normalstress is essentially equal to the yield stress of the asperity which is dependenton both the material and the temperature. The presence of contaminants,oxides, or lubricants does not affect the value of the normal stress but doesconsiderably reduce the shear stress. Any substance which is present impedesbonding between asperities reducing the strength of the junction and, therefore,decreases the required shear force for rupture.Friction also results due to ploughing. Ploughing shown in Figure 2.22occurs when the harder asperity scratches the softer surface during sliding. Theenergy supplied by the ploughing force plastically deforms the surface and cansignificantly contribute to the friction force. Lubricants are employed to reducethe shear strength of an interface and, therefore, to decrease the frictioncoefficient.37Surface DistortionFIN 2.22: a) PIcughbQ on a rIS and b) dtormItfl witt an abrasN (351.2.5.1 LubricatIng MechanismsSeveral modes of lubrication are recognized as operative in the reductionof shear stress at the interlace between two sliding surfaces namelyhydrodynamic (thick-film) lubrication, mixed lubrication, boundary layerlubncation, extreme pressure lubrication, elastohydrodyriamic andplastohydrodynamic lubrication, and solid film lubrication. Hydrodynemic lubricationIn hydrodynamic lubrication, a film of fluid prevents contact between thetwo surfaces (34,35]. The thickness of the film is approximately one order ofmagnitude larger than the roughness of either surface. Because the film can be38developed either hydrostatically by entrapping the fluid or hydrodynamically by awedge effect involving the sliding surfaces and the viscous fluid at the interface,the bulk properties of the lubricant are most important in the design for effectivelubrication. In thick film lubrication the viscosity of the fluid, the geometry of thesurfaces, and the relative motion of the surfaces generate sufficient pressures toprevent solid contact maintaining low friction coefficients of 0.001 to 0.02. Inmoving or sliding systems such as in the reciprocating mould system of acontinuous casting machine, a surface such as the wetted mould drags oil into anarrow opening similar to the gap which develops due to the thermal contractionof the thin solidified steel shell at the meniscus [34,35].The single most important bulk property of a lubricant operating in thismode of lubrication is viscosity. Newton in the seventeenth century proposed amechanism for quantifying the viscosity of a fluid. He developed a model of fluidflow based on the concept of layers of fluid sliding over each other as shown inFigure 2.23. The internal friction of the fluid produces shear stresses betweensliding layers of fluid which essentially retards the faster layer and acceleratesthe slower layer. Newton postulated that the viscous shear stresses were directlyproportional to the shear strain rate where the rate of shear is equivalent to thevelocity gradient. The constant of proportionality is the dynamic viscosity of thefluid [34].The viscosity of a fluid is strongly dependent on temperature andpressure. As temperature increases, the molecules in the fluid, having moreenergy, move farther apart causing the liquid to expand and the viscosity todecrease. Increases in pressure force the molecules of a fluid togetherincreasing both the intermolecular forces and the viscosity.39VFigure 2.23: Newtonian method of fluid flow [34). Boundary lubricatIonIn boundary lubrication, a thin layer of lubricant physically adheres to thesurfaces due to van der Waals forces or chemically reacts with the surfaces. Theimportant properties of a fluid operating in this mode is not so much its bulkproperties but its reactivity with the metal surfaces. Friction coefficients rangebetween 0.1 and 0.4 depending on the strength and thickness of the boundaryfilm. Under conditions of excessive temperatures, the film or layer is destroyeddue to wear or desorption of molecules from the surfaces resulting in theevolution of larger friction forces.The constituents of a lubricant conducive to boundary lubrication are liquidfatty acids as they are highly reactive with the metal surfaces. One or twomolecular layers have been shown to be effective in reducing the wear at theinterface by as much as 1000 times [34,35].Sliding velocity, u402.5.1.21 Fatty acidsFats and oils are esters of fatty acids and glycerol. Fatty acids are long,straight-chained hydrocarbons having a carboxylic acid attached at one end whileglycerol is a trihydroxy alcohol. Triglycerides are esters formed from one mole ofglycerol and three moles of fatty acid. The structural arrangement of each isshown in Figure 2.24. Oils contain a higher percentage of glyceryl esters ofunsaturated fatty acids than do fats [37,38].The properties of various fatty acids depend on chain length of themolecule, on the degree of unsaturation, on the geometric isomerism, and ondouble bond position with respect to the carboxyl group and with respect to eachother. As chain length increases, melting point increases and water solubilitydecreases [39]. Saturation raises the melting point. Fatty acids have higherboiling points when compared with substances of similar molecular weight due tothe strong polarization of the carboxyl group and consequent hydrogen bonding[40]. Melting and boiling points for some of fatty acids are tabulated in Table Oxidation of fatty acidsWhen certain fatty acids are exposed to air, the oils absorb oxygen andcan form a hard, elastic film. Oils can be categorized according to theirtendencies to dry: nondrying oils, semidrying oils, and drying oils. Olive orpeanut oil are examples of nondrying oils; cottonseed, corn, castor or rape are allsemidrying oils; linseed ortung oil are considered drying oils [38]. Drying oilscontain a high proportion of unsaturated acids: linolenic, linoleic, licanic, andeleostearin [40].41H HHHHHHHHH HHH H HHHOHSteanc Acid - Cl 8H3602 carboxylic acida) a long chain fatty acidH-O-CH2H-C-CI1H-O-CIl2b) glycerol0R-g-0-CH20R..g.0.C 3H+ciR--0-CII2C) a triglycerideFigure 2.24: The structura’ arrangement of a fatty ac,a glyceride, and a tnglyceride [37].TABLE2.1:MeltIngandboilingpointsforsomefattyacids.-COO4ONSCIENTIFICCARBONMOLECULARMELTINGBOILINGNAMENAMECHAINWEIGHTPOINTPOINT(8)(C)(C)LauricacidDodecanoicC12:O200.344131kMyriscicacidTetradecanoicC14:O228.454—PalniticacidHexadeceanoicC16:O256.462.5390760/268100StearicacidOctadecanoicC18:0284.573.1360d/23215Oleicacid9—OctadecenoicC18:1282.55.5286100/22815Linoleicacid912—OctadecadienoicC18:2280.5—5200_250u/22916Linolenicacid91215—OctadecatrienoicC18:3278.4—1123017/125005ArachidicErcosanoicC20:O312.5—328d/2031Erucic13doccosenoicC22:1338.6—265Superscriptsindicatepressuresatwhichboilingoccurred.d:indicatesdeconposition.43The glycendes of fatty acids, having double bonds, absorb oxygenproducing peroxides which catalyze the polymerization of the unsaturatedportions [39,41]. Further reactions result in cross linkages between different acidchains producing giant three-dimensional molecules [40] solidifying the oil[39,41]. The peroxides form primarily from linolenic and linolic glycerides [42].Table 2.2 lists the relative rates of oxidation of four unsaturated fatty acids [43].TABLE 2.2: Relative rates of oxidation of unsaturatedfatty acids.Fatty Acid Number of Relative OxidationDouble Bonds RateStearlc 0 0.6Oleic 1 6Llnoleic 2 64Linolenic 3 100Oxidation of the oil proceeds at three different rates. Induction is relativelyslow proceeding more rapidly as peroxide formation increases. The peroxidereaction is autocatalytic but as the available double bonds are reduced, thereaction decreases [37,42]. The rate of oxidation is greatly dependent onsaturation and structure. The effect of temperature on this reaction is as great orgreater than for most chemical reactions exponentially decreasing the inductionperiod [37].Oil oxidizes more rapidly in shallow layers as the induction period is morerapid than in thicker films. Lead, manganese, and cobalt are strong catalysts44increasing the oxidation reaction. Fatty acids and lubricationThe addition of small amounts of fatty acid to mineral oils dramaticallylowers friction values. The decrease in friction has been generally accepted asthe direct result of the fatty acid adhering to the surface of the metal substrate.The bond produced has sufficient strength to resist removal from the rubbinginteraction of two surfaces.Adherence of the acid to the metal can either be physical or chemical.Physical adsorption allows the molecules to attach continuously and detachthemselves to and from the surface respectively. As the temperature rises, thethermal oscillations increase causing a net de-adsorption of molecules reducinglubricity. In a boundary lubricated system dependent upon physically adsorbedpolar molecules, friction rises, wear increases, and sticking occurs attemperatures near the bulk melting point of the absorbate as molecules becomedisorientated and the solid film melts.Work done in 1949 by Tabor & Tingle [36] and Bowden & Moore [50]showed that physical adsorption cannot be entirely responsible for the lowerfriction values obtained by fatty acids over paraffins or alcohols. Both sets ofauthors asserted that it is the ability of the fatty acid to react with the metalsubstrate and form a soap which provides superior lubrication. Formation of themetallic soap results in lateral adhesion between molecules and this adhesion isnot destroyed until the soap film softens or melts. When the lateral adhesion isstrong, the lubricant reduces the contact area of the metal surfaces. Fatty acid45lubrication persists up to the melting points of the metallic soaps which are abovethe bulk melting point of the fatty acid.The rate at which fatty acids were adsorbed was found to be metaldependent. In adsorption experiments using solutions of saturated acids, Daniel(1955) [44] found that chromium, platinum, gold, silver, nickel, iron, tin andaluminum powders all adsorbed 90% of the total adsorbed within 5 minutes andreached equilibrium within 1 hour whereas a second group of powdered metals,copper, lead, zinc, and cadmium adsorbed continually with time [45].If physical adsorption has occurred on a metal and valence electrons havebeen exchanged, then chemisorption has taken place. Chemisorption increaseswith increasing temperature and is governed by an activation energy. Comparedwith physical adsorption, chemisorption has higher heats of adsorption (10,000 to100,000 cal/mole). The soaps formed due to this reaction are low shear strengthsolids and, therefore, are good lubricants as long as the temperature remainsbelow the melting point of the soap (120 0 C) [46]. Greenhill [47] showed thatbreakdown of lubrication occurs at the bulk melting point of the lubricant except inthe case of reactive metals which have formed metallic soap films [40]. Oxygen,water, and oxides present on a metal surface enhance the chemisorption of fattyacids to form a soap film and further reduce the friction coefficient.In their paper on oleic acid and friction measurements, Miller andAnderson (1957) [48] support the findings of Tabor & Tingle [36] and Bowden &Moore [50]. They found that without any oleic acid additions, the coefficient offriction increases from room temperature to 95 0 C and then decreases due tothe oxidation of the lubricant. When 1 % oleic acid is added the coefficient of46friction decreases until 120 C and then rises rapidly. They explain the frictionalbehaviour of the 1% oleic acid solution as a combination of physical andchemical adsorption processes. As the temperature increases, the adsorbedoleic acid molecules bound by the physical adsorption forces absorb sufficientthermal energy enabling them to react with some of the surface metal atoms ofthe substrate to form a soap. As the temperature exceeds 120 0 C, thermalagitation together with an increase solubility of both the soap and the oleic acidresults in destruction of the adsorbed layers and, thus, the increase in friction[36]. Mixed lubricationMixed lubrication becomes operative under the conditions when the filmthickness in hydrodynamic or hydrostatic lubrication is reduced by decreasingviscosity, decreasing sliding speed, or increasing the load. The load between thetwo surfaces becomes supported by hydrodynamic pockets and by metal-to-metal contact [34,35].Under these conditions, the film thickness is reduced from 10 times theheight of the asperities to 3 times and the friction coefficient increases from 0.001to 0.4 [34].In the study by Short et al. [9], mixed lubrication may be the operatingmode of lubrication, as the presence of carboxylic acid esters would provideboundary lubrication while the synthetic oil base would support hydrodynamiclubrication.472.5.1.4 Extreme-Pressure (EP) lubricationChemical reactions alter the surface irreversibly in EP lubrication.Sulphur, chlorine, and phosphorous in EP fluids form salts on the surfacespreventing the formation of microwelds. Temperature has a great effect on thesefilms destroying them as the temperature increases [34]. Elastohydrodynamic/Plastohydrodynamic lubricationWhen stresses are encountered, particularly in drawing and rollingapplications, which are sufficient to distort the tool or die either plastically orelastically, the geometry of the interface is altered. Changes in the pressuredistribution affect the viscosity of the lubricant helping to develop hydrodynamicfilms which cause a decrease in friction and wear. Solid-film lubricationLayer lattice compounds such as graphite and molybdenum disulfide aredeposited on the surfaces through a liquid medium such as water or oil. Solidlubricants have low shear strength due to their layer structure reducing theinterface friction. Complete covering of interfaces is essential to providesufficient lubrication. The strength of the film depends upon application methodsand metal surfaces. Currently, the effectiveness of polymers such as Teflon as alubricant is being studied [34].48The current literature on mould-strand interaction deals with frictionmeasurements employing accelerometers, strain gauges, or load cells primarilyin slab casting. Little research has been directed towards friction measurementsin billet casting. Correlation between the oscillation cycle and the load cellresponse has generally been neglected with the two exceptions of Mairy andWolf [6,7,8] and Schacht [11,12]. The load cell response shown by Wolf andMairy [8] closely resemble the results obtained in this study while the results fromthe study conducted by Schacht are similar.Researchers, particularly Wolf and Mairy [6,7,8] and Schacht [11,12], haveattempted to relate surface defects to force behaviour. In this study, an attempthas been made to elucidate the load cell response in terms of the mould-strandinteraction during an oscillation cycle in billet casting and finally, consideration isgiven as two which modes of lubrication are operating at the meniscus during thecasting of steel billets.49CHAPTER 3. SCOPE AND OBJECTIVESThe principal objectives of this investigation were to elucidate themechanism of lubrication in billet casting and to evaluate the effect of lubricatingoils on mould-strand interaction. The following methodology was adopted toachieve the objectives of the research programme.i) The mould assembly of an operating billet caster was instrumentedwith load cells and during casting the signals were continuously monitored forselected periods of time. Two campaigns of trials were conducted.ii) Linear variable displacement transducers (LVDTs) were also placed onthe mould table to monitor the oscillation cycle; and output from the LVDTs andthe load cells was fed to the data acquisition system. Load behaviour could thenbe correlated with the oscillation cycle.iii) Steel composition, casting speed, type of lubricant, lubricant flow,mould temperature distribution, bulk inlet and outlet water temperatures, andwater temperatures at the top of the four faces were also monitored during thetrials.50iv) The influence of the following variables on loads was examined:1. position of mould during the oscillation cycle2. type of mould lubricant3. lubricant flow4. carbon content of the steel and5. casting speed.v) Each mould lubricant tested was analyzed for viscosity-temperaturebehaviour, flash point, and fatty acid content.vi) Samples of billets cast during the data acquisition times were collectedand shipped to UBC where they were examined for surface defects such as pinholes, oscillation mark depth, cracks, and depressions. Correlation of lubricantand lubricant flow with billet quality was attempted.51CHAPTER 4. EXPERIMENTAL PROCEDURETwo plant trials were conducted successfully during this study. The firsttrial at Western Canada Steel (WCS) in Richmond, British Columbia duringMarch 1988, provided valuable insight, experience, and confidence in theproposed method for sensing the mould-strand interaction. In November 1988the second trial at IVACO Rolling Mills in L’Orignal, Ontario proved conclusivelythat mould-strand interaction could be detected with load cells. Furthermore, thistechnique could be utilized on-line on an ongoing basis as a quality control tool.Described in this chapter are the relevant particulars of, each industrialoperation, the mould instrumentation, and data acquisition. Informationpertaining to the mould oils tested and billet analysis are also included. Finally inthe last section, details on the two plant trials are provided.4.1 Industrial Operations4.1.1 Details of castersBoth operations employ four-strand curved billet casters having a 7.92metre (26 foot) radius. The caster at WCS was constructed by Rokop while thecaster at IVACO is a Concast machine. WCS casts 108 by 102 mm (4.25 by 4in) to 140 by 127 mm (5.5 by 5 in) rectangular section sizes; IVACO typicallycasts sections of 120 mm square (4.7 by 4.7 in). Both mini-steel mills melt scrapin an electric furnace of 65 tonne capacity. The section sizes of the billets castduring the trials at WCS and IVACO were 140 mm square and 120 x 122 mm,respectively. Details of the casters can be found in Table 4.1.524.1.2 Mould Assembly4.1.2.1 Western Canada SteelWCS supplied a mould housing for the installation of the load cells. In thispreliminary trial, the instrumentation of the mould was limited to these measuringdevices only.A double-taper mould tube having a 2.2 %Im upper taper and a 0.6 %Imlower taper together with a wall thickness of 12.7 mm was supplied. The lengthof the tube was 762 mm (30 in) with the break point in the taper occurring at 348mm (13.7 in) below the top of the mould.4.t2.2 Ivaco Rolling MillsA complete mould assembly furnished by IVACO was shipped to UBC forinstrumentation. The mould housing was retrofitted by P&S Engineering ofVancouver. A new top plate was constructed to specified tolerances to ensurethat the mould would be uniformly constrained on all four sides. The plate wassplit along the diagonal for ease of assembly and dowel pins were utilized tosecure the mould precisely in position. The water baffle manufactured byAccumold also met specified tolerances to ensure a uniform water gap aroundthe mould periphery. Original spacers near the meniscus on the inside walls ofthe baffle were removed to prevent constraint and permanent distortion of themould wall which expands thermally outwards during operation. The water bafflewas positioned precisely into the top plate and secured employing a split-platewhich also supported the mould on all four faces to a tolerance of less than 553thousandths of an inch.TABLE 4.1: Details of the caster and mould parametersduring the plant trials at Western Canada Steeland IVACO Rolling Mills.WESTERN IVACOCANADA ROLLINGSTEEL MILLSType of Machine: curved billet castersMachine Designed by: ROKOP CONCASTRadius of Machine: 7.9m (26ft) 7.9m (26ft)Mould Assembly:Taper type: double parabolicTaper: 2.2 %/m upper as shown in0.6 %/m lower Figures 4.1 ,4.2Mould Thickness: 12.7mm(1/2in) 12.7mm(1/2in)Mould Length: 762mm(3Oin) 711 mm(28in)Support: two-sided four-sidedMeniscus Level: 167 mm 100 mmFoot Rolls: none noneCooling Water Velocity: 12 rn/s 14 rn/sTundish Capacity: 14 tons 6 tonsBreak point: 0.35 m 0.33 mMould Oscillation Characteristics:Stroke Length: 9.5mm(3/8in) 9.5mm(3/8in)Frequency: 90 cpm 120 cpmType of oscillation: sinusoidal sinusoidalCasting speed: 36-40 mm/s 34-42 mm/s85-95 in/mm 80-100 in/mmNegative strip perIod: 0.097-0.1 36s 0.125-0.1 875sReformed parabolic mould tubes from Accumold were used in both thetest strand and the control strand. The tubes were of copper-chromium-zirconium and were 12.7 mm (0.50 in) thick. Tables 4.2 and 4.3 list the internaldimensions and corresponding tapers of the parabolic tubes at 2.54 cm intervalsfor the curved and straight walls respectively. Figures 4.1 and 4.2 show tapermeasurements for the internal curved and straight wall dimensions of the mould54tube employed as the test mould. These traces were completed at UBC prior tothe plant trial. Details of the casters and operating conditions are summarized inTable 4.1.TABLE 4.2: Parabolic tapers of the curved sides of the mouldtube employed in the Ivaco plant trial.Distance Below Mould Dimensions TaperMeniscus(mm) (in) (mm) (in) (%/m)0.0 0 124.4 4.897 4.9025.4 1 124.2 4.891 4.0850.8 2 124.1 4.886 3.2776.2 3 124.0 4.882 2.45101.6 4 123.9 4.879 2.45127.0 5 123.9 4.876 2.45152.4 6 123.8 4.873 2.45177.8 7 123.7 4.870 1.63203.2 8 123.6 4.868 1.63228.6 9 123.6 4.866 2.45254.0 10 123.5 4.863 1.63279.4 11 123.5 4.861 0.82304.8 12 123.4 4.860 1.63330.2 13 123.4 4.858 1.63355.6 14 123.3 4.856 1.63381.0 15 123.3 4.854 0.82406.4 16 123.3 4.853 1.63431.8 17 123.2 4.851 0.82457.2 18 123.2 4.850 1.63482.6 19 123.1 4.845 0.82508.0 20 123.1 4.847 0.82533.4 21 123.1 4.846 1.63558.8 22 123.0 4.844 0.82584.2 23 123.0 4.843 0.82609.6 24 123.0 4.842 0.82635.0 25 123.0 4.841 0.82660.4 26 122.9 4.840 0.82685.8 27 122.9 4.839 0.82711.2 28 122.9 4.838 0.8255TABLE 4.3: ParabolIc tapers of the straight sides of themould tube employed In the Ivaco plant trial.Distance Below Mould Dimensions TaperMeniscus(mm) (in) (mm) (in) (%lm)0.0 0 122.4 4.819 4.9025.4 1 122.3 4.813 4.0850.8 2 122.1 4.808 2.4576.2 3 122.0 4.805 3.27101.6 4 121.9 4.801 2.45127.0 5 121.9 4.798 2.45152.4 6 121.8 4.795 1.63177.8 7 121.7 4.793 2.45203.2 8 121.7 4.790 1.63228.6 9 121.6 4.788 1.63254.0 10 121.6 4.786 1.63279.4 11 121.5 4.784 1.63304.8 12 121.5 4.782 1.63330.2 13 121.4 4.780 0.82355.6 14 121.4 4.779 1.63381.0 15 121.3 4.777 1.63406.4 16 121.3 4.775 0.82431.8 17 121.3 4.774 0.82457.2 18 121.2 4.773 1.63482.6 19 121.2 4.771 . 0.82508.0 20 121.2 4.769 0.82533.4 21 121.1 4.767 1.63558.8 22 121.1 4.766 0.82584.2 23 121.1 4.765 0.82609.6 24 121.0 4.765 0.82635.0 25 121.0 4.764 0.82660.4 26 121.0 4.763 0.82685.8 27 121.0 4.762 0.82711.2 28 120.9 4.761 0.821 —.—124.1,.-,.—-1‘I?Ei._..E U) C.2—, C 0 E14——-.ti1.•‘.—-‘1 2-.11‘-i.-).,—.7—I—I.—.—..—4••.•DistanceFromTop(mm)Figure4.1IVACOparabolictaperforthecurvedwall.1‘‘21“11‘21q p121.7U) C-U)121.’EE121.4•0121.’121.2C1.l1‘2I27.’n.-.-t(p -JDistanceFromTop(mm)4X’&o:.Figure4.2IVACOpartolictaperSmthetmlgt*waN.584.1.3 Lubrication Distribution SystemCurrently, industrial billet casting operations employ a mould lubricatingsystem which consists of a top plate having a single oil channel. This system isschematically shown in Figure 4.3. The oil is pumped into this channel usuallythrough a single inlet and is assumed to distribute and lubricate all four wallsevenly. Oil distribution measurements made by Walker and Hemingway [51] atsix steel plants clearly indicated that the oil distribution was distinctly nonuniformaround the mould. A subsequent investigation of the oil distribution in the mouldwas completed by Perri and Bakshi [52] and resulted in a new lubricator platedesign. The new design (patents applied for) provides uniform oil flow to the fourfaces of the mould. Another improvement to the oil distribution system was thedevelopment of a new gasket which reduced the feeding area in the cornerregions providing a more equal distribution of oil around the mould periphery.This revised design was employed in the successful second plant trial at IVACOwhile the original design was tested in the preliminary WCS trial.An mROY electric pump manufactured by Milton Roy Industries, Ltd.,Model Number 5R1 1OA-1 17 was acquired to pump the test oils from theircontainers into the oil system. The pump was checked prior to the trial to ensurethat it was in good working order and that the oil distribution was uniform.Recalibration of the system was made at the plant by collecting the volume of oilflowing down each face during a specific time interval.FrontOil lrutFIgure4.3OrigInalplatedesignPar UdcatlonsystemusedIntheWesternCanadaSteel plar4trial.Oil ChannelWlh:1InchDeØh:318InchReplaceableSteelRing01604.2 Mould InstrumentationTo monitor mould behaviour and mould-strand interaction, the test mouldat Ivaco was instrumented as follows:• Load cells were located between the mould housing and the oscillator table tomeasure the loading on the mould dunng operation.• Linear variable displacement transducers (LVDTs) were placed on the mouldtable to record the oscillation characteristics of the mould system.• Thermocouples were installed in the walls of the copper mould tube to measurethe temperature distribution within the mould wall, during operation.• Inlet and outlet water temperatures at the centre of each of the four faces werealso measured using thermocouples.4.2.1 Load CellsMould loading was measured with time employing strain gage load cellssupplied by OMEGA. The load cells are of the LCG series capable of handling44.5 kN (10 000 lbf) of compressive loading. As illustrated in Figure 4.4, theseload cells are 3.81 cm (1.5 in) in diameter and have a total height of 1.575 cm(0.62 in). The compressive loading is sensed by the centrally positioned loadbutton.Specifications supplied by the manufacturer indicate that the load cells are61designed to handle 150% of full scale load, operate to temperatures of 121 C(250 0 F), and have a repeatability value of ± 0.05% Full Scale (± 5 lbs or .± 22.7N).Figure 4.4 SchematIc ifluatration o4 load cell. Al dlmeneions are Wi miltlmetres.The load cells were positioned between the mould housing and the mouldoscillating table as illustrated in Figure 4.5. Four circular recesses, schematicallyshown in Figure 4.6, were machined into the mould housing plate into which anidentified load cell was placed. Load cells 1 and 2 were installed in the front ofthe mould system while load cells 3 and 4 were set in the rear positions. Sincecontact between the housing and oscillation table occurred only through the fourload cells, any changes in loading on the mould would be sensed by the loadcells.Due to the coarse threads on the hold down bolts which connect thehousing to the oscillating plate, a bolt-spring assembly, represented in Figure 4.5,was designed to control the initial torquing load on the load cell by partitioningany external load applied to the system. Appendix A explains in detail the381p981Figure4.5SchematicUhis*raUonatpolionkigatbadciibetwenmouldhouskigandmouldoscUlntbandtheboi-ipdngaangem.usedkitheIvacopb,trial.HoldDownBoltAdiustableSpocerLoodCellOscilloto,ToNeI.63BOTTOM V(EWFijre 4.6 SchematiC ilkjstratlon of mou cut out toI iCad cellplacement.Eas f/WestSide79.375mmEast/WestSide262mmSiDE VIEW64partitioning effect of the spring in this system. The 19.5 mm (3/4 in) innerdiameter springs of length 50 mm (1.9685 in) were obtained from the Extra HighPressure Series produced by Producto. The XHP Series springs are made fromSAE 6150 (0.5% C, Chrome and Vanadium additions) steel and have a springconstant of 22.25 kN per 25.4 mm (5000 lbf per inch). The ends of the springswere ground flat ensuring even contact with the top washer and the mould plate.During installation of the mould housing on the oscillator table, the loadcells were secured within the recessed circular hollows to prevent overloading.Once the mould housing had been lowered onto the oscillating plate, the load cellpositioner shown in Figure 4.8 was turned until the load cell button contacted theoscillating table. The bolts were then torqued until the springs were fullycompressed at which point the load cells registered loads of approximately 8.9kN (2000 lbf).The 10 volt DC excitation voltage required by the load cells was suppliedby stepping down a 12 volt DC battery with a simple resistance circuit shown inFigure 4.9. The excitation voltage could be set to exactly 10 volts DC by theadjustment of the 200 ohm potentiometer.Prior to the plant trials, the load cells were calibrated on an lnstronmachine by loading them in compression. Each load cell was calibrated from4.45 kN to 44.5 kN (1000 to 10000 lbf) in 4.45 kN (1000 lbf) increments andcalibration curves were established.65Figure 4.7 Load cell positioner used to ensure contact of loadcell button on oscfllating table.635mm635mm1/2-20 N.E - 381mm9525mmH 3823mm 166200OhmAdj.Pot.SPST+.....IIIIII12V+O—2OmV OutputFigure 4.8 Step down circuit for load cefls.674.2.2 Linear Variable Displacement TransducersCorrelation between mould cycle characteristics and mould loadingresponse was achieved by the placement of two linear variable displacementtransducers (LVDTs) on the front and rear of the mould plate. LVDT #1 was aSelabs SE 381115 while LVDT #2 was a Sangamo ACR 15. The analog signalof these LVDTs was sampled at a frequency of 40 Hz. Both devices wereattached to the static main frame of the caster at a point which allowed theretractable sensor to rest on the mould table. The sensor had a lengthappropriate to the oscillation stroke. Placement of the oscillation monitors isshown in the plan view illustration of the mould system in Figure 4.9.Two Daytronic 3230 signal conditioners were employed to operate theLVDTs. The resistive network shown in Figure 4.10 was used to step down theLVDT output voltage. To remove high frequency noise the signals were passedthrough a low-pass, resistance-capacitance filter circuit, also illustrated in Figure4.10, prior to being fed into the data acquisition system.Each LVDT and its signal conditioner were calibrated to provide an outputof + 2.500 volts for a + 10 mm displacement and + 10.0 my output from the stepdown and resistance-capacitance filter circuit.4.2.3 ThermocouplesSeventy-two OMEGA Type T copper-constantan intrinsic thermocoupleswere embedded in three of the four mould walls to monitor the heat-extractioncharacteristics of the mould. Details of the thermocouple insertion procedureOscillatorOscillatorTableHousingTableMouldI-I______LoodCell4—[17OLoodCell2LVDT±1____LoadCell3j’0iI°oj—LoodCellI1LI.LVDTTopViewofMouldHousingFigure49PlacementolLVDTSandloadcellsemployedintheIVACOplanttdal.691 kilo ohm220 kilo ohm+1- 10 millivolt dc 0/p+1- 2.5 volts dc i/pFigure 4.10 Rsilve neiwod’ for signa’ condltlonrs used inconpnctiofl with LVDTs.+0.68 microfarad2 kilo ohm220 kilo ohm+70have been described by Brimacombe et al. [53]. Figure 4.12 shows thepositioning of the thermocouples in one of the straight walls: one column of 18thermocouples was placed along the centreline with a second column similarlypositioned but offset from the centreline by 35 millimeters. The inside andoutside curved walls had 18 thermocouples only along the centreline. Allthermocouples were nominally set 6 millimeters from the cold face of the 12.7mm (0.5 in) thick mould wall and held in place by threaded copper plugs.Two-wire, copper-constantan thermocouples were employed to monitorbulk water temperatures at the inlet and outlet sites and the temperature of thewater at the top, centre of each wall.4.2.4 Casting speedCasting speed was monitored by tapping into the signal from the 40-voltDC casting speed meter. The 40-volt signal was stepped down and filtered toproduce a 10 millivolt DC signal as shown in Figure 4.13 which was thentransmitted to the data acquisition system.4.3 Data Acquisition SystemAcquisition of all data was achieved through an anaglog-to-digitalconverter board installed in a portable Toshiba 3200 PC. The AID board systemwas made by Metrabyte and was Model DAS 8, EXP-ENC. The analog signalsfrom all the electronic devices were passed through the analog-to-digital interfacewhich had a sensitivity of± 0.0122mV. The resulting digital output was stored onI.2.•—3..7.I.•—10•—II.•__.85mm115mm145mm195mm245mm310mm340mm370mm400mm450mm500mm550mm600mm700mmMeniscus115mmPlenumDivider265mmTaperBreakpl355mm100mm130mm170mm220mm12•II••14•Is.16 Ii.IIIIII35mmI-.Figure4.11PosmoningofthermocouplesInastraighiwall.721 kilo ohm1 mega ohm+10 millivolts dc 0/p40 volts dc i/pStep down circuit for the casting speed signal.+0.68 microfarad2 kilo ohm1 mega ohmFigure 4.1273the hard disk of the computer using the software program, LABTECKNOTEBOOK developed by Laboratory Technologies Corporation inMassachusetts.Thermocouple and casting speed signals were sampled at 1 Hz while a 40Hz sampling rate was used for the load cells and LVDTs. Using NOTEBOOKfour data acquisition files were set up:1. TCHI: Thermocouple, casting speed, and time channels were collectedsimultaneously at 10 Hz for 1 minute up to three times a heat.2. TCLOW: Thermocouple, casting speed, and time channels werecollected simultaneously at 1 Hz for 10 minutes two to three times a heat.3. LCA: Signals from the four load cells, the two LVDTs, and the internaltime channels were collected simultaneously at 40 Hz for 120 or 180 secondstwo to three times during the heat.4. LCB: Signals from the 5 thermocouples set along the centreline nearthe meniscus, which was assumed to be 115 millimetres down from the top of themould, and the casting speed were collected simultaneously at 1 Hz. Both loadcell signals were acquired concurrently.744.4 Lubricating OilsThree casting oils were chosen for the second trial to evaluate their effecton mould-strand interaction. Selection was based on two physical properties,namely flash point and viscosity, and on the chemical composition of the oil.Lubricants A and B were mineral/synthetic blends. The third oil, Lubricant C wasalso a mineral based oil with a small addition of rapeseed. The oils wereanalyzed for viscosities, flash temperatures, and fatty acid content at anindependent laboratory in Vancouver and will be discussed in Chapter 5.To observe the effect of oil flow on the mould-strand interaction, four rateswere selected for each oil: 54, 44 , 34 , and 24 mI/mm.4.5 Billet SamplesIn the Ivaco trials, billet samples of 25 to 30cm (10 to 12 in) in length werecut from both the instrumented test strand and the adjacent control strand. Usingfacilities at UBC these samples were metallographically prepared and examinedfor cracks, pinholes, oscillation marks, and depressions. The billet surfaces weresand-blasted and surface roughness was measured with an automatedprofilometer. Transverse slices were etched in an aqueous solution of 50:50 HCIat 85 0 C for as long as necessary, typically 30 minutes, to reveal the as-caststructure. Longitudinal sections were etched in picric acid and examined formacrostructure, porosity, and segregation. The subsurface structure beneath the75oscillation marks was photographed and examined. Thus an attempt was madeto correlate lubricant type, flow, steel grade, and billet quality.4.6 Plant Trials4.6.1 Trial at Western Canada SteelBecause the preliminary trial at WCS was conducted simply to explore theuse of load cells to measure the loading on the mould during casting, the mouldwas instrumented only with load cells and not with thermocouples. Five heatswere periodically monitored at several data acquisition rates. The steelcompositions in these heats are tabulated in Table 4.4 while the acquisitionconditions are summarized in Table 4.5. The lubrication conditions during thetrial corresponded to normal practice and, therefore, oil type and flow were notvaried at this time. Measurements made in December, 1987 by members of TheCentre for Metallurgical Process Engineering, indicated that the distributionaround the four mould walls was nonuniform and that the total flow was high (100mI/mm). The trial essentially consisted of load cell measurements under normalcasting conditions to test technique and equipment.4.6.2 Procedural changesThe importance of simultaneously monitoring the mould oscillation cycleand the load cell response became apparent from this plant trial. Only throughcorrelation of mould position and, therefore, mould velocity, with load cellresponse is it possible to determine how oscillation cycle events such as76upstroke, downstroke, and negative strip affect the loading of the mould.Therefore, two LVDTs to sense mould position were incorporated intoinstrumentation for the second trial; one was placed on the front of the housing,and the second on the rear.TABLE 4.4: Steel compositions of the heats monitored at WCS.Heat C Mn S P Si Cr NI Cu Al Mn/S Mn/SiRatio Ratio304400.31 1.17 0.026 0.010 0.16 0.16 0.18 0.36 0.006 45.0 7.330441 0.32 1.16 0.045 0.012 0.16 0.16 0.20 0.43 0.005 25.8 7.3304420.34 1.19 0.027 0.013 0.16 0.14 0.18 0.35 0.004 44.1 7.4304430.37 1.13 0.020 0.015 0.14 0.14 0.17 0.35 0.004 56.5 8.1304440.20 0.66 0.023 0.010 0.21 0.10 0.14 0.33 0.005 28.7 TRIAL AT IVACO ROLLING MILLSThe project objectives were realized during the plant trial conducted atIVACO. Computerized data from both load cells and LVDTs were accumulatedduring several heats together with mould temperatures. Data from 22 heats wereaccumulated during the plant trial at IVACO. The carbon content of the steelvaried from 0.035 to 0.42 per cent as can be seen from Table 4.6 which lists theheats and the composition of the grades cast; Table 4.7 summarizes the castingconditions; and the lubricating conditions are tabulated in Table 4.8. The77Tabli 4.5 CondItions .mploy.d for data acquisitionduring th Wt•m Canada StI plant trial.AT RUN DATA ACQUISITION TINE 0? RUNRATE(Hz) (.111)30440 3 20 start up4 50 215 30 346 40 487 25 6330441 9 50 2410 25 3511 20 4512 50 6430442 13 20 start up14 20 1515 20 3216 20 5117 20 no load; oscillations30443 18 20 start up19 20 dummy bar disconnect20 100 1221 20 69; end30444 22 20 start up23 50 2024 20 4525 20 6026 20 end78lubricant, CC-b, at a flow of 54 mI/mm, was employed for the majority of heats.However, the flow of Lubricant C was varied during the casting of a 0.15%carbon heat when values of 44 to 34 and 24 mI/mm were set. The flow ofLubricant A was varied during a sequence of 0.09% carbon heat while for theLubricant B the flow was reduced from 54 to 24 mI/mm during the casting of a0.12% carbon steel heat.Two or three billet sections were cut from each heat monitored. Attemptswere made to take the cast sections during periods of data collection. Each billetwas identified with a two digit number: the first number represented the strand (2for the control strand and 3 for the test strand); the second number indicated theapproximate time at which the billet was cast in the mould. Table 4.9 lists thebillet samples, the time during the heat at which they resided in the mould, andthe corresponding lubricating conditions.Table4.6ComposItionof heatsmonitomdduringtheIVACOplanttrial.Heat Composition(%)HeatNa/SHa/StNo.CHoSPStCrNICuHoSnAllattolatto-824636.046.530.—900.036—80—900.034—29.029.0152927841663302515302786360.0380.0381581287763113155928384174156128424378154128062342B2463885—900.036—80—850.034—29.229.2152627781670303815912896661180.0380.03629.229.2157728705172156828544276B2463995—1000.040—87—930.037—29.229.2152927841652300515942902651180.0420.0391572286143771567285315492820B2464080—820.034—820.03529.229.216062923761370.035153027861685306515792875498915662850366482464190—920.038—87—900.036—29.229.21581287752930.0390.0381529278416733043156728523868157428661566285182464285—900.036—83—900.035—29.229.115292784166930361567285338690.0380.038156728523869154728161832824643900.038900.03829.229.215142758169030741572286258104157128595710115832882TALZ4.1.CastIngConditionsofHeatsMonitoredatIVACOPlantTrial.(cout’d)HeatCASTINGSPEED:MOULDWATER:LiquldusLadLeTundtshNo.ControlStrand:TestStrand:ControlTestTemp.Temp.Temp.Superheat(In/mm)(mis)(In/min)(mis)(tie)(t/s)CFCFCFCP824644900.038900.03829.329.21513275516813057155028223161155528314216155228253970824645900.038900.03829.229.315122754163329711559283847841547281715342794224082464688—890.037—85—900.036—29.329.414952754166730331557283562LII0.0380.038154428124988153427943868153327923868824647900.038900.03829.229.214942721L6242956153227893767153127883767153828004479153327913868A234081000.042900.03828.828.41528278316803056156628513869156128423360156428473665155528312749824649900.038850.036——1528278316643027156628503867156128413358155928383L55A23409800.034800.03428.828.61518276416282962154728172953154628142853153828002036824650800.034850.03628.828.5152127691628296215362798154528132444154028041935153928031834TABLE4.7.CastingConditionsofHeatsMonitoredatIVACOPlantTrial.(cont’d)HeatCASTINGSPEED:MOULDWATER:LiquidusLadleTundtehNo.ControlStrand:TestStrand:ControlTentTemp.Temp.Temp.Superheat(In/mm)(a/a)(In/mb)(rn/n)(1/n)(1/n)CFCFCFCPA2341278—820.033—78—830.033—28.728.515232773165230051553282830550.0350.035155328273055154128051832824653NANA28.628.5152327741651300415412805183115612841386715472816244282465488—950.03)—85—920.036—28.828.615212769167630491555283134620.0400.039156928574888156628514582A23413850.036830.03528.828.71520276916273049156128424173155928383969155328273368824655950.0401000.04228.828.7152227711655301115682854468115782873561021573128645193824657NANA28.828.714962724165430101512275415082747122383Tabi. 4.8 Lubilcatlng conditions for runs at IVACO plant trial.REAT RUN CARSON MOULDFL arzCONTENT LUSRICANT() (.1/am)0.046824636 NOOIL.PRNcC—b 541 CC—1034824637 2 0.040cc—iO824638 4 0.068cc—b 545 CC—la 546 cC—lO 54824639 7 0.045c—o 548 CC—10 54824640 9 0.035cc—tO 5410 CC—tO5411 CC—tO54824641 12 0.041 CC—10 5413 CC—105414 CC—1054826642 15 0.043cc—tO 5416 CC—L054• 17 CC—tO34824643 21 0.170CC—tO 5422 CC—tO5423 CC—b34824644 24 0.180CC—10 5425 CC—tO34324645 26 0.180CC—tO 5427 CC—].0 54324646 28 0.400CC—b 5429 CC—105430 CC—tO34324647 31. 0.420cc—b 5432 CC—1054824648 NDOIL1 — —NOOIL2A - -A23408 34 0.051cC—la 5435 CC—1034A23409 36 0.150cC—jO 4437 CC—jO 3438 CC—ID24826650 39 0.120 ——A23412 40 0.090STEELSKIN 3441 STULSKIN4442 STEELSKIN34824653 43 0.094STE!LSKIN 3444 STEELSKIN2445 — —46 SrEELSKIN24824654 47 0.12051—LN 5448 51—tN54A234t3 50 0.12051—LN 5451 51—tN5452 51—tN5484TibIa 43 Ust of bIfl.ts coU•ctd during th• IVACO plant trIal.Grade unit Seaples Straad rundisb-rp. (‘C)Cotte1 test (ei) (‘C)Strati Strati324636 10083—1 1546 190.046% 3—215313—3 1530 3$24637 1008 2—3 3—313 1559 410.04% 2—1 3—64 1541 23326638 1008 2—33—3 13 157, 530.068% 2—7 3—738 1561 4282463 1008 2—3 3—3 12 1572 ‘30.045% 27 3—737 1549 20324640 1008 3—Z7 1.594 640.035% 237 41 1566 36324641 1008 2—33—3 12 1368 390.04L%3—7 37 1567 38324642 1008 2—3 3—31.3 1567 382—7 3—7 38 1.550 23.324643 1018 2—33—3 12 1572 580.17% 2—73—7 37 1582 68324644 1018 2—3 3—3 12 1555420.1.8%3—7 37 1.552 39324645 1018 2—3 3—3 12 1.559470.1.8%3—7 37 1.53927824646 1039 2—3 3—313 1.548 53o.oz27 37 381.534 38324647 1019 2—7 3—737 1.534 39A23408 1001 2—3 3—312 1.561 330... 051.1.324649 QL 2—3 3—312 1.564 36k23409 1015 2—3 3—314 1546 2$0 2—63—6 35 1538 20“ 2—1 3—1 48TabI4.9LIstofblll•tscoliactdduringfliIVACOplanttrialcontlnud.GradeStUetSa.pl.sStrandruadtsbSupeiatBeatTine(C)No.ControlTest(em)(‘C)StraStread3246501010NoneNonefl-lizk2341210102—33—314155330o0973—5281553302—73—74215411882465310102—33—3U1546230.09%2—73—73315613852465410122—33—3121557320.12%2—73—736156645A23413101.2NoneNonefi8246551.0102—33—3U157032465760372—3N.A.1508120186CHAPTER 5. RESULTS OF INDUSTRIAL TRIALSIn this chapter the load cell data obtained in the two plant trials ispresented. The load cell data collected at the Western Canada Steel (WCS)plant trial displayed the same periodicity as the mould oscillation cycle. However,because the oscillation signal was not logged while load cell data was beingmonitored, it was difficult to correlate mould position with the load variation. Animportant outcome of this first trial at WCS was that it confirmed that load cellscould be employed to detect load changes during mould oscillation. The load celldata accumulated during the IVACO plant trial was extensively studied, however,successful correlation between mould position and load variation during theoscillation cycle enabled a detailed analysis of the upstroke, downstroke, andnegative-strip periods.The physical properties of the mould lubricants tested during the IVACOplant trial were determined and compared. Some differences were noted whichmay account for the observed variation in the load cell response from one oil tothe next.An evaluation of billet quality was undertaken by the UBC BilletQuality Research Team which included the characterization of the surface andinternal quality.875.1 Western Canada Steel ResultsThe stored millivolt signals from the load cells were transferred to the mainframe UBC computer where the signals were converted to loads using theestablished calibration relationships. Figure 5.1 illustrates the typical, periodicresponse of a rear load cell obtained during both trials. The waveformcharacteristically has a flat, broad upper peak containing secondary peaks and asingle, narrow lower peak. The magnitude of the average maximum loadappears to be a constant whilst the lower peaks vary with time.The total change in load for each oscillation cycle was determined toexamine the effect of carbon on load cell response. Averages and standarddeviations were obtained and the results are grouped according to heats andsummarized in Table 5.1. Clearly, there is no apparent carbon effect in thecarbon range 0.20 to 0.37 percent. Percent deviations are significant indicatingconsiderable fluctuations within individual runs. During the analysis of the IVACOdata, it became evident that the variation in minimum load was due to onespecific operating parameter which will be discussed later.Examination of the front load cell response indicated possible mechanicalinterference. During some runs the output from load cell 1 appears to bedampened and in some cases the periodicity of the waveform vanishes. Figure5.2 illustrates this effect on load cell 1 in Run 22. The irregular loading cyclesobtained during the WCS trial could result either from the larger 0-rings utilizedto seal the inlet and outlet water orifices or from failure to fully compress thesprings. Redesigning the water seal system prior to the IVACO trial was notfeasible so for the second plant trial the springs were fully compressed.12000-255011000--235010000-ciooo•2150z 4195048000-•17507000LubrIcantAatioomi/mm-1550Heat30444;Run22;Carbon:0.20%6000-—1350110111112113114115116117118119120121122123124125TIME(s)Figure5.1:Typical,periodicresponseof arearloadcellobtainedduringbothtrials.89Table 5.1: Peak-to-peak averages and standard deviationsfor the Western Canada Steel plant trialresults.EAT# REAR LOAD CELLSLC1 LCZRUN L SD SD L SD SD(N) (lbf) (N) (lbf) (Z) (N) (lbf) (N) (lbf) (t)30440 3 966 (217) 396 ( 89) 41 1722 (387) 534 (120) 31(0.31% C) 4 1633 (367) 307 ( 69) 19 2483 (558) 418 ( 94) 175 1086 (244) 427 ( 96) 39 1615 (363) 329 ( 74) 206 1210 (272) 427 ( 96) 35 2189 (492) 409 ( 92) j7 1509 (339) 263 ( 59) 17 1584 (356) 254 ( 57) 16AVE 1282 (288) 280 ( 63) 22 1918 (431) 396 ( 89) 2130441 9 4134 (929) 743 (167) 18 1482 (333) 356 ( 80) 24(0.32% C) 10 3022 (679) 445 (100) 15 1682 (378) 316 ( 71) 1911 886 (199) 289 ( 65) 33 1179 (265) 320 ( 72) 2712 1606 (361) 245 ( 55) 15 1722 (387) 258 ( 58) 15AVE 2412 (542) 1451 (326) 60 1517 (341) 249 ( 56) 1630442 13 1117 (251) 659 (148) 59 1486 (334) 948 (213) 64(0.34% C) 14 1219 (274) 583 (131) 48 1562 (351) 409 ( 92) 2615 636 (143) 169 ( 38) 27 1202 (270) 236 ( 53) 2016 797 (179) 187 ( 42) 23 983 (221) 218 ( 49) 2217 957 (215) 369 ( 83) 39 770 (173) 325 ( 73) 42AVE 943 (212 236 ( 53) 25 1202 (270) 334 ( 73) 2830443 18 2982 (670) 1188 (267) 40 3155 (709) 1055 (237) 33(0.37% C) 19 2808 (631) 783 (176) 28 3168 (712) 868 (195) 2720 1798 (404) 556 (125) 31 1825 (410) 623 (140) 3421 703 (158) 169 ( 38) 24 619 (139) 174 ( 39) 28AVE 2074 (466) 1050 (236) 51 2194 (493) 1224 (275) 5630444 22 1104 (248) 356 ( 80) 32 1562 (351) 543 (122) 35(0.20% C) 23 1135 (255) 285 ( 64) 25 1589 (357) 289 ( 65) 1824 943 (212) 236 ( 53) 25 1006 (226) 205 ( 46) 2025 1032 (232) 392 ( 88) 38 1015 (228) 218 ( 49) 21AVE 1055 (237) 85 C 19) 8 1295 (291) 325 C 73) 25SD • STANDARD DEVIATION90Table 5.1 coWt: Peak-to-peak averages and standard deviationsfor the Western Canada Steel plant trialresu Its.HEATLC3 FRONT LOAD CELLS LC4RUN 61. SD SDSD SD(N) (lbf) (N) (lbf) (%) (N) (lbf) (N) (lbf) (%)30440 3 3618 (813) 1531, (344) 42 1206 (271) 360 ( 81) 30(0.31% C) 4 4744(1066) 819 (186) 17 4081 (917) 739 (166) 185 2283 (513) 454 (102) 20 2185 (491) 463 (104) 21.6 4143 (931) 703 (158) 17 3182 (715) 547 (123) 177 3013 (677) 623 (140) 21 1887 (424) 592 (1.33) 31AVE 3560 (800) 957 (215) 27 2510 (564) 1130 (121)4530441 9 2741 (616) 530 (119) 19 8869(1193)1727 (388) 19(0.322 C) 10 2715 (610) 427 ( 96) 16 2212(497) 414 ( 93) 1.911 1949 (438) 303 ( 68) 16 1985 (446) 334 ( 75)1712 3111 (699) 498 (112) 16 1825 (410) 525 (118)29AVE 2630 (591) 490 (110) 19 372.i (837) 3435 (169 9230442 13 2198 (494) 1170 (263) 53 1099 (247)699 (157) 64(0.342 C) 14 151 ( 34) 583 (131) 385 1041 (234) 31.2 ( 70) 3015 2247 (505) 449 (101) 20 1615 (363) 436 ( 98) 2716 1673 (376) 267 C 60) 16 5772 (1297) 1411 (317)2417 623 (140) 218 ( 49) 35 1032 (232) 427 ( 96)41AVE 1380 (310) 948 (213) 69 2114 (475) 2060 (463)9730443 18 1526 (343) 485 (109) 3.. 1402 (315)641 (144) 46(0.37% C) 19 2817 (633) 596 (134) 21 1896 (426) 699 (157) 3720 2332 (524) 543 (122) 23 1428 (321) 574 (129) 4021 1113 (250) 307 ( 69) 28 1277 (287) 312 ( 70) 24AVE 1949 (438) 770 (173) 39 1500 (337) 271 ( 61) 1830444 22 2875 (646) 1802 (405) 63 1184 (266)392 C 88) 3323 2857 (642) 427 C 96) 15 1593 (358) 298 ( 67) 1924 2383. (535) 409 ( 92) 17 1980 (445) 596 (134)3025 2261 (508) 401 ( 90) 18 2212 (497) 650 (146) 29AVE 2594 583 320 C 72) 12 1753 (394) 445(100) 25SD - STANDARD DEVIATION‘1000z100009000D800047000600050002125jc;.,—..•...,‘4.—17251525Figure5.2:Dampenedwaveformof thefrontloadcellfromthefirstplanttrial.LubricantAat100mI/mmHeat 30444;Run22;Carbon:0.20%2325110Ui1121)31)4115116117118119120121122123124125TIME(s)1325112592Since the mould oscillation cycle was not monitored simultaneously duringthe trial, it was not possible to correlate mould position with load cell response.However, the trial was successful in that it did confirm the ability of the load cellsto measure or monitor the loading pattern of the mould housing unit on theoscillating table of a billet casting machine.5.2 IVACO Results5.2.1 Analysis of LubricantsSamples of the three oils, Lubricants A, B, and C, examined in the mouldlubrication experiments during the trial undertaken at IVACO, were analyzed by alocal testing laboratory in Vancouver. Flash points, viscosities at 0 and 100 0 C,together with their total free fatty acid content and fatty acid composition, weredetermined. Flash pointsThe flash point for each lubricant was determined employing twotechniques: Cleveland Open Cup Method (COG) and Pensky-Martens ClosedFlash Tester (PMCF) [55]. The Cleveland Open Cup Method can be employed todetermine the flash and fire points of petroleum products and other liquids exceptfuel oils while the Pensky-Marten Method is employed for fuel and lube oils. Bothmethods are similar in that oil is placed in the test cups to a specified level andthen heated slowly. At specified intervals, a test flame is passed across the topof the cup and the flash point is reached when the test flame causes the vapoursabove the liquid to ignite. The difference between the two methods is that in the93CCC method, the liquid surface is exposed to the atmosphere and depending onthe existing ventilation, some of the vapour may escape. Thus, the resultingreadings may be high. in the Perisky-Martens Method, the liquid is covered and,therefore, this method is more controlled and dependable in determining a valueof the flash point. PMCT values are lower than those obtained using the COOmethod. In this section, the flash point values are given for both methods butbecause the flash points supplied by the manufacturers are determined by theCOC method, emphasis is placed on it.Table 5.2 tabulates the flash point for these oils as obtained from thetesting laboratory and from supplier specifications. The results from thelaboratory indicate a 27 degree difference between the highest (Lubricant A) andlowest (Lubricant C) flash points; the supplier specifications show a 32 degreevariation.TABLE 5.2: Flash points of the thr.. mould lubricants testedat WACO.FLASH TEMPERATURE(OC)MOULD LUBRICANT GENERAL TESTINd SUPPLIER SPECLABORATORIES SHEETSLubricant C 210 232Lubricant B 227 200Lubricant A 237 22794These flash points were obtained using the specified ASTM [551 procedurefor the Cleveland Open Cup Method which stipulates that for judging theacceptability of results with a 95% confidence the following limits, should beconsidered:a) duplicate results by the same operator should be considered suspect if theydiffer by more than 8 0 C andb) results of two laboratories differing by more than 16 0 C should be consideredsuspect.Also illustrated in Figure 5.3 are the laboratory flash point values with theacceptability limits indicated by arrows. Clearly, the flash points for Lubricants Band A are probably equal on the basis of statement a) while Lubricant C isslightly lower. The flash point values as supplied by the specification sheets arerepresented by open circles. Only the supplier value for Lubricant A falls withinthe ASTM of ±16 0 C limits for acceptability. The oils, B and C, appear not tomeet supplier specifications. At best, it can be surmised that Lubricant A has thehighest flash point and the remaining two have similar but lower flash pointvalues.Flash points obtained using the Perisky-Martens Closed Tester [55] aresignificantly different from the Cleveland Open Cup Method as shown in Table5.3. The Pensky-Martens method generates lower values of flash point thandoes the Cleveland Open Cup Method as mentioned earlier. The percentdifference between the Pensky-Martens method and the Cleveland Open CupMethod is also included in Table 5.3. It is unclear at this point which of the two95FLASH TEMPERATURES OF MOULD LUBRICANTSAS DETERMINED BY LABORATORY AND SUPPLIER ANALYSES260• LABORATORY ANALYSISo SUPPLIER SPECIFICATIONS250 4790 (_J_)w 240458D230°rn437>220 CwI 416 pq210200 0190 I I 374Lubricant C Lubricant B Lubricant AMOULD LUBRICANTFigure 5.3: Flash temperatures of mould lubricants asdetermined by laboratory and supplieranalyses.96values is a reliable parameter for determining the temperature at which a givenoil would flash in the mould. However, the values from the Cleveland Open CupMethod which are obtained in an open environment are probably more suitablethan those obtained from the Pensky-Marteris Closed Flash Method forcomparison to mould wall temperatures.TABLE 5.3: Flash points as determined by the Pensky-MartensMethod.LUBRICANT C B AFLASH POINT 182 208 184(0 C)Difference fromCleveland Open Cup (%) -13 -8 - ViscositIesThe kinematic viscosity of the oils at for 0 and 100 0 C was determined inthe laboratory as well. Table 5.4 tabulates the viscosities determined from boththe laboratory analysis and the supplier specification sheets. Depicted in Figure5.4 is the viscosity variation with temperature for each oil. The semilogarithmicgraph shows that at temperatures of less than 50 C, there exists a largedifference between the viscosities of the three oils. This disparity is significant ifthe mould wall temperature at steady state is less than 50 0 C as the velocity ofthe oil down a vertical wall is inversely proportional to its absolute viscosity.Examination of several computer runs [60] indicates that temperatures at the topof the mould wall are about 30 0 Celsius and remain less than 50 0 C until four-fifths of the distance to the meniscus.97TABLE 5.4 KinematIc viscosities of the three mould oilstested at the IVACO plant trial.TEMPERATURE KINEMATIC VISCOSITY FROM ANALYSIS(0 C) Lubricant C Lubricant B Lubricant A(cSt)0 662.0 367.0 280.038* 125 76 76100 8.6 7.7 11.5KINEMATIC VISCOSITY FROM SUPPLIERLubricant C Lubricant B Lubricant A(cST)38 60 46 44100 9* Values estimated from Figure 5.4At 30 0 C, both Lubricant A and Lubricant B have similar kinematicviscosities of 110 cSt while Lubricant C has a viscosity of 180 cSt. At themeniscus level where temperatures range between 95 and 110 0 C, theviscosity-temperature behaviour of the three lubricants is similar. However, thesupply of Lubricant C at the meniscus could be less than the other two lubricantsowing to its higher viscosity in the 30 to 50 0 C temperature range.-(I-)(-)>-Cl)0C-)>C-)wzMEASURED KINEMATIC VISCOSITY VERSUS TEMPERATURETEMPERATURE ,0200Figure 5.4: Measured kinematic viscosity versustemperature.1021011000 50 100 150995.2.1.3 Fatty acid contentFatty acids present in the oil may aid lubrication particularly at operatingmould temperatures. Table 5.5 lists the fatty acid content of the three oils asanalyzed by the laboratory. Lubricant A has 30 times the fatty acid contentrelative to the other two oils. Of the fatty acid present, approximately 92% of theTABLE 5.5: Fatty acid content of the oils used in the IVACOplant trial.ACID CARBON CHAIN PERCENT OF TOTAL FA ACIDSC B ALauric acid C12:0 0.1 0.1 0.1Myristic acid C14:0 0.1 0.1 0.6Palmitic acid C16:0 5.1 4.2 2.6Palmitoleic acid C16:1 0.1 0.1 1.5Stearlc acid C18:0 24.1 1.5 2.0Oleicacid C18:1 58.3 61.5 63.1Linolelcacid C18:2 11.1 19.6 26.1Linolenic acid C18:3 0.3 9.7 1.1Arachidic acid C20:0 0.1 0.5 0.5Elcosenoic acid C20:1 0.1 0.7 0.8Behenic acid C22:O 0.1 0.5 0.1Erucic acid C22:1 0.1 0.6 0.1Lignoceric acid C24:4 0.1 0.1Freefattyacid 0.10 0.13 3.88as oleic acid)hydrocarbons in the Lubricants A and B have double bonds while only 70% of thefatty acid content found in Lubricant C have double bonds. The fatty acids whichstrongly undergo oxidation are linoleic (Cl 8:2) and linolenic (Cl 8:3). Again bothLubricants A and B have similar amounts while Lubricant C has the least.1005.2.2 Load Cell Results5.2.2.1 Load cell profilesLoad cell signals obtained during the IVACO plant tnal were converted tocompressive loads based on the Instron calibrations. The load response wasplotted against time using the available Houston plotter at UBC which allows theprinting of large graphs enabling a more accurate evaluation of load changes. Aswith the WCS data, possibly due to the large 0-rings or inlet water interference,dampening of the front load cells was evident. Therefore, as indicated by theresponses of the front load cells, no further analysis of the front load cells wasundertaken.The load cell responses resemble those obtained during the WCS trial ascan be seen typically from Figure 5.5. Several important features requirecomment. Firstly, the response is periodic having a 2 Hz frequency. Secondly,minimum loading (the lower peak) undulates at a low frequency while maximumloading (the upper peak) remains relatively constant. Thirdly, the upper peaksare distinctly different from the lower peaks. The lower peaks exhibit a relativelysmooth response with the appearance of intermittent double peaks and ofoccasional jags located on the downstroke (left side) irrespective of oil flows.The upper peaks are broader than the lower peaks and, as will be shown, appearto be more influenced by oil flow. At the higher oil flows, the upper peaks tend tobe flat with secondary peaks varying in number and height. As oil flow isreduced, the upper peaks increase markedly and exhibit sudden smalldecreases.11007100LoadCell3o 4 0 -J1000,0011001100°°HeatA23412;Carbon:0.09%Bun40;LubricantA;54mI/mmTIME(s)Figure5.5:Responseofreartoadcellsfromthesecondptanttrial1025.2.2.2 Load cell profIles and oscIllation cycleTo correlate the position of the mould with the load cell response, thesignal from the LVDTs placed on the mould assembly was fed into the dataacquisition system together with the load data. Figure 5.6 compares the keycomponents of mould oscillation with the load cell response for a 0.09% carbonheat cast with the A lubricant fed at 54 mI/mm. Also included on Figure 5.6 is therepresentative load cell responsewhich would be obtained during the oscillationof an empty mould. From the empty mould loading profile, it is evident that theload cell response obtained when casting reflects the frequency of the oscillationcycle.Indicated on the load cell response cycle are the theoretically calculatedperiods of negative-strip, when the mould travels downward faster than thestrand, mould upstroke and downstroke, and the points at which upward anddownward maximum velocity of the mould occur. It is apparent that as the mouldmoves downward, a sudden decrease in compressive load occurs as thenegative-strip period begins. This decrease continues smoothly until themaximum downward velocity of the mould is reached or shortly thereafter atwhich point the load begins to increase smoothly until upstroke begins. Duringupstroke the load fluctuates as indicated by the presence of spikes. Thesespikes, as suggested earlier, can reflect the phenomena associated with mouldlubricant, carbon content, and lubricant flow. A detailed examination of thesespikes will be presented subsequently.Irrespective of oil type, flow, or steel carbon content, the onset of negativestrip typically initiates decompression of the load cell. Figures 5.6 and 5.7103II0—In—W.®<.0 .0._I•(0IL,U-Load (l,f)eJ cJI----------_--.-::=:::::::-::::::::02a.LI(1q) ayo,—_‘EEZ_— =(mi) avoI I! I104iF;,_CE8u,>%a0 .—:-EEZZE::= E——— C’) — —— a_--h—____- .3_•-__-a:-.•-—.‘E3j:::::::::EEE:::::II)U.a 0 -J1400I$001100110010001q00100•?00*00TINE(s)Figure5.8:Keycomponentsof themouldoscillationcycleimposedontheloadcellresponse;0.05%carbonwithLubricantCfedat54mI/mm.fr.(‘I106illustrate the decompression for Lubricants A and B. Even though the shape ofthe upper peaks is different with different conditions, the loading decreases forboth oils at the beginning of negative-strip time. Figure 5.8 shows a load cellprofile for reduced Lubricant C flow revealing the same relationship betweendecompression and negative-strip. Carbon content does not affect thisdecompression and negative-strip. Carbon content does not affect thiscorrelation since similar results can be seen in Figures 5.9, 5.10, and 5.11 whichrepresent load cell responses and oscillation cycles for 0.051, 0.17, and 0.42 %carbons, respectively. MinImum load and casting speedCasting speed has been imposed on the graphs corresponding toLubricant A fed at 54 and 24 mI/mm. As shown in Figures 5.12 and 5.13, castingspeed strongly correlates with the undulating minimum compressive loadingpattern, which consequently, affects the average delta toads. Delta load isdefined as the difference between successive minima and maxima in the loadcell response plot. Minimum load increases as casting speed increases anddecreases as speed decreases. This visual correlation was confirmed throughsimple statistical analysis.Minimum loads have been graphed against casting speed in Figures 5.14to 5.17 for Lubricant B at flows of 54, 44, 34, and 24 mI/mm respectively. Theresults clearly indicate that minimum load increases linearly with casting speed.The results for the oils C and A support this conclusion and are given in AppendixB.ll1IIlIIIIIIIIIIIIIII;’I‘IfI II’IIIi‘\IIIIIIIIIIiiIIIII IIIIIIII‘IIIIIIIIIIIIii!a‘‘V.I’IIIIIiiIIflUE(s)Figure5.9:Keycomponentsofthemouldoscillationcycleimposedontheloadcellresponse;0.15%carbonwithLubricant Cfedat54mI/mm.‘4..1J••‘me0 -JI1001II,...eonHe123409Cibon: 0.15%Run35; Li*idcanlC;24nhlln*iI I I-__- -———- ——————— ———.—— _— ——.1 -i1—— -——. ) d -ii”1 IOqi) avoi—aa —_:zzjzzZEE——— _— —0S.5qC108ii>.—0IS,0----::::=:-::::---------I109• Se- 2 1(Jq)-_a4CSI> C.)VaCo‘flu,0‘C510CCU,.2’‘aLLp..FITIME(s)Figure5.12:CastingspeedimposedontheloadceOresponseforLubricantAtedat54mLlmin(0.09%C).castingspeed--••$IIS1133134415,31.3’’$••3IIS6Run40;LubdcantA;54mIImln6)1U),?•I3133)4444566$44$44#1•)•.1)$I.?ISIII?$I1II1I4N)I$r$$LoadCell41E anLoadCell3Heat824653;Carbon:0.09%Hun44;LUbriCantA;24mIImInan—,———————,—-—1—-•1—-—-—-—T——-—---T—,——--—---.—.--——t——-.•1-)°.S‘‘••a..i..TiME(a)Figure5.13:Castingspeedimposedontheloadcellresponsefor Lubricant Afedat24mI/mm(0.09%C).DEPENDENCEOFMINIMUMLOADONCASTINGSPEEDLubricantBat54mIImin10000—12250HEATB24654RUN48O.12CARBON-210029000--—z1950c——18000 —800001650z—-7000.-—1594.7+69.5X/—-1STAM)ARBDEVIATtON-1500--2STANDARODEVIATIONS6000135080859095100105CASTINGSPEED(in/mm)r’)Figure5.14:DependenceofminimumloadoncastenforLubricantBfedat54mI/mm(0.12%DEPENDENCEOFMINiMUMLOADONCASTINGSPEEDCASTINGSPEED(in/mm)Figure5.15:DependenceofminimumloadoncastingspeedforLubricantBfedat44mI/mm(0.12%C).C-.)LubricantBat44mI/mmHEATA264130.12xCARBONRUN502250z CD 0 -J z100009000800070006000_e-.——_.——_.-_z C1800cr1575—2672.9.58.0X—-iso--2SD808590951001350105DEPENDENCEOFMINIMUMLOADONCASTINGSPEEDLubricantBat34mI/mm90002025HEATA23413RUN510.i2’CARBON--8500-.1890_rgi’•z__w•..__••——-I.——z—8000-1•,.—-cm--b—-1755---,1___•••-——-•1-17500-----•——•_•_-•-.—--1620ZJ—7000———_•__•——•0—•_—14856500—28455÷56.2*X-—-iso--2SD600011350708090100CASTINGSPEED(in/mm)Figure5.16:DependenceofminimumloadoncastinspeedforLubricantBfedat34mI/mm(0.12%).CASTINGSPEED(in/mm)Figure5.17:DependenceofminimumloadoncastingspeedforLubricant Bfedat24mI/mm(012%C).OFMINIMUMLOADONCASTINGSPEEDLubricantBat24mI/mmHEAT0.12zA26413RUN52CARBONDEPENDENCE100009000CD 0 —J8000z70006000———————-a-———22502100z1950C1800131650‘—‘0 -,15001350110—3048.8+56.8X—-iso--2S0708090100fr fr 01116The best fit straight line for all mould lubricant conditions was determinedemploying the technique of least-squares regression. The results are tabulatedin Table 5.6 and the regression lines are plotted in each of Figures 5.14 to 5.17.The correlation coefficients listed in the table provide some statistical support forthe minimum load-casting speed relationship. With the exception of Lubricant Cat 44 mI/mm, all coefficients are above 0.6 while most are above 0.75.The coefficients of determination indicate how much of the vanation inminimum load can be attributed to casting speed. For the B mould oil, thedetermination coefficient indicates that 60 to 70% of the sample variation inminimum load can be explained by variation in casting speed. For the middleflows of Lubricant A, over 70% of the variation in minimum load is attributable tocasting speed. Over 50% of the minimum load variation is the result of castingspeed fluctuations for Lubricant C at the lower flows. Carbon content of the steelmay be responsible for the differences in the dependence of minimum load oncasting speed..The slopes of the plots of minimum load versus casting speed arepresented by heat in Figure 5.18. Clearly, Lubricant A in Heat A23412, the first inthe sequence, demonstrates the strongest dependence while Lubricant Bdemonstrates the weakest dependence. Figure 5.18 also indicates the sensitivityof flow with respect to casting speed.Maximum load was also graphed versus casting speed as shown inFigures 5.19 and 5.20. Because no correlation could be established, fluctuatingcasting speed does not appear to affect maximum load.TABLE51:LInearregressionresuibletminimumloadandNEAT11912CARSONURIICAarVWUU1SLOfl1NTEICETalIIgIAT10NCOIVF1CIIIflCO(tFICIZWICt1IET191I1T1OI(.1I.I)(N•(N)A7340936C4444.440000.3%0.151313458.122400.8040.646382453.625700.1320.536*23412400.0,A33.845600.6060.374)4451.023400.8310.101423411.613200.8190.1138246534)0.09A343q000.8490.171442437800.6910.486124654480.12B5469.515900.8100.689A2341)500.12B4458.026700.1130.599SI3456.228500.1620.511522456.830500.181I-.-.1118CEz0LUz-Jz000LUCDLULUIIC,,LUa0-J0SLOPE DEPENDENCE ON CASTING SPEED FOR THE THREEMOULD LUBRICANTS TESTEDFigure 5.18: Slope dependence on casting speed for thethree mould lubñcants tested.(A234 13)L.ibricant B—4. — — —807060504030(B24654)Lubricant B\20 30FLOW40 50RATE (rn/mn)60MAXIMUMLOADDEPENDENCEONCASTINGSPEEDCASTINGSPEED(in/mm)Figure5.19:MaximumloaddependenceonCastingspeedforLubricantBat34mI/mm.LubricantBat34mI/mmHEATA26413O.12’RUN51CARBON2360aa..aa.asaaa’a.•.a•••_lI••••a••!IC...•..9’a..a...z cm 0 -J ED ><229310500102009900960093009000aa •a •Ia.•a•..tarn..?.a••a••..•aI2226aa a•a.a>< 021590a...Ia•aaa7080902025100MAXIMUMLOADDEPENDENCEONCASTINGSPEEDCASTINGSPEED(in/mm)Figure5.20:MaximumloaddependenceonCastingspeedforLubricantCat24mI/mm.HEATLubricantCat24mI/mm0.15,A26409RUN3CARBON:.2180..a..a... aaaz cm 0 -J ED ><a2153a9700960095009400930092009100a.a. .a..a••_aa—•._.a..•aa-2126...—a>< C 020990aa.a•a• aa.aa aa a.a—802072.859095100r\)1215.2.2.4 The Influence of lubricant type, flow, and carbon content onload cell responseTraces of the rear load cell data for the three mould oils at different flowsare shown in Figures 5.21 through 5.32. Figures 5.21 to 5.23 present the loadcell response obtained with mould oil C for the flows of 44, 34, and 24 mI/mm for0.15% carbon heat. The load cell response for the 44 mI/mm flow (Figure 5.21)differs in some respects from those for the two lower flows (Figures 5.22 and5.23). Maximum loads are higher and the upper peaks exhibit more intensespiking which drop in load almost 1000 N (250 lbf). The pounng temperature andthe superheat were lower at the beginning of the heat when the data for thehigher flows was accumulated which may account for these peculiar results.Minimum loads from the 44 mI/mm data reached similar values as for the 34 and44 mI/mm rates but double lower peaks are more prevalent. The 34 and 44mi/mm tests have similar load cell responses.The load cell response for the mould oil A at rates of 54, 44, 34, and 24mI/mm for a 0.09% carbon heat are shown in Figures 5.24 to 5.28. With theexception of the 34 mI/mm run during Heat A23412 (Figure 5.26), the upperpeaks for the other rates exhibit analogous load cell behaviour. The upper peaksare relatively constant and to various degrees have broad, flat and severelyspiked peaks. The run from Heat A2341 2 which differs from the others, haspeaks which increase by as much as 1000 N (250 lbf). In the more severe upperpeaks, the load increases with time. All rates showed lower double peaks.Figures 5.29 to 5.32 show the load cell response for the third mouldlubricant, B, tested at rates of 54, 44, 34, and 24 mI/mm during the casting of a0.12% carbon steel. It is apparent that the load cell response for this lubricant isFigure5.21:TracesoftherearloadcellresponseforLubricantCat44mI/mm(0.15%C).265025502450235022502150—.-o20501950185011501650•55014601350r’)r’JA23409;Run36;Carbon:0.15%Lubricant Cat44mI/mmz 4120003*00030000900080007000600040414243444546474849505*52535455TIME(s)12000A23409;Run37; Carbon:0.15%Cat34mI/mm110001000090002 48000700026502550245023502250215020501950$850$750$650$5501450$350-o 4 0 -JLoadCell3100101102103104105$06107$08i0110Iii112113114115TIMC(s)Figure5.22:TracesoftherearloadcellresponseforLubricantCat34mI/mm(0.15%C).()(o%s10)u!w/Iwiueouqn-jojesuodseiiieopeoieeieqi joseoej:gein6ij(s)cc,ccccicocivcv,pçvivovocci-10009EII3peon0M-occi0001OSLI-AoceiI::cLç0006ocizocu-00001occzOcPi-%SLO:uoqJe‘gunj6OtCVWSHoccz-111W/lwp18)iueojqnioci-0001112000LubricantAat54mi/mmA23412;Run40;Carbon:0.09%—Sz 41100010000900080007000600026502550245023502250215020501950185017501650155014501350.0 4LoadCell370717273747576777879808182838485TIME(s)Figure5.24:TracesoftherearloadcellresponseforLubricantAat54mI/mm(0.09%C).I— r’.)u-i120002650LubricantAat44mI/mm255000HeatA23412;Run41;Carbon:0.09%II0245023501000022502150.—z2050ci900044o19500_J1850800017501650700015501450LoadCell36000--_1——-1____I__i-ii-1350979899100101102103104105106107108109110Iii112TIME(s)Figure5.25:Tracesof therearloadcellresponseforLubricant Aat44mI/mm(0.09%C).265025502450235022502150—s-Q2050Q1950185017501650155014501350LubricantAat34mi/mEnHeat A23412;Run42; Carbon:0.09%z Q 493949596979899100101102103104105106107108TIME(s)Figure5.26:TracesoftherearloadcellresponseforLubricant Aat34mI/mm(0.09%C).12000Figure5.27:TracesoftherearloadcellresponseforLubricantAat34mI/mm(0.09%C).2650255024502iSO225021502050Q19501850175016501550$450$350I-.t\)Aat34mI/mmHeat B24653;Run43;Carbon:0.09%z 4 S11000100009000800070006000LoadCell3$60161162163164165166167168$69170Ill172173174$75liME(s)$20002650255024502.SS022502150—.s-a2050Q1950$850$750$650$550$450$350Lubricant Aat24mI/mmHeat B24653;Run44; Carbon:0.09%2 411000$0000900080007000LoadCell36000-Iii8586878889909)929394•9596979899$00TIM[(s)Figure5.28:TracesoftherearloadcellresponseforLubricant Aat24mI/mm(0.09%C).1’.).012000-Lubricant Bat54mI/mm2650-2550-Heat B24654;Run48; Carbon:0.12%24502350-2150-205090004-19500 -J-18508000-1750•16507000-15501450LoadCell36000-1-i—1i—---j---—r——,-rr1135032333435363738394041424344454647TIM[(s)Figure5.29:TracesoftherearloadcellresponseforLubricantBat54mI/mm(0.12%C).12000LubricantBat 44mI/mm26502550Heat A23413;Run50;Carbon:0.12%1100024502350225010000-t Ij215oc%RIzIf-20S0’’90001M/Jyyyyy4 SI950-I85080001750-7000-1550•1450LoadCell36000----13503536373839404!424344454647484950T(M[(s)Figure5.30:TracesoftherearloadcellresponseforLubricant Bat44mI/mm(0.12%C).12000-LubricantBat34mI/mm2650-2550HeatA23413;Run51;Carbon:0.12%11000--2450-2350-2250100002150-oz205009000-0419507000118508000•1750-1550-1450LoadCell36000-—135040414243444546474849505152535455liME(s)Figure5.31:TracesoftherearloadcellresponseforLubricantBat34mI/mm(0.12%C).12000z 0 4ci 4 0 -JLubricant Bat24mI/mmHeat A23413;Run52; Carbon:0.12%110001000090008000700060002650255024502350225021502050195018501750165015501450135015161718192021222324252627282930TIME(s)Figure5.32:TracesoftherearloadcellresponseforLubricantBat24mI/mm(0.12%C).I- w w134different from that produced when Lubricant C and Lubricant A are employed withdifferent carbon grades cast. All flows exhibit strong increases in load in theupper portion of the cycle and the severity of this increase becomes morepronounced as flow decreases. Maximum loads attained are similar for the threelower oil rates being slightly lower for the higher rate.Figures 5.33 to 5.36 present the load cell response heats with carboncontents of 0.035, 0.05, 0.18, and 0.42%, respectively when the flow of C ismaintained at 54 mI/mm. These figures clearly reveal that the load cell responseis sensitive to the carbon present in the steel cast. The low-carbon heats,Figures 5.33 and 5.34, exhibit large variations in upper peak response havingfluctuations within and between peaks. In sharp contrast is the response of themedium-carbon grades, Figures 5.35 and 5.36. In the load cell responses of the0.18% and 0.42% carbon steels, maximum peaks are essentially constant andthe severity in the magnitude of the spikes is reduced. Double lower peaks aremore prevalent in the 0.051 % carbon than in the other three load cell responsesof this grade. The 1039 grade also has a large number of lower double peakspresent. QualitatIve and quantitative analysis of load cell responseA qualitative and quantitative analysis of load cell response wascompleted for each mould oil condition and steel grade. The qualitative analysisinvolved a visual examination of the individual cycles in each run and anassessment of the upper peak characteristics. Tables 5.7, 5.8, and 5.9summarize the classification of the upper peaks for each mould lubricant.Included in the tables are examples of the classifications.2775z 0 425752375;:‘-Q 0217519751775157515161718192021222324252627282930TIME(s)Figure5.33:TracesoftherearloadcellresponseforLubricantCat54mI/mm(0.035%C).01(A)Lubricant Cat54mEIminHeat A23408;Run35; Carbon:0.051%z 42775257523752175197517751575TIME(s)Figure5.34:Tracesof therearloadcellresponseforLubricant Cat54mI/mm(0.051%C).13000z Q 4 2TiME(s)$975Figure5.35:Tracesof therearloadcellresponseforLubricant Cat54mi/mm(0.18%C).(A)LubricantCat54mI/mmHeatB24644;Run25;Carbon:0.18%11000100009000aooo7000277525752375—. ¶32175LoadCell395969798991001011021031041051061071081091101715277525752375..02175197517151575TIME(s)Figure5.36:TracesoftherearloadcellresponseforLubricant Cat54mI/mm(0.42%C).LubricantCat54mi/mmHeatB24647;Run31;Carbon:0.42%11000c1000047000LoadCell9596979899100101102103104105106107108109110139The data for Lubricant A was collected during “piggy-back” heats. The oilflow was varied from 54 to 44 and 34 mI/mm in the first heat and from 34 to 24mi/mm in the second. Visual assessment is summarized in Table 5.7 andindicates that the two highest flows have similar upper wave patterns having onlyconstant peaks with a slightly higher percentage of major rather than minorsecondary peaks. When examining the results for the two runs of 34 mI/mm,there appears to be some discrepancy. The results do not agree but both showincreasing jagged upper peaks. When re-examining the initial Houston plots ofthese two runs, the second set of data is undeniably more constant. Mould-strandinteraction is much more apparent from the upper peak responses in the first setof data obtained for the 34 mI/mm rate. To this date, the discrepancy betweenthese two runs are unexplainable as differences in casting conditions have notbeen identified. The upper peaks for the 24 ml/min oil flow are all steep havingmainly jagged secondary spikes.140TABLE 5.7: QualItative classification of load cell outputduring the upstroke for Lubricant A mould oil.Percentage of peaks Is Indicated In each of theclassifications. The carbon content Is 0.09%.WAVE DESCRIPTION FLOW(mi/mm)Heat A2341 2 Heat 82465324 34 34 44 54A: 4 55 10 4740B: 17 21 31 53 60C: 29 24 25 0 00: 50 0 35 0 0EXAMPLES:A B C DUpper peaks: Plateau Plateau Ascending AscendingSecondary: Small Large Small Large141The qualitative results for Lubricant B are summarized in Table 5.8. Theupper peaks in this case generally ascend rapidly resulting in a classificationbased on peek steepness and spike seventy. The evaluation indicates that theremay exist a correlation between flow and peak shape: as flow decreases, themagnitude of the peak increases. For each rate all peaks have been groupedtogether within two adjacent classifications. The peaks for the 54 mI/mm oil flowall lie within the two middle classifications; only steep peaks are present for thetwo rates of 44 and 24 mI/mm. The peaks of Lubricant B differ from those ofLubricant A as there are no upper peaks which exhibit only minor secondarypeaks.During a heat of 0.15% carbon, the flow of mould Lubricant C was reducedfrom 44 to 34 and 24 mI/mm. The load profiles measured were classified intothree groups as shown in Table 5.9. The higher flows of 34 and 44 mI/mm gaverise to a more jagged upper peak than was the case of the 24 mI/mm rate.142TABLE 58: QualitatIve classification of load cell outputduring the upstroke for B mould lubricant.Percentage of peaks is Indicated in each of theclassifications. The carbon content is 0.12%.WAVE DESCRIPTIONEXAMPLES:FLOW(mI/mm)24Heat A2341 334 44Heat B2465454A: 0 2 0 0B: 0 14 3 59C: 75 80 51 41D: 25 4 46 0A B C 0Upper peaks: Plateau Plateau Ascending AscendingJagged: Small Large Small Large‘I143TABLE 5.9: QualitatIve classification of load cell outputduring the upstroke for C mould lubricant.Percentage of peaks Is indicated in each of theclassifications. The carbon content Is 0.15%.FLOW(mi/mm)Heat A23409PEAK DESCRIPTION 24 34 44A: 23 6 8B: 19 42 57C: 58 52 35EXAMPLES:A B CUpper peaks: Plateau Plateau PlateauSecondary peaks: Small Several LargeNumber of peaks: < 3 3-4 multipleThe quantitative analysis consisted of determining the averages ofmaximum açd minimum load and calculating the average load change duringeach upstroke cycle and the corresponding standard deviations. Alsodetermined was the load change distribution during the upstroke. Results are144shown in Tables 5.10, 5.11, and 5.12 for Lubricants A, B, and C, respectively.Figure 5.37 illustrates collectively the maximum-minimum and change in load foreach lubricantlsteel grade condition.Maximum load for the two lubricants, A and B, tends to decrease as flowincreases particularly within heats. The maximum load values for C mouldlubricant shows the reverse trend having lower values at the lower flow.At the two lowest flows, Lubricant C had the lowest maximum loads andLubricant B had the highest. At the highest flows, Lubricant A had the lowestloads.A comparison of the standard deviations was undertaken in an attempt toquantify the upper peak fluctuations. Variation in maximum load for Lubricant Aappears to increase slightly as flow decreases. The 54 mI/mm rate run has thesmallest deviation of 1.2% while the 34 mI/mm of the first heat in the sequencehas the largest deviation of 3.0%. Mould lubricant B had the largest variationsoverall as all lay within 2.2% to 2.6%. The 24, 34, 44, and 54 mI/mm runs havesimilar variations of 2.2% or 2.3%. The maximum load variations for C varybetween 1.4% and 2.3% having no significant trend. Due to the low number ofheats monitored when reducing mould lubricant flow, trends may be obscured.Within each set of rates for each lubricant, there exist similar maximumvalues for at least two flows. The 54, 44 and 34 mI/mm (Heat B24653) values forLubricant A (0.09% carbon) differ by less than 53 N (12 lbf). The 24 mI/mm rateis higher by 700 N (157 Ib) indicating greater mould-strand interaction. Themaximum loads for Lubricant B (0.12% carbon) are higher than for the other two145lubricants for any flow. The two flows which have similar maximum load valuesare the middle rates of 34 and 44 mI/mm. The maximum loads which are similarfor Lubricant C (0.15% carbon) are the lowest rates of 24 and 44 mI/mm.The A mould lubricant tends to have the lowest change in load during theupstroke portion of the oscillation cycle but a fairly high variability. Lubricant Bexperiences the greatest load increases during upstroke as the change in loadirrespective of flow is above 500 N (113 lbf). However, the load increases aremore consistent than the other two oils as indicated by the substantially lowerpercent deviations. Lubricant C demonstrates the greatest stability for change ofload during the upstroke as little difference in the load increases exists. Due tothe similarity in the magnitude of load increases during upstroke, a statisticalevaluation employing the F-test indicates that there are essentially no differencesin delta load during upstroke for the following mould lubricants and flows:a) C at 24,34, or44 mi/mmb) B at34or44 mi/mmc) A at 44 or 54 mi/mmThe F-test method is summarized in Appendix B.146TABLE 5.10: QuantItative analysis of upstroke period forLubricant A.Heat 824653 Heat A2641 2Flow:(miImIn) 24 34 34 44 54Number ofCycles: 358 359 359 359 358Average CastingSpeed:(mmls) 40 37 38 38 38(In/mm) 94 87 89 89 89Maximum Load:(N) 10048 9376 9683 9372 9323(ibf) 2258 2107 2176 2106 2095Standard Deviation:(N) 190 150 291 134 115(lbf) 43 34 66 30 26%Devlation: 1.9 1.6 3.0 1.4 1.2Delta Load:(N) 596 316 668 396 387(Ib) 134 71 150 89 87Standard Deviation:(N) 158 124 224 118 175(ibf) 36 28 50 27 39% DeviatIon: 26 39 34 30 45147TABLE 5.11: Quantitative analysis of upstroke period forLubricant B.Heat A23413 Heat B24654Flow:(mi/mm)Number ofCycles:Average CastingSpeed:(mm/s)(in/mm)Maximum Load:(N) 10239 9861 9906 9714(lbf) 2301 2216 2226 2183Standard Deviation:(N) 230 254 214 207(lbf) 52 57 48 47% Deviation: 2.2 2.6 2.2 2.1Delta Load:(N) 1032 783 832 503(Ibf) 232 176 187 113Standard Deviation:(N) 187 199 192 167(ibf) 42 45 43 38% Deviation: 18 25 23 3324359 V3890343593788443583891542393890148TABLE 5.12: QuantitatIve analysis of upstroke period forLubricant C.Flow:(mi/mm)Number ofCycles:Average CastingSpeed:(mm/s)(mi/mm)242393724Heat A23409342393734442393544Maximum Load:(N) 9358 9398 10061(lbf) 2103 2112 2261Standard Deviation:(N) 156 130 229(Ibf) 35 29 52%Deviatlon: 1.7 1.4 2.3Delta Load:(N) 547 543 592(lbf) 123 122 133Standard Deviation:(N) 203 174 228(Ibf) 46 39 51% Deviation: 37 32 39As would be expected, the previous results are reflected in the distributionof the loading increases during the upstroke. The change in load during upstrokefor each cycle was categorized and summarized in Tables 5.13, 5.14, and 5.15.149Maximum values: •Minimum values: oOile statistically equivalent based on t- testFigure 5.37: Maximum, minimum, and change in load duringthe upstroke cycle.Lubricant A00 %Ce653 A412I I I—Lubricant C015%CA409I10500IOIOC97009300850O8900’z0)EECC(UEEzC24003aaoo:180040001900300aoo24344454Flow Rate (mi/mm)243444150Most load changes for Lubricant A lie below 890 N (200 lbf) and above 223 N (50lbf). In fact, the greatest percentages lie below 668 N (150 lbf). The Lubricant Bhas a substantial percentage of its cycles having upstroke loading changesabove 890 N (200 lbf). The distribution of upstroke loading for Lubricant C isfairly consistent as independent of flow the majority of cycles have upstroke loadchanges within 223 N to 890 N (50 to 200 lbf), slightly higher than Lubricant A.TABLE 5.13: DistributIon of delta loads during upstroke forLubricant A - 0.09% CARBON.FLOW:(mI/mm) 24 34 34 44 54DELTA LOAD DISTRIBUTiON (%)(N)HEAT B24653 HEAT A234120- 223 0.6 27.3 0.3 3.6 6.4(0 - 50 lbf)223-445 17.0 58.2 18.1 66.5 51.7(50- 100 lb)445 - 668 53.1 13.9 34.5 27.7 37.7(100- 150 ibf)668 - 890 26.0 0.6 29.5 2.2 3.9(150-200 lbf)890-1113 3.4 0.0 13.9 0.0 0.3(200 - 250 Ib)1113-1335 0.0 0.0 3.3 0.0 0.0(250 - 300 lbf)1335 - 1558 0.0 0.0 0.3 0.0 0.0(350 - 350 lbf)151TABLE 5.14: DistrIbution of delta loads during upstroke forLubricant B - 0.12 % CARBON.FLOW:(mI/mm) 24 34 44 54DELTA LOAD DISTRIBUTION (%)(N)HEAT A23413 HEAT B246540-223 0.3 0.3 0.3 1.30 - 50 bf)223 - 445 0.0 5.0 2.0 41.0(50 - 100 lbf)445 - 668 2.5 23.7 18.2 42.2(100- 150 Ibf)668 - 890 16.7 39.3 40.2 11.7(150-200 Ibf)890- 1113 47.6 26.7 31.6 3.8(200 - 250 Ibf)1113- 1335 27.0 5.0 7.5 0.0(250 - 300 Ibf)1335-1558 5.8 0.0 0.3 0.0(300 - 350 lb)152TABLE 5.15: DIstribution of delta loads during upstroke forLubricant C -0.15 % CARBON.FLOW:(mI/mm) 24 34 44DELTA LOAD DISTRIBUTION (%)(N) HEAT A234090-223 5.9 3.3 1.30- 50 lbt)223-445 22.6 23.0 29.3(50-lOOlbi)445 -668 45.2 51.5 40.6(100 -l5Olbt)668-890 21.8 18.8 16.3(150-200 lbs)890-1113 4.2 3.3 10.5(200 - 250 lbf)1113-1335 0.4 0.0 1.7(250 - 300 lbf)1335 - 1558 0.0 0.0 0.4(300 - 350 lbf)Similar qualitative and quantitative analyses as was conducted to studythe effect of oil flow on the mould-strand interaction was completed for differentgrades cast with Lubricant C at 54 mI/mm. The qualitative results are tabulatedin Table 5.16. The 1008 upper peaks have the greatest variation having peaks inevery classification. The load cell response for the 0.05% carbon gradesconsisted of peaks with spikes which are mainly intermediate or major. The othertwo grades, 1018 and 1042, show essentially only two peak types.153TABLE 5.16: Qualitative classification of load cell outputduring the upstroke for several carbons whenC mould lubricant Is used at 54 mI/mm.CARBON WAVE DESCRIPTION:(%) A: B: C: D: E:(%) (%) (%) (%) (%)0.035 5 35 5 50 50.04 14 63 23 0 00.045 2 25 22 46 50.051 0 1 0 22 770.17 48 52 0 0 00.18 0 44 54 0 20.42 33 67 0 0 0Upper peaks Secondary peaksA: Plateau SmallB: Plateau LargeC: Ascending SmallD: Ascending IntermediateE: Ascending LargeThe quantitative results are tabulated in Tables 5.17 and 5.18. Theaverage change in load during upstroke is significantly higher for the 1008 steelgrade compared to that for the two higher steel grades. The 1 008s also exhibithigher maximum loads and greater standard deviations during upstrokecompared to the two other steels which had been cast under similar lubricationconditions. Table 5.18 summarizes the maximum loads and standard deviationsfor the three steel grades. From this table, it is evident that the 1008s havemaximum loads which are as much as 21% greater than the other two carbongrades. Also found in Table 5.18 are average percent deviations for each steelgrade and it is clear that as the carbon content of the steel increases, variationdecreases. The 1 008s vary by over 2%: the 101 8s vary by 1% half that of the1 008s; and the 1 039s vary by 0.7%.154TABLE 5..17:Quantltatlve analysis of upstroke periodduring casting of several carbon grades whenmould Lubricant C Is used at a flow rate of54 mI/mm.GRADECARBON 10081018 1039NUMBER OF 0.035 0.040.045 0.051 0.17 0.18 0.42CYCLES 335 239 239357 239 239 239AVERAGECASTING SPEED(mJs) 38 37 40 35 39 33 40i.rL/mir1 89 88 9583 93 78 94MAX LOAD (N) 11779 10862 1121011072 10449 9714 10551so (N) 344 206 187 313 951 10]. 73ZSD (%) 2.9 1.9 1.7 2.8 0.9 1.0 0.7DELTA LOAD (N) 641 445 414 1268343 294 365SD (N) 214 142 141 287 122.8 77 118%SD (Z) 33.3 32.0 34.1 22.635.8 26.2 32.3155TABLE 5.18:Variation In maximum load as a functionof carbon content of steel.HEAT RUN Z CARBON ?IAXIMUMLOAD(N)STANDARDDEVIATION(N)Z STANDARDDEVIATION824640 9 0.035 11779 3442.9211 0.035 10880 2522.32• 824637 2 0.04 10862 2061.9324639 7 0.045 11210 1871.67A23408 35 0.051 11072 3152.85AVERAGE: 2.33%B24643 21 0.17 10898 1211.1122 0.17 10449 950.9123 0.17 9946 1401.41B24644 24 0.18 9692 650.6825 0.18 9714 1011.04AVERAGE: 1.03%324647 31 0.42 10551 730.6932 0.42 10627 710.67AVERAGE: 0.68%156The loading distribution for the different grades is summarized in Table5.19. Three of the tour 1008 runs show results similar to the 1018s and the1 039s as most, If not all, the upstroke load changes for these carbons lie under670 N (150 lbf). The 0.051% carbon shows a much broader distributionspectrum of loading changes during upstroke. Almost 95% of the load changeduring upstroke for this carbon ranged between 890 and 1780 N (200 - 400 lbf).TABLE 5.19: Distribution of delta loads during upstroke forseveral carbon steels when C mould lubricant isused at a flow of 54 mi/mm.1008 1018 1039CARBON (%) 0.035 0.04 0.045 0.051 0.17 0.18 0.42DELTA LOAD (N) DISTRIBUTION (%)0-223 60.0 3.3 7.1 0.3 6.3 13.8 8.4(0-50 lbf)223- 445 38.8 50.2 54.0 0.0 64.4 82.0 70.7(50-100 Ibf)445- 668 1.2 39.3 34.3 2.8 17.6 4.2 19.7(100-150 lbf)668-890 0.0 6.7 4.6 5.6 1.7 0.0 0.8(150-200 lbs)890-1113 0.0 0.4 0.0 18.8 0.0 0.0 0.4(200-250 lb)1113-1335 0.0 0.0 0.0 34.2 0.0 0.0 0.0(250-300 lb)1335-1558 0.0 0.0 0.0 23.8 0.0 0.0 0.0(300-350 lbf)1558-1780 0.0 0.0 0.0 11.8 0.0 0.0 0.0(350-400 lbt)1780-2003 0.0 0.0 0.0 2.8 0.0 0.0 0.0(400-450 lbf)1575.2.2.6 Peak-to-peak change In rear load cellsInitially, correlation between the total change in load during one oscillationcycle and oil type and flow was attempted. The loads measured on the two rearload cells were summed and the maximum and minimum loads during each cyclewere determined. From the results presented in Table 5.20 and shown in Figure5.38, it is evident that no definitive relationship exists between delta loads and oilconditions. The percent standard deviations for Lubricant C all range between13% and 15% indicating that the variation from the average delta load isindependent of flow. The two lowest rates of the mould lubricant B have similartotal load changes but different standard deviation: the 34 mI/mm has a higherdegree of fluctuation than the 24 mI/mm. The highest rate of 54 mI/mm showsthe greatest fluctuation but the lowest magnitude in total load change. LubricantA has the lowest total load changes than the other two mould lubricants at anyrate but tends to have the highest degree of variability. The variability appearsnot to be flow dependent as the rate of 24 and 54 mI/mm have similar percentdeviations (17.0% and 18.6%, respectively). The middle rates have similarvalues of 11.0, 12.7, and 14.2%.158AVERAGE PEAK-TO-PEAK CHANGE IN REAR LOADFOR THE THREE MOULD LUBRICANTS6000I55004412005000z C50 =450034 24T44 1 90004000 f 44 542500 I I 24 1__‘SDCC0I I 1 6CC2500450Lubricant C Lubricant B Lubricant AMOULD LU9RCANFigure 5.38: Average peak-to-peak change in rear load forthe three mould lubricants at different flow rates.159TABLE 5.20: Average peak-to-peak change In rear load forthe three mould lubricants used in the IVACOplant trials.Lubricant Flow(mi/mm)Run Delta Standard StandardLoad Deviation Deviation(N) (lbf) (N) (Ibf) (%)As with the Western Canada Steel results, a carbon effect on delta load isnot evident. Consider the averages and standard deviations listed in Table 5.21.The 1008 grade includes the 0.035, 0.04, 0.045, and 0.051 percent carbons. Thedelta loads and standard deviations are inconsistent as the average delta load forthe rear load cells for the 1008 grade lie between 2528 to 6128 N while thestandard deviations range between 11.3% and 30.4%. The medium carbonshave a smaller range of delta loads and lower percent deviations.C 24 38 3667(824) 463(104) 12.6(0.15% C) 34 37 3680 ( 827) 516 (116) 14.044 36 4486 (1008) 681 (153) 15.2B 24 52 3889 ( 874 334 ( 75) 8.6(0. 12% C) 34 51 3880 ( 872 512 (115) 13.244 50 3609(811) 365(82) 10.154 48 3502(787) 490(110) 14.0A 24 44 2799(629) 476(107) 17.0(0.09%C) 34 43 3409(766) 374(84) 11.034 42 3573(803) 454(102) 12.744 41 3102 ( 697) 441 ( 99) 14.254 40 3030(681) 565(127) 18.6160TABLE 5.21: Average peak-to-peak change In rear load forseveral carbon contents when Lubricant C wasused at 54 mI/mm.Carbon Heat Run Delta Standard StandardContent Load Deviation Deviation(%) (N) (lbt) (N) (lbf) (%)0.035 B24640 9 4214 ( 947) 676 (152) 16.10.040 B24637 2 3671 (825) 583 (131) 15.90.045 B24639 7 2528 (568) 690 (155) 27.30.045 B24639 8 2603 (585) 792 (178) 30.40.051 A23408 35 6128 (1377) 690 (155) 11.30.17 B24643 21 3582 (805) 401 (90) 11.20.17 B24643 22 3782 (850) 294 (66) 7.80.18 B24644 24 2884(648) 405(91) 14.00.18 B24644 25 3106 (698) 730 (164) 23.50.42 B24647 31 3289 (739) 610 (137) 18.50.42 B24647 32 2755 ( 619) 627 (141) 22.8The results for the delta loads give no support to the premise that thecarbon content of the steel influences the total delta load of the load cells. Theimportant parameter from this analysis is the high degree of variability asindicated by the large standard deviations. With the exception of the 0.17%carbon steels and one run of 0.18% carbon steel, all runs have standarddeviations which are more than 580 N (130 lbf). Variability can reach as high as27 to 30% as for the 0.045% carbon. In the following section, it will be shownthat variation in minimum load, seen as a low frequency undulation, can becorrelated directly to casting speed changes. This effect appears to mask theinfluence of other variables on the absolute delta loads.1615.2.2.7 Frequency analysisUsing a Fourier transfer program developed by Edmumd Osinski (64], thepower spectrum or frequency distribution in the load cell response for several ofthe runs accumulated was determined. Figure 5.39 illustrates the results for a0.17% carbon cast under Lubricant C at 54 mI/mm. These results arerepresentative of the other runs. The most significant frequency for all runsoccurs at 2 Hz reflecting the oscillation frequency of the mould. Harmonics occurat decreasing strengths at frequencies of 4, 8, 10, etc Hz. The third strongestfrequency appears as a low frequency between 0.05 and 0.1 indicating a periodof 10 to 20 seconds. This frequency may be reflecting both the upper or lowerpeak fluctuations. BIllet quality assessmentA billet quality evaluation was completed by The Centre for MetallurgicalProcess Engineering on the collected billet sections sampled during the IVACOtrial. Inspections were performed on 37 billet sections sampled from the teststrand (Number 3) and 34 from the control strand (Number 2). Fourassessments were conducted on each billet section:1. After removal of scale by shotblasting, a visual examination wasconducted to determine the presence of surface cracks, depressions, bleeds orlaps, roughness, and pinholes. Uniformity of oscillation marks was alsodetermined.2. Measurement of shape was made to assess bulging and rhomboidity.J\JJIiJJ1fC’.t6•‘Frequency(us)Figure5.39:Frequencyanalysisfortheloadcellresultsof a0.17%carboncastwithLubricant Cat54mI/mm.1633. Profilometer measurements quantified the surface roughness and theoscillation mark characteristics.4. Billet sections were prepared for inspection of internal quality bysulphur printing and macroetching. Each billet section was examined to evaluatethe degree of internal cracking, the presence of inclusions, location of pinholes ifpresent, and the solidification structure.The surface inspection resuLts for both the test and control strand billetsare summarized in Table 5.22. Those billets cast on the test strand (3) exhibitedbetter defined oscillation marks than those on the control strand (2). The teststrand produced more billets having oscillation marks which were parallel andevenly spaced which may reflect the different mould taper employed. Morebillets from this strand had smooth surfaces and were inclusion free. Conversely,a larger number of the billets from the test strand contained depressions,pinholes, and bleeds/laps compared to billets from the control strand.Visual assessment of the billets from the test strand indicated that all the1010 and most of the 1008, 1012, and 1018 billets experienced surfaceroughness. Only those billets from the 1015 and 1039 grades exhibited smoothsurfaces.The profilometer measured the billet profiles along the unrolled faces atthree locations namely the midface and along two off-corner positions on thestraight walls as shown in Figure 5.40. Table 5.23 summarizes the degree ofsurface roughness and the corresponding standard deviations for the threetraces. The roughness measurements along the centreline are significantlyTABLE5.22:BIlletevaluationsummarybysteelgrade.PFICENTAGOFIlLLTSDISPLAYINGEACH0FECTBleed.Irreg.Non-Non—OIC/Lap.SpacingParallelLinearloughee..Depre...Inctue.Pinhole.a.pre...StrandNo.ofSid.SideSideSideSideSideSideSideSid.GradeNo.Billet.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.40:Orientationofthebilletsidesandprofilometertraces.01ProfilometerSideATraces’ICwoCw166TABLE 5.23: Billet profilometer measurements.SURFACE ROUGENESSoilType!GradeRate_______________________Heat ICW Std. CL Std. OCW Std.No. Billet Side (mm) 0ev. (mm) 0ev. (mm) 0ev.B24641 1008 Lomis 2—3 A 0.175 NA 0.404 NA 0.154 NA(0.041% C) B 0.16 NA 0.433 NA 0.184 NALomis 2—7 A 0.188 0.095 0.388 0.293 0.221 0.103B 0.176 0.088 0.398 0.223 0.144 0.075CC 10 3—3 A 0.233 0.115 0.316 0.229 0.215 0.12453 ml B 0.205 0.092 0.291 0.2 0.207 0.07853 ml 3—7 A 0.232 0.123 0.289 0.199 0.204 0.0988 0.207 0.09 0.252 0.197 0.192 0.097B24642 1008 Lomis 2—3 A 0.149 0.052 0.368 0.243 0.169 0.074(0.043% C) B 0.18 0.083 0.443 0.258 0.167 0.086Lomis 2—7 A 0.179 0.086 0.314 0.311 0.129 0.085B 0.211 0.109 0.327 0.229 0.13 0.074CC 10 3—3 A 0.158 0.088 0.314 0.284 0.166 0.09353 ml B 0.137 0.077 0.293 0.21 0.152 0.07553 ml 3—7 A 0.16 0.091 0.289 0.252 0.134 0.087B 0.172 0.086 0.27 0.156 0.217 0.106A23412 1010 Lomis 2—3 A 0.33 0.136 0.324 0.196 0.328 0.158(0.09% C) B 0.326 0.15 0.372 0.179 0.301 0.127Lomis 2—7 A 0.336 0.137 0.356 0.22 0.298 0.125B 0.333 0.144 0.331 0.181 0.332 0.148Steelsk 3—3 A 0.349 0.16 0.364 0.253 0.274 0.15153 ml B 0.308 0.18 0.394 0.248 0.241 0.15940 ml 3—5 A 0.339 0.159 0.325 0.195 0.257 0.147B 0.366 0.173 0.407 0.227 0.271 0.15230 ml 3—7 A 0.319 0.166 0.347 0.17 0.264 0.155B 0.302 0.154 0.29 0.17 0.3 0.135B24653 1010 Lomis 2—3 A 0.343 0.143 0.329 0.199 0.305 0.146(0.09% C) 8 0.335 0.151 0.375 0.213 0.376 0.123Lomis 2—7 A 0.321 0.136 0.338 0.186 0.362 0.16B 0.314 0.144 0.362 0.176 0.355 0.134Steelsk 3—3 A 0.296 0.151 0.452 0.214 0.296 0.17830 ml B 0.448 0.191 0.361 0.209 0.38 0.19320 ml 3—7 A 0.322 0.144 0.448 0.218 0.347 0.185B 0.319 0.168 0.389 0.207 0.28 0.157B24654 1012 Lomis 2—3 A 0.219 0.134 0.317 0.171 0.33 0.13(0.12% C) B 0.282 0.17 0.33 0.197 0.339 0.133Lomis 2—7 A 0.26 0.114 0.265 0.165 0.267 0.139B 0.274 0.114 0.31 0.171 0.284 0.115167TABLE 5.23: Billet protilometer measurements con’t.OSCILLATION KARK DEPTESoilType!Grade RateBlach53 ml53 mlHeatNo._____B24654 1012(0.12% C)A23409 1015(0.15% C)824644 1018(0.18% C)824647 1039(0.42% C)ICW Std. CL Std. OCW Std.Billet Side (mm) Dcv. (mm) Dcv. (mm) Dcv.3—3 A 0.246 0.116 0.402 0.206 0.273 0.149B 0.333 0.193 0.35 0.197 0.241 0.1263—7 A 0.235 0.113 0.339 0.219 0.251 0.138B 0.365 0.171 0.361 0.193 0.246 0.125Lomis 2—3 A 0.158 0.079 0.215 0.117 0.302 0.167B 0.199 0.099 0.238 0.148 0.159 0.08Lomis 2—6 A 0.161 0.077 0.269 0.152 0.197 0.123B 0.214 0.09 0.296 0.16 0.164 0.096Lomis 2—8 A 0.184 0.085 0.231 0.11 0.211 0.104B 0.19 0.092 0.274 0.137 0.186 0.104CC 10 3—3 A 0.162 0.092 0.266 0.099 0.158 0.07640 ml B 0.201 0.11 0.224 0.11 0.15 0.0730 ml 3—6 A 0.191 0.077 0.226 0.11 0.164 0.094B 0.251 0.13 0.221 0.135 0.192 0.084Lomis 2—3 A 0.195 0.093 0.157 0.074 0.15 0.08B 0.184 0.096 0.199 0.104 0.119 0.053Lomis 2—7 A 0.128 0.056 0.171 0.082 0.144 0.081B 0.16 0.105 0.17 0.089 0.121 0.052CC 10 3—3 A 0.177 0.066 0.15 0.08 0.12 0.05853 ml 8 0.162 0.091 0.189 0.087 0.1 0.05253 ml 3—7 A 0.146 0.079 0.181 0.106 0.157 0.07B 0.143 0.072 0.238 0.12 0.149 0.174Lomis 2—7 A 0.101 0.041 0.124 0.058 0.118 0.071B 0.155 0.058 0.146 0.069 0.115 0.043CC 10 3—7 A 0.121 0.053 0.17 0.099 0.121 0.04653 ml B 0.174 0.057 0.112 0.069 0.1 0.046168larger than those at the off-corner of the billet for all grades except the two highercarbon steels. This effect is also evident in Figures 5.41 through 5.48 where theaverage roughness values and the standard deviations for two 1008 grades(0.04% and 0.043% carbons), the 1010 (0.09% carbon), the 1012, 1015, 1018,and 1039 steel grades are plotted.The three highest carbon grades, 1015, 1018, and 1039, all displaysignificantly reduced surface roughness as indicated by the results in Table 5.24and in Figures 5.46, 5.47, and 5.48. The 1008 and 1010 billets all have anaverage roughness along the centreline above 0.252 mm with the majority above0.300 mm. In the 1015 grade, the ceritreline roughness measurements are lowerranging between 0.215 mm and 0.296 mm while the 1018 and 1039 grades havethe lowest values mainly between 0.110 mm and 0.190 mm. Comparison of datafrom Figures 5.41 and 5.42 for the 1008 and Figures 5.47 and 5.48 clearly showthe effect of carbon on roughness.Figures 5.43 and 5.44 show the degree of roughness for a 0.09% carbonwhen Lubricant A was employed at reduced flows. Centreline roughnessremains largest and has the greatest variation. Insufficient data preventsconclusions from being made regarding the effect of reduced oil flow on thequality of billet surfaces.Table 5.24 lists the average centreline roughness for each grade. Thisnumber represents the average roughness when considering all billets fromStrand 3 at all flows within each heat. Clearly these values suggest that surfaceroughness is carbon dependent.169I F I —Grade: 10087OSide I05 DSide204’ 82-303 O:L02’ 75m10•1(est.)0•5’04’ 2-703•0•4•03’ 3-3Oil:C02’ 53m101’0•4’0•3 3-702’0•I’00ff—Corner Off—CornerFigure 5.41: Average roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant C at 54mi/mm. Grade was a 1008.170SCU014cI((I0•403020•I0•40•302012-30l:L75m1(est.)2-70403020•I3-353m1040•3020•I3-70 0ff—Corner 0ff—CornerFigure 5.42: Average roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant C at 54mi/mm. Grade was a 1008.171I__________I Grode:t0l&lI OSidel05’ [Side2 IO.4 2-303’ OiI:L?5m102 (est.)04’ I 2-703’02’ -05•3-3C 0iI:A a04 H 53m1a10302’ a0C) ,-— -Q.4, H() 4QmI03’ a.02’04’ 3-703 30m102”01 I0ff-Corner Off—CornerFigure 5.43: Average roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant A at 44 and 34mi/mm. Grade was a 1010.1720’40302050•40’3.02Figure 5.44: Average roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant A at 34 and 24mi/mm. Grade was a 1010.0504030223Oil: L75m1(est.)0•40•327II 05 C( II 3.-3Oil:A30m1FI3-720m10•I Off— Corner Off— Corner173mC0CC)#44‘4C’)020•403022-3Oil: L75m1(est)0•l2-70•5040•3020•l3-30il:B53m10•5020•l3-70 Off— CornerFigure 5.45:Off-h CornerAverage roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant B at 54mi/mm. Grade was a 1012.174IIU0‘4400402012-3Oil: L75m1(est.)0•40•30201H2-60’30•I2-8030I3-3Oil: C40m10•40•3020i0I3-630m10ff-Corner Off- CornerFigure 5.46: Average roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant C at 44 and 34mI/mm. Grade was a 1015.175I I IGrade: 10180’ Side IC Side 20 Ofl:L01 75m1(estj0•302 2-7CL• 0U0•302’ 3-3CI’ Oil:CC’53m10•3‘3-702’01’C 1Off—Corner Q 0ff—CornerFigure 5.47: Average roughness depth for strands 2(control) and 3 (test) from billet samples.Test strand employed Lubricant C at 54mi/mm. Grade was a 1018.176I I IGrade: 1039OSide ISide 203’02 2”701 0iI:L75m10‘(est.)IIO 03o 3-7‘i:: 02 OiI:C53m10•1’I I_0 0ff-Corner 0ff-CornerFigure 5.48: Average roughness depth for strands 2 (control)and 3 (test) from billet samples. Test strandemployed Lubricant C at 54 mI/mm. Grade was a 1039.177TABLE 5.24: Averaqe surface roughness along thecentreilne for several steel grades.GRADE HEAT ROUGHNESS STANDARD DEVIATION(mm) (mm) (%)1008 B24641 0.287 0.0264 9.21008 B24642 0.292 0.0181 6.21010 A23412 0.355 0.0436 12.31010 B24653 0.413 0.0448 10.91012 B24654 0.363 0.0275 7.61015 A23409 0.234 0.0213 9.11018 B24644 0.190 0.036 19.21039 B24647 0.141 0.041 29.1Depressions were present only on two grades, 1008 and 1010. Table5.22 reveals that 88% of the low carbon 1 008s and some of the 101 Os exhibitedsevere surface depressions. From Table 5.23 and Figures 5.41 and 5.44, thedeeper profilometer trace along the centreline reflects the presence of the deepmid-face depressions which occurred virtually on all retrieved billet sections in the1008 grade and on some of the 1010 billets. This effect is not apparent in thehigher carbon steels seen in Figures 5.46 to 5.48. The 1012 did not exhibitdepressions but having a carbon content near the peritectic, it is not surprisingthat the surface of these billets was relatively rougher than the higher carbongrades examined.Surface cracking was not apparent on any of the billets cast on the teststrand. The only billet which had a visible crack was from Strand 2 during thecasting of Heat 824637 which was a 1008 grade. The surface of this billet is178shown in Figure 5.49 and the crack is obvious. A longitudinal section was slicedfrom this billet and is shown in Figure 5.50. The depth of this crack extendedinward 10 mm with secondary cracking extending a further 10 mm.From the prepared transverse and longitudinal sections, subsurfacecracking is apparent at the base of the depressions formed in the test mould.The cross section of a transverse slice from a 1008 billet is shown in Figure 5.51and illustrates these cracks. The longitudinal sectic n Figure 5.52 is of a 1008billet cast at the same time as the billet from Strand 2 shown in Figures 5.49 and5.50 with the surface crack along the depression. Two cracks at the base ofeach depression are apparent in this figure. Crack initiation occurs 3 to 4 mmfrom the billet surface suggesting that the depression was formed in the upperregion of the mould.Figures 5.53, 5.54, and 5.55 are photographs of the surfaces of 1008,1010, and 1039 billet samples. On the photograph of the 1008 billet, Figure 5.53,two depressions are indicated and the oscillation marks shown are uneven.Table 5.22 indicates that all grades demonstrated irregular spacing andnonparallel marks but does not consider the degree of the variability. Thesurface of the 1010 steel grade billet shown in Figure 5.54 is typically rough withobscured oscillation marks. The 1039 surface shown in Figure 5.55 clearly hasbetter defined oscillation marks which are more linear and occur almost uniformlyacross the face. Any aberrations present on the oscillation marks are minutecompared to the lower grade steels.Figure5.49:Photographofthesurfaceofa1008gradebilletfromstrand2(control)showingdepressionsandasurfacecrack.1 0Figure5.50:Longitudinalsectionofthe1008gradebilletfromstrand2(contrOl)showingthepresenceofacrackatthebaseofadepression.Scale:0.8mm=1.0mmC181Figure 5.51: Transverse section of a 1008 grade billetfrom strand 3 (test) showing the presence ofcracks beneath a depression.Figure5.52:Longitudinalsectionofa1008gradebilletfromstrand3(test)showingthepresenceofcracksbeneathadepression.Figure5.53:Photographofthesurfaceofa1008gradebilletshowingtheirregularspacingandnonparallelismoftheoscillationmarks.Figure5.54:Photographofthesurfaceofa1010gradebilletshowingthesurfaceroughnesstypicalofthisgradeofsteel.185‘+ly3.36’I:.‘ s ——r - - — —-:-- -:: -Figure 5.55: Photogrpah of the surface of a 1039 grade billetshowing shallower and even oscillation marks comparedto the other two grades (1008 and 1010)186The profilometer traces for the 1008, 1018, and 1039 billets cast undersimilar mould lubricant conditions are shown in Figures 5.56 through 5.58. Onthe surface of the 1008 billet, Figure 5.56, two depressions appear: 70 mm and45 mm in length. The trace of this billet reflects the irregular spacing of theoscillation marks already seen in this grade by the photograph in Figure 5.53.The 1018 and 1039 billet surfaces have better defined oscillation marks whichdecrease in depth as carbon content increases.Figures 5.59 to 5.63 show the profilometer traces for the 1010 grade castwith Lubricant A at reduced flows. The traces, Figures 5.61 and 5.62, indicatethat depressions are more apparent on the surfaces of those billets cast underthe higher flows of 54 and 44 mI/mm. Just as the load cell signals for the secondheat of this 1010 steel grade reflected less mould-strand interaction, the twoprofilometer traces, Figures 5.60 and 5.61, show that the billet cast in thesequence heat under a flow of 34 mI/mm had a less erratic topography than didthe billet cast slightly earlier cast under similar lubricant conditions.Using the profilometer traces, the average oscillation mark depths andstandard deviations were computed and are listed in Table 5.25. In order toprovide some perspective, the largest and smallest depths measured for eachbillet are also included. The 1008 grade has the deepest oscillation markaverage of all grades examined. The smallest average for an oscillation markdepth is 0.239 mm which occurs for the 1039 grade. The values given for the1010 grade suggest that some tendency may exist for decreasing oscillationmark depth with reduction of oil flow within heats. With the exception of the 44mI/mm oil rate, lower flows produce smaller oscillation mark depths.OSCILLATIONMARKDEPTHGRADE1008o-WACO6403-2SIDEACLLubricantCat54mI/mm-0.1--0.2--0.3--0.4--0.5--0.6--0.7--0.8--0.9--1--11--1.2--1.3--1.4--1.5--1.6--1.7--1.8-—1.9— 04080120160200240DISTANCE(mm)Figure5.56:Profilometertracesfora1008steelgrade.OSCILLATIONMARKDEPTHGRADE1018IVACOPLANTTRIALB246433-3SIDEACLLubricantCat54mI/mm—Mi\!\/IIIIIIIIIt1I020406080100120DISTANCE(mm)CD w0.10-0.1-0.2-0.3-0.4-0.5--0.6140160180Figure5.57:Profilometertracesfor a1018steelgrade.OSCILLATIONMARKDEPTHGRADE1039IYACOPLANTTRIALB246473-7SIDEACL0.3-LubricantCat54mI/mmIM3020406080100120140160180DISTANCE(mm)Figure5.58:Profilometertracesfor a1039steelgrade.OSCILLATIONMARKDEPTHGRADE1010IVACO6533-7SIDIACLOIL20mI/mm0-—LubricantAat24mI/mm-0.1--0.2--0.3--0.4--0.5--0.8--0.9-—1——1.1—-1.2--1.3-—1.4—-i04080120160200240DISTANCE(mm)Figure5.59:Profilom€tertracesfora1010steelgrade0whenLubricantAisfedat24mI/mm.OSCILLATIONMARKDEPTHGRADE1010-0.6-0.7-08--0.9-—IDISTANCE(mm)Figure5.60:Profilometertracesfora1010steelgradeatwhenLubricantAisfedat34mt/mEn.IVACOPLANTTRIALB246533-3SIDEACL0.2 0.10-0.1-0.2-0.3-0.4-0.5jLubricantAat34mI/mm:1/:VIIIIIII04080120160200240280OSCILLATIONMARKDEPTHGRADE1010WACO4123-7SIDEACLOILFLOW30ml0-LubricantAat34mI/mm-0.1--0.2--0.3--0.4--0.5- IIIIIIII04080120160200240DISTANCE(mm)Figure5.61:Profilometertracesfora1010steelgradeatwhenLubricantAisfedat34mI/mm.OSCILLATIONMARKDEPTHGRADE1010IVACO4123-5SIDEACLOILFLOW40m0--0.1--0.2--0.3--0.4--0.5--1.3-LubricantAat44ml/mjn—1.4— 04080120160200240DISTANCE(mm)Figure5.62:Profilometertracesfora1010steelgradeatwhenLubricant Aisfedat44mI/mm.OSCILLATIONMARKDEPTHGRADE1010WACO4123-3SIDEACLOILFLOW54ml0--0.1--0.2--0.3--0.4--0.5- El*—1.1—-1.2--1.3-LubricantAat54mI/mm—1.4—(III1III020406080100120140160180200DISTANCE(mm)Figure5.63:Profilometertracesfor a1010steelgradeatwhenLubricant Aistedat54mI/mm.195TABLE 5.25: OscIllation mark depth and standard deviationfor several oils and flows.STEEL OIL FLOWAVERAGE STANDARD DEEPEST SHALLOWESTGRADE TYPE DEPTH DEVIATION(rrdlmin) (mm) (mm) (mm) (mm)1008 C 54 0.693 0.606 1.8220.2801010 A 24 0.484 0.165 0.8370.23334 0.537 0.170 0.860 0.24434 0.575 0.282 1.068 0.26744 0.467 0.109 0.712 0.29754 0.627 0.245 1.027 0.2311015 C 44 0.330 0.080 0.4750.2501018 C 54 0.400 0.141 0.6120.2651039 C 54 0.239 0.119 0.4800.100All grades, with the two exceptions of the 101 8s (C at 54 mI/mm) and the1 039s, contained surface inclusions. In Figure 5.53, inclusions areevident andare indicated on the photograph at several sites on the 1008 billet surface.Pinholes were present on 12% of the 1008 steel grade billets cast with LubricantC at a rate of 54 mI/mm, 33% of the 1010 steel grade billets cast with Lubricant Aunder several flows, 50% of the 1012 steel grade billets cast with Lubricant Bunder several flows, and 33% of the 1015 steel grade billets cast with Lubricant Cat several flows.5.3 Lubrication of Meniscus During the Oscillation CycleTheoretical calculations were conducted to determine the velocity of theoildown the mould wall for several wall temperature distributions. These profileswere then utilized to determine how well the meniscus is lubricated duringan196oscillation cycle.Given a viscosity-temperature profile and a flow, the velocity profile of thelubricant was found employing the software program, TEACH-2E [63]. Theprogram was run by V. Sahajwalla [65] using a constant flow of 45 mI/mm, theviscosity-temperature profile for Lubricant A, and an assumed mould temperatureprofile obtained from heat flux measurements made at Western Canada Steel inRichmond, British Columbia [66]. The assumptions for the model are:1. Flow is laminar.2. Steady state conditions exist.3. The initial momentum is neglected. The verticalmomentum at the entrance point is zero and due tothe low horizontal velocity of the oil, thehorizontal momentum is minimal.Cooling water effects were also examined by increasing the mould walltemperature distribution accordingly. The predicted average oil velocities alongthe mould wall for several cooling water conditions are shown in Figures 5.64 to5.68. Assuming the meniscus level is 100 mm below the top of the mould, theaverage oil velocity remains constant until three-fifths of the way down, at whichpoint, the velocity dramatically increases. Velocities range between 1.05 and1.42 cm/s for a mould water temperature range of 10 to 50 0 C.For a stroke length of 10.0 mm and a meniscus level at 100 mm below thetop of the mould, it is reasonable to assume the velocity of the oil in this region tobe linear. Integrating these linear relationships, the distance the oil travels in anyAVERAGEOILVELOCITYvsDISTANCEINLETWATERTEMP-10C1.8-1.7-1.6-(0) ()0.6-0.4-020406080100120VERTICALDISTANCEALONGTHEMOULD(mm)Figure5.64:Averageoilvelocityalongthemouldwallwhentheinletwatertemperatureis100C.2 1.91.8 1.7 1.2 1.11 0.9 0.5AVERAGEOILVELOCITYvsDISTANCEINLETWATERTEMP-20CU) 0020406030VERTICALDISTANCEALONGTHEMOULD(mm)Figure5.65:Averageoilvelocityalongthemouldwallwhentheinletwatertemperatureis200C.100120IDU) C.) EAVERAGEOILVELOCITYvsDISTANCEINLETWATERTEMP-30C2.2 1.6 1.4 1.2 0.8 0.6 0.4 0.2020406080100VERTICALDISTANCEALONGTHEMOULD(mm)Figure5.66:Averageoilvelocityalongthemouldwallwhentheinletwatertemperatureis300C.120AVERAGEOILVELOCITYvsDISTANCE22INLETWATERTEMP-40C2- 1.8=016 1.4-01.2 0.8-wi..UUU•U0.6-0.4-0.2-0—IIIII020406080100120VERTICALDISTANCEALONGTHEMOULD(mm)Figure5.67:Averageoil velocityalongthemouldwallwhentheinletwatertemperatureis400C.AVERAGEOILVELOCITYvsDISTANCE2.2-INLETWATERTEMP-50C2- 1.8-16-1.4-01.2-1 o1-LU (90.8-••LU....0.6-0.4-0.2- 0—II1020406080100120VERTICALDISTANCEALONGTHEMOULD(mm)I”)Figure5.68:Averageoilvelocityalongthemouldwallwhentheinletwatertemperatureis500C.202time period can be determined. Figures 5.69 and 5.70 schematically representthe flow of a cool and warm oil relative to the displacement of a mould oscillatingat a frequency of 2 Hz. The lower solid horizontal line represents the bottom ofthe oscillation stroke 100 mm down the mould wall. At the top of the stroke, themould will have moved upwards placing the meniscus 110 mm below the top ofthe mould. The principal assumption made in this analysis was the existence ofa constant oil supply at the 100 mm position of the mould.In Figures 5.69 and 5.70, the schematic (a) represents the beginning ofupstroke where the oil supply is constant. During the upstroke period, themeniscus effectively moves downward relative to the 100 mm position, ZM, of themould. By 0.06 seconds after upstroke begins as shown in (b), the non-lubricated mould surface extends 0.653 mm and 0.412 mm for the cool and warmoils, respectively. Illustrated in (e) is the area of rionelubrication when the top ofthe stroke is reached. The dry region extends 6.798 mm for the cooler oil whilethat for the warmer oil is slightly less at 5.575 mm. Even though the warmer oilprovides 12% more lubricated surface area than does the cooler lubricant, theunlubricated surface is extensive.As downstroke begins, the mould is bringing the supply of oil towards themeniscus. Theoretical calculations indicate that oil once again reaches themeniscus after 0.345 seconds from the beginning of the upstroke while the cooleroil is delayed by a further 0.04 seconds. Halfway through the downstroke, theexcess supply of warm oil is now 55% larger than that for the cooler oil.Similar analyses were completed for stroke lengths of 6.35, 12.7, 19.05,CoolMouldLubricont(10°C)Z5Osin4lTtfrequency2HzI.-UpstrokeDownstroketO.375stO.40Sf0.435stO.5Ost:O625s10.675stO.725st:O.75s(a)(b)(c)(d)(e)(f)(g)(h)ZMZZMzMeniscushiZhiFigure5.69:Schematicillustrationoftheoillubncationatthemeniscusdunnganoscillationcyclewhenthewatertemperatureis100C.excessoileWormMouldLubricant(50°C)Z:50sin47Ttfrequency2Hz[I—UpstrokeDownstroket:O.375stO.40stO435st:O.50stO625st=O.675s1=O.725sI=O.75s(a)(b)(c)(d)(a)(f)(‘)(h)ZhlZMCZzUUQ2 0 h.. U)zLubricantnotreachingmeniscusUMeruscuszz__________________________U’ZUFigure5.70:Schematicillustrationoftheoillubricationatthemeniscusduringanoscillationcyclewhenthewatertemperatureis500C.excessoil205and 25.4 mm (1/4, 1/2, 3/4, and 1 inches) and for mould frequencies of 1.333,1.667, 2.0, and 2.333 Hz (80, 100, and 140 cpm). The effective periods of alubricated meniscus are graphed in Figures 5.71 to 5.74 for several strokelengths at the four mould frequencies. The ordinate represents the duration ofthe oscillation cycle and stroke length is on the abscissa. The duration periodwhen oil is feeding the meriiscus is represented as the area between thecorresponding curves. Consider Figure 5.71. The frequency is 80 cpm and,therefore, the period of the oscillation cycle is 0.75 seconds. If the oil is warm (50o C) and the stroke length is 6.35 mm, oil feeds the meniscus for 67% of theoscillation cycle, initiating at 0.2 seconds, after downstroke begins, andterminates at 0.72 seconds, almost halfway through the upstroke. The resultsgraphed in Figures 5.71 through 5.74 indicate that as stroke length increases, oilreaches the meniscus for an appreciably smaller fraction of the oscillation periodand that the downstroke period of the oscillation cycle receives more lubricantthan the upstroke period. For 80 cpm, the warmer oil reaches the meniscus for65% of the downstroke period and remains in contact only for 8% of the upstrokeperiod. At 120 cpm oi(-meniscus contact is substantially lower with contactestablished for 52% of downstroke but a contact time of 10% during upstroke.As frequency increases from 80 to 140 cpm, the effect of the lubricanttemperature decreases. At frequencies of 80 and 100 cpm, the disparitybetween the 10 and 50 degree heated oil is more significant than at the twohigher frequencies of 120 and 140 cpm. At the highest frequency of 140 cpm, nodifference exists between the duration of oil contact time between the cooler andwarmer oil temperatures.The three Figures 5.75, 5.76, and 5.77 illustrate the period of meniscusU, 0 C) > C) C 0 0 C) U, 0 0,C L. 0 0 E I-6351271905254StrokeLength(mm)Figure5.71:Durationof timeoilispresent atthemeniscusfor different strokelengthswhentheoscillationfrequencyis1.333Hz(80cpm).09 08 07 060 U’50°C05Time otoilpresentmeniscus0410°C0 f=45m1/minf=80cpm(1333Hz)PO75s0187509375f/•\Q.3750350°Cz 00 b. U’ C 0 00205625tC) 0 0 C) 0 0 C 0 a, E I.Stroke(mm)Figure5.72:Durationoftimeoilispresentatthemeniscusfordifferentstrokelengthswhentheoscillationfrequencyis1.667Hz(100cpm).F) C) —108 06 050 f45mI/mmf:IOOcpm(1667Hz)0 a D04:O6s015z 0-07502a, 0 a C 0045t0•I 0635127190525407-Durationoftimeoilispresentatthemeniscusfor different strokelengthswhentheoscillationfrequencyis2.0Hz(120cpm).0 C) 0 C 0 0 0 U, 0 0, C z 0 0 E I-0 4- ‘A06-05.-04 03 02 01Time0 f=45m1/m,nfl2Ocps(200Hz)P05s01250625(\0.251Coilpresent01meniscusz 0-0 4- ‘A C 0 0I0375635II1271905254StrokeLength(mm)Figure5.73:(‘3f°45m1/minC) >1 C) 0 C) U, 0 0, z 0 4 E I-06-05-040302 01 00 4- U,DTimeoilpresenl0150°CmeniscusfI4Ocpm(2333Hz)P=04286SOH053(\.20.4286032z19-50°Ca,.0 U, C 0 0II_6351271905254StrokeLength(mm)Figure5.74:Durationoftimeoilispresentatthemeniscusfor differentstrokelengthswhentheoscillationfrequencyis2.333Hz(140t\) Ccpm).210lubricant contact for the two mould frequencies of 1.333 and 2.0 Hz (80 and 120cpm) and at 5 stroke lengths. The first two figures show the contact period whenthe warm oil is employed for the two mould oscillation frequencies. The lastfigure illustrates the difference in meniscus-oil contact time for a cool oil fordifferent stroke lengths when employed during a mould frequency of 2.0 Hz (120cpm).From Figures 5.75 and 5.76, it is evident that the period of lubrication atthe meniscus is longer for a slower frequency. Also apparent is that themeniscus receives lubricant for a longer period for shorter stroke lengths. For astroke length of 6.35 mm (1/4 in), the meniscus is lubricated for almost 3/4 of thetotal oscillation cycle; from the beginning of the downstroke to halfway though theupstroke while for the 25.4 mm (1 in), lubrication occurs for only an 1/8th of thecycle.Comparison of the warm and cool oils at the same frequency of 2.0 Hz(120 cpm) (Figures 5.76 and 5.77) indicate that the initial contact of the oil andthe meniscus is delayed while the termination of the contact period is initiatedearlier. Again the shorter stroke lengths are mostly affected. This analysisclearly indicates that the meniscus region is only receiving lubricant during onequarter to half the oscillation cycle corresponding to the downstroke period. InChapter 6, where the load cell response is characterized and a proposedexplanation of the mould-strand interaction is provided, it will become obvious,that the meniscus region is being lubricated during a period when the mould ismechanically bearing down on the strand which masks the effects of anylubrication.211LUBRICATION AT MENISCUSCEC-.)0(flbFOR AN OSCILLATIONSTROKE LENGTh25.40 mm ( 1 in)19.05 mm (/4 in)12,70 mm (1/2 in)9.5 mm (3/8 in)Figure 5.75: Lubrication at the meniscus for anoscillation frequency of 1.333 Hz (80 cpm)when cooling water temperature is 50 0 C.FREQUENCY OF 1.333 HERTZ151050-D/0.1875 0.. 0.9375-15—6.35 mm (1/4 r)me (s)E0(.1bFigure 5.77: Lubncatiori at the meniscus for anoscillation frequency 012.0 Hz (120 cpm)when cooling water temperature is 100 C.212FOR AN OSCILLATIONOF 2.0 HERTZ/LUBRICATION AT MENISCUSFREQUENCY1510SC-5-1Q-0.125 C. 0.625‘wi)n)in)STROKE LENGTH—25.40 mrn( I—19.05 mm (/4—. 12.70 mm (1/2--9.5 mm (/8—6.35 mm (1/4Time213CEc-)0(fJFigure 5.76: Lubrication at the meniscus for anoscillation frequency of 2.0 Hz (120 cpm)when cooling water temperature is 50 0 C.FOR AN OSCLLATONOF 2.0 HERTZ/LUBRICATION AT MENISCUSFREQUENCY151050-5-10-150.125 0. .5ST0KE LENGTH25.40 mm (19.05 mm (3/412.70 mm (1/29.53 mm (3/86.35 mm (1/40.625r)ir)ri)Time /NS)214The analysis of the plant trial data completed in this chapter illuminatedthe interaction between the mould and billet. It became evident that mould oil, oilflow, and carbon content of the steel strongly influences the load cell responseduring the upstroke period. Casting speed fluctuations indicative of possiblemould-strand interactions affects the downstroke period. This interaction mostprobably is mechanical.Summarized in point form are the important findings of the load cellanalysis:1. The results from the initial plant trial at Western Canada Steel provideddirection for a second plant trial at IVACO as it was seen that the load cellresponse was indeed characteristic and reproducible. However, during this trial itbecame evident that the front load cells gave inconsistent signals and that norelationship between carbon and change in load existed.2. A comparison of oil properties indicated that Lubricant A had the highest flashpoint but variation between supplier and laboratory values is sufficient to reducethe significance of this parameter. At temperatures less than 50 0 C, Lubricant Chas the highest viscosity thereby increasing the time needed for the oil to reachthe meniscus. At operating temperatures when steady state has been reached,all three mould lubricants have similar viscosities. Lubricant A has the highestfatty acid content by 30 times. Of the fatty acid present, Lubricants A and B havedouble bonds in 90% of the hydrocarbon molecules present.3. The upper load cell response is mould lubricant and flow sensitive.Decompression of the load cells occurs generally at the beginning of negative-215strip time. There exists a strong correlation between minimum loading andcasting speed. Maximum load for Lubricants A and B decreases as oil flowincreases; no such trend exits for Lubricant C. Lubricant B demonstrates theleast variability in delta load during upstroke but the greatest variability inmaximum load. Lubricant C has the least sensitivity in delta load duringupstroke. Lubricant A at 44 and 54 mI/mm has essentially equal delta loads asdoes Lubricant B at 34 and 44 mI/mm. Lubricant B shows the most susceptibilityto increased delta loads as oil flow is decreased; Lubricant C the least. LubricantA is the most consistent having the smallest range of delta loads. Upper load cellresponse is somewhat carbon sensitive: the lower carbon grades show abroader loading distribution spectrum than do the higher carbons.4. The 1008 steel grade demonstrated deep depressions on the billet surface.Oil flow did not significantly affect the surface quality of the billet.5. The period of lubrication occurred during the downstroke period and notduring upstroke. Shorter stroke lengths provided more lubrication at themeniscus as did slower oscillation frequencies.216CHAPTER 6. DISCUSSION6.1 IntroductionChapter 5, which was primarily an analysis of the results of the plant trial,included an examination of the properties of the lubricants, particularly in thecontext of the mechanisms of lubrication. Theoretical calculations were alsoperformed to assess whether there was a continuous supply of lubricant to themeniscus under all conditions. In this chapter, a mechanism for mould-strandinteraction will be presented that is consistent with the results in Chapter 5. It isworthwhile, however, to first discuss the salient findings from Chapter 5 which arepertinent to the mechanism that will be proposed.Lubricant properties and behaviour for different temperatures werediscussed in Section 5.2.1. Through the testing of different properties, it becameapparent that there were several significant differences between the three oilstested in the second plant trial. Lubricant A had the highest flash point relative toLubricants B and C. Lubricant A also had the lowest viscosity for temperatures ofless than 50 0 C indicating that for three-fifths of the distance down the mould tothe meniscus, this oil would flow the fastest. However, at temperatures whichexist near the meniscus, the viscosity behaviour of all three oils was similar. Onthe other hand, if boundary layer lubrication is significant, then the fatty acidcontent is important. The chemical analysis revealed that Lubricant A had thirtytimes the amount of fatty acids relative to the other two lubricants.The second and the largest section in Chapter 5 is the load cell data whichamounted to over 21 Megabytes of binary data. After plotting the load cell data217versus time, several observations were apparent. Responses from the load cellswere periodic, not sinusoidal as was the oscillation motion, and werereproducible. The load cell response was characterized by two distinctly differentregions: an upper part where the load varied between being relatively flat tobeing steeply peaked; and a lower portion which smoothly decreased to aminimum load and then smoothly increased. Also apparent were the effects ofdifferent oils and reduced flow on the upper portion of the load cell cycle. Carboncontent distinctly affected the upper peaks: the casting of lower carbon contents(1008, 1010, and to some extent 1012 steel grades) produced undulatingmaximum loads while the maximum loading pattern of the 1015, 1018, and 1039steel grades were invariant with time. Minimum load values undulatedirregardless of lubricant, flow, or carbon content. Initially it was not obvious whichparameter was responsible for the high degree of fluctuation in the minimum loaduntil casting speed was plotted on the same time scale as the load cell data. Atthat point, it became apparent that minimum load was highly dependent on themagnitude of casting speed.Correlation of the oscillation cycle with the load cell response showed thatthe smooth decrease in load occurred essentially during the period of negativestrip while the upper peak characteristics occurred during the mould upstrokeperiod. This observation, in conjunction with the influence of casting speed onminimum load and the smooth decrease in load during the negative-strip, periodwill be related to the mechanical interaction between the mould and the strandlater in this chapter.An important aspect of mould lubrication was the theoretical calculationson meniscus lubrication presented in Section 5.3 which were conducted with the218properties of Lubricant A. It became apparent that at a flow of 45 mI/mm, themeniscus would not receive any new lubricant during most of the upstrokeperiod. With this point in mind, it will be shown that the characteristics of the loadcell response during the upstroke reflect the lubricating conditions in the mould.The fourth topic in Chapter 5 was the assessment of the billet samples.The evaluation of the surface quality of the billet samples indicated that only twogrades of steel, the 1008 and 1010, exhibited transverse depressions. Surfaceprofile measurements showed that the billets of the 1010, 1012, and 1018 steelgrades were rougher than the 1039 steel grades. Other observed surfacedefects were minor.Finally based on current knowledge, it appears that there may be twomechanisms of lubrication operating during an oscillation cycle. Considerationhere must be given to both hydrodynamic and boundary layer lubricationmechanisms as discussed in Chapter 2. Hydrodynamic lubrication requires afinite film of oil between the mould and the strand and the performance of the oilis highly dependent on the viscosity. The mechanism of boundary layerlubrication invo’ves long-chain hydrocarbons which adhere to a substrate. Thesemolecules may oxidize at high temperatures to form a lacquer layer whichreduces friction by facilitating sliding between the mould and the strand.Pyrolysis of the hydrocarbons may also occur which increases the hydrogencontent of the gases in the air gap, and thereby, affects the heat transfer due tothe high conductivity of hydrogen. In the following sections these seeminglydisparate observations will be linked through a mechanism for mould-strandinteraction which is also consistent with earlier mechanisms proposed foroscillation mark formation.2196.2 Mould-Strand InteractionThe load cell response during an oscillation cycle is characterized by twodistinctly different regions. The downstroke period is marked by a smoothdecompression of the load cell while the load registered during the upstrokeperiod can be level or increasing accompanied by multi-secondary peaksdepending on the lubricating conditions. Evidence of the effect of lubricatingconditions on a load cell response has been presented in Figures 5.21 to 5.36. Itis proposed that the smooth response during the downstroke is dominated by themechanical interaction between the mould and the strand during the negativestrip period and that the upper peak response reflects mould lubricationconditions or carbon content. In the following sections, evidence will bepresented to support the hypothesis that the downstroke is dominated bymechanical interaction. In a subsequent section on mould lubrication, theupstroke period will be shown to reflect the lubricating conditions which is seen inthe load cell response.6.2.1 Mould-Billet DynamicsDuring the oscillation cycle which is sinusoidal, the mould.interacts withthe strand that is moving downward. The mould oscillates at a predeterminedfrequency with a preset stroke length while the strand is removed at a constantrate. The IVACO operation utilizes a 2 Hz mould frequency in conjunction with a9.5 mm (3/8 in) stroke length while the casting speed is in the range of 0.034 and0.042 rn/s (80 and 100 inches per minute) (Table 4.3). These parameterscharacterize the dynamics of the mould and billet interaction.220Figure 6.1 illustrates the displacement, velocity, and accelerationbehaviours of the mould during a complete oscillation cycle. The mould travels9.5 mm (3/8 in) from the top to the bottom of its stroke in 0.250 seconds and themaximum velocity attained by the mould is 60 mm/s which is reached halfwaythrough both the upstroke and downstroke portions of the oscillation cycle.The period in which the mould is travelling downward at a velocity greaterthan that of the strand is referred to as negative strip. Because thecharacteristics of the oscillation cycle are fixed, an increase in strand velocityreduces the duration of the negative-strip period. Figure 6.2 graphicallyillustrates this effect and Table 6.1 lists the times at which negative-strip isinitiated and terminated and its duration. If casting speed is increased by 4.2mm/s (10 ipm), the onset of negative-strip is delayed and the period is shortenedby approximately 10%. The importance of the effect of casting speed on thelength of the negative-strip period will become apparent in the discussion onminimum load.TABLE 6.1: Initiation and length of negative-strip timefor several casting speeds for a mould oscillationcycle having a 2 Hz frequency and a 9.5 mm stroke.Strand tneg tneq tneaVelocity begin end duraDon(mis) (1pm) (s) (s) (s)0.034 (80) 0.1729 0.3271 0.15420.038 (90) 0.1799 0.3201 0.14020.042 (100) 0.1875 0.3125 0.1251221zC-)>-C-)0-J>z0-JC-)TME (s)MOULD DYNAMICS FOR ThE IVACO CASTING OPERATION(I)(-)z>--JD00 0.2 0.4 0.6 0.8Figure 6.1: Mould dynamics for the IVACO casting operation.222J-)IMOULD VELOCITYIVACO PLANT TRIALTME (s)--60DFigure 6.2:ID—FREQuENCY: 0.5 SSTR0K: 9.525 mm(0.375 ) 12060>-U000.9>-45809000cmpmpm—-120L. — -180Influence of casting speed on the length of the negative strip period.2236.2.2 Net Force Between Billet and MouldThe direction of the reaction force on the mould resulting from theinteraction between the mould and billet will be upward or downward dependingon the relative velocity between the mould and the strand. Figure 6.3 illustratesthe mould-strand relative velocities considering a strand velocity of 0.038 rn/s (90ipm). When the mould is travelling downwards more rapidly than the strand, thestrand tends to push upwards on the mould, and decompresses the load cells,reducing the net compressive force. Conversely, during the positive strip period,any sticking or friction between mould and strand results in a downward force onthe mould and, therefore, an increase in the compressive load. The effect of therelative velocity on the force direction is illustrated in the schematic drawing inFigure Load Cell DecompressionThe smooth decompression of the load cells during the downstroke is,therefore, due to the upward reaction force generated by the mould bearing downon the strand as discussed in the preceding section. This event is closely linkedto the mechanism of oscillation mark formation in billets proposed bySamarasekera et al. [18] who postulated that during the negative-strip period,when the mould is moving down faster than the strand, the former squeezes onthe newly solidified strand at the meniscus causing the shell to buckle. Theresulting depression is an oscillation mark. This event is reflected in the loadresponse as a smooth decompression. The sequence of events is illustrated inFigure 6.5, on the assumption that the mould has assumed a negative dynamicoperating taper. The mould bulges out during operation and the location of theE>-HU0-J224RELATIVE MOULD VELOCITYIVACO PLANT TRIALNE UPWARDS-240>1007525-25-FREQUENCY: 0.5 sSTRQi<E: 9.525 mmC(C.D75 2)TNEG60w>nV.. --K0.4 0.6- -DCNET FORCE: DOWNWARDSLvi_Figure 6.3: Relative mould and strand velocities.Mouldvu’1 FM,4<4a) During t,, the bflhet pihe upwsrdon the moJid eat4ng dompeuaon01 the load cells.b) During tpos. the bithet pushes downwardon the moulo causing comprsselon of theload calls.euivFiguri 6.4: The resulting directiOfl of re net orce dLrnng thedynamic osc:Haton of me mou’d.226(a) (b) (c)Figure 6.5: The formation of an oscillation mark due to the mechanicalinteraction between mould and strand during the negative stripperiod.eMouldDisplacement TimestrokepstrokeLoad CellResponseFormation ofOscillationMark227maximum bulge occurs 90 to 100 mm below the meniscus. Thus, a straight,untapered mould would acquire a negative taper at the meniscus. However, thepositive taper imparted to the mould compensates for this effect, therefore, thedynamic operating taper depends on the relative magnitude of the two. Figure6.5 schematically shows the mould bearing down on the meniscus forming anoscillation mark and illustrates the concurrent response of the load cell. Thebeginning of the downstroke is illustrated in Figure 6.5a. The distorted mould,represented by the solid, cross-hatched line, does not begin to bear down on themeniscus until the beginning of the negative-strip period as shown in Figure 6.5b.The mould increasingly bears down on the solidified strand, shown in Figures6.5b and 6.5c, creating a larger decompression of the load cell due to the largerupward reaction force of the strand as discussed in section 6.2.2. During thisshort time period (0.12 to 0.15 seconds), the oscillation mark is formed. As theend of negative-strip is approached, reflected in the steadily increasing load, thefaster moving strand gradually disengages itself from the mould until the upstrokeperiod begins and the second mechanism of lubrication becomes operational.This will be discussed later in a section on mould lubrication. PoInt of Initiation of the downward Interaction of the mouldwith the strandLoad cell responses from the same run and, therefore, similar mouldlubricating and pouring conditions are reproduced and illustrated in Figures 6.6and 6.7. Both indicate that decompression begins at the start of the negativestrip period continuing until just after the maximum downward velocity istheoretically reached. It is evident from these two figures that no significant eventoccurs at the end point of negative-strip. The loading continues to increase until11111111iii111111111IIIIIliiiIiiiH,I1HIIIIIIIIIIliiiIIIIIllIliiiIIIIIIIIIIliiiliiiliiiIIIIIIIIIIllI!!!!!IIIIIIIII111IIIIIL.oadceIl’iIIIIIIII11111IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIllIIIIIIIIIIliiiiiI‘HIliiiIIIIiIIIHIIIIiiIIIilI”4IIIiiIIiIIiIIIll_________IIIIIiIIIIIIIIIIIIIIIIIIIIIIIIIIIIillIIiiIIIIIIIIIIIIIIIIIII—liiIIIIi\,IIiiIIIIIIIIIIIIiIIIIII’..IIIIIIIIIiIiiIIIillIII‘‘‘‘iiIIIIIIIIIIIIIIII_oadceiI3iiiIIIIIIillIIllIiiiIIIIIIIii‘IIIIIiiIIIIIIIIIIIIIIIIIIfliiiiiIiIIHIIIIIIiiIIIIiiIIiIIiIillIIIIIIIIIIIIIIIIIIiiIillIIIIIIIIIIIIIIllIIIIIIIIIIIIIHiII__LVIIIIIIillIIIIIIIIillIIliiIIIIiiiIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIiIIliiiIiliiiIIIIIiIIIIIIIiiIIIIIIIIIiiII_____iIII_____IIIIIIIIIIIIIIIIII_,L1114LIII-j•i..ø•eml01IIWNIN,U•U)USUIIFigure6.6:Thekeyoscillationfeaturesimposedonaloadcellresponsewhenstrandvelocityis36.0mm/s(85iom.4‘I I’IIIIIIIIIIIIIIIIIIIi\H IIII11111111 IIIIIIIIIIIIIIIHIJIIIIIIIIIIIIIIIIII\\II‘\II’I‘i”iJ1IIIIIIIIIII.IIIIIIIIIIIIIIIIIIIIIIIIIfIIIIIIIIIIIIIIIIIIIIIIIIIII’IIfIIIIIIIIIIIIIIIIIIIIIIIIIIIII III•iI.IIllIJ.iI_L_.1L.‘aso302304)0so.3iu:(sec)Thekeyoscillationfeaturesimposedonaloadcellresponsewhenstrandvelocityis42.3mm/s(100ipm).I...iso.1400is..IIIIIILoadCell4IHIIIIIIIIIIIIIIIIIIII’ LoadIIIIIIIICell311000IIIIII...jj1111fIIIII*0IIIIIIIIIIIIIIEAT—121Ii’I‘II——I--102S2242O00t\) r\)Figure6.7:230the beginning of upstroke, at which point the second mechanism, that oflubrication becomes operative. Time delay between minimum load and maximum downwardvelocityIt had been assumed that when the mould reached its maximumdownward velocity, the interaction between the mould and the strand would bealtered. This assumption proved to be incorrect. The time delay. between themaximum downward velocity of the mould and the minimum load is also afunction of the casting speed. In Figure 6.6, the casting speed is 36.0 mm/s (85in/mm) and the delay in minimum load is 0.05 seconds while for the castingspeed of 42.3 mm/s (100 in/mm) shown in Figure 6.7, the delay is 0.03 seconds.These time delays are consistently regular for specific casting speeds. Table 6.2compares the difference between the strand and the mould velocities when themould reaches its maximum downward velocity and when minimum load occurs.For comparison, the velocity at 0.03 and 0.05 seconds after the time minimumload occurs is also included. Under casting conditions of 42.3 mm/s (100 ipm),the mould travels 17.5 mm/s faster than that of the strand. This velocitydifference is increased by 36% to 23.8 mm/s under the slower casting speed of36 mm/s (85 ipm). In both cases examined, minimum load occurs at the pointwhere mould-strand velocity differs by 12 - 13 mm/s, after which the load beginsto increase, It is the difference between the strand and the mould velocitieswhich dictates the continuance of the load cell decompression; as the mouldcontinues to overtake the strand, decompression continues.231TABLE 6.2 Relative velocity difference between mould andstrand during the downstroke period for specifictimes.STRAND VELOCITY: -42.3 mm/s (1001pm)TIME MOULD VELOCITY DIFFERENCE(5) (mm/s) (mmls)MaximumMould velocity 0.25 -59.8 17.5Minimum load 0.28 -55.6 13.30.03 s later 0.31 -43.6 1.3STRAND VELOCITY: -36.0 mm/s (85 1pm)TIME MOULD VELOCITY DIFFERENCE(s) (mm/s) (mm/s)MaximumMould velocity 0.25 -59.8 23.8Minimum load 0.30 -48.4 12.40.05 s later 0.35 -18.5 - SeverIty of load cell decompression as a function of castingspeedThe sensitivity of minimum load with respect to casting speed is dictatedby the magnitude of the negative-strip since it is during this period thatmechanical interaction occurs. The load cell responses in Figures 6.6 and 6.7are from the same data but differing by 20 seconds. Under similar castingconditions but significantly different casting speeds, the minimum load changesfrom 7565 N (1700 lbf) at 40.0 mm/s (85 in/mm) in Figure 6.6 to 8455 N (1900 lbf)at 42.3 mm/s (100 in/mm) in Figure 6.7. Figure 6.8 also shows the response of0-JFigure6.8:Thefluctuatingcastingspeedimposedonloadcellresponsetoindicatetheeffectof speedonminimumload.1”)U)t\)Time(s)233minimum load with casting speed. At 42.3 mm/s (100 1pm) the minimum load forload cell Number 3 is about 8010 N (1800 lbf) while near casting speeds of 33.8mm/s (80 ipm), the load has decreased to 7565 N (1700 lbf). The calculatedlinear regression curves shown in Figures 5.14 to 5.18 and in Appendix A, alsosubstantiated by the high correlation coefficients in Table 5.6, indicate thatminimum load increases with casting speed. When casting at slower rates, notonly does the duration of the negative-strip period increase as shown in Figure6.2 but the relative velocity between mould and strand also increases. Thecombination of these two factors means that the mould begins to bear down onthe strand sooner and tries to overtake the strand to a greater degree than whencasting speeds are higher. As a result the minimum load reached by slowercasting conditions is lower than that reached by faster casting speeds.6.3 LubricationAs previously discussed, the decompression of the load cell occurs duringdownstroke at the time of formation of an oscillation mark. The effects of mouldlubricant, on the other hand, appear not to have any influence on thedecompression of load cells but markedly affects load changes during theupstroke.When observing the casting operation, sticking at the meniscus is visibleand is manifested by small areas along the mould wall at which solid steel isadhering to the chrome plating. In some instances, these areas appear to beclimbing upwards and then are released. These events are reflected on the billetsurfaces as the deformations which are evident along the oscillation mark as234seen in Figures 5.53 and 5.54 and to a lesser extent in Figure 5.55. Thepresence of secondary peaks during the upstroke may result from these adheringand releasing phenomena at the meniscus.6.3.1 The Effect of Changing Acceleration on Total LoadBecause the mould system is dynamic in that the acceleration componentis continuously changing, the downward force of the mould is also changing.Assuming a mould weight of 1112.5 N (250 lbf) per load cell, the force seen by aload cell will fluctuate with the acceleration of the mould and is shown in Figure6.9. Assuming equal distribution of the mass of the mould on each load cell, themass per load cell is 113.4 kg and the force varies according to Equation 6.1.FM=113.4(arn+g) 6.1where am is the mould accelerationg is gravityAt the beginning of upstroke, a quarter of the mould force is approximately 1030N (232 lbf) which increases to 1200 N (270 lbf) at the end of the upstroke period.Therefore, the force of the mould increases by approximately 170 N (38 lbf). Thisdifference suggests that during upstroke any increase in the force greater thanthe 170 N (38 lbf) can be attributable to other events in the mould such as theadherence of steel to the mould wall or some mechanical interaction betweenmould and strand. As the initial mould force is estimated, it is reasonable toassume a slightly higher value for changes in force which would represent mouldforce changes. Therefore, changes of loads less than 223 N (50 lbf) will beattributed to the force changes of the mould due to the variations in acceleration.235zwC-)cD-JD0>-H0wEMPTY MOULD-1000-1050-1100-1150-1200FORCE DURING AN OSCILLATION CYCLE-J0-JD0>-HFigure 6.9: Empty mould force during an oscillation cycle isshown.0 0.2 0.4 0.6 0.8TME (s)2366.3.2 Shear Stresses at the MeniscusTo determine whether or not the steel at elevated temperatures couldsupport the load changes seen during upstroke, a shear stress analysis wasundertaken. Changes of 1560 N (350 lbf) on load cell Number 3 were observedduring the upstroke period as indicated by Tables 5.10, 5.11, and 5.12. Due tothe distribution of load between the bolt, the members, and the load cell, thisforce of 1560 N is effectively 25% of the real load on the mould as described inAppendix A. A force of 6240 N (1400 lbf) is then actually being applied to themould. To determine whether sticking alone could account for this loadingchange, the shear stress produced within the strand was compared to theultimate tensile strength of continuously cast steel at the appropriatetemperature.Billet surface temperatures were assumed to lie between 1350-1450 0 Cas obtained from values calculated for billets by the mathematical model of heattransfer formulated by Brimacombe [61]. For steels at 13700 C, the maximumtensile stress of 10 MPa was obtained from studies conducted on the tensilestrength and ductility of continuously cast steels above 800 0 C by Weinberg [62].Using the value for the tensile stress, a resolved shear strength of 14 MPa forsteel at hot temperatures was then calculated.The area of sticking was detemined from the surface of a 0.09% carbonbillet shown in Figure 5.54. Some of the oscillation marks are non-uniformacross the face as is expected for low carbon steels. The oscillation marksexhibit waviness which reflects the shape of the meniscus at the time the marksformed. Operators refer to the clinging of these steel sections as skulling and237recognize that these localized portions of steel are adhering to the mould wall.The area of sticking was assumed to be the area of the deformation alongthe centre oscillation mark as indicated on the photograph in Figure 5.54.Therefore, the total area would be 6.45 x 10-04 m2.The shear stress resulting from the actual force on the steel at thedeformation site of 6240 N and an area of 6.45 x 10-04 m2 is 8.3 MPa whichrepresents only 69% of the theoretical shear stress for hot tearing at 1370 0 C[62]. Therefore, it is certainly plausible that the change in loading during upstrokemay indicate the degree of sticking and may reflect the ability of the lubricant tolubricate the mould and prevent sticking.Calculations were also made to determine the required area that wouldadhere and then release to produce the observed jagged load cell responseduring upstroke. These load changes typically range between 100 N (25 lbf) and500 N (110 lb). Using the 14 MPa shear strength value determined earlier, thearea that would have to stick and release in order to cause changes of this orderwould be approximately 30 square millimetres.6.3.3 Mould LubricationIn the following sections, the influence of lubrication on sticking at themeniscus and the concomitant response of the load cells will be analyzed.2386.3.3.1 Mould lubricant typeIn sections and, it was reported that the upper peakresponse was strongly influenced by lubricant type and flow rate. Comparing theupper peak response from Figures 5.25, 5.30, and 5.21, which are for a flow rateof 44 mI/mm, Lubricants A, B and C respectively, it is seen that the difference inresponses are due to the differences in lubricant type. All three responses arefor flow rates of 44 mI/mm. The upper peaks for Lubricant C in Figure 5.21exhibit consistently three or four sharp secondary peaks. Over 70% of the loadcell cycles experience load changes greater than 445 N (100 lbf) as indicatedfrom Table 5.15.The upper peaks shown in Figure 5.25 for Lubricant A are significantlydifferent than those of Lubricant C. Some of the cycles contain secondary peaksbut they are much less pronounced. From the distribution in Table 5.13, 70 % ofthe peaks observed with Lubricant A lie below 445 N (100 lbf) and only 2.2% areabove 668 (150 lbf). Lubricant B demonstrated again a different response duringupstroke than did the other two mould lubricants. The most significant feature ofthe response is the steady increase in load. These increases in load are thegreatest of the three oils tested since 98% of the peak distribution lie above 445N (100 lbf) as shown in Table 5.14.From the above comparisons, the behaviour of the upper peak responseswhich reflects the mould-strand lubrication reveals Lubricant A to have the bestlubricating efficiency. The upper peak responses for Lubricant B showed thegreatest increase in load with no release occurring during the upstroke. Inextreme cases (1558 N (350 lbf)), the change in load during upstroke can239represent over 8% of the total strand weight. Mould lubricant flowThe analysis of meniscus lubrication completed in Section 5.3 clearlyindicates that lubricant is present at the meniscus specifically during thedownstroke period. These calculations were based on a flow of 45 mI/mm;therefore, if flow is reduced the duration of the lubrication period will also bereduced. With this in mind, the effects of flow on mould loading are nowexamined.Comparison of the load cell response shown through Figure 5.21 to Figure5.32 illustrates the tendency for increases in the distribution in the change of loadduring upstroke with decreasing flow irrespective of lubricant type. Thequalitative analyses summarized in Tables 5.7 through 5.9, involving theexamination of peak shape and the development of categories incorporatingpeak height and number of secondary peaks, substantiates this claim. All threeoils tested produced similar load cell behaviour as oil flow was reduced butLubricant A and C were most alike. Lubricant B demonstrated greater peakheights (loading) than did the other two lubricants under reduced flow conditions.Of the three lubricant properties examined, the only one that may explain thisbehaviour during casting was the low flash point given by its supplier, Itsviscosity was the lowest above a temperature of 50 0 C and second lowest below50°C. The total free fatty acid content was 0.13% which is similar to that ofLubricant C while the fraction of the number of molecules having double carbonbonds present was greater than that of Lubricant C.240The load typically increases from beginning to end of the upstrokeparticularly for the lower flows as depicted in the series of Figures 5.21 through5.36. The oil-meniscus analysis discussed in Section 5.3 provides theexplanation. As illustrated by the Figure 5.71 through 5.77, new lubricant is onlyavailable to the meniscus region during the initial moments of upstroke. Evenallowing for faster oil velocities, the meniscus is without new lubricant for themajority of the upstroke period. The lubrication, therefore, which occurs underthese circumstances is due to the presence of remnant oil or a residue layer andnot due to the continual replenishment of oil during every cycle. It is alsopossible that some oils completely break down while other oils only partiallydecompose. This increase in load during upstroke may reflect the diminishingavailability of lubricant, whether the mechanism of lubrication is hydrodynamic orboundary layer, to lubricate the meniscus region. As upstroke proceeds, theremnants of oil are being broken down due to the severe conditions at themeniscus. Consequently, lubricity is reduced allowing loads to graduallyincrease. Reduction of peaks occurs when oil is once again able to lubricate themeniscus or when the mould moves downward onto the strand.The maximum loads reached by the three oils are shown in Figure 5.37and are shown in Tables 5.10, 5.11, and 5.12. Clearly evident from Figure 5.37,is the strong dependency of maximum load for Lubricant A on flow, the weakerdependency for Lubricant B, and the lack of a flow dependency for Lubricant C.Lubricant A increased by 800 N (180 lbf) between 54 and 24 mI/mm while theincrease for Lubricant B was only 500 N (112 lbf). Lubricant B producedmaximum loads ranging between 9700 N (2183 lbf) and 10200 N (2292 lbf) whileunder Lubricant A, maximum loads ranged between 9300 N (2090 lbf) and 10100 N (2270 lbf). These differences may reflect the fatty acid content as241Lubricant A contains 30 times more than Lubricant B. As discussed in Chapter 2,not only do the fatty acids provide long, polarized hydrocarbon molecules whichaid in reducing friction between two sliding surfaces but in the environment at themeniscus, oxidation of the thin layer of oil is likely, forming a layer of lacquerwhich also contributes to the reduction of friction at the interface between themould and the strand. Oxidation of the oil would not occur as readily if thetechnique of shrouding is employed; boundary layer lubrication may still be theoperating mode of lubrication due to the presence of the hydrocarbons.However, under shrouding conditions, decomposition of the constituents of the oilmay occur providing a rich atmosphere of hydrogen and carbon which would beconducive to the mechanism of hydrodynamic lubrication. Possibly all threeconditions are operating giving rise to a mixed method of lubrication.Further support for the boundary lubrication mechanism is indicated by theobservations of the loading distributions listed in Tables 5.13, 5.14, and 5.15.Assuming that a change in load during upstroke of less than 223 N (50 lbf) is dueto mould weight (Section 6.3.1), only the 34 mI/mm of Lubricant A had asignificant number of cycles within this range. The other lubricants had anegligible number of cycles within this loading range. Lubricant B was thepoorest as indicated by the larger percentage of the cycles above the range of890 N (200 Ibf).Lubricant A maintained lubricity at all flows as the majority of load cellcycles had changes in load of less than 558 N (150 lbf) during the upstroke(Table 5.13). Lubricant C also shows similar behaviour in load changes asindicated in Table 5.15. The lubricants A and C were employed during thecasting of a 0.09% and 0.15% carbon steels, respectively. The 0.09% carbon242steel initially solidifies in the delta solid phase and does not complete itstransformation, and therefore contraction, to the gamma phase until about 1410 0C. The 0.15% carbon steel begins its transformation to the gamma phase at thehigher temperature of 1490 0 C. This difference in solidification behaviour meansthat the higher carbon steel completes its contraction earlier, higher up in themould creating a larger air gap between the mould and the strand enhancingreduced contact. From this discussion, if Lubricant C has equal lubricationcapabilities as Lubricant A, it would be expected that Lubricant C employed for a0.15% steel should demonstrate less interaction between the mould and thestrand and, therefore, produce lower load changes during the upstroke than thevalues attained during the casting of a 0.09% carbon steel. But, this is not thecase. From the Tables 5.13 and 5.15, it is seen that this expectation is notrealized and, therefore, it is concluded that Lubricant A is the better lubricant.The similarity of maximum loads for at least two flows shown in Tables5.10 through 5.12 indicates that decreasing the flow by 10 mI/mm does notnecessarily reduce the lubricating capability of the oil and, therefore, thedetermination of the optimum lubricating conditions in billet casting is possible.From the statistical evaluations, it is apparent that changes in load duringupstroke are similar whether Lubricant A is fed at 44 or 54 mI/mm, whetherLubricant B is fed at 34 or 44 mI/mm, or whether Lubricant C is fed at 24, 34, or44 mI/mm. This statistical test suggests that flow can be reduced withoutsignificantly increasing the interaction between the mould and the strand.An attempt was made to correlate the variation in load cell response forthese oils to their physical and chemical properties. Lubricant A provided thelowest and most consistent changes in loads during the upstroke period of the243oscillation cycle as shown in Tables 5.13 through 5.15. The comparison of themagnitude of the mould-strand interaction suggests that Lubricant A at 34 mI/mmlimits this interaction to the greatest degree when considering the three oils at allflows. Examination of its properties indicated that of these three oils, Lubricant Ahad the highest flash point, the lowest viscosity at the cooler temperatures, andthe highest fatty acid content. These three factors jointly contribute to thedevelopment of a better mould lubricant.Inherently, comparison of the effectiveness of these mould lubricants isdifficult due to limited data collected within this project. Each lubricant wasemployed during the casting of different carbon grades which have differentsolidification characteristics and, therefore, different behaviour within the mould.6.4 Depressions6.4.1 The Effect of Carbon Content on Load Cell ResponseThree carbon grades were cast under similar mould lubricating conditionsenabling a comparison of load cell results. Differences in load cell responsewere clearly evdent in the visual assessment. Figures 5.33 through 5.36 are theresponses for three grades cast under similar lubricating conditions. The changein response between the 1008 and the higher carbon grades is dramatic. Thelowest carbon grade 1008 exhibits the greatest degree of mould-strandinteraction during upstroke while the upper peaks of the two higher carbongrades, 1018 and 1039, are remarkably constant. Qualitative and quantitativeresults in Tables 5.16, 5.17, and 5.18 also substantiate these visual observations.Maximum load which occurs during upstroke is greatest for the 1008 grade asshown in Table 5.19 whereas the two higher grades show lower maximum loads.244Significantly more important is the degree of fluctuation indicated by the averagepercent deviations. The 1008 steel grades exhibit the largest average deviationof 2.33% which infers that maximum load fluctuates to the greatest degree forthis steel grade. The maximum load fluctuations are reduced to 1.03% for the1018 steel grade and a further reduction to 0.68% for the 1039 grade.6.4.2 Depressions and the Load Cell SignalBinding of the 1008 and 1010 steel grades was evident from themetallographic examination performed on the billet sections. The billetevaluation, summarized in Table 5.22, revealed that 88% of the grade 1008 and33% of the grade 1010 billet samples from the Ivaco plant trial exhibitedtransverse depressions. Surface profilometer measurements for the 1008 and1010 grade billets, shown in Figures 5.56 and 5.57, also reflect the presence ofthese depressions. Profilometer measurements for, and visual evaluation of, thehigher carbon grades did not show the presence of depressions. However, theaverage surface roughness values along the centreline, as given in Table 5.24,do indicate that the surfaces of the 1010 and the 1012 grade billets are rougherthan those surfaces of the three higher grades (1015, 1018, 1039) examined.Samarasekera and Bnmacombe [21] have proposed a mechanism for theformation of transverse depressions which was discussed in Section 2.4.1 ofChapter 2. Essentially, the mechanism involves the deformation of the solidifyingshell due to a tensile force. When binding of the strand occurs in the mould forwhatever reason, the solidified shell experiences an axial tensile stress due tothe mechanical pulling of the strand by the withdrawal rolls. The magnitude ofthis tensile stress is a function of the severity of the binding. As steel has high245ductility between 1150 and 1430 C, the solidified shell near the surface mayplastically flow or neck. This locally necked region is manifested as a depressionon the surface of the billet. The region of low ductility at the solidification front willoften exhibit stress cracks running perpendicular to the applied force, as shownin Figure 5.52.It is surmised that as binding occurs and the depression forms, the frictionbetween mould and the strand increases and than as conditions in the mouldchange causing the disengaging of the strand from the mould, the force on themould would then decrease. The load cell responses for the 1008, the 1010, andthe 1012 steel grades are consistent with this loading force behaviour.This undulating behaviour of the upper peaks is demonstrated in Figures5.13, 5.21, 5.22,5.26,5.29,5.30,5.34, and 5.35 which are representative of theload cell signals for the 1008, 1010, and 1012 steel grades. Note that the upperpeak fluctuation is not related to casting speed in the same way as the lowerpeak fluctuation as indicated by the regression analysis and the plots ofmaximum load and casting speed shown in Figures 5.12 and 5.13.Examination of the upper peak loading undulations indicates that thoughthese cycles are repeated frequently the amplitude varies from slight to severe.The grades which exhibit this behaviour are the 1008, 1010, and to some extentthe 1012 steel grades. The load cell response for the 1018 and 1039 steelgrades did not exhibit any of these fluctuations as shown in Figures 5.35, 5.36,and 5.37. These qualitative results are supported by the statistical analysis of thedegree of fluctuation in the maximum load. Table 5.18 indicates that themaximum load fluctuation for the 1008 grade is twice that for the 1018 grade and246almost three times that of the 1039 steel grade.The increase in loading on the mould to a maximum and then returning tonormal for these short periods of time suggests that some sort of mechanicalinteraction between the mould and the strand is taking place. The duration ofthese loading cycles varied from a few (3 to 4) to several (16 to 20) mouldoscillations. This time period corresponds roughly to the distance between theobserved depressions on the surface of the retrieved 1008 and 1010 steel gradebillets. This correlation confirms that the load cell system was detecting bindingin the mould and in extreme cases, the formation of transverse depressions.247CHAPTER 7. SUMMARY AND CONCLUSIONS7.1 SummaryThis project which investigated mould lubrication in a billet caster of acontinuous casting machine involved industrial measurements, analysis of someof the physical and chemical properties of mould lubricants, a metallographicassessment of billet sections, and an evaluation of the lubrication of themeniscus during an oscillation cycle. An operating billet mould was instrumentedwith four load cells positioned at the rear and the front of the mould to measurethe loading behaviour on the mould during casting; two linear variable differentialtransducers were placed on the top of the mould housing to monitor theoscillatory motion of the mould unit; thermocouples were installed in the mouldwall in order to record the copper temperatures. The data obtained from thethermocouples is the subject of a separate study.The three lubricant properties of flash point, viscosity, and fatty acidcontent were assessed and obtained from a local laboratory. The billet samplescollected during the main trial were metallographically examined to determine thepresence of cracks, pinholes, inclusions, and oscillation marks. Profilometermeasurements were completed which indicated the degree of surface roughnessand recorded the depth of the oscillation marks. Furthermore, a theoreticalcalculation for the oil flow down the mould wall was also completed to ascertainthe degree of lubrication during an oscillation cycle.248The load cell data, in conjunction with the LVDT signals, indicated thatthere were two parts to the loading of the mould during each oscillation cycle.During the downstroke, particularly at the beginning of the negative-strip period,the load on the mould decreased smoothly until just after the point at which themould reached its maximum downward velocity; then the load smoothlyincreased until the beginning of the upstroke period, It became apparent duringthe analysis of the data that casting speed highly influenced the minimum point;the higher the casting speed, the lower the minimum load attained. Owing to theresulting force direction during the negative-strip period, the greater the relativevelocity between the mould and the strand, the greater the decompression of theload cells as the mould squeezes down on the solidifying shell. During theupstroke period, the loading on the mould behaved in a different manner thanduring the downstroke period. Influenced by the mould lubricant, the lubricantflow, and the carbon content of the steel, the upper peaks exhibited differenttrends. Use of Lubricant B and reduced lubricant flows (34 and 24 mI/mm)created ascending peaks which substantially increased in load relative to theother two lubricants, A and C, and to the higher flows (54 and 44 mI/mm)employed. Carbon content influenced the behaviour of the upper maximumpoints. Lower carbon contents less than 0.1 % created upper peak fluctuationswhile the two steel grades of 1018 and 1039 produced constant maxima.The laboratory analysis indicated that Lubricant A had the highest flashpoint. Lubricant C had the highest viscosity at temperatures less than 50 0 C butall three Lubricants, A, B, and C, had similar viscosities at operatingtemperatures. Of the three lubricants, Lubricant A had 30 times the fatty acidcontent than did the other two mould lubricants tested. Lubricants A and C hadthe largest percentage of hydrocarbon molecules containing double bonds. Fatty249acids are responsible for boundary lubrication while double bonding of carbonatoms in a hydrocarbon molecule promotes the formation of a lacquer layer athigh temperatures.The metallographic examination indicated that surface defects such aspinholes, inclusions, and cracks were minor. Only two grades of steel, 1008 and1010, exhibited the presence of transverse depressions. The 1039 steel gradeexhibited smoother surfaces than, any other grade sampled during the plant trial.The analysis of the lubrication at the meniscus during an oscillation cyclerevealed that technically lubricant is present at the meniscus during thedownstroke period and does not feed the meniscus during the upstroke period.Shorter stroke lengths and slower oscillation frequencies provide betterlubrication at the meniscus during an oscillation cycle than do longer strokelengths and higher oscillation frequencies.7.2 ConclusionsThe following conclusions may be drawn from the investigation andanalysis of the data obtained from the plant trial at lvaco:1. The response of the load cells was periodic and reproducible. Upper peakresponses reflect changes in mould lubricant and in lubricant flow rate as well ascarbon differences.2502. The direction of the reaction force resulting from the interaction betweenthe mould and the billet is upward when the mould is moving downward morerapidly than the strand and the resultant direction is downward when thedownward motion of the strand is greater.3. The load cell response clearly indicated that the interaction between themould and the strand during an oscillation cycle was composed of two parts: thesmooth decrease in loading particularly during the negative-strip period and thejagged increase in loading during the upstroke period. This observation infersthat there are two types of interactions occurring between the mould and thestrand during each oscillation cycle.4. The smooth decrease in loading during the negative-strip period has beenshown to be a function of the casting speed which reflects the degree at whichthe mould attempts to overtake the strand. The faster the relative velocitybetween the mould and the strand, the greater the interaction. This period of theload cell response reflects the mechanical interaction of the mould with the strandduring which any effect of lubrication is masked.5. The decompression of the load cells due to the mould bearing down onthe strand has been linked to the mechanism of oscillation mark formation inbillets.6. The upper peak response of the load cell signal is influenced by thelubrication conditions and by the carbon content of the steel. As indicated fromthe analysis of the presence of lubricant at the meniscus and from the observedascending upper peak behaviour of the load cell response at reduced flows, the251mode of lubrication may be a combination of both hydrodynamic and boundarylayer mechanisms. Hydrodynamic lubrication, due to either a finite liquid film or agaseous layer, may initially be operating with boundary layer lubricationbecoming operative as the lubricant oxidizes leaving only residues and/or alacquer layer to provide an interface between the mould and the strand.7. From the metallographic examination completed on the billet sections fromthis plant trial, reduction of oil flows from 54 mI/mm to 24 mI/mm does notdetrimentally affect the quality of the billet surface. The physical assessment ofthe billets infers that major dollar savings can be achieved by lubricant reductionif uniform flow around the periphery of the mould is ensured.,8. The fluctuation of the maximum load was shown to be steel gradedependent. Medium-carbon steels experienced less mould-strand interactionduring upstroke as shown by their smoother surfaces and as reflected by theconsistent upper peak response. Low-carbon steels cast under similar mouldconditions as the medium-carbon steels interacted with the mould to a greaterdegree. Binding of the 1010 and 1008 carbon grades was evident due to thepresence of depressions on the billet surfaces. The fluctuation in upper peakmaxima support these visual observations possibly reflecting the effect ofexcessive taper.9. The utilization of load cells in the mould of a billet caster can potentially beemployed on-line to detect casting difficulties as they occur.2527.3 Future WorkSince this project was an initial attempt to explore the interaction of themould and the strand in billet casting through the utilization of load cells, severalfactors were overlooked and, therefore, require further assessment. Firstly andmost imperative, is the necessity to test several mould lubricants during thecasting of steels having similar carbon contents. Better characterization of mouldlubricants would then be possible.Further reduction of oil rates may be possible with negligible deleteriouseffects on billet surface quality. Reduced oil flow will potentially eliminate pinhole formation due to excessive lubricant. Several steel grades can now be castwith reduced flows and on-line observations will quickly determine theoptimization of lubricant flow. Conversely, data should be accumulated underexcessive flows and correlated to surface quality.Further examination of lubricant properties is required. Through thisstudy, it became apparent that mould lubricants had similar viscosity-temperaturebehaviour, that flash temperatures were relevant only in comparison with oneanother and, may have a wide range of possible values; and finally that severallubricants having various amounts of free fatty acids present are necessary todetermine whether this component affects mould lubrication.Taper effectiveness could also be determined through employment of theload cell system.253REFERENCES[1] LA. Bakshi and co-workers: unpublished work, 1988.[2] P.E. Efremov, V.S. Rutes, P.G. Shmidt, and G.F.Konovalov: Stalin English, 1970, pp 692-694.[3] M. Komatsu, T. Kitagawa, and K. Kawakami: TransactionsISIJ, Vol. 23, No. 10, 1983, p F-i.[4] H. Nakato, S. Omiya, Y. Habu, T. Emi, K. Hamagami, andT. Koshikawa: Journal of Metals, Vol. 36, March, 1984.[5] H. Yamanaka, M. lkeda, T. Nishitani, and T. Ando:Transactions ISIJ, Vol 23, No. 10, 1983, p F-4.[6] M. Wolf: AIME Electric Furnace Conference, Vol. 40,1983, pp 335-346.[7] B. Mairy, D. Ramelot, M. Dutrieux, L. Deliege, M.Nourricier, J. Dellieu: Proceedings of the 5th ProcessTechnology Conference, Detroit, April 1985, pp 101-117.[8] B. Mairy and M. Wolf: Fachberichte, Vol. 20, No. r,April, 1982, pp 222-227.[9] B. Short, F. Faires, and T. Horton: Concast TechnologyNews, Vol. 26, No. 3, 1987, p5.[10] G. Foussal et al: Rev. Metall., Vol. 4, 1984, pp 299-307.[11] C. Schacht: Steelmaking Proceedings, Atlanta, Vol. 66,1983, pp 269-275.[12] C. Schacht: Steelmaking Proceedings, Pittsburgh, Vol.70, 1983, pp 443-448.[13] patent: Werke Klockner and Klockner-Humboldt-Deutz,Continuous casting breakout early warning process-involving continuous acoustic monitoring of mouldregion; 07 January 1984.[14] J. Callahan: Journal of the American Society ofLubrication Engineers, Pittsburg, May, 1966, pp 9-118.[15] W. Ritter: Iron and Steel Engineer, Vol. 44, No. 2,February, 1967, pp 113-118.[16] I. Saucedo, J. Beech, and G. Davies, Proc. 6th IntlVacuum Metallurgy Confer, 1979, pp 885-904.254[17] H. Tomono, W. Kurz, and W. Heinemann: MetallurgicalTransactions B, Vol. 1 2B, June, 1981, pp 408-411.[18] l.V. Samarasekera, J.K. Brimacombe, and R. Bommaraju:ISS Transactions, Vol. 5, 1984, pp 79-94.[19] R. Bommaraju, J.K. Brimacombe, and l.V. Samarasekera:ISS Transactions, Vol. 5, 1984, pp 95-1 05.[20] S.N. Singh and K.E. Blazek: Open Hearth Proceedings,AIME, 59, pp 264-283.[21] LV. Samarasekera and J.K. Brimacombe: W.O. PhilbrookMemorial Symposium Conference Proceedings, 1988, pp157-171.[22] l.V. Samarasekera and J.K. Brimacombe: InternationalMetals Reviews, Vol. 23, No. 6, 1978, pp 286-300.[23] l.V. Samarasekera and J.K. Brimacombe: MetallurgicalTransactions B, Vol. 13B, pp 105-116.[24] I.V. Samarasekera and J.K. Brimacombe: Ironmaking andSteelmaking, No. 1, 1982, pp 1-15.[25] I. Saucedo, J. Beech, G.J. Davies: Metals Technology,Vol. 9, July, 1982, pp 282-291.[26] S.N. Singh and K.E. Blazek: Journal of Metals, October,1974, pp 17-27.[27] A. Grill and J.K. Brimacombe: lronmaking andSteelmaking, 1976, Vol. 2, pp 76-79.[28] E. Takeuchi and J.K. Brimacombe: MetallurgicalTransactions B., Vol. 15B, No. 3, September 1984, pp493-509.[29] E. Takeuchi and J.K. Brimacombe: MetallurgicalTransactions B., Vol. 16B, No. 3, September 1985, pp605-625.[30] l.G. Saucedo: Steelmaking Conference Proceedings, Vol.7, 1987, pp 449.[31] J. W. Donaldson: Journal of Metals, December 1965, pp1-6.[32] D.l. Brown: Journal of Metals, April 1965, pp 2-8.255[33] F.P. Bowden, D. Gregory, and D. Tabor: Nature, No.3952, July, 1945.[34] E.S. Nachtman and S. Kalpakjian: LUBRICANTS ANDLUBRICATION IN METALWORKING OPERATIONS, Marcel Dekker,inc., 1985.[35] PRINCIPLES OF TRIBOLOGY, edited by J. Hailing,MacMillan Education, 1987.[36] D. Tabor and E.D. Tingle: Research Surface Chemistry,1949, pp 217-220.[37] G.E. Mortimor: CHEMISTRY: A CONCEPTUAL APPROACH, D. VanNostrand Company, 1979.[38] L.J. Desha: ORGANIC CHEMISTRY, McGraw Hill, 1952.[39] E. Eckey: VEGETABLES, FATS, AND OILS, Reinholt, 1954.[40] J.Packer, and J. Vaughan: ORGANIC CHEMISTRY, ClarendonPress, 1958.[41] C.R. Noller: TEXTBOOK OF ORGANIC CHEMISTRY,Philadelphia, W.B. Saunders Co., 1966.[42] B. Levitt: OILS, DETERGENTS, MAINTENANCE SPECIALITIES,Vol. 1, Chemical Pub. Co., 1967.[43] FUELS AND CHEMICALS FROM OILSEEDS, edited by E.B.Schu Its and R.P. Morgan, Westview Press for theAmerican Association for the Advancement of Science,1984.[44] S.G. Daniels: Faraday Society Transactions, Vol. 47,1951.[45] R.S. Marrell and H.R. Wood: CHEMISTRY OF DRYING OILS,Van Nostrand, 1925.[46] F.J. More: ORGANIC CHEMISTRY, John Wiley & Sons, 1933.[47] E.B. Greenhifl: Faraday Society Transactions, Vol. 45,1949.[48] A. Miller and A. Anderson: Journal of the AmericanSociety of Lubrication Engineers, April, 1957.[49] R.A. Knights, D.E. Humphreys, and A. Perkins:Ironmaking and Steelmaking, Vol. 13, No. 1, 1986, pp32-39.[50] F.P. Bowden and A.C. Moore: Research 2, 1949, pp 585-586.256[51] N. Walker and N. Hemingway: unpublished work, 1987-1988.[52] A. Peril and I. Bakshi: unpublished work, 1988.[53] l.V. Samarasekera: unpublished work, 1988.[54] J.E. Shigley and Larry 0. Mitchell: MECHANICALENGINEERING DESIGN, 4th Edition, McGraw-Hill BookCompany, 1983.[55] 1969 Book of ASTM Standards, American Society ForTesting and Materials, Philadelphia, Vols 16 and 17 1969.[56] THE MAKING, SHAPING, AND TREATING OF STEEL, Chapter 21:Continuous Casting of Semi-Finished Steel Products.[57] R.J. Fruehan: I&SM, March, 1983, pp 33-37.[58] J.L. Devores: PROBABILITY STATISTICS FOR ENGINEERINGAND THE SCIENCES, Brooks/Cole Publishing Company,California, 1982.[59] W.H. Press, B.P. Flannery, S.A. Teukoesky, W.T.Vetterlinq: NUMERICAL RECIPES: THE ART OF SCIENTIFICCOMPUtING, Cambridge University Press, 1986.[60] I.V. Samarasekera: unpublished work, 1987.[61] R.J. Dippenaar, I.V. Samarasekera, and J.K. Brimacombe:ISS Transactions, Vol 7, 1986, pp 31-43.[62] F. Weinberg: Metallurgical Transactions B, Vol 1 OB,June 1979, pp 21 9-227.[63] A.D. Gosman and F.D.K. lderiah: “TEACH-2E, A GeneralComputer Programme for Two-dimensional TurbulentRecirculating Flow”, mt. Rep., Dept. of Mech. Eng.,Imperial College, Univ. of London, 1976.[64] Edmond Osinski: unpublished work, 1989.[65] V. Sahajwalla: unpublished work, 1989.[66] l.V. Samarasekera: private communication, 1989.257APPENDIX ALOAD PARTITIONING OF EXTERNAL FORCES258Force Distribution in the Mould AssemblyThe schematic illustration in Figure 4.5 depicts the mould assembly unit.The unit consists of a hold-down bolt tightened in tension, a spring completelyclosed and in compression, and a load cell also in compression. Due to theseparation of the two members by the load cell button, the members are notinvolved in the transference of any load experienced by the mould. Any externalload applied to the assembly is partitioned between the bolt-spring componentand the load cell; the load seen by each component is a function of its stiffnessconstant.Consideration must be given for two cases of loading, depending on thedirection of the external load [54,551. In one case, the external load is attemptingto lift the housing off the mould table. The bolt has been preloaded to an initialtensile load. The preload has placed the load cell and the oscillating table incompression which effectively increases the resistance to the upward externalload. With application of the external load, the bolt initially in tension extends byan amount d,db= Pb (1).kbThe load cell originally in compression decreases by the amountdic = (2). The increase in boltKLcdeformation is equal to the decrease in deformation of the load cell(3).kb KLcThe total external load is distributed between the bolt-spring and the load cellunits:P=Pb+PLC (4).The load partitioning is then:Solving (3) for PL and substituting in (4) obtains an expression for the loadin terms of the load on the bolt:P=Pb(1 + Ku) (5)KbSolving (5) for the force on the bolt gives (6):Pb= KhP (6)Kb + KIcThe load partitioned for the load cell is:259= KhPKLC Kb+KLCLC Kit’, P (7)b+KLCPLC S the change in load resulting from the newly applied external load.Stiffness Constants1. Load Cells: complete compression of the load cell button would be 0.004thousandth of an inch and the load required would be 10 000 lbf. Therefore, thestiffness constant would be 2.5 E06 lbf per in.2. Bolt: Kb=AE/lE 30 E06 psi [56]= the total thickness of the members fastened together= 3.25 inA = 0.7610 in2 (based on major diameter)Therefore, kb = 7.025 E06 lb/inThe amount of the external load seen by load cell is:PL = 2.5 E06 P = 0.2625 PT025 E06+2.5 E06OR P = 3.81 PL260APPENDIX BMINIMUM LOAD VERSUS CASTING SPEEDFOR MOULD LUBRICANTS A AND CIDEPENDENCE OF MINIMUM LOAD ON261CASTING SPEEDz900008500zLubricant A at 24 mI/mmCASTING SPEED (in/mm)100009500HEATQ.Q97B24653 RUN 44CARBON2250—— 2115—.- .— I—— .• —aazC——1980—• —— -I800075007000•_ —018453781 .8+50.5*X1 SD2 SD801710•90 100 1101575262DEPENDENCE OF MINIMUM LOAD ON CASTING SPEEDCASTNG SPEED (in/mm)Lubricant A at 34 mI/mmHEAT A23416 RUN 42O.O9’ CARBONz0-Jz20251000090008000700060005000——— .——1800— ——— ——zC015750—— ———1322.3+71.6*X—- 1 SD--2 SD7013509080 100 1101125DEPENDENCE OF MINIMUMI263LOAD ON CASTING SPEEDCASTNG SPEED (in/mm)Lubricant A at 34 mI/mmz0-Jz900085008000750070006500600020251890z17550>162001485135010070 80 901DEPENDENCE OF MINIMUM264LOAD ON CASTING SPEECLubricant A at 44 mI/mmCCASTING SPEED (in/mm)HEAT A26412 RUN 410.097. CARBON——189090008500z8000a7500700065006000— . .1—• —•——_i.I •1755zC0>1620—U— 2544.7÷57.0*X1 SD2 SD148575 80135085 90 95 100 105DEPENDENCE OF MINIMUM265LOAD ON CASTING SPEEDSPEED (n/mn)Lubricant A at 54 mI/mmHEAT A26412 RUN4O0.09 CARBON—z0z—.-—9000850080007500700065006000—— .—— .. ——••_.• ——• —— _•20251890z175501620 \.i014851350110— 4563.8÷33.81*X—- 1 SD-- 2 SD70 80 90 100266MINIMUM LOAD DEPENDENCE ON CASTING SPEEDCASTING SPEED (in/mm)Lubricant C at 24 mI/mnHEAT A26409 RUN 380.15 CARBONz0-Jz9000850080007500700065006000———•-44——•——— —• ——• ——zC00-202518901755162014851350100— •— —•— 2573.1÷53.6*X—- 1 SD-- 2 SD80 85 90 95267DEPENDENCE OF MINIMUM LOAD ON CASTING9000Lubricant C at 34 mI/mmSPEEDCASTNG SPEED (in/mn)HEAT0.15zA23409 RUN 37CARBON 1875— —zC-Jz__. —17258000700060005000———__..—.——zC1575>1425-cCTh1275—2229.4+58.1*X—- 1 SD--2 SD70 80 901125100DEPENDENCE OF MINIMUM268LOAD ON CASTING SPEEDCASTI NC SPEED (in/mm)Lubricant C at 44 mI/mmHEATO.15A23409 RUN 36CARBONz0-Jz100009000800070006000- ‘I————..__——••. —-1• • —I‘I.’-.——_••_•I22502100z195018001650 ‘—015001350•._—I_ •— 40027+44.4EX70—- 1 SD-- 2 SD80 90 100DEPENDENCE OF MINIMUM LOAD ON269CASTING SPEEDz1000009500zLubricant C at 54 mI/mmCASTING SPEED (in/mm)1100010500HEAT 824640 RUN90.0357. CARBON75a—‘2340a—a———.•• aazC2205aj.a .-a••a• —— a•a ———— aa90008500800002070aa_—‘ •I5316.O÷49.3*X1 SD7519352 SD80 85 95 100901800DEPENDENCE OF MINIMUM LOAD ONI270CASTING SPEED‘—‘ 9000z08000zz1800>CAST I NC; SPEED (in/mm)10000Lubricant C at 54 mI/mmHEAT A23408 RUN 350.051% CARBON 2100.—— —— .__-— —— ,I.__ ——— .-1I—— ——1950700060002553.6÷63.2*X1 SD1650 ‘—“C2 SD 150070 75 80 85135090 95 100271DEPENDENCE OF MINIMUM LOAD ON CASTING SPEEDCASTING SPEED (in/mm)Lubricant C at 54 mI/mmHEAT0.17B24643 RUN 21CARBONII.zC-Jz2160100009500900085008000——.——I— II___w_ .1—II——I2070—.—zC0>19800--I4997.2+43.8*X1 SD8018902 SD85 90 951800100CASTING SPEED (in/mm)272DEPENDENCE OF MiNIMUM LOAD ON CASTING SPEEDLubricant C at 54 mI/mm20251950HEAT B24644 RUN 250.18z CARBONI-—————— ..—.——zC-Jz•• —— ••aS90008500800075007000——•a.I•a• —j-gaa‘S•.1.az1875C1800>1725 ‘—‘01650157590-w• ——— •4600.4+42.7*X1 SD702 SD75 80 85DEPENDENCE OF MINIMUM LOAD ON273CASTING SPEEDCASTING SPEED (in/mm)Lubricant C at 54 mI/mmHEAT B24647 RUN 310.42 CARBONa-zaz——--— .—10000950090008500800075007000• —• —-—22502115z19800>1845017101575110—5800.8+31.1*X—- 1 SD--2 SD80 90 100274APPENDIX CANOVA: A STATISTICAL COMPARISONOF AVERAGES AND STANDARD DEVIATIONSFOR MULTI-SAMPLES DRAWN FROM DIFFERENTPOPULATIONS275ANOVA: A STATISTICAL COMPARISON OF AVERAGES ANDSTANDARD DEVIATIONS FOR MULTI-SAMPLES DRAWN FROMDIFFERENT POPULATIONS [59,60]One method available for determining whether data sampled from twoor more numerical distributions are different or consistent is the analysis of thevariance (ANOVA). The method determines whether or not there are differencesin true averages associated with different levels or treatments of a factor. In thiscase, the factor is oil flow and the levels are the rates of flow. The nullhypothesis assumes that there are no difference between any of the means; thatdifferent levels of the factor have no effect on the average responses. Disprovingthe null hypothesis proves that the data sets are from different distributions; thatdifferences do exist between the true average responses for the different levels.A common statistic measuring device in determining the significanceof a difference of means is the F-test. When the null hypothesis is true, thestatistic F has a probability distribution called an F distribution; traditional levels ofsignificance are 0.05 and 0.01. Therefore, after determining F, the nullhypothesis would be rejected in favour of the alternative hypothesis if F is greaterthen or equal to the F distribution of a given significance level.276Proposition: Let X,. and S i = I..... 1 denote the sample mean andvariance of the i th sample. Define the between-samples estimator ô byJ(X, —X..)2,=1 =Xand the within-sample estimator ô& by:s r’‘ I l[-J—lI I —-1 Z 2 (X,, —Xi(X _X.)I=j l(J—1)Then âj is an unhiaed estimator of cr when H0 is true. but E(oj) > o whenH0 is false, while â, i, unbiased for cr whether or not H0 is true.F =± = JZ(A —X..)/(l —1)21 (X, — X.)/I(J — 1)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items