UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Heat transfer, oil lubrication and mould tapers in steel billets casting machines Chandra, Sanjay 1992

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

Item Metadata

Download

Media
831-ubc_1992_fall_chandra_sanjay.pdf [ 11.32MB ]
Metadata
JSON: 831-1.0078515.json
JSON-LD: 831-1.0078515-ld.json
RDF/XML (Pretty): 831-1.0078515-rdf.xml
RDF/JSON: 831-1.0078515-rdf.json
Turtle: 831-1.0078515-turtle.txt
N-Triples: 831-1.0078515-rdf-ntriples.txt
Original Record: 831-1.0078515-source.json
Full Text
831-1.0078515-fulltext.txt
Citation
831-1.0078515.ris

Full Text

HEAT TRANSFER, OIL LUBRICATION AND MOULD TAPERS INSTEEL BILLETS CASTING MACHINESBySanjay ChandraB.Tech, Institute of Technology, Banaras Hindu University, India, 1983A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FORTHE DEGREE OF DOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Metals and Materials Engineering)._—W accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAMay 1992© Sanjay Chandra, 1992In 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.Department of Hi TLS Ai’jD 1EJPL &7CCThe University of British ColumbiaVancouver, CanadaDate_______________DE.6 (2/88)ABSTRACTThis study examines in detail the factors that influence mould-billet interaction and heattransfer during the continuous casting of steel billets. In an extensive three-year project, majorindustrial trials were held in three Canadian steel plants involving in each case an operating mouldinstrumented with arrays of thermocouples to record mould wall and mould cooling water temperatures. Additionally, load cells were installed between the mould housing and the oscillator tableto record mould-billet interaction. Linear variable displacement transducers were attached to themould wall in order to monitor mould displacement. Measurements were made under differentcasting conditions- steel grades, types and flow rates of lubricating oils and mould tapers- andwere recorded on a computer controlled data acquisition system. The liquid steel surface in themould was also filmed during casting.Two existing mathematical models of the mould were modified and used to calculate theaxial heat flux profiles and the dynamic distortion of the mould during service. A two-dimensional,finite-difference, heat-flow, mathematical model of the billet was developed to simulate solidification and shrinkage as a function of axial position in the mould. The coefficient of thermal contraction of steel was estimated as a function of steel carbon content and temperature fromexperimental data in the literature Ofl the lattice parameter of and ‘ unit cells; this was particularlyimportant to model the shrinkage of low-carbon steels. It has been shown that in theory, the lowcarbon steels (C <0.15%) should experience the largest contraction due to 6- yphase transformation;but in practice, they shrink less because heat transfer to the mould is low compared to higher carbongrades. A computer programme was developed to analyse the load cell response as a function of(ii)mould displacement. Finally billet samples collected during the trials were metallographicallyexamined to study the different aspects of the solidification in the mould e.g., cracks, oscillationmark depth and rhomboidity.The most important result of the research work has been the finding that the heat transfer inthe mould is significantly influenced by the taper of the mould wall in the meniscus region. A highinitial taper (2.5-3.0%/rn) in the meniscus region can compensate for the outward bulging of themould wall during operation preventing it from acquiring a negative taper. This absence of negativetaper has been shown to decrease mould-billet interaction during the negative strip period therebyleading to a decrease in the heat extracted in the meniscus region. This finding has been corroboratedby an analysis of the load cell signals. It has been shown unambiguously that, for high mould heattransfer, a shallow initial taper of the mould, that permits the wall to acquire a bulged shape, isrequired. High heat transfer in the mould is likely to result in adverse lubrication condition forcasting high-carbon steel billets.Filming of the steel surface has shown that only some of the lubricating oil flowing down themould wall reaches below the meniscus while the remainder collects on the liquid steel surface andburns. As a result an increase in the flow rate of the oil is not reflected in a commensurate increasein lubrication or heat transfer. In fact the industrial trials have clearly revealed that the existing flowrate of oil at all three plants could be reduced at least by half without any visible deleterious effecton billet quality.It has also been possible to link various sensor signals to the generation of defects in the billet,in particular to the formation of off-corner internal cracks, transverse depressions and billetrhomboidity. This together with the linkages between mould heat transfer and operating variables(iii)now makes it possible to conceive of a control system consisting of an instrumented mould and anexpert system that not only can asses billet quality on-line but can also initiate corrective action bychanging operating conditions that alter the heat transfer in the mould.(i \!)Table of ContentsAbstract.jjList of Tables xiList of Figures xiiiList of Symbols xxxiAcknowledgements xxxivCHAPTER 1: INTRODUCTION 1Chapter 2: LITERATURE REVIEW 102.1 Thermomechanical Behaviour of the Mould 102.2 Mould Lubrication with Oil 122.2.1 Mechanism of lubrication 132.2.2 Factors affecting mould friction 142.2.2.1 Oil flow rate and oil type 142.2.2.2 Steel grade 152.2.2.3Mouldtaper 162.3 Heat Transfer in the Mould 162.3.1 Effect of oil on heat flux 172.3.2 Effect of mould powders on heat flux 172.3.3 Effect of steel grade on heat flux 182.3.4 Mould taper 182.4 Mould-Friction Measuring Devices 192.4.1 Accelerometers 192.4.2 Strain gauges 202.4.3 Load cells 20Iv12.5 Load Cell Response and its Analysis 222.6 Mould-Strand Interaction and Billet Quality 232.6.1 Oscillation Marks 232.6.2 Transverse Depression and Transverse Cracks 242.6.3 Billet Rhomboidity and Internal Cracks 242.6.4 Off-Corner Internal Cracks 272.6.5 Pinholes 28Chapter 3: SCOPE AND OBJECTIVES OF THE PRESENT WORK 47Chapter 4: EXPERIMENTAL 494.1 Pre-Trial Preparations 494.1.1 Retrofit of mould housing 494.2 Measurement of Mould Wall Temperature 504.3 Measurement of Mould-Billet Friction Forces 514.4 Measurement of Mould Displacement 524.5 Other Miscellaneous Measurements 534.5.1 Casting Speed 534.5.2 Metal Level 534.5.3 Internal Dimensions of the Mould 544.5.4 Filming of the steel surface 544.6 Data Acquisition 544.6.1 EXP-16 544.6.2 DAS-8 554.6.3 General 55[vi]4.7 Details of the Trials 564.7.1 Casting machines and casting practice 564.7.2 Oils used in the trials 574.7.3 Thermocouple arrangements 574.7.4 Chemical compositions and casting conditions of different heats 574.8 Laboratory Work 584.9 Analysis of Mould Temperature Measurement 594.9.1 Conversion of thermocouple measurement to mould wall temperatures 594.9.2 Data Filtration technique 604.10 Analysis of Load Cell Response 61Chapter 5: RESULTS OF PLANT TRIALS 925.1 Mould Temperature Data 925.1.1 Effect of carbon content 925.1.2 Effect of oil type 935.1.3 Effect of oil flow rate 935.1.4 Effect of oscillation frequency 945.1.5 Mould temperature at the three Plants 945.2 Mould-Billet Friction Forces 945.2.1 Effect of oil type and flow rate on load cell response 955.2.2 Effect of carbon content 955.2.3 Effect of mould oscillation frequency 965.2.4 Difference in the load cell signals from the three Plants 965.3 Billet Quality Evaluation 975.3.1 Transverse depressions 97[vii]5.3.2 Off-corner internal cracks.985.3.3 Midway cracks 1005.3.4 Surface roughness 1015.3.4.1 Effect of carbon 1015.3.4.2 Effect of oil type 1015.3.4.3 Effect of oscillation frequency 1035.3.5 Rhomboidity 1045.3.6 Other defects 1045.3.6.1 Craze cracks 1045.3.6.2 Zipper marks 105Chapter 6: MATHEMATICAL MODELLING OF MOULD BEHAVIOUR 1776.1 Mathematical Model of Heat-Flow in the Mould 1776.1.1 Characterisation of heat transfer in the water channel 1786.2 Mathematical Model of Mould Distortion 179Chapter 7: MATHEMATICAL MODELLING OF BILLET SOLIDIFICATIONAND SHRINKAGE 1847.1 Mathematical Model of Billet Contraction 1847.2 Shrinkage Calculation 1857.3 Transverse Variation of Heat Flux 1877.4 Calculation of (Carbon- and Temperature-Dependent) Coefficient of ThermalLinear Expansion of Steel 1887.4.1 Variation of lattice parameter of delta phase with carbon content andtemperature 1897.4.2 Variation of lattice parameter of pure gamma iron with temperature 190[viii]7.4.3 Variation of lattice parameter of gamma phase with carbon content 1907.5 Model Verification 192Chapter 8: MODEL PREDICTIONS 1998.1 Heat-Flux Profile and Mould Wall Temperatures 1998.1.1 Typical heat-flux profile 1998.1.2 Effect of carbon content of steel 2008.1.3 Effectof oil flow rate 2008.1.4 Effect of oil type 2028.1.5 Effect of mould-oscillation frequency 2038.1.6 Difference in the mould heat extraction rate at the three plants 2048.2 Variables Affecting Distorted Mould Shape 2058.2.1 Steel composition 2068.2.2 Pre-existing mould taper 2068.3 Billet Solidification and Quality 2078.3.1 Billet mid-face temperature 2078.3.2 Billet shell thickness 2078.3.3 Billet shrinkage profile 208Chapter 9: MECHANISM OF MOULD HEAT TRANSFER 2349.lHeatTransferinZonel 2359.1.1 The role of mould shape in the heat transfer in Zone I 2359.2 Heat Transfer in Zone II 2419.3 Heat Transfer in Zone III 2449.4 Interaction of Zones 244[ix]Chapter 10: DESIGN OF MOULD TAPERS 26410.1 Conventional Design of Mould Taper 26410.2 New Approach to Design of Mould Taper 26510.3 Other Design Parameters 26810.3.1 Shaping of mould walls 26810.3.2 Mould water pressure 26810.3.3 Material of the mould 269Chapter 11: SENSOR SIGNALS AND BILLET DEFECTS 27011.1 Thermocouples 27011.1.1 Off-corner Cracks 27011.1.2 Rhomboidity 27111.1.3 Bleeds and Laps 27211.2 Load Cells 27311.2.1 Transverse Depressions and Cracks 27311.2.2 Adverse lubrication Conditions 27411.2.3 Mould-Billet Interaction at the meniscus 27411.3 Towards a “Smarttt Mould 274Chapter 12: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS FORFUTURE WORK 281REFERENCES 286[xjLIST OF TABLESTable 2.1 Coefficient of friction obtained with different lubricants at room 29temperature [241Table 2.2 Effect of taper of the narrow face of a slab mould on the average 29heat flux in the mould 1361Table 4.1 Details of the casting practice at Plants B, B and C 64Table 4.2 Details of the test and control strands used in the trial at Plants B, E 65and CTable 4.3 Property of various lubricating oils used in the trials at the three 66PlantsTable 4.4 Depth and axial position of thermocouples used to monitor mould 67wall temperature at Company BTable 4.5 Depth and axial position of thermocouples used to monitor mould 68wall temperature at Company ETable 4.6 Depth and axial position of thermocouples used to monitor mould 69wall temperature at Company CTable 4.7 Chemical compositions ol the heats monitored at Company B 70Table 4.8 Chemical compositions of the heats monitored at Company E 71Table 4.9 Chemical compositions of the heats monitored at Company C 72Table 4.10 Important casting conditions for heats monitored at Company B 73Table 4.11 Important casting conditions for heats monitored at Company E 75Table 4. 12 Important casting conditions tor heats monitored at Company C 77Table 4.13 Summary of the different types of oils and the flow rates at which 79they were used at the three Plants(xi)Table 5.1 Oscillation characteristics of the oscillator at Plant C for two differ- 107ent oscillation frequenciesTable 5.2 Difference in the maximum load attained during the upstroke and 107the load at the beginning of tN while casting 1045 grade steel atCompany CTable 5.3 Difference in the maximum load attained during the upstroke and 107the load at the beginning of tN while casting different steel grades atCompany CTable 5.4 Decompression of the load cell during the negative strip time for 107which the decompression lasts for two oscillation frequencies atCompany CTable 5.5 Percentage of the negative strip time for which the load cell is 107decompressed (during negative strip period) at Company B, E andCTable 7.1 Enthalpy and specific heat functions used in the heat-flow model 193[73]Table 7.2 Thermal conductivity functions for low carbon steel used in the 193heat-flow model [73]Table 7.3 Results of regression analysis on the experimental data of Ridley 194and Stuart [77]Table 7.4 Mid-face shell thickness at the bottom of the mould at Company B 194for billets of different grades(xii)LIST OF FIGURESFigure 1.1 Schematic diagram of a casting set-up for the continuous casting of 5steel billetsFigure 1.2 Schematic diagram of heat removal in a continuous casting 6machineFigure 1.3 Sinusoidal oscillation cycle of the mould 7Figure 1.4 The effect of cooling rate on billet microstructure 8Figure 1.5 Relationship between negative strip time (tN) and oscillation mark 9depth [12]Figure 2.1 Schematic diagram of a typical mould used for continuous casting 30of steel billetsFigure 2.2 Schematic diagram of fatty acid molecules adhering to the solid 31surface [24]Figure 2.3 Load cell response at different flow rate of mould lubricant B during 32casting of low-carbon billets [13]Figure 2.4 Effect of carbon content of steel on mould friction during continu- 33ous casting of steel [261Figure 2.5 Specific heat extraction as a function of distance from top of mould 34for different lubricants [30]Figure 2.6 Effect of mould flux on the heat-flux profile for low-carbon steel 35[32]Figure 2.7 Effect of mould flux on the heat-flux profile for high-carbon steel 36[32]Figure 2.8 A typical load cell profile [13] 37Figure 2.9 Change in the minimum load with casting speed [13] 38Figure 2.10 Load cell signal for a billet binding in the mould [13] 39(xiii)Figure 2.11 Mechanism for the formation of an oscillation mark due to interac- 40tion of mould with billet during negative strip time [13]Figure 2.12 Mechanism for the formation of transverse depressions and trans- 41verse cracks [28]Figure 2.13 Longitudinal corner cracks in continuously cast steel billets [181 42Figure 2.14 Off-squareness in continuously cast steel billets [18] 43Figure 2.15 Schematic diagram showing the formation of sub-surface crack on 44the diagonal at the obtuse-angle corners of a off-square billet [18].Figure 2.16 Schematic diagram showing the generation of an internal crack due 45to bulging of the billet shell in the mould and a hinging action in theoff-corner region [58]Figure 4.1 Schematic diagram of the set-up for measurement of mould wall and 80mould cooling water temperatureFigure 4.2 Schematic illustration of the load cell 81Figure 4.3 Schematic illustration of the positioning of the load cell between the 82mould housing and mould oscillating table and the bolt-springarrangementFigure 4.4 Schematic diagram showing the placement of LVDTs and the load 83cellsFigure 4.5 Schematic diagram of the construction of a LVDT 84Figure 4.6 Photograph of MetraByte’s Universal Expansion Interface; expan- 85sion multiplexer/ampilifier systemFigure 4.7 Photograph of the 8 channel high speed A/D converter and Timer 85counter interface from MetraByteFigure 4.8 Schematic diagram showing the main parts of the data acquisition 86set-upFigure 4.9 Flow chart showing the procedure for inspection of billets 87(xiv)Figure 4.10 Circuit diagram of the thermocouple connection 88Figure 4.11 Flow chart showing the models used, and the steps involved, in the 89analysis of mould thermocouple dataFigure 4.12 A typical unfiltered mould-thermocouple response at Company B 90Figure 4.13 Effect of metal level fluctuation on the temperature recorded by the 90meniscus thermocoupleFigure 4.14 Various components of a mould oscillation cycle 91Figure 5.1 Unfiltered response of selected thermocouples at Company C while 108casting a 1045 grade billetFigure 5.2 Typical standard deviation values for temperature data collected by 109mould-thermocouples at Company CFigure 5.3 Time-averaged axial temperature profile for different carbon con- 110tents at Company B. (Note that the temperatures are those recordedby thermocouples located approximately midway between the coldand hot faces of the mould)Figure 5.4 Time-averaged axial temperature profile for a 1018 grade of steel 111billets cast at 25 ml/min of Canola, Mineral_S, HEAR and Mineral_O lubricating oils at Company C. (Note that the temperatures arethose recorded by thermocouples located approximately midwaybetween the cold and hot faces of the mould)Figure 5.5 Time-averaged axial temperature profile for a 1018 grade of steel 111billets cast at 100 mI/mm of Canola, Mineral_S, HEAR and Mineral_O lubricating oils at Company C. (Note that the temperatures arethose recorded by thermocouples located approximate iy midwaybetween the cold and hot faces of the mould)Figure 5.6 Time-averaged axial temperature profile for a 1045 grade of steel 112billets cast at 25 ml/rnin of Canola, Mineral_S, HEAR and Mineral_O lubricating oils at Company C. (Note that the temperatures arethose recorded by thermocouples located approximately midwaybetween the cold and hot faces of the mould)(xv)Figure 5.7 Time-averaged axial temperature profile for a 1045 grade of steel 112billets cast at 100 mi/mm of Canola, MineralS, HEAR and Mineral_O lubricating oils at Company C. (Note that the temperatures arethose recorded by thermocouples located approximately midwaybetween the cold and hot faces of the mould)Figure 5.8 Time-averaged axial temperature profile for a 1018 grade steel 113billets cast at 25, 70 and 100 mlfmin of Mineral_S oil at CompanyC. (Note that the temperatures are those recorded by thermocoupleslocated approximately midway between the cold and hot faces of themould)Figure 5.9 Time-averaged axial temperature profile for a 1018 grade steel 113billets cast at 25,70 and 100 mi/mm of HEAR oil at Company C.(Note that the temperatures are those recorded by thermocoupleslocated approximately midway between the cold and hot faces of themould)Figure 5.10 Time-averaged axial temperature profile for a 1018 grade steel 114billets cast at 25, 70 and 100 ml/min of Mineral_O oil at CompanyC. (Note that the temperatures are those recorded by thermocoupleslocated approximately midway between the cold and hot faces of themould)Figure 5.11 Time-averaged axial temperature profile for a 1018 grade steel 114billets cast at 0, 25,70 and 100 mI/mm of Canola oil at Company C.(Note that the temperatures are those recorded by thermocoupleslocated approximately midway between the cold and hot faces of themould)Figure 5.12 Response of selected thermocouples at Company C with change in 115oil flow rate from 0 mI/mm (no oil) to 100 mI/mm of Canola oil for1018 grade steel billet at Company C. (Note that the temperaturesare those recorded by thermocouples located approximately midwaybetween the cold and hot faces of the mould)Figure 5.13 Time-averaged axial temperature profile for a 1045 grade of steel 115cast at 144 and 96 cpm of mould oscillation at Company C. (Notethat the temperatures are those recorded by thermocouples locatedapproximately midway between the cold and hot faces of themould)(xvi)Figure 5.14 Time-averaged axial temperature profile obtained at Company B, E 116and C while casting a 1018 grade steel billet. (Note that the temperatures are those recorded by thermocouples located approximatelymidway between the cold and hot faces of the mould)Figure 5.15 Time-averaged axial temperature profile obtained at Company B, E 116and C while casting a 1045 grade steel billet. (Note that the temperatures are those recorded by thermocouples located approximatelymidway between the cold and hot faces of the mould)Figure 5.16 Load cell response for a 1045 grade billet cast with Canola oil at 25 117mi/mm at Company CFigure 5.17 Load cell response for a 1045 grade billet cast with Canola oil at 117100 milmin at Company CFigure 5.18 Load cell response for a 1045 grade billet cast with Mineral_S oil at 11825 mi/mm at Company CFigure 5.19 Load cell response for a 1045 grade billet cast with Mineral_S oil at 118100 mI/mm at Company CFigure 5.20 Load cell response for a 1045 grade billet cast with HEAR oil at 25 119mi/mm at Company CFigure 5.21 Load cell response for a 1045 grade billet cast with HEAR oil at 100 119mi/mm at Company CFigure 5.22 Load cell response for a 1045 grade billet cast with Mineral_O oil at 12025 mI/mm at Company CFigure 5.23 Load cell response for a 1045 grade billet cast with Mineral_O oil at 120100 mI/mm at Company CFigure 5.24 Load cell response for a 1018 grade billet cast with Canola oil at 25 121mi/mm at Company CFigure 5.25 Load cell response for a 1045 grade billet cast with Canola oil at 25 121mi/mm at Company CFigure 5.26 Load cell response for a 5160 grade billet cast with Canola oil at 25 121mi/mm at Company C(xvii)Figure 5.27 Load cell response for a 1045 grade billet cast with HEAR oil at 25 122mi/mm and at 144 cpm of mould oscillation at Company CFigure 5.28 Load cell response for a 1045 grade billet cast with HEAR oil at 25 122mi/mm and at 96 cpm of mould oscillation at Company CFigure 5.29 Load cell response for a 1045 grade billet cast with HEAR oil at 100 123mi/mm and at 144 cpm of mould oscillation at Company CFigure 5.30 Load cell response for a 1045 grade billet cast with HEAR oil at 100 123mi/mm and at 96 cpm of mould oscillation at Company CFigure 5.31 A typical load cell response for Company B. (Note: Cross marks 124indicate start and end of negative strip time; square marker corresponds to end of load cell decompression)Figure 5.32 A typical load cell response for Company E. (Note : Cross marks 124indicate start and end of negative strip time; square marker corresponds to end of load cell decompression)Figure 5.33 A typical load cell response for Company C. (Note: Cross marks 124indicate start and end of negative strip time; square marker corresponds to end of load cell decompression)Figure 5.34 Load cell response at Company B (enlarged). (Note : Cross marks 125indicate start and end of negative strip time; square marker corresponds to end of load cell decompression)Figure 5.35 Load cell response at Company E (enlarged). (Note: Cross marks 125indicate start and end of negative strip time; square marker corresponds to end of load cell decompression)Figure 5.36 Load cell response at Company C (enlarged). (Note: Cross marks 125indicate start and end of negative strip time; square marker corresponds to end of load cell decompression)Figure 5.37 Macro-etch of a longitudinal section of a 1008 grade billet from 126Company B showing transverse depressions and cracks at the baseof these depressionsFigure 5.38 Macro-etch of a longitudinal section of a 1039 grade billet from 127Company B showing a ‘smooth’ surface (Mag. 0.8 X)Figure 5.39 Schematic diagram showing the eight off-corner sites of a billet 128(xviii)Figure 5.40 Macro-etch of a transverse section of a 1090 grade billet from 129Company E showing typical off-corner, internal cracks (Mag. 0.8X)Figure 5.41 Bar graph showing average depth of off-corner, internal cracks at 130the eight off-corner sites on the test and control strand for a 1018grade billet from Company E. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.42 Bar graph showing average depth of off-corner, internal cracks at 130the eight off-corner sites on the test and control strand for a 1050grade billet from Company E. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.43 Bar graph showing average depth of off-corner, internal cracks at 130the eight off-corner sites on the test and control strand for a 1080grade billet from Company B. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.44 Bar graph showing average depth of off-corner, internal cracks at 130the eight off-corner sites on the test and control strand for a 1090grade billet from Company E. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.45 Bar graph showing average depth of off-corner, internal cracks at 131the eight off-corner sites on the test and control strand for a 1146grade billet from Company E. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.46 Bar graph showing longitudinal off-corner depressions on the test 131and control strand for 1018, 1050, 1080, 1090 and 1146 grades ofbillets from Company EFigure 5.47 Macro-etch of a transverse section of a 1045 grade billet from 132Company C showing typical off-corner, internal cracks (Mag. 1.0X)Figure 5.48 Bar graph showing average depth of off-corner, internal cracks at 133the eight off-corner sites on the test and control strand for a 1018grade billet from Company C. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)(xix)Figure 5.49 Bar graph showing average depth of off-corner, internal cracks at 133the eight off-corner sites on the test and control strand for a 1045grade billet from Company C. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.50 Bar graph showing average depth of off-corner, internal cracks at 133the eight off-corner sites on the test and control strand for a 5160grade billet from Company C. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.51 Bar graph showing average depth of off-corner, internal cracks at 133the eight off-corner sites on the test and control strand for a 1084grade bifiet from Company C. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.52 Bar graph showing average depth of off-corner, internal cracks at 134the eight off-corner sites on the test and control strand for a L-325grade billet from Company C. (Note: Numbers on top of barsrepresent percentage of the billets with cracks)Figure 5.53 Macro-etch of a transverse section of a 1018 grade billet from 135Company E showing typical mid-way cracks (Mag. 0.8 X)Figure 5.54 Bar graph showing average depth of mid-way cracks at four loca- 136tions on the test and control strand for a 1018 grade billet fromCompany E. (Note: Numbers on top of the bars represent percentageof the billets with cracks)Figure 5.55 Bar graph showing average depth of mid-way cracks at four loca- 136tions on the test and control strand for a 1050 grade billet fromCompany E. (Note: Numbers on top of the bars represent percentageof the billets with cracks)Figure 5.56 Bar graph showing average depth of mid-way cracks at four loca- 136tions on the test and control strand for a 1080 grade billet fromCompany B. (Note: Numbers on top of the bars represent percentageof the billets with cracks)Figure 5.57 Bar graph showing average depth of mid-way cracks at four loca- 136tions on the test and control strand for a 1090 grade billet fromCompany E. (Note: Numbers on top of the bars represent percentageof the billets with cracks)(xx)Figure 5.58 Bar graph showing average depth of mid-way cracks at four loca- 137tions on the test and control strand for a 1146 grade billet fromCompany E. (Note: Numbers on top of the bars represent percentageof the billets with cracks)Figure 5.59 Macro-etch of a transverse section of a 1018 grade billet from 138Company C showing typical mid-way cracks (Mag. 1.0 X)Figure 5.60 Bar graph showing average depth of mid-way cracks at four loca- 139tions on the test and control strand for a 1018 grade billet fromCompany C. (Note: Numbers on top of the bars representpercentage of the billets with cracks)Figure 5.61 Bar graph showing average depth of mid-way cracks at four loca- 139tions on the test and control strand for a 1045 grade billet fromCompany C. (Note: Numbers on top of the bars representpercentage of the billets with cracks)Figure 5.62 Bar graph showing average depth of mid-way cracks at four loca- 139dons on the test and control strand for a 5 160 grade billet fromCompany C. (Note: Numbers on top of the bars representpercentage of the billets with cracks)Figure 5.63 Bar graph showing average depth of mid-way cracks at four loca- 139tions on the test and control strand for a 1084 grade billet fromCompany C. (Note: Numbers on top of the bars representpercentage of the billets with cracks)Figure 5.64 Bar graph showing average depth of mid-way cracks at four loca- 140tions on the test and control strand for a L-325 grade billet fromCompany C. (Note: Numbers on top of the bars representpercentage of the billets with cracks)Figure 5.65 Photograph of the surface of a 1008 grade billet from Company B 141(Mag. 1.0 X)Figure 5.66 Photograph of the surface of a 1012 grade billet from Company B 142(Mag. 1.0 X)Figure 5.67 Photograph of the surface of a 1039 grade billet from Company B 143(Mag. 1.0 X)(xxi)Figure 5.68 Graph showing the influence of oil flow rate on oscillation mark 144depth of a 1018 grade test strand billet cast with Canola oil atCompany EFigure 5.69 Graph showing the influence of oil flow rate on oscillation mark 144depth of a 1018 grade test strand billet cast with HEAR oil atCompany EFigure 5.70 Graph showing the influence of oil flow rate on oscillation mark 144depth of a 1018 grade test strand billet cast with Mineral_S oil atCompany EFigure 5.71 Graph showing the influence of oil flow rate on oscillation mark 144depth of a 1018 grade test strand billet cast with Soybean oil atCompany EFigure 5.72 Photograph of the surface of a 1018 grade billet cast on the test 145strand with Canola oil at 65 mI/mm at Company E (Mag. 1.0 X)Figure 5.73 Photograph of the surface of a 1018 grade billet cast on the test 146strand with HEAR oil at 65 mi/mm at Company E (Mag. 1.0 X)Figure 5.74 Photograph of the surface of a 1018 grade billet cast on the test 147strand with Mineral_S oil at 65 mI/mm at Company E (Mag. 1.0 X).Figure 5.75 Photograph of the surface of a 1018 grade billet cast on the test 148strand with Soybean oil at 65 mI/mm at Company E (Mag. 1.0 X).Figure 5.76 Graph showing the influence of oil flow rate on oscillation mark 149depth of a 1080 grade test strand billet cast with Canola oil atCompany EFigure 5.77 Graph showing the influence of oil flow rate on oscillation mark 149depth of a 1080 grade test strand billet cast with HEAR oil atCompany EFigure 5.78 Graph showing the influence of oil flow rate on oscillation mark 149depth of a 1080 grade test strand billet cast with Mineral_S oil atCompany E(xxii)Figure 5.79 Graph showing the influence of oil flow rate on oscillation mark 149depth of a 1080 grade test strand billet cast with Soybean oil atCompany EFigure 5.80 Photograph of the surface of a 1080 grade billet cast on the test 150strand with Canola oil at 45 mi/mm at Company E (Mag. 1.0 X)Figure 5.81 Photograph of the surface of a 1080 grade billet cast on the test 151strand with HEAR oil at 45 mi/mm at Company E (Mag. 1.0 X)Figure 5.82 Photograph of the surface of a 1080 grade billet cast on the test 152strand with Mineral_S oil at 45 mi/mm at Company EFigure 5.83 Photograph of the surface of a 1080 grade billet cast on the test 153strand with Soybean oil at 45 mI/mm at Company E (Mag. 1.0 X).Figure 5.84 Graph showing the influence of oil flow rate on oscillation mark 154depth of a 1090 grade test strand billet cast with Canola oil atCompany BFigure 5.85 Graph showing the influence of oil flow rate on oscillation mark 154depth of a 1090 grade test strand billet cast with HEAR oil atCompany EFigure 5.86 Graph showing the influence of oil flow rate on oscillation mark 154depth of a 1090 grade test strand billet cast with Mineral_S oil atCompany EFigure 5.87 Photograph of the surface of a 1090 grade billet cast on the test 155strand with Canola oil at 45 mI/mm at Company E (Mag. 1.0 X)Figure 5.88 Photograph of the surface of a 1090 grade billet cast on the test 156strand with HEAR oil at 45 mi/mm at Company E (Mag. 1.0 X)Figure 5.89 Photograph of the surface of a 1090 grade billet cast on the test 157strand with Mineral_S oil at 45 mI/mm at Company E (Mag. 1.0 X).Figure 5.90 Graph showing the influence of flow rate of Canola oil for a 1146 158grade billet at Company EFigure 5.91 Graph showing the influence of flow rate of HEAR oil for a 1146 158grade billet at Company B(xxiii)Figure 5.92 Photograph of the surface of a 1146 grade test strand billet cast with 159Canola oil at 65 mllmin at Company E (Mag. 1.0 X)Figure 5.93 Photograph of the surface of a 1146 grade test strand billet cast with 160HEAR oil at 65 mi/mm at Company E (Mag. 1.0 X)Figure 5.94 Graph showing the influence of oil flow rate on oscillation mark 161depth of a 1018 grade billet cast with Canola oil at Company CFigure 5.95 Graph showing the influence of oil flow rate on oscillation mark 161depth of a 1018 grade billet cast with HEAR oil at Company CFigure 5.96 Graph showing the influence of oil flow rate on oscillation mark 161depth of a 1018 grade billet cast with Mineral_S oil at Company C.Figure 5.97 Graph showing the influence of oil flow rate on oscillation mark 161depth of a 1018 grade billet cast with Mineral_O oil at Company C.Figure 5.98 Photograph of the surface of a 1018 grade billet cast with Canola oil 162at 25 nil/mm at Company C (Mag. 1.0 X)Figure 5.99 Photograph of the surface of a 1018 grade billet cast with HEAR oil 163at 25 mI/mm at Company C (Mag. 1.0 X)Figure 5.100 Photograph of the surface of a 1018 grade billet cast with Mineral_S 164oil at 25 mI/mm at Company C (Mag. 1.0 X)Figure 5.101 Photograph of the surface of a 1018 grade billet cast with Miner- 165al_O oil at 25 mi/mm at Company C (Mag. 1.0 X)Figure 5.102 Graph showing the influence of oil flow rate on oscillation mark 166depth of a 1045 grade billet cast with HEAR oil at Company C(Mag lOX)Figure 5.103 Graph showing the influence of oil flow rate on oscillation mark 166depth of a 1045 grade billet cast with Mineral_O oil at Company C.Figure 5.104 Photograph of the surface of a 1045 grade billet cast with HEAR oil 167at 100 mi/mm at Company C (Mag. 1.0 X)(xxiv)Figure 5.105 Photograph of the surface of a 1045 grade billet cast with Miner- 168al_O oil at 100 mi/mm at Company C (Mag. 1.0 X)Figure 5.106 Graph showing the influence of oil flow rate on oscillation mark 169depth of a 5160 grade billet cast with Canola oil at Company CFigure 5.107 Graph showing the influence of oil flow rate on oscillation mark 169depth of a 5160 grade billet cast with HEAR oil at Company CFigure 5.108 Graph showing the influence of oil flow rate on oscillation mark 169depth of a 5160 grade billet cast with Mineral_S oil at Company C.Figure 5.109 Graph showing the influence of oil flow rate on oscillation mark 169depth of a 5160 grade billet cast with Mineral_O oil at Company C.Figure 5.110 Graph showing the oscillation mark depth for a 1045 grade billet 170cast with HEAR oil and with a mould oscillation of 96 cpm atCompany CFigure 5.111 Graph showing the oscillation mark depth for a 1045 grade billet 170cast with Mineral_O oil and with a mould oscillation of 96 cpm atCompany CFigure 5.112 Graph showing the oscillation mark depth for a 1080 grade billet 170cast with HEAR oil and with a mould oscillation of 130 cpm atCompany EFigure 5.113 Bar chart showing off-squareness of billets for four steel grades cast 171with four different oils at two flow rates on the test strand and withCanola oil at 70 mI/mm on control strand at Company C. (CCanola, SS = Mineral_S, H = HEAR, 0= Mineral_O)Figure 5.114 Macro-etched longitudinal section of a 1018 grade billet showing a 172typical example of craze cracking on the inside radius at CompanyB (Mag. 1.0 X)Figure 5.115 Plot of craze crack index versus copper content of billets collected 173at Company CFigure 5.116 Plot of craze crack index versus nickel content of billets collected at 173Company C(xxv)Figure 5.117 Plot of craze crack index versus carbon content of billets collected 173at Company CFigure 5.118 Plot of craze crack index versus Ni/Cu ratio of billets collected at 173Company CFigure 5.119 Surface of a 1080 grade control strand billet cast with Soybean oil at 17465 mi/mm at Company E showing evidence of beads (Mag. 1.0 X)...Figure 5.120 Surface of a 1080 grade test strand billet cast with Canola oil at 65 175mi/mm showing severe “zipper marks” at Company E (Mag. 0.8 X).Figure 5.121 Transverse macro-etch through a bead on a 1080 grade billet at 176Company B (Mag. 14 X)Figure 6.1 Mesh used for modelling a longitudinal section of the mould wall 181[181Figure 6.2 Forced convection - heat transfer curve [18] 182Figure 6.3 Mesh used to model mould distortion [62] 183Figure 7.1 Mesh used for modelling one quarter of a transverse section of a 195billetFigure 7.2 Lattice parameter of austenite as a function of carbon content at 196various temperature [77]Figure 7.3 Effect of temperature on the dilation coefficient for austenite 197Figure 7.4 Variation of mean-linear coefficient of thermal expansion of steel 198with temperatureFigure 8.1 Heat-Flux profiles at Company B for billets with different carbon 211contents(xxvi)Figure 8.2 Heat-Flux profiles at Company C for billets with different carbon 211contentsFigure 8.3 Heat-Flux profiles for 1018 grade steel billets cast with Canola oil at 2130, 25, 70 and 100 mI/mm at Company C. (Note: plot for 70 mi/mmof oil lies in between those for 25 and 100, but has not been shownfor clarity)Figure 8.4 Heat-Flux profiles for 1018 grade steel billets cast with Canola oil at 2140, 25, 70 and 100 mI/mm at Company C (enlarged in the meniscusregion). (Note: plot for 70 mi/mm of oil lies in between those for 25and 100, but has not been shown for clarity)Figure 8.5 Heat Extracted in the mould for 1018 grade steel billets cast with 215Canola oil at 0, 25, 70 and 100 mi/mm At Company CFigure 8.6 Mould hot face temperature during casting of 1018 grade steel 216billets with Canola oil at 0, 25 and 100 mI/mm at Company CFigure 8.7 Heat-Flux profiles for 1018 grade steel billet cast with Canola, 217HEAR, Minerai_O and Mineral_S lubricating oil at 25 mi/mm.(Note: the plots for Mineral_O and HEAR lie between those forCanola and Mineral_S oils but have been omitted for clarity)Figure 8.8 Heat-Flux profiles for 1045 grade steel billets cast at mould oscilia- 218tion frequencies of 96 and 144 cpm at Company CFigure 8.9 Heat-Flux profiles for 1045 grade steel billets cast at Companies B, 219EandCFigure 8.10 Graph showing the match between predicted heat extraction rate in 220the mould and the heat extracted by the mould cooling waterFigure 8.11 Specific heat extraction rate (kj/kg), for the moulds at Companies B, 221EandCFigure 8.12 Calculated distortion of the mould during the casting of low and 222high carbon heats at Company BFigure 8.13 Computed distortion of the mould during service at Company E 223Figure 8.14 Computed distortion of the mould during service at Company C 224(xxvii)Figure 8.15 Surface temperature at the mid-face of the billet for different grades 225at Company BFigure 8.16 Shell thickness at the mid-face of the billets for different steel 226grades at Company BFigure 8.17 Shell thickness at the mid-face of the billets for different steel 227grades at Company EFigure 8.18 Shell thickness at the mid-face of the billets for different steel 228grades at Company CFigure 8.19 Billet shrinkage profile, at Company B, for “medium” and “high” 229carbon steelsFigure 8.20 Billet shrinkage profile, at Company B, for low carbon steels 230Figure 8.21 Surface roughness of a “high” carbon steel billet cast at Company B. 231Figure 8.22 Surface roughness of a low carbon steel billet cast at Company B 232Figure 8.23 Billet shrinkage profile for 1045 grade steel cast at Company C 233Figure 9.1 Axial profiles of gap and shell thermal resistances at Company B 246Figure 9.2 Axial profiles of gap and shell thermal resistances at Company E 247Figure 9.3 Axial profiles of gap and shell thermal resistances at Company C 248Figure 9.4 The three zones of heat transfer in the continuous casting billet 249mouldFigure 9.5 Shape acquired during service by a mould with a steep initial taper. 250Figure 9.6 Shape acquired during service by a mould with a shallow initial 251taper(xxviii)Figure 9.7 Graph showing the effect of negative strip time on the oscillation 252mark depth on 1008 grade billets cast through parabolic tapermoulds. (Negative strip time at Strand 3 and Strand 4 are 0.14 and0.19 s respectively)Figure 9.8 Graph showing the effect of negative strip time on the oscillation 253mark depth on 1008 grade billets cast through double taper moulds.(Negative strip time at Strand 3 and Strand 4 are 0.14 and 0.19 srespectively)Figure 9.9 Effect of‘decompression” load on the mould heat transfer during 254the casting of 1018 grade billets at Company CFigure 9.10 Effect of negative strip time on the mould heat transfer during the 255casting of 1045 grade billets at Company CFigure 9.11 Effect of decompression period (measured as a percentage of tN), on 256the mould heat transfer during the casting of 1045 grade billetsFigure 9.12 Axial heat-flux profiles on two adjacent mould walls at Company C. 257Figure 9.13 Computed mould shape during operation for two adjacent mould 258walls at Plant CFigure 9.14 Measured peak heat flux for different initial mould tapers 259Figure 9.15 Hot face temperature of the mould wall while casting 1045 grade 260steel billets at Plants B, E and CFigure 9.16 Variation in mould heat transfer on account of different amounts of 261Hydrogen gas in the mould-billet gapFigure 9.17 Axial profile of gap and shell thermal resistances while casting 1045 262grade billets at 144 cpm of mould oscillation at Company CFigure 9.18 Axial profile of gap and shell thermal resistances while casting 1045 263grade billets at 144 cpm of mould oscillation at Company CFigure 11.1 Heat Flux, billet shrinkage and mould wall profile at Company C 276(xxix)Figure 11.2 Mould thermocouple response while casting billets with off-corner 277cracksFigure 11.3 Mould hot face temperature indicating unequal heat transfer on 278adjacent facesFigure 11.4 Response of mould thermocouple on adjacent mould walls indicat- 279ing unequal heat transferFigure 11.5 Load cell response at Company C obtained while casting billets 280without mould lubricating oil(xxx)LIST OF SYMBOLSk Thermal conductivity of steel, kW/m°C.T Temperature, °C.t Time below meniscus, s.p Density of steel, kg/rn3.Heat capacity of steel, JIkg°C.T Pouring temperature °C.T, Liquidus temperature of steel, °C.Solidus temperature of steel, °C.x, y, z Transverse directions, m.X, Y Width and thickness of billet, m.q0 Heat flux from the surface of billet to the mould, kW/m2.V3 Specific volume of delta unit cell, cm3/g.V Specific volume of gamma unit cell, cm3/g.a37’ Lattice parameter of delta iron at temperature T and carbon content C, angstroms.a7’ Lattice parameter of gamma iron at temperature T and carbon content C, angstroms.(xxxi)Carbon content of delta phase, atomic percent.W Carbon content of gamma phase, weight percent.MT Dimension of the mould wall at the top of the mould, m.MB Dimension of the mould wall at the bottom of the mould, m.ML Total length of the mould, m.Km Thennal conductivity of mould, kW/m°C.Pm Density of mould, kg/rn3.Cpm Specific heat of mould, J/kg°C.Density of water, kg/rn3.V4, Velocity of cooling water in channel, ni/s.d,4, Water channel gap width, m.C,, Specific heat of water, JIkg°C.T Temperature of water, °C.Heat-transfer coefficient at the mould/cooling water interface, kW/m2°C.QFC Forced convection heat flux.(xxxii)QFD Heat flux in the fully developed nucleate boiling region.QTR Heat flux in the transition region between the point of incipient and the fully developednucleate boiling.QFN Heat flux at the inception of boiling.Heat flux at the inception of boiling using the equation for fully developed boilingregion.Viscosity of fluid, N.s/rn2a Surface tension of liquid! vapour interface, N/rn.CSf Emperical constant that depends on the nature of the heating surface/fluid combination.g Gravitational acceleration rn/s2.hfC Forced convection heat-transfer coefficient.Saturation temperature of water °C.Subscriptsf fluid.I liquid.m mould.v saturated vapour.w water.(xxxiii)ACKNOWLEDGEMENTI would like to express my sincere thanks to my supervisors, Professors I.V. Samarasekeraand J.K. Brimacombe, for their invaluable guidance and advice during the course of the researchwork. Mr. lan Bakshi organised the industrial trials at the three steel plants and his role in this andother areas of the work is deeply appreciated. The assistance of Mr. Robert Hapke and Mr. NeilWalker in the instrumentation of the nioulds and the metallographic examinations of the billets issincerely appreciated. I would also like to thank Mr. Walker for his help in printing the photographsused in the thesis.I am grateful to the management of TATA STEEL for granting me the necessary study leaveto complete my thesis.Finally, I would like to thank the manufacturers of the mould at Company C; their poormachining helped create ideal conditions for high mould heat transfer!(xxxiv)CHAPTER 1: INTRODUCTIONThe continuous casting of steel billets involves the pouring of molten steel at a controlledrate into a water-cooled copper mould and the continuous withdrawal of a partially solidified billetas shown in Figure 1. 1. The withdrawal of the steel strand is aided by the oscillation of the mouldand a constant supply of a lubricating oil onto the mould wall. The mould extracts heat and freezesa solid shell which thickens progressively as it moves down the mould. As a result of the cooling,the solid shell shrinks and pulls away from the mould wall thereby opening up an air gap. Heattransfer from the molten steel to the mould wall takes place, as shown in Figure 1.2, by the followingfive steps in series [1](i) Convection in the liquid steel pool.(ii) Conduction in the solid shell.(iii) Conduction and, to a lesser extent, radiation across the air gap(iv) Conduction in the mould wall.(v) Convection at the mould-cooling water interface.Heat transfer from molten steel to the walls of the continuous casting mould is controlled, inlarge part, by conduction across the air gap. In the upper region of the mould the air gap isconsiderably less than a millimeter wide but, in many cases, accounts for as much as 80-90% ofthe total resistance to heat flow [2]. To compensate for the billet shrinkage, the mould walls aretapered inwardly; the resulting reduction in the air gap improves the rate of mould heat extractionand decreases the surface temperature of the billet at the mould exit, thereby reducing the tendencyof the billet surface to reheat [3-5] and form sub- surface cracks. Lack of sufficient taper, additionally,[1]can lead to the formation of off-corner cracks. On the other hand, an excessive taper can causedifficulty in the withdrawal of the strand [3,6] which promotes mould wear [3,7] and, in extremecases, causes the billet to jam in the mould [1].The quantification of the strand-mould gap is thus a primary step towards defining mouldtaper. The gap is, however, a complex function of several variables and its width changes in boththe longitudinal and transverse directions which renders it extremely difficult to characterize. Theshrinkage of the billet is affected significantly by the grade of steel being cast, particularly the lowcarbon grades where the contraction accompanying the solid-state transformation from the delta toaustenite phase must be taken into account. Another factor contributing to the complexity of theanalysis is the distortion of the mould which can be significant [8].Mouldoscillation is probably the most outstanding feature of the conventional casting processwith an upright (curved or straight ) mould. Historically mould oscillation was applied for the firsttime in 1949 on two pilot plants constructed independently by S. Junghans and I. Rossi [9,10]. Thistechnique helped to minimize casting problems and surface defects due to shell sticking which weretypical in continuous casting with a stationary mould. Consequently mould oscillation was adopted,with little delay, on further installations.The oscillation is typically sinusoidal and for some period of the downstroke, the mouldvelocity is faster than the strand withdrawal speed. This period is referred to as negative strip (tN)and when this is expressed as a fraction of the total period of oscillation it is called NSR (negativestrip ratio). Some aspects of mould oscillation are shown in Figure 1.3. Oscillation of the mouldleads to the formation of ‘oscillation marks’ which give the billet surface a rough appearance. Thesolidification structure beneath an oscillation mark- on account of the locally reduced heat transfer- is much coarser than the subsurface structure between the marks [11] as shown in Figure 1.4. The[2]negative-strip time is known to correlate strongly, as shown in Figure 1.5 with the depth ofoscillationmarks [12]. This has, of course, to be balanced against the positive effect that mould oscillation hason preventing shell sticking.Traditionally oil has been used as a lubricant for casting small sized billets while slab andbloom moulds have been lubricated by mould powders. In the case of billets the liquid lubricant ispumped to the top of the mould into an oil channel from which it weeps through a narrow slot ontothe hot mould face. The velocity of the oil down the mould wall is dictated by the oil feed rate andthe viscosity which is a function of the mould wall temperature [13]. Proper lubrication by oil isessential to ensure good billet quality in terms of surface defects and cracks. Additionally thepresence of oil at and below the meniscus is likely to affect heat transfer rates in the mould and isan important factor in billet shrinkage and, therefore, mould taper. On the other hand, excessive oilflow is uneconomical and contributes to the formation of pinholes on the billet surface. Originallyoils for mould-strand lubrication were derived from animal and vegetable sources; however, morerecently, mineral oils, are being utilised in mould lubrication.The Continuous Casting process has been established worldwide due to its higher yield,enhanced productivity and more uniform quality compared to the conventional ingot casting process.As much as 65% of the current steel production in the world is continuously Cast whilst in Japanthis figure exceeds 90%. Significant savings in energy can be realised if the Continuously cast steelis directly hot charged into the reheat furnace prior to rolling [14]; but for direct charging to besuccessful it is necessary to produce good quality billets.After nearly two decades of research at UBC it has been conclusively shown that heat transferand solidification in the water-cooled mould are of prime importance to billet quality. In view ofthe fact that steel has low ductility and mechanical strength close to the solidus temperature [15,16],[31it is not surprising to note that a wide variety of casting problems, ranging from breakouts to shapedefects, are directly related to the events in the mould. Thus mould-strand interaction controls billetquality to a large extent.The present work was undertaken to understand the various factors that influence mould-strand interaction. With the experience at UBC of organising numerous successful plant trials,measurement of mould wall temperature and mould-billet friction forces were carried out onoperating moulds at three different steel plants in Canada. Data was collected under a variety ofoperating conditions and then processed through several mathematical models developed for thepurpose. Predictions from the models were verified by several independent observations includingexamination of surfaces and internal structures of billets collected during plant trials. Dramaticallydifferent mould-heat transfer rates observed at the three plants could be successfully linked to theshape of the distorted mould during service.This study has led to a very comprehensive understanding of the factors that influence mouldheat transfer and thereby has laid the grounds for the design of mould tapers to cast a wide rangeof steel grades. It has also been possible to identify an important property of mould lubricating oilsnecessary for proper lubrication and to recommend oil flow rate. Furthermore, signals from sensorsattached to the mould have been related to the formation of defects in the billet. With the knowledgeof the various operating variables and their impact on mould heat transfer the foundation has beenlaid for the design of a control system capable of identifying and correcting billet defects duringcasting.[41MOULD COOLING WATERLIQUID STEEL INSOLID SHELr.BILLET (partially solidified) OUTMOULD WALLMENISCUS LEVELMOULD COOLING WATERAIR GAPFIGURE 1.1 Schematic diagram of a casting set-up for the continuous casting of steel billets.[51I I II I II i 1I I IWaterI I II I I MouldI I I GapI I II I II I I ShellFIGURE 1.2 Schematic diagram of heat removal in a continuous casting machine.[6]Oscillation e1ocityy 2irafcos (2irft)2________________________sting speedtp irccos (—v/2xaf)/wftN 1/f—tpNSR(2ftN)X lOO()FIGURE 1.3 Sinusoidal oscillation cycle of the mould.[7]11 C) H! D C CD C CD B C) C 1 D0.6100.2 c00.200.250.301N. SecFIGURE 1.5 Relationship between negative strip time and the depth of oscillation marks.[9]Chapter 2: LITERATURE REVIEWThis chapter gives an overview of the knowledge available in the literature on billet qualityand mould-billet interaction. Notwithstanding the many years that billet casting has been practiced,the mechanism of oil lubrication has never been examined in detail. While there is a rich collectionof literature on mould-strand interaction in slab casting, such is not the case in the field of oillubrication in billets. There are a few minor references to lubrication in the billet mould but it isclear that these are not based on rigorous analysis. Not surprisingly then some of the conclusionsare contradictory.2.1 Thermornechanical Behaviour of the MouldThe billet mould is one of simple design- a copper tube having a wall thickness of 12-18 mmand constrained within a steel jacket with water flowing through the annulus as shown in Figure2.1. The copper tube is secured in the mould assembly by spilt plates fitting into slots close to thetop of all four mould tube faces. The mould has been the subject of extensive research by Samarasekera and Brimacombe [17-22]. The following paragraphs summarize their findings.The temperature distribution in the wall of a billet casting mould depends on the amount ofheat extracted from the solidifying billet, the rate at which it is conducted through the mould walland the rate of heat transfer to the cooling water. During the casting operation the mould distortsand changes shape in response to internally generated thermal stresses. Distortion of the mould,only part of which is permanent (plastic), arises from the combination of the differential thermalexpansion due to non-uniform heating of the mould wall, the restraint of the free expansion of thecopper by the mould support system and the geometric configuration of the mould itself. Thus themould tube bulges where it is hottest near the meniscus to give a negative outward taper as opposed[10]to the normal inward taper. Furthermore, the bulge in the mould is not static but changes dynamicallyin response to mould temperature variations caused by metal level fluctuation and, in some cases,by nucleate boiling in the cooling channel.The variables that influence the magnitude of the bulge, and its position relative to the topof the mould are cooling water velocity, metal level fluctuation, position of the mould tube constraintrelative to the top of the mould, and the type of the constraint. It has been found that a four-sidedconstraint to support the mould tube within the housing is superior to a two-sided constraint. Asregards cooling water velocity, it has been established that a drop in its value leads to an increasein both the magnitude of the negative taper below the metal level and the peak distortion.With the aid of mathematical models, based on data collected in industrial trials, Brimacombe,Samarasekera and co-workers have been able to link mould distortion and dynamic mould wallmovement due to nucleate boiling to several mould-related quality problems such as rhomboidity,non-uniform oscillation mark depth, etc. The impact that mould-strand interaction has on billetquality is covered in a subsequent section.While the effect of several operating variables on the distortion of the mould has beenestablished by Brimacombe, Samarasekera and co-workers, the impact of a few other variableshave not been quantified. Thus the effect on mould distortion, of casting low-carbon billets asopposed to high-carbon billets has not been established. Similarly, the impact that the pre-existingmould taper may have on the shape the mould acquires during service, is an area that needs furtherinvestigation.[11]2.2 Mould Lubrication with OilAs mentioned earlier, oils used for mould-strand lubrication were originally derived fromanimal and vegetable sources. Occasionally, mineral oil is blended with vegetable oil and usedduring casting. There are a multitude of physical and chemical tests which yield useful informationon the characteristics of lubricating oils. Some of the most common tests are outlined below.(a) The Carbon Residue of a lubricating oil is the amount of deposit, in percentage by weight, leftafter evaporation and pyrolysis of the oil under certain prescribed conditions.(b) The Flash Point of an oil is the temperature at which the oil releases enough vapor to ignitewhen an open flame is applied while the temperature at which vapors are released rapidly enoughto sustain combustion is called the Fire Point. For any specific product, both flash and fire pointswill vary depending on the apparatus and the heating rate. The flash point ofoils varies with viscosity- higher viscosity oils have higher flash points.(c) The Pour Point of a lubricating oil is the lowest temperature at which it will pour or flow whenit is chilled without disturbance under prescribed conditions. Most oils contain some dissolved waxand, as an oil is chilled, this wax begins to separate as crystals that interlock to form a rigid structurewhich traps the oil in small pockets preventing it from flowing.(d) Probably the single most important property of a lubricating oil is its viscosity. Viscosity canbe determined by measuring the force required to overcome fluid friction in a film of knowndimensions. Viscosity determined in this way is called dynamic or absolute viscosity with units ofpoise (P) or Pascal seconds (Pa s). Dynamic viscosity is a function only of the internal friction ofa fluid. Kinematic viscosity combines the effect of oil density and viscosity and can be obtainedfrom division of dynamic viscosity of an oil with its density giving units of Stokes (St) or squaremillimeters per second (mm2/s). The viscosity of any fluid changes with temperature - increasingas the temperature is decreased.[12]A good lubricant, apart from the properties discussed above, needs to be non-toxic and notsmoke excessively when used.2.2.1 Mechanism of lubricationThree common modes of lubrication encountered in operation of an industrial process wheretwo metallic surfaces slide past each other are hydrodynamic (thick film or fluid), boundary layerand mixed lubrication [231.In hydrodynamic orjluid lubrication the surfaces in relative motion are separated by a lubricantlayer of appreciable thickness and under ‘ideal’ conditions there is no wear of the solid surface.The thickness of the film is approximately one order of magnitude larger than the roughness ofeither surface. The resistance to motion is due entirely to the viscosity of the interposed lubricantlayer and the coefficient of friction encountered are in the range 0.001 - 0.002.If the sliding speeds are low or the loads are high then it is often impossible to obtain fluidlubrication and the thick lubricant layer breaks down leading to boundary layer lubrication. In thismode the surfaces are separated by a lubricant film of only a few molecular dimension and underthis condition the friction is influenced by the nature of the underlying surface as well as by thechemical constitution of the lubricant. The bulk viscosity plays little or no part in the frictionalbehaviour and the coefficient of friction encountered is in the range 0.1 - 0.4.Mixed lubrication becomes operative under conditions when the film thickness is reducedfrom 10 to 3 times the height of the asperities on the surface and the coefficient of friction increasesfrom 0.001 to 0.4.In the case of boundary layer lubricants, addition of a small amount of fatty acid to mineraloils significantly lowers friction values [24]. This has generally been attributed to the adherence ofthe fatty acid to the surface of the metal substrate. It has been shown that the fatty acid moleculesorient themselves with the carbonyl groups at the solid surface. This results in the formation on the[13]surface of a film of fatty acid molecules all attached to the surface as shown in Figure 2.2. Theselayers actually isolate the surface of the two metals and thereby reduce friction. The forces ofadhesion are strong enough to resist removal of the fatty acid and there are indications that a chemicalreaction actually takes place at the surface resulting in the formation of a soap film that is chemicallybound to the metal surface.Indirect evidence of the formation of a soap film can be observed from the fact that fatty acidsare most effective as friction reducers where the nature of the metal permits a definite chemicalreaction. Table 2.1 shows that non-reactive surfaces are almost unaffected by fatty acids as comparedto more reactive surfaces. Another piece of indirect evidence is that the temperature at which thelubricant film breaks down is considerably higher than the melting point of the fatty acid temperatureand corresponds approximately to the stage at which metallic soap, formed by chemical reaction,softens and melts. It has been found that the greater the number of carbon atoms and the longer themolecule, the lower the coefficient of friction. This is expected as the longer hydrocarbon chainswould provide more effective separation of the two surfaces.2.2.2 Factors affecting mould friction2.2.2.1 Oil flow rate and oil typeThe only published work on the effect of oilflow rate on lubrication is that of Brendzy [13.In an industrial trial with a mould instrumented by load cells, Brendzy showed that the reductionof oil flow (from 54 mi/mm to 24 mi/mm) for three different types of lubricant studied resulted inincreased interaction between the strand and mould as shown in Figure 2.3. That such an interactioncould be seen regardless of the oil type or the effect of carbon is an indication of how stronglyfriction is affected by flow rate. This enhanced interaction, however, did not seem to have anysignificant effect on billet quality. The relevance of the different regions of the sensor signal isexplained in a subsequent section.[141It has been shown [131 that the three oils under study exhibited a different degree of lubricationat the same flow rate. It has been suggested that this may, in part, be a reflection of the differencesin the fatty acid content of the different lubricants. Additionally, the lubrication by these oils showeda varying degree of dependency on flow rate. However, as changes in the lubricant type could notbe separated from the changes in the grade of steel cast, this finding remains inconclusive. Furthermore, in light of the prevailing high temperatures during continuous casting, it is unrealistic toexpect that the amount of fatty acid in the oils, can significantly alter the performance of lubricants.2.2.2.2 Steel gradeMairy et al. [25] have shown that the lubrication is different with each steel grade and that,at least in case of slab casting, the flux practice must be adapted to the different steel grades. It wasshown by them that steels with 0.13% carbon produce lower friction signals than those with 0.40%carbon. This has been attributed to the peritectic transformation in low-carbon steels giving rise tomaximum contraction and non-uniform shell growth. It is debatable, however, if the shrinkage oflow-carbon steels would necessarily be higher than high-carbon grades. While there is an “extra’contraction in low carbon grades arising from the to y phase transformation, the heat transfer forthese grades is known to be low. Without carrying out appropriate calculations, it is difficult to saywhich factor dominates.Singh and Blazek [26] plotted mould friction as a function of carbon content of steel, Figure2.3, from their work on an experimental caster. The stationary experimental mould was lubricatedby Swift 1011 oil at 8.5 mi/mm. While acknowledging a wide variation in values the authorsconcluded that steel with a high carbon content (more than 0.4%) tended to have lower mouldfriction which they attributed to the higher carbon in steels acting as a lubricant. The low frictionvalue for the 0.1 % carbon steel (compared to lower carbon grades) was thought to be related tothe amount of rippling on the surface of the billet.[15]Since Singh and Blazek have not discussed how friction measurements were carried out andin light of their ‘experimental mould”, it is difficult to comment on how much of their results applyto an industrial caster. In any case a recent work at Hoogovens, Holland [27), where an operatingmould was instrumented by accelerometers, has shown that higher friction values were obtainedwhile casting low-carbon grade steel billets. This result contradicts that obtained by Singh andBlazek. The higher friction values for low-carbon grades have been explained [271 on the basis oflarger contact area of the billet with the mould arising out of a smaller shrinkage for these grades.2.2.2.3 Mould taperKomatsu et al. [28] have studied mould friction on a small scale experimental caster usingmould powder as a lubricant and found that excess tensile stress is applied to the solidified shellby a tapered mould during positive stripping periods. It will be shown in a subsequent section thatthe heat flux in the mould which governs the magnitude of solid shell contraction, is taper dependent.2.3 Heat Transfer in the MouldIn its simplest form, the transfer of heat from the liquid steel to the mould cooling water takesplace by conduction through the solid shell, across the billet-mould gap and through the mould wallfollowed by convection at the mould cooling water interface, as was explained in the previouschapter. From the heat transfer point of view, the mould can be divided into two zones [29] : anupper region in which heat extraction can be influenced by factors altering the gap width or gapconductivity like taper, mould distortion, lubricant type and flow rate, and a lower region of gapand shell resistance dominance in which the heat extraction can be influenced by factors like castingspeed (alters the shell thickness) and taper (changes gap). The influence of the lubricating mediumon the heat transfer is examined in the following section.[16]2.3.1 Effect of oil on heat fluxTo lubricate the mould with oil, an oil film is created on the mould wall. This oil film flowsdown to the meniscus and pyrolyses in contact with the steel meniscus. Part of the oil escapes asgas while the rest may be pushed into the gap between the mould and the solidifying shell duringthe downstroke of the mould oscillation [27,30]. Mould powders, on the other hand, melt and wetthe steel, with the extent of wetting being controlled by interfacial forces. The difference in behaviourbetween the two types of lubricant gives rise to different patterns of heat extraction. A plot ofspecific heat-flux profiles (total heat extracted per unit weight) for an oil and two different mouldfluxes is shown in Figure 2.5. Klipov et al. [31] have reported that the oil heat flux is greater thanthe powder in the upper part of the mould by 15-20% while it is the reverse in the lower part of themould by 20-25%. The higher upper-mould heat flux with oil is probably on account of ahydrogen-rich atmosphere (from the pyrolysis of oil) between the shell and the mould whichincreases the thermal conductivity of the gap. In the lower part of the mould spray water is believedto penetrate and decompose in the mould/strand gap to from a hydrogen-rich gas [32]. Taylor [32]suggests that oil wets the strand surface more effectively than a high melting point powder andreduces the production of hydrogen.There are some difficulties in understanding the explanations offered by the authors above.Firstly, it is well known that heat transfers in billet casting (using oil) are higher than those obtainedin slab casting (using powders) and secondly, it is unclear how a layer of oil can ‘wet” the steelstrand which is at a temperature in excess of 1100 °C.2.3.2 Effect of mould powders on heat fluxSingh and Blazek carried out experiments in a continuous casting mould with horizontal waterpassages [33]. They found, using mould powder as a lubricant, that[171(i) In the case of low-carbon steel (C — 0.10%) the heat transfer increased just below themeniscus but decreased thereafter as shown in Figure 2.6.(ii) In the case of high-carbon steel (C 0.40%) the heat transfer was significantly lowerover the entire mould length as shown in Figure 2.7.The behaviour of low-carbon steel is believed to be caused by the molten flux filling in the‘ripples’ on the steel skin just below the meniscus thereby increasing the thermal conductivity ofthe gap. In the lower regions of the mould the ferrostatic force, in the opinion of the authors, causesthe steel shell to be in contact with the mould wall and the presence of the molten flux then acts asan insulator. In case of high-carbon steels, the authors suggest that the billet surface, being smooth,are in good contact with the mould wall and the introduction of the mould flux causes the latter tobehave as an insulator thereby decreasing heat transfer. It needs to be mentioned, however, thatthese experiments were on a stationary mould the length of which was almost half that of a typicalindustrial mould (0.8 m).2.3.3 Effect of steel grade on heat fluxSeveral researchers [17,34] have reported a drop in the heat transfer rate while casting steelswith carbon content of around 0.10%. This reduction is thought to stem from the large volumeshrinkage accompanying the 6 to ‘yphase transformation for these grades of steel. The phase changeand the resulting shrinkage occur when the solid shell of the billet is thin causing the billet surfaceto ripple. The increased surface roughness of the billet locally increases the mould-billet gap causinga drop in the heat transfer from the billet to the mould.2.3.4 Mould taperAs mentioned in the previous chapter, in the case of billets, some researchers have observedthat mould taper improves heat transfer 13,4] and also decreases the surface temperature at the strandexit [5], presumably because it reduces the gap width over the lower region of the mould. However,[18]details of the actual measurements are not available and it is difficult to guess whether heat transfermeasurements were carried out with a systematic change in mould tapers. In the case of continuouscasting of slabs, mould design, in particular the taper of the narrow plates [34,35], is known toinfluence the heat transfer in the mould. Deshimaru et al. [34] have enhanced heat extraction ratesnear the corner by using a plate with a higher taper at the corner. Wolf [36] has shown that the heatflux in the mould is enhanced with increase in the taper of the narrow face as shown in Table 2.2.2.4 Mould-Friction Measuring DevicesFriction in the continuous casting mould has been monitored by several techniques. In almostall cases there is little knowledge on the exact nature of the friction forces so that usually a relativemeasure of the friction is made, making it difficult to compare results from different sources. Inaddition, certain devices employ measuring techniques from which the effect of friction cannot becleanly separated.2.4.1 AccelerometersShort et al. [36] replaced linear variable displacement transducers (LVDTs) used for routinechecking of mould oscillation with accelerometers as the latter were found to be simpler and moresensitive than the LVDTs and, additionally, could be used as an indicator of mould friction. Theseresearchers blended esters with commercially available mould lubricating oils to optimize wettingbehaviour, mould heat transfer and mould friction. Their experience shows that good mechanicalstability of the oscillator is required to avoid interference with what they called “metallurgicaleffects of solidification” in the mould. They found that commercial oils could be blended with estersto lower friction in the mould to levels obtained with powder lubrication. No further details areavailable on their work.More recently, Stel et al. [27] have also used accelerometers to measure mould friction. Theirwork complements the results obtained by Brendzy [13].I 19]2.4.2 Strain gaugesYamanaka et al. [37] have measured the frictional forces in their experimental caster bymounting strain gauges on the centre rod which pulls the solidified shell. Wolf [38] refers to asimilar technique where the net frictional force was measured by piezo-quartz transducers mountedon the mould support. The observed increase in frictional force at higher casting speed was attributedto a decrease in mould flux thickness in the mould-shell gap. Foussal et al. [39] positioned straingauges on the coupling rod of a mould and found that the signal generated was periodic althougha difference in phase exists between mould displacement and mould friction. In the case of theJapanese workers, the frictional force was found to lag behind the mould displacement by 90 degreesat slow casting speeds (1.2 rn/mm). However, this phase difference became negligible at high castingspeeds. They attributed this behaviour to the rheological characteristics of mould flux. In the caseof the French workers, it is the mould displacement which lags behind the friction force for whichthey offer no explanation.The apparent contradiction in the observation of the two researchers reinforces the importanceof the location at which installation strain gauges are installed and the difficulties of simulating theactual behaviour of the mould in a laboratory. Furthermore, it is unclear what exactly is meant by“lag” or “lead” of the friction signal over the mould displacement signal.2.4.3 Load cellsThe third and probably the most popular method of monitoring mould friction has been theuse of load cells. Several researchers have used load cells mounted on the mould oscillating table[13,25,40-43], of which Brendzy’s [13] work conducted on (industrial) billet moulds, is the onlyone that discusses the results obtained in some detail.[20]Komatsu ‘S [40] measured, the friction force in terms of the difference in apparent mouldweight and mould inertia weight and found it to decrease along a straight line during casting andto be higher for tapered moulds. Additionally the friction force during the negative strip perioddecreased as the NSR decreased.Gloor [41] has developed a commercial mould friction measuring system (MFM) that employsa load cell mounted on the connecting rod between the eccentric drive and the short-lever oscillationmechanism. Data acquisition and evaluation by a computer equipped with special hardware andsoftware makes it possible to calculate and deliver the friction values virtually in real time. Hiswork shows that boron-alloyed steel (C 0.19%) has a substantially higher friction than a 0.48%carbon steel when cast on the same strand, with the same mould at casting speeds of 1.8 rn/mm and1.5 m/min respectively.Mairy and Wolf [25] studied the friction acting on a slab mould and found that friction forceswere much higher and tended to fluctuate more when oil was used as the lubricant instead ofpowder.This observation is similar to that of Short et al. [35] in their investigation of lubricants in billetcasting. Additionally, 0.12 % carbon steels yielded a significantly higher friction level than that forlower carbon (0.08%, 0.06%) when consecutively cast with the same mould powder. Stel et al. [27]have also reported higher friction forces when billets are cast with oils rather than with mould flux.Brendzy [13], in an industrial trial, placed load cells between a mould housing and theoscillator table. Linear Variable Displacement Transducers (LVDTs) were placed on the mouldtable to record the oscillation characteristics of the mould system. A copper mould and mouldcooling water were instrumented with an array of thermocouples. In view of the detailed nature oftheir work and its relevance to the current proposal their results are described in somewhat greaterdetail in the next section.[21]2.5 Load Cell Response and its AnalysisAs mentioned earlier, a billet mould with an empirically designed parabolic taper (4.9%/rnin the meniscus region, 1.8%/rn in the middle of the mould and 0.8%/rn toward the end of the mould)was instrumented with load cells placed between the mould housing and oscillator table to measureloading on the mould during casting. LVDTs were located on the mould table to record the oscillationdisplacement of the mould system.A typical load cell profile is shown in Figure 2.8. The response of the load cells is periodicand consists of two distinct modes of mould-strand interaction in each oscillation cycle. The firstoccurs during the upstroke and the second during the downstroke. The minimum loads (valleys)vary at a low frequency while the maximum loads (peaks) remain relatively constant. Additionallythe nature of the peaks is distinctly different from that of the valleys. While the minimas exhibit arelatively smooth appearance, the maxima are broader and appearjagged. This phenomena has alsobeen observed by Stel et al. [271.It was further seen that as the mould moves downwards (as indicated by the LVDT signals),there is a sudden decrease in the compressive load as the negative-strip period begins. This decreasecontinues smoothly until the maximum downward velocity of the mould is reached at which pointthe load begins to increase. During upstroke the load cells signals exhibit a slip-stick behaviour.By superimposing the casting speed on the load cell response Brendzy showed (Figure 2.9)that there was a clear relationship between the variation of the minimum load and the casting speed.It was seen that minimum load increases as casting speed increases and decreases as the speeddecreases. This visual correlation was subsequently verified through regression analysis. The loadmaxima, on the other hand, were found to be a function of oil type and flow rate. In an interesting[221analysis of the load cell data, it was postulated that the variation of the maximum loads (as seenfor low carbon grades, in Figure 2.10) were a manifestation of binding. This conclusion was subsequently corroborated by an examination of surface of the billet cast during the test period.2.6 Mould-Strand Interaction and Billet QualityOf the fundamental processes taking place during the casting of steel, heat transfer andsolidification in the water-cooled mould are among the most important. It is recognised that a widevariety of casting problems, ranging from breakouts to shape defects and surface quality, are relateddirectly to events in the mould. Clearly then mould-strand interaction controls billet quality to alarge extent. Some of the common defects in billets that can be traced to adverse mould strandinteractions are discussed below.2.6.1 Oscillation MarksSeveral mechanisms have been proposed to explain the formation of oscillation marks inbillets. Conceptually these can be divided into the following two categories:(i) The solidifying shell at the meniscus “sticks” to the mould wall such that on the upstroke ofthe mould, the shell ruptures allowing liquid steel to partially fill the gap created. Subsequentlywith the mould moving downwards, there is a period of “healing” when the ruptured shellreforms [32,44-49).(ii) The billet mould distorts during operation so as to acquire a negative taper at the meniscus.During downstroke the distorted mould jams down on the solidified shell causing it to buckleleading to the formation of an oscillation mark [17] as shown in Figure 2.11 .This mechanismcan adequately explain the effect of several operating variables like mould cooling water flowrate, mould wall thickness and mould material on the depth of oscillation marks. Further,recent work [13] involving load cells has confirmed the interaction of the mould and the billetduring the negative-strip period. Strong support for this mechanism has also been seen in the[23]work of Stel et al. [27], who have observed a sudden decrease in mould acceleration in themiddle of negative strip period. The authors explain this by considering the interactionbetween a negatively tapered mould with the billet surface during the negative strip period.The obstruction of the mould movement by the strand causes a drop in the downwardacceleration of the mould.There is general consensus that factors enhancing meniscus solidification promote formationof oscillation marks and that the pitch of these marks can be obtained by dividing the casting speedwith mould oscillation frequency.2.6.2 Transverse Depression and Transverse Cracks.A mechanism for the formation of transverse depression has been proposed by Samarasekeraand Brimacombe [291 as shown in Figure 2.12. Here a schematic representation of a longitudinalsection of the billet in the mould is shown and, owing to inadequate shrinkage of the billet or/andexcessive taper of the mould, the billet binds in the mould and the solid shell is then subjected totwo opposing forces - a withdrawal force pulling it downward and a friction force, on account ofbinding, resisting withdrawal. Under these conditions, the solid shell behaves somewhat like aspecimen in a tensile test and forms a “neck” in the ductile section adjacent to the surface region.This “neck” is what is seen as a transverse depression on the surface of the billet. The less ductileregions of the shell, close to the solidification front, may break open forming a transverse crack atthe base of the depression.2.6.3 Billet Rhomboidity and Internal Cracks.Two of the most common mould-related quality problems encountered in billet casting arerhomboidity and longitudinal corner cracks as shown in Figures 2.13 and 2.14. Detailed researchby Samarasekera, Brimacombe and co-workers [18] has established that surface cracks and shapedefects are related. Indirect evidence of this also appears in work by other researchers who have[24]shown that when rhomboidity is reduced through corrective measures for adverse mould conditions,it also leads to a reduction in the severity of longitudinal corner cracks [50-53]. Further when thetwo defects occur together, the cracks tend to appear at the obtuse-angle corners of the rhomboidbillet [49,50]. When the cracks form in the absence of rhomboidity they are usually a result ofimproper corner radius [3,54] or mould distortion and wear [50,51].By analysing heat flux data obtained from industrial trials with several mathematical models,Samarasekera and Brimacombe [18] have shown that low mould cooling water velocities lead tointermittent boiling asynchronously on different mould faces. The result is that the different facesof the billet cool at unequal rates which causes nonuniform shrinkage and rhomboidity as the colderfaces contract more than hotter faces (which may even expand if the surfaces reheat). Further thesituation may be aggravated by the mould itself assuming a rhomboidal shape in response to theasynchronous nature of intermittent boiling of the mould cooling water.The dynamic billet rhomboidity may give rise to longitudinal cracks at the obtuse-anglecorners of the shell because a tensile strain acting parallel to the diagonal joining the acute-anglecorners is generated at the solidification front as shown in Figure 2.15. Tensile strains may also begenerated at the solidification front due to surface reheating if the obtuse-angle corner of the billetpulls away from the mould, thereby creating a locally wide corner gap and reducing heat flow tothe mould. Depending on the crack depth, the extent of reheating, and the magnitude of the tensilestrains generated by the ensuing shrinkage of the shell as it cools deeper in the mould, the crackmay penetrate to the surface and become a visible defect at the corner. In extreme cases this couldlead to a breakout.The proposed mechanism is consistent with several plant observations listed below.(i) The simultaneous occurrence of longitudinal corner cracks and billet rhomboidity.(ii) Change in the orientation of billet rhomboidity during a cast.[25](iii) The presence of longitudinal crack at the obtuse angle corners.(iv) The interdendritic nature of the crack [50,55,56].(v) Greater severity of these defects in steels containing 0.18% - 0.25% carbon [5 1,57]and higher carbon (>0.4%) grades [52,58].(High heat transfer rate during the casting of high-carbon grade billets is likely tocause the water to boil while the low ductility of steels with carbon content between0.18% and 0.25% may play an important role in the formation of cracks).(vi) The improvement in billet quality with reduction in mould cooling water flow ratesin high-carbon grades to levels lower than that maintained for lower carbon steels[58].(As the water velocity is reduced, boiling becomes more vigorous and less intermittent,such that cooling is more uniform around the periphery of the mould, and rhomboidconditions are less likely).(vii) The increase in severity of rhomboidity with decrease in section size [51].(Smaller size billets (100-130 mm square) are sometimes cast through moulds thathave wall thickness of 6-9.5 mm compared to 12.7 mm wall thickness for larger billets(150-180 mm square). Thinner mould walls would cause the water to boil leading torhomboidity in small billets).(viii) Reduction in rhomboidity by machining horizontal serrations on the outside surfaceof the mould wall in contact with the cooling water [57].(The roughness of the outside wall promotes sustained boiling and effectively eliminates the boiling hysteresis that triggers thermal cycling in the mould wall and causesintermittent boiling in the cooling channel).[26]The mechanism proposed above may be too crude to explain the important effect of the othervariables such as corner radius, superheat and casting speed on rhomboidity and longitudinal cornercracks. There is one common factor linking these variables viz., all affect corner shell thickness. Itis probable that a thinner corner shell (on account of increased casting speed, higher superheat andincreased corner radius) may simply be more susceptible to surface cracking in the zone of intermittent boiling because they are considerably hotter, and cracks can more easily propagate to thesurface.Off-squareness can be linked to the spray cooling as well. It has been observed by Bommarajuet al. [59] that the obtuse angle corners of off-square billets usually had the deepest oscillationmarks, and hence reduced local heat extraction in the mould, relative to other areas around theperiphery of the billet. Under such conditions these corners, emerging from the mould, would bethin and hot, as is often observed in the spray chambers of continuous casting machines, Crackscould form most easily adjacent to these corners owing to the locally weakened shell. On the otherhand, corners with shallow oscillation mark experience higher heat extraction in the mould andwould have thicker and cooler solid shell at the exit of the mould. Thus the solid shell profile ofsuch a billet at the bottom of the mould may be as shown in Figure 2.16. The off-squareness at theexit from the mould cannot exceed that of the mould itself but, once the billet reaches the spray,the cooling of the shell again would be non-uniform due to the varying shell thickness. The differential contraction can then cause the billet to assume a rhomboid shape.2.6.4 Off-Corner Internal CracksIn a study on mould behaviour and billet solidification, Bommaraju et al. [59] found thatinternal cracks were observed within 15 mm from any corner at a depth of 5 mm from the surfacein many of the transverse billet sections. The authors postulated that deep oscillation marks formin the off-corner regions and locally reduce heat extraction and shell growth. Towards the exit of[27]the mould where the mould-billet gap is large especially with an empirically designed single taper,bulging and subsequent hinging of a face or faces of the billet in the off-corner regions would takeplace. This would result in the generation of tensile strains at the solidification front in the off cornerregions leading to internal cracks as shown in Figure 2.17. Thus cracks can form and continuegrowing inwards following the solidification front, as it advances, as long as the strain is maintained.Based on this mechanism, off-corner cracks should appear at the sites that have the deepest oscillation marks as was found by the authors.Thus the achievement of uniform, shallow oscillation marks should reduce the incidence andseverity of off-corner cracks. Also measures to minimize shell bulging in the lower region of themould should decrease the cracking problem. Thus a properly tapered mould could have a beneficialeffect by reducing the mould-billet gap.2.6.5 PinholesDonaldson [601, in his examination into the quality of billets, found that pinholes occur morefrequently on the sides of the billets than at the corners and that the pinholes have a tendency toform in zones. Excessive lubricant has often been thought to promote the formation of pinholes.The oil pyrolyzes due to the elevated temperatures providing excess hydrogen responsible forcreating pinholes. Recently researchers have related pinholes to the oxygen content of the liquidsteel [27]. Other researchers, like Brown [611 have shown that the presence of excess moisture inthe ladle or tundish and high tapping temperature may cause pickup of the hydrogen and nitrogenwhich would also lead to pinholes.I 28jTable 2.1 Coefficient of friction at room temperature [24].Paraifin Paraflin Oil ÷ 1%Surfaces Clean Oil Laurie AcidNickel (17 0.3 0.28Chromium 0.4 0.3 0.3Platinum 1.2 0.28 (1.25 Non-reactiveSilver 1.4 0.8 0.70.9— 0.4Copper 1.4 0.3 0.08Cadmium 0.5 0.45 0.0,5Zinc 0.6 0.2 0.04Magnesium 0.6 0.5 0.08Iron 1.0 0.3 0.2 1Alununum 1.4 0.7 0.3 Lees reacüveTable 2.2 Effect of taper of the narrow face of a slab mould on the average heat flux in themould [36].Taper (%jm) 0 1.3 2.2 2.6Mould heat flux 15008 16704 19575(kcal/mmirz)Strand is sticking in the mould.1291HousingLiner iMouldU)001v1Figure 2.1 Schematic diagram of a typical mould used for continuous casting of steel billets.1.30]Figure 2.2 Schematic diagram of fatty acid molecules adhering to the solid surface [24].[3 1II—I LuCilcantBlI 54 mLfnn I IIo.12%C I Io.i2%c tto I I_____________I: P[/yh\fyd%\flflI I$ 1o 1 2 4Time (s) Tim. (s)I LLI11t 8II34mUnIn I0.1 2% C jYVA\/YVVV‘LttdcaNB’ ‘°24mimiin IL0.12%c I7 • 1513 7 IS• 2 2 4 5 1 2 3Time (s) Time (s)Figure 2.3 Load cell response at different flow rates of mould lubricant B during casting of low-carbon billets [131.[32]240g 160‘ Sulfw “H•at120SuHur e“ I 1H.at I 180 i40 jC00 .1Carbon, weight percentFigure 2.4 Effect of carbon content ot steel on mould fricition during continuous Casting of steelbillets. Data obtained using an “experimental” mould lubricated by Swift 1011 oil at 8.5 mI/mm[26].1.33]Figure 2.5 Specific heat extraction as a function of distance from the top of the mould for different lubricants [311.mm[34jI I- I—— —IUUCRS(D TUSE AND NO FLUX—INIdERSED TUSE AND FLUX—O.I0%CAReCNCASTING SPEED • 50 191000 -LEVELg $04—I”zI-.I-401—0v/I020( —M0 * i I I I0 2 4 4 4 10 12 14 11 iiO4STANCE FROM TO OF Id0LD inch.sFigure 2.6 Effect of mould flux on the heat-flux profile for low carbon steel 33].I 35}I I-______________________________________________LIQUID UVEL1200—0.40% CARSON• WITH FLUXI —0— WITHOUT FLU*IwSII—— X(FOR CAST WITHOUTFLUXI - 510 kId(hr)(ft3)X (FOR CAST WITHLUX)—417kSd(hrIIft2)2000 2 4*0 12 14 16 ii- OISTANE SELOW P OF HOLD. indFigure 2.7 Effect of mould flux on the heat-flux profile for high-carbon steel 1331.1361Empty Moukt LoadIng Profilea aLUI II I iI I II I,UII’: iaIll iII I hiI I I VII‘ I I II I II I II I I INeal A23412; Ca.bon: 0.09% I I II I IRun 40; L.dcanl A; W1gn1. I II ‘‘I I LH., ri )I III II I I II I f II I1 I II II I II I I IIHq’t.’JII I I)JIII i. ItI I II I II Ill I II. I I I I II iii Ilii I IiII I It— — ..—I— _J ____tj___If270t2Figure 2.8 A typical load cell profile [13].0-J‘I..H..I...ii..a..a..a..a..‘is‘finLoad Cell 3II:II II I! I3•i •iTIME (s)‘I,[37]Load Cell 4Load Cell 3H.a 824653; Carbon: 0.09% Run 44; LUbdCanI A; 24 mUndnI 4 I I a S S• • S • II • S S I SI — S SI II p aTIME (s)Figure 2.9 Change in the minimum load with casting speed [13].[38]12000C.= 0.05%11500-11000- iii i It10500 -10000-U)IIz09500-wz9000-8500-8000-7500 -7000-60 64 68 72 76 80 84TIME ELAPSED DURING CASTING (secs)Figure 2.10 Load cell response for a billet binding in the mould [13].[39]Mould 4 \ rimeDisplacement-________________Load Cell aResponseFormation ofOscillationMarkMould(a) (b) (c)Figure 2.11 Mechanism for the lormation of an oscillation mark due to interaction of the mouldwith the billet during negative strip period Ii1401FRICTIONAL FORCEBIIG MO I- -(HIGH FRJCrION)CR.ACIC INLOW DUCrIUTY REGIONOFSH&LNEcKINGOFDUCrUSHLTOmRM IDEPRESSION \\_____suEu. \ UQUID S1tH.-11*WITHDRAWAL FORCEFigure 2.12 Mechanism for the formation of transverse depressions and transverse cracks (not toscale) [29].[411Figure 2.13 Longitudinal corner cracks in continuously cast steel billets [18].[42]fFigure 2.14 Off-squareness in continuously cast steel billets [18].[43]Figure 2.15 Schematic diagram showing the formation of sub-surface crack on the diagonal atthe obtuse—angle corners of a off—square billet [1$].StrOmOtt-squarebd.t COntouiingott-corneran$ernol CracksUpperSproyt//////\\\\Figure 2.16 Schematic diagram showing billet with non—uniform shell thickness being distortedinto off—square shape by spray cooling [591.1451Mould wallFigure 2.17 Schematic diagram showing the generation of an internal crack due to bulging of thebillet shell in the mould and a hingin action in the off—corner region F59 I.Of f - cornerCold crackcornerI I I ItpressureFerrostotcLiquid poolSolid shell146]Chapter 3: SCOPE AND OBJECTIVES OF THE PRESENT WORKAs has been discussed in the previous chapters the genesis of quality problems of the billetlies in the mould. Though sketchy in details, it is clear from the chapter on literature review, thatresearch workers are in consensus that heat transfer and lubrication in the mould are the most crucialparameters that control billet quality.It is also clear that, owing to dissimilar shrinkage characteristics of different grades of steel,there cannot be a universal mould taper through which all grades of steel can be successfully cast.Furthermore, while the need for a lubricating oil is well established, little is known about theinfluence the various properties of oil have on lubrication and heat transfer.Considering the complicated nature of the continuous casting process and the interplay ofvarious components of the casting machine it is highly unlikely that any useful knowledge can begained from laboratory experiments. Clearly plant trials, in which operating moulds can beinstrumented by sensors, need to be carried out. It was felt that the measurement of mould temperature and mould-billet friction forces would reveal the effect of various variables on mould-billetinteraction and provide insight into design of mould taper for different grades of steelWith a view to ultimately understand variables that affect mould taper design and the role ofoil in mould lubrication, three plant trials in which operating moulds were instrumented bythermocouples and load cells, were organised. It was hoped that the analysis of the data collectedfrom trials would be able to fulfill the following objectives[1] To determine the mechanism by which heat is extracted in the mould as well as determinethe heat extraction rate for a range of steel grades.[2) To modify and use a mathematical model of the mould wall capable of calculating heat-fluxprofile down the mould wall.[471[3] To develop a mathematical model to simulate the solidification of steel and its shrinkageas a function of its position in the mould.[4] To develop a computer program to analyse load cell signals as a function of mould displacement.[5] To ultimately specify oilflow rates and mould tapers necessary to successfully cast a widevariety of steel grades.[6] To link sensor signals to formation of defects in the billet.[48]Chapter 4: EXPERIMENTALPlant trials were conducted at three plants identified as Companies B, C and E. In order tosystematically study the effects of various operating parameters, it was decided to collect mouldtemperature and mould-billet friction data for different grades of steels. For each grade of steel,data was collected at least at three different flow rates of the mould lubricating oil. A total of sixdifferent types of oils were used - four vegetable based oils and two mineral oils. The oils werechosen so as to give a range of boiling points, eg. the mineral oils have low boiling point (< 200°C) while the boiling points of vegetable oils are higher. In order to asses the impact, if any, of thefatty acid content of the oil a High Erucic Acid oil (HEAR), was used. Data could be collected forbillets with carbon contents ranging from 0.05% to 0.80%. For some period of the trial, the oscillationfrequency of the mould was changed from the one normally used. The details of the trials and theequipments used to collect data are discussed below.4.1 Pre-Trial PreparationsIn preparation for the trial the condition of the mould system was determined at each of thethree plants. To this end, a series of checks were made on the oscillator, mould water flow, mouldoil distribution system and mould design. Additionally the three companies were responsible forchecking the machine alignment and monitoring the quality of the mould water carefully to ensurethat it met desired standards.4.1.1 Retrofit of mould housingAt Company B and C, the original mould constraint system had consisted essentially of twokeeper plates which fitted into slots machined into the straight sides of the mould wall. This is thesame system that has been extensively studied and found to result in non-uniform distortion aroundthe mould periphery and subsequent billet quality problem [621, particularly if a shallow metal level[491(<100 mm) is maintained. Therefore, the constraint system was changed to one incorporatingtight-toleranced four-sided constraints, known to be superior from a billet-quality stand point [63].This required the manufacturing of new split plates and the cutting of new slots on the four mouldwall faces. The tolerances achieved by custom machining to match the split plates to mould tubeslots was less than 0.076 mm. Similar work had to done on the top plates at Company E.A new oil plate was also manufactured, prior to the trials, based on the UBC Oil DistributionSystem [64]. This system ensures uniform oil flow on all faces of the mould wall. Finally, recesseswere machined into the mould housing to accommodate the load cells and several holes were drilledand tapped in the housing wall for thermocouple wires to pass through.4.2 Measurement of Mould Wall TemperatureBrimacombe, Samarasekera and co-workers have, for the last two decades, successfullyinstrumented operating moulds with thermocouples to measure mould wall temperatures[8,13,17,191. The same time-tested technique of thermocouple installation was adopted in thepresent work.The installation procedure, in its simplest form, consists of drilling holes through the steelbaffle tube and the copper mould wall such that when the thermocouple wire is inserted, the tip ofthe thermocouple rests approximately mid-way between the hot and the cold faces of the mould.Care is taken to ensure that all the holes that are drilled are at the mid-face of the mould and to thesame depth in the mould wall. A flat bottom drill is used to flatten the drilled hole which is thentapped using a bottom tap to ensure threading to the bottom of the hole. The hole depth is measuredand recorded.Single wire, Type T (Copper-Constantan) intrinsic thermocouple, is used to measure mouldwall temperature. A bead is created on the Constantan (55% Cu- 45% Ni) thermocouple wire(diameter = 0.81 mm) by using a TIG welding machine. The bead is filed to produce a flat foot like[50]end approximately 0.30-0.40 mm thick. Heat shrinkable tube (1.6 mm in diameter) is then shrunkonto the bare Constantan wire. This wire is then inserted through the baffle into the mould wall andheld in place by a threaded copper plug screwed into the copper mould. Use of silicone sealentensures a water tight fit. Shielded copper wires are joined with the Constantan wire in the waterchamber and the former is then connected to the Data Acquisition System by bringing out the wiresthrough holes cut in the mould housing. The mould is pressure tested to ensure that there are nowater leaks. To monitor the bulk inlet, the bulk outlet and the outlet water temperature at each faceof the mould two wire, type T thermocouple were used. The temperature of the data acquisition orjunction box (cold junction temperature) was measured by a mercury-in-glass thermometer. Aschematic diagram of the set up is shown in Figure 4.1. All thermocouple wires were tested forcontinuity and calibrated.Type T thermocouple can be used to measure temperature from -270 °C to 400 °C and theuse of an inthnsic thermocouple ensures that the arrangement has a low thermal inertia. The timetaken by a 1 mm Constantan wire on a copper substrate to reach 95% of the steady state e.m.f is ofthe order of micro-seconds [65].4.3 Measurement of Mould-Billet Friction ForcesBrendzy in her thesis [13] has described in detail the installation procedure for the load cellsfor measurement of mould-billet friction forces and the same method was followed.The load cell is a transducer that converts a load acting on it into an analog electrical signal.This conversion is achieved by the physical deformation of strain gauges which are bonded to theload cell button and wired in a Wheatstone bridge configuration. Weight applied to the load cellthrough compression produces a deflection of the button which introduces strain to the gauges. Thestrain produces an electrical resistance change proportional to the load. The load cells are of theLCG series (of OMEGA) and capable of withstanding 44.5 KN of compressive loading. The cells[51]are 38.1 mm in diameter and have a total height of 15.8 mm and are shown in Figure 4.2. Manufacturers specifications indicate that these can handle 150% of the full scale load, operate accuratelyin temperatures up to 121 °C and a have repeatability value of +1- 0.05% of full scale (- 22.7 N).The load cells were calibrated prior to use on an Instron machine. The 10 volts power supplynecessary for the excitation of the load cell was obtained by connecting three 6 volts batteries inseries. The resulting 18 volts supply was stepped down to 10 volts via a self compensating voltagecircuit designed to maintain a constant output voltage. Filters were placed in the circuit to eliminateany noise pick up in the 10 volts line.The load cells were positioned between the mould housing and the mould oscillator table asshown in Figures 4.3 and 4.4. Four circular recesses were machined into the mould housing plateinto each of which a load cell was placed. Due to the coarse threads on the hold down bolts, whichconnect the housing to the oscillating table, a bolt spring assembly was designed to control theinitial torquing load on the load cell. As the mould housing was lowered onto the oscillating table,the load cell positioner was turned until the load cell button contacted the oscillating table. Thebolts were then tightened until the spring became fully compressed.4.4 Measurement of Mould DisplacementTo record mould displacement, linear variable differential transformers (also called linearvariable displacement transducers) were used.The LVDT is an electromechanical device that produces an electrical output proportional tothe displacement of a separate moveable core. It consists of a primary coil and two secondary coilssymmetrically spaced on a cylindrical frame as shown in Figure 4.5. A free-moving, rod-shapedmagnetic core inside the coil assembly provides a path for the magnetic flux linking the coils. Whenthe primary coil is energized by an external ac source, voltages are induced in the two secondarycoils. These are connected series opposing so that the voltages are of opposite polarity. Therefore,[52]the net output of the transducers is the difference between these voltages, which is zero when thecore is at the centre or null position. When the core is moved from the null position, the inducedvoltage in the coil towards which the core is moved increases, while the induced voltage in theopposite coil decreases. This action produces a differential voltage output that varies linearly withchanges in core position.The LVDTs’ are attached to the housing in such a way that they register the movement ofthe mould. Since it is not necessary to know the absolute displacement of the mould, the LVDTs’do not have to be calibrated on the oscillator. It is only necessary to ensure that the full rangemovement of the LVDT, when in use, is within the linear output range of the circuitry.Two Daytronic (model 3130) signal conditioners were used to operate the LVDTs. TheLVDTs’ were calibrated to provide an output of 2.5 volts per 10 mm displacement.4.5 Other Miscellaneous Measurements4.5.1 Casting SpeedThe signal from the withdrawal roll tachometer which is proportional to the casting speed,was used to record the casting speed. The signal from the tachometer is typically 0-40 or 0-200volts D.C and was stepped down (0-20 millivolts) and filtered for any electrical noise ( 8 or 45 Hz)before being sent to the data acquisition system. The calibration of the signal was done on site bymeasuring the output voltage relative to the casting speed gauge and adjusting a variable resistorso as to produce a convenient ratio (e.g 10 my = 100 ipm).4.5.2 Metal LevelThe 4-20 milliampere signal from the metal level controller was modified to produce a 0-10millivolt output before it was connected to the data acquisition setup. This signal was calibrated bylowering a steel billet into the mould and measuring the output signal for different lengths of thetest billet in the mould.t5314.5.3 Internal Dimensions of the MouldTo measure the internal mould dimensions (taper trace) of a continuous casting billet mouldan apparatus consisting of 3 LVDTs and a 10 turn potentiometer are assembled and calibrated insuch a way as to enable a constant measurement of the mould profile as the equipment is passedthrough the mould. Taper traces are thus obtained for three locations on each mould wall. Theaccuracy of the measurements are +1- 0.024 mm.4.5.4 Filming of the steel surfaceThe meniscus region of the steel surface was filmed by a hand held camcorder focussed onthe steel surface in the mould. It was thus possible to see and record the motion of the oil down themould wall.4.6 Data Acquisition4.6.1 EXP-16Metrabyte’s Universal Expansion Interface, Model No EXP- 16, is an expansion multiplexer/ amplifier system as shown in Figure 4.6 that can be used with any data acquisition system.Each EXP-16 concentrates 16 differential analog input channels into one analog outputchannel and also provides signal amplification, filtering and conditioning. Additionally, theinstrumentation amplifier provides gains of 0.5, 1, 2, 10, 50, 100 as well as programmable gaincapability. The 16 differential input channels are selected by a solid state 4 bit TTL/CMOS compatible address. Provision is made on the board for filtering, attenuation and measuring currentinstead of voltage. All analog input connections are conveniently made on miniature screw connectorstrips. Cold-junction sensing and compensation circuitry as well as a biasing resistor for openthermocouple detection also exists in the system. The EXP-16 can be connected directly to[54]MetraByte’s DAS-8 or sets of EXP- 16s can be cascaded by identical cables to a total of 128 channels(16x8 = 128) of standard voltage, 112 (16x7= 112) of thermocouple measurement. When usedwith DAS-8, channel selection is via the OPI-4 digital outputs of the DAS-8.4.6.2 DAS-8MetraByte’s DAS-8 is an 8 channel, 12 bit high speed, A/D converter and timer/counterboard, shown in Figure 4.7, for the IBM PC. The DAS-8 board is 5 inches long and can be fittedinto a “half-slot” of a PC. All connections are made through a standard 37 pin D male connectorthat projects through the rear of the computer.DAS-8 is a successive approximation AID converter with sample/hold. The full scale inputof each channel is ±5 volts with a resolution of 2.44 millivolts and the inputs are single ended witha common ground. The A/D conversion time is typically 25 microsecond (maximum 35 microsecond). The 8254 programmable counter timer which provides periodic interrupts for the AIDconverter has 3 separate 16 bit down counter one of which is connected to the sub-multiple of thesystem clock and all the 1/0 functions of the remaining two are accessible to the user. Input frequencies of up to 2.5 MHz can be handled by the 8254. The 7 bits of TTL digital I/O provided arecomposed of one output port of 4 bits and one input port of 3 bits. Each output handles 5 standardTTL loads ( 8 mA sink current). A precision +10.00 v (.1v) reference voltage output is derivedfrom the A/D converter reference. This output can source/sink 2 mA.4.6.3 GeneralTo ensure that ground loops are not created all signals were grounded at the source (mould).The shielding was also done appropriately by grounding all shields at one point (mould). Theinstrument end of the data acquisition setup was “floated” by the use of an isolation transformer.Electrical noise was inadvertently introduced in signals from the sensors at Company E making thetask of data analysis extremely difficult, Figure 4.8 is a schematic diagram showing the main[551components of the data acquisition set up.The computer based data acquisition system by controlled by Labtech Notebook Softwarefrom Laboratory Technologies Corporation, USA. The signals from the load cells and the LVDT’salong with signals from a few selected thermocouples were sampled at 50 Hz for 120 seconds;thermocouple signals were collected at 1 Hz. for 600 seconds (10 mm) and at 30Hz. for 100 seconds.After the completion of the trial all thermocouple wires were tested to identify those that hadfailed during operation.4.7 Details of the TrialsThis section sumniarises in the fom of several tables the different conditions under whichthe data was collected at the three plants.4.7.1 Casting machines and casting practiceThe details of the casting machines and the moulds used in the control and test strands at eachof the three Plants are shown in Tables 4.1 and 4.2. As can be seen in Table 4.1, the moulds atCompanies B & C were square while the mould at Company E was rectangular. Shrouding of thesteel stream between the mould and the tundish with nitrogen gas was carried Out at all threecompanies to minimize oxygen pick up by the steel steam. Such a shrouding practice also decreasesthe chances of the mould lubricating oil burning on the mould wall above the meniscus. The negativestrip time for the companies varied, undei sandard oscillation frequency between 0.13- 0. 16seconds. Table 4.2 is a comparison of the control and test strands at each of the three companies.The important features are the following[a] The test mould at Company B & E had steep upper tapers while the test mould at CompanyC was almost untapered near the meniscus.[b] All moulds except for the one used in the control strand at Company C were an alloy ofCu-Cr-Zr.[56J{c] The mould cooling water flow rate was in excess of 10 m/s in all three companies[d] The test moulds at all three Companies were equipped with the UBC design oil distributionsystem.4.7.2 Oils used in the trialsThe properties of the various lubricating oils used in the trials are summarised in Table 4.3.As can be seen, the temperature at which 20% of the oil boils off is the lowest for the mineral oilMineral_S (230 °C), and approximately 280 °C for the other oils. In addition HEAR oil has 45.2%of Erucic fatty acid which is well in excess of the value for the Erucic acid content of the other oils.4.7.3 Thermocouple arrangementsThe arrangement of the thermocouples in terms of their depth and axial position on the mouldwall are given in Tables 4.4, 4.5 and 4.6 for Companies B, E & C respectively. At Company Bthermocouples were installed on the inner curved wall (ICW), the right side wall (RSW; at bothcentre and off-centre locations) and the outer curved wall (OCW). At Company E & C thermocoupleswere mounted on the inner curved wall and the right side wall only. Additionally at Company E,five thermocouples (6 A,B,C,D,E) were placed at the meniscus level from one end to the other ofthe right side wall.4.7.4 Chemical compositions and casting conditions of different heatsThe chemical compositions of the heats monitored at the three plants are tabulated in Tables4.7, 4.8 and 4.9 while the casting conditions for the same heats are shown in Tables 4.10, 4.11 and4.12 for Companies B, E and C respectively.At Company B (Table 4.7), data was collected for billets with carbon contents between 0.046%to 0.42%; at Company E from 0.17% to 0.89% (Table 4.8) and at Company C, the carbon contentsof the billets were between 0.19% to 0.86% (Table 4.9).[571At Company B data was collected, as shown in Table 4.10, with flow rates of oil varyingbetween 24- 53 mi/mm. As shown in Table 4.11, at Company E, the oil flow rate was variedbetween Oml/min (no oil) to 110 ml/min. Additionally the oscillation frequency of the mould waschanged form the normal value of 170 cpm to 130 cpm. At Company C, the frequency of mouldoscillation was changed, for selected heats, from 144 cpm to 96 cpm. Additionally, data was acquiredfor flow rates of oil varying between Oml/min to 100 ml/min. To clearly identify the effect of oilflow rate on heat transfer, the flow rate of Canola oil was changed from 0 mi/mm to 100 mI/mmin the middle of a data collection sequence. Table 4.13 summarizes the different types of oil andthe flow rates at which they were used at the three plants.4.8 Laboratory WorkThe billet samples collected during the plant trials were subjected to a rigorous inspectionprocedure which is outlined in the form of a flow chart in Figure 4.9.Sulphur printing of transverse and longitudinal sections was carried out after the surfaceswere ground, washed with soap and water and then dried. The printing was done on resin coatedphotographic paper which had been immersed for 3 to 4 minutes in a 4% sulphuric acid solutionin water. The soaked paper was placed on the section, emulsion side down and rolled to ensuregood contact and left there for approximately 4 minutes. The paper was removed, labeled, washedin water to remove the sulphuric acid, fixed and dried. Care was taken to orient the billet sectionwith the ICW at the top of the paper.Macro-etching was done after the completion of the sulphur printing. A solution of 50% HCLand 50% water was heated to 85 °C for around thirty minutes and then scrubbed to remove the blackoxide. The surface was covered with glycerin to prevent rusting, photographed, washed in warmwater, dried with alcohol and sprayed with a clear lacquer.58]When the subsurface structure was to be examined an appropriately prepared sample wasetched with a solution of 5% Picric acid in water at 80-85 °C.The various kinds of inspections carried out can be broken into the following broad categories:Dimensional checks Internal inspection Surface Inspection(i) Distance between opposite (i) Internal cracks (i) Profilometer measurementsfaces (ii) Inclusions of oscillation-mark depth(ii) Off squareness (iii) Porosity (ii) Longitudinal and transverse(iv) Cast structure depressions(iii) Surface cracks(iv) Bleeds, laps, pinholes andother surface imperfections4.9 Analysis of Mould Temperature Measurement4.9.1 Conversion of thermocouple measurement to mould wall temperaturesIf the slope of the thermocouple voltage output versus the temperature curve (the Seebeckcoefficient) is plotted against temperature it becomes quite obvious that the thermocouple is anon-linear device. It is thus necessary to fit a polynomial, as shown below, to convert the thermocouple voltage to temperature.T=a0+a1x23 +a/where T= temperature, x = thermocouple voltage, a0,a1 a are coefficients unique tothe type of the thermocouple and n is the order of the polynomial. The coefficients are availablefrom Omega Temperature Measurement Handbook and Encyclopedia which notes that the accuracyfor the type T thermocouple is +1-0.5°C if a 7t1 orderpolynomial is used. During the actual calculation[591ofthepolynomial an alternative form of it, as suggested by Homer, is used to speed up the calculation.This form, shown below, converts the time consuming exponentiation operation to one involvingonly multiplication.T = a0 +x(a1 +x(a2+x(a3+x(a4+x(a5+x(a6+a7x))))))A schematic diagram of the temperature measuring set up and the equivalent circuit to whichit can be reduced is shown in the Figure 4.10. It needs to be noted that a second thermocouple iscreated by the junction of the Copper and Constantan wire in the water chamber and, as shown inthe circuit, the voltage generated at this junction (V2) opposes the voltage being measured (Vi).The measured voltage (VM) thus needs to be increased by V2 to get Vi. The value of the opposingvoltage (V2) is obtained by the two wire thermocouples used to measure the water temperature.An important point here is that the junction referred to above lies in the outlet water for the mouldthermocouples that are above the plenum divider, and the inlet water for those thermocouples thatare below the plenum. The reference junction temperature is, as mentioned before, measured by amercury-in-glass thermometer place at the appropriate location. In keeping with the correct procedure, the reference junction temperature is converted to a voltage, added to the sum of VM andVi and the resultant voltage is reconverted to a temperature value.4.9.2 Data Filtration techniqueFigure 4.11 is a flowchart showing the different stages through which the analysis of mouldthermocouple data proceeds. The filtered thermocouple response is input to a mathematical modelof the mould to obtain the axial heat-flux profile down the length of the mould and the temperaturedistribution in the mould wall. The latter is used in a elasto-plastic model of the mould to computethe distortion of the mould wall while the heat-flux profile is used in a heat flow model of the mouldto calculate billet shrinkage profile and shell thickness. The details of the models are discussed inthe next few chapters.[601A typical (unfiltered) mould-thermocouple response is shown in Figure 4.12. The fluctuationsin the temperature are believed to be caused largely by metal level variation in the mould due, inpart, to casting speed change [17]. This is confirmed in Figure 4.13 where an increase in the metallevel (metal level drops lower in the mould), is seen to cause a drop in the temperature sensed bythe meniscus thermocouple. Clearly, such effects have to be isolated from the variation in temperature caused by strand-mould interaction and associated gap changes and the method of data-filtering has been explained in an earlier publication [17]. The first step is the identification of thethermocouple just above the meniscus (for Company B, it was found to be the second thermocouple,TC2, on the inner curved wall) at which signal fluctuations are predominantly due to metal levelchanges rather than strand-mould interaction. Each temperature recorded by TC2 corresponds to aparticular metal level and, therefore, by isolating data (for all thermocouples) for only those timeperiods during which the temperature of TC2 is within a narrow range, it is possible to select datacorresponding to a fixed metal level. With the availability of the casting speed signal it was alsopossible to further refine this process of data filtering by extracting data for time periods whentemperatures recorded by TC2, as well as the casting speed, were both within a narrow range. It isthought that this stringent criteria of data extraction better reflects the effect of gap width on thethermocouple response. The extracted data was time averaged for each thermocouple to givetemperature profiles for different heats.To maintain consistency in data extraction from heat to heat the reference temperature andcasting speed range were selected to be within 5 C° and 2 mm/s of the mean value recorded by TC2and the casting speed sensor respectively.4.10 Analysis of Load Cell ResponseSince the load cell signal are sampled at 50 Hz. an enormous amount of data is collected evenin a short period of 10 seconds. To be able to analyze the load signal correctly it is necessary to[611plot the same on a paper 1524 mm by 1224 mm- a size that cannot be handled by conventionalplotters. Thus a special plotter was acquired to carry out the plotting and a software to do the samewas developed. Also plotted alongside were the casting speed and metal level signals. With plotsof signals from LVDTS, load cells and other related sensors all available on the same page, analysisof mould billet interaction could be carried out conveniently.Consider Figure 4.14 which shows the various components of the mould oscillation. Asexplained by Brendzy [13], during the negative strip period of the down stroke the mould pusheson the billet and the billet pushes back thereby reducing the load sensed by the load cell. The shapeof the load cell signal during the upstroke of the mould is important from the stand point of mouldfriction. It is imperative to break up the load cell signal into pre and post negative strip periods andanalyse the load changes in those periods. A computer programme to do so was developed thatidentifies the beginning and end of negative strip periods in each oscillation cycle and calculatesload changes in these periods. This calculation is carried out for the entire period for which the datahas been collected. The output of the programme was so designed as to enable quick and easycomparison of load cell signals under different conditions. In particular the programme quantifiesthe following:(1) The difference in load at the beginning and end of negative strip period.(2) The maximum amount of decompression that the load cells experience during the negative stripperiod. This is the difference in the load at the start of the negative strip time and the minimum loadattained during negative strip period.(3) The percentage of the negative strip time for which the decompression of the load cell occurs.This is the time elapsed between start of negative strip period and the attainment of minimum load,expressed as a percentage of the negative strip period.(4) The difference in the maximum load during the upstroke and the load at the beginning of the[621negative strip time.(1)- (3) are different ways of quantifying mould-billet interaction during the negative stripperiod while (4) is an indicator of friction during the upstroke of the mould. On account ofdifferencesin pre-loads of the load cells as well as the distribution of load between the bolts, 0-rings, springand the load cells at the three plants, it is not possible to compare absolute values of load acrosstrials. This is overcome by the use of parameter (3) which only compares the time for which theload cell is decompressed.[63]Table 4.1 Details of the casting practice at Plants B, E, and C.COMPANY B COMPANY E COMPANY CMachine type Curved mould Curved mould Curved mouldMachineradius 7.9m(26’) 3.9m(13’) 7.9m(26’)MouldLength 812.8 mm 812 mm 835 mmHeat size 60 tonne 60 tonne 150 tonneSequence Yes Yes YesBillet size 120 x 120 mm 127 x 178 mm 140 x 140mm(4.7 x 4.7) (5 x 7”) (5.5 x 5.5”)Nominal casting speed 2 - 2.5 m/min 1.65 rn/mm 1.90 rn/mm(80 - 100 ipm) (65 ipm) (75 ipm)Reoxidation protection N2 shrouding N2 shrouding N2 shroudingOscillation type Sinusoidal Sinusoidal SinusoidalStroke length 9.5 mm (0.37”) 6.4mm (0.25”) 11.2mm (0.44”)Oscillation frequency 2 Hz. 2.8 & 2.2 Hz. 2.4 & 1,6 Hz.(120 cpm) (170 & 130 cpm) (144 & 96 cpm)Negative strip time 0.15 s 0.12 & 0.13 s 0.16 & 0.19 sMouldlead 2.1mm 2.2&1.3mm 5.3&3.Omrn[64]Table 4.2 Details of the test and control strands used in the trials at Plants B, E and C.Features COMPANY B COMPANY E COMPANY CTest Control Test Control Test Control(#3) (#2) (#4) (#6) (#4) (#5)Material Cu-Cr-Zr Cu-Cr-Zr Cu-Cr-Zr Cu-Cr-Zr Cu-Cr-Zr DHP CopperThickness 12.7 mm 12.7 mm 19 mm 12.7 mm 16 mm 12.7 mmCorner Radius -- 3.2 mm 3.2 mm 3.2 mm 4.8 mmConstruction Tube Tube Reformed Reformed Tube TubeTube TubeMouldLength 812.8 mm 812.8mm 812mm 812mm 835 mm 835 mmMould Taper Parabolic Parabolic Double Triple Multiple Double(4.9 at top (4.9 at top (2.7 &, 0.8 (3.6, (0.4, 2.5, (1.4, 0.6and 0.8 and 0.8 %/m) 0.9,0.55 %/m etc.) %/m)%/m near %/m near %/m)end) end)Water Gap 4.76 mm 4.76 mm 3.2 mm 4.8 mm 3.2 mm 6.3 mmWater Flow rate 44 1/s 44 1/s 28 1/s 31 1/s11 rn/s 18 rn/s 12 m/s 12.4 ni/s rn/sConstraint Type Four-sided Two-sided Four-sided Two-sided Four-sided Two-sidedDistribution Sys- UBC Conventio UBC Conventio UBC Conventiotem for Oil design nal design nal design design nalOil Channel Vol 167 ml 120 ml 200 ml 140 ml 280 ml 90 mlOil Gap 0.41 mm 0.76 mm 0.38 mm 0.35 mm 0.38 mm 0.51 mmGasket Type Cross Conventio Cross Conventio Cross Conventional nal nalNo of Oil Lines 1 1 1 1 1 2Gap Protector No No Yes No Yes No[65]Table 4.3 Property of various lubricating oils used in the trials at the three Plants.Canola HEAR Mineral_S Mineral_O Soybean 51-LNType of Oil Vegetable Vegetable Mineral Mineral Vegetable VegetableViscosity 160 185 200 250 160 N.A(SUS) @38 °CFlash Point (oC) >315 >300 226 >320 327 227Fire Point (°C) >360 >350 252 N.A 343 N.ABoiling Point (°C)Start 205 215 170 205 180 20520% 280 280 230 270 275 30050% 315 320 300 315 320 33590% 335 335 330 335 335 350Fatty Acid ContentsPalmitic (c16:0) 5.3% 3.2% 2.9% 5.1% 9.9% 2.1%Oleic (c18:1w9) 57.7% 14.9% 40.3% 22.1% 18.1% 47.3%Linoleic (c18:2w6) 23.6% 15.1% 27.1% 65.5% 55.9% 24.6%Linolenic (c18:3w3) 9.2% 9.3% 26.0% 2.7% 11.3% 18.3%Eicosenoic 0.7% 8.2% 0.3% 0.3% 0.4% 0.6%(c20: 1w9)Erucic (c22:lwll) 0.3% 45.2% 0.1% 0.1% 0.1% 0.1%[66]Table 4.4 Depth and axial position of thermocouples used to monitor mould wall temperatureat Company B.T.C. No. Distance from Hole Depth T.C. No. Distance from Hole Depthmould top (mm) mould top (mm)(mm) (mm)ICW - Centreline Right Side- Centreline1 85 7.44 1 85 6.172 100 5.87 2 100 6.123 114 5.77 3 116 6.104 130 5.715 4 131 6.155 146.5 5.61 5 145 6.106 170 5.715 6 170.5 5.947 195 5.84 7 196 5.848 220 6.02 8 221 5.899 243 6.35 9 246 5.8710 312 6.25 10 313 5.8911 342 5.89 11 343 6.0712 372 5.77 12 373 5.9213 402 5.64 13 403 6.0214 452 5.79 14 453 6.0715 505 6.35 15 503 5.9716 552.5 5.59 16 553 5.9217 603 5.74 17 603 6.0918 712 5.69 18 703 6.35Right Side - Off-centre OCW - Centreline1 84 6.12 1 86 5.842 100 6.17 2 101 5.263 115 6.12 3 115 5.494 130 6.12 4 130 5.495 145 5.92 5 146 5.616 170 5.99 6 170 5.697 196 5.97 7 196 5.728 220 5.89 8 221 5.799 246 5.82 9 244 5.7210 312 5.87 10 312 5.7411 342 5.97 11 343 5.6912 372 5.87 12 372 5.6613 402 6.04 13 402 5.6914 452 5.92 14 453 5.5215 502.5 5.99 15 502.5 5.7416 552.5 6.10 16 553 5.9217 602 6.10 17 603 5.8718 713 6.30 18 713 5.82[671Table 4.5 Depth and axial position of thermocouples used to monitor mould wall temperatureat Company E.T.C. No. Distance from Hole Depth T.C. No. Distance from Hole Depthmould top (mm) mould top (mm)(mm) (mm)ICW - Centreline Right Side- Centreline1 85 9.90 1 84.4 10.002 99 9.85 2 99 10.153 114 10.00 3 115 10.104 129 10.00 4 130 10.005 144 9.85 5 145 10.006A 159 9.90 6A 160 9.80B 159 9.65 B 160 9.90C 158 9.90 C 159 9.80D 158 9.80 D 158 9.90E 159 9.75 E 158 10.257 188 9.80 7 190 9.608 218 9.70 8 218 9.509 248 9.75 9 250 9.3510 277 9.70 10 277 9.4511 327 10.10 11 327 9.1512 358 9.95 12 358 9.9013 388 9.85 13 390 9.2014 418 9.90 14 419 8.8015 448 9.60 15 448 8.7516 479 10.20 16 478 9.8017 509 10.20 17 509 9.6518 569 10.00 18 569 9.6519 629 10.15 19 629 9.6520 690 9.95 20 690 9.8521 750 9.45 21 750 9.90[68]Table 4.6 Depth and axial position of thermocouples used to monitor mould wall temperatureat Company C.T.C. No. Distance from Hole Depth T.C. No. Distance from Hole Depthmould top (mm) mould top (mm)(mm) (mm)ICW - Centreline Right Side - Centreline1 67 8.0 1 65 7.752 85 7.9 2 80 7.83 101 7.7 3 100 7.84 116 7.9 4 116 7.85 131 7.9 5 129 7.86 146 7.8 6 144 7.87 162 7.9 7 160 7.78 177 7.9 8 174 7.89 191 7.9 9 189 7.7510 216 7.85 10 214 7.7511 241 7.9 11 299 7.812 300 7.9 12 318 7.8513 320 7.9 13 349 7.814 350 7.8 14 378 7.5515 381 7.8 15 409 7.6516 411 7.9 16 438 7.817 441 7.8 17 468 7.918 471 7.9 18 499 7.6519 501 7.8 19 549 7.6520 551 7.9 20 609 7.721 602 7.8 21 649 7.7522 652 7.9 22 699 7.823 702 8.1 23 749 7.824 751 8.1[69]Table 4.7 Chemical compositions of the heats monitored at Company B.Heat Grad C Mn S P Si Cu Cr Ni Mo Nb V Sn Pb Zn AlNo. e (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)B24636 1008 .046 .530 .038 .012 .160 .110 .070 .080 .020 - - .006 -- .004B24637 1008 .040 .420 .039 .017 .120 .090 .070 .080 .020 - - .006 -- .004B24368 1008 .068 .460 .026 .009 .140 .090 .040 .060 .020 -- .005 -- .003B24369 1008 .045 .360 .022 .009 .080 .100 .030 .080 .020 -- .006 -- .003B24640 1008 .035 .370 .026 .011 .090 .120 .040 .070 .020 -- .006 - - .004B24641 1008 .041 .480 .020 .008 .090 .100 .030 .070 .020 -- .005 - - .004B24642 1008 .043 .440 .020 .008 .120 .090 .030 .070 .020 -- .005 -- .003B24643 1018 .170 .830 .018 .018 .200 .100 .080 .080 .020 - - .006 -- .004B24644 1018 .180 .750 .013 .010 .240 .170 .070 .100 .020 - - .010 -- .005B24645 1018 .180 .850 .018 .018 .280 .140 .070 .090 .020 - - .009 - - .005B24646 1039 .400 .780 .015 .014 .240 .130 .100 .100 .020 - - .008 - - .005B24647 1039 .420 .810 .019 .012 .210 .130 .100 .090 .020 -- .008 - - .004A23408 1008 .051 .350 .021 .014 .100 .110 .040 .090 .020 - - .007 - - .004B24649 1008 .050 .400 .020 .011 .110 .090 .070 .090 .010 - - .007 - - .005A23409 1015 .150 .350 .026 .017 .100 .270 .070 .110 .020 - - .010 - - .004B24650 1010 .120 .470 .018 .013 .100 .210 .050 .100 .020 - - .011 - - .004A23412 1010 .090 .410 .025 .014 .120 .350 .060 .100 .020 - - .011 - - .004B24653 1010 .094 .460 .020 .010 .110 .190 .040 .080 .020 - - .009 - - .004B24654 1012 .120 .400 .022 .008 .100 .240 .050 .100 .020 - - .010 - - .004A23413 1012 .120 .420 .020 .008 .120 .230 .040 .100 .020 - - .009 - - .004B24655 1010 .110 .350 .023 .007 .080 .240 .040 .110 .020 - - .012 - - .004B24657 4037 .400 .730 .018 .011 .240 .110 .070 .060 .021 - - .005 - - .004[701Table 4.8 Chemical compositions of the heats monitored at Company E.Heat Grad C Mn S P Si Cu Cr Ni Mo Nb V Sn Pb Zn AlNo. e (%) (%) (%) (%) (%) (%) (%) (%) (cf,) (%) (%) (9,) (%) (%) (%)26493 5160 .57 .79 .025 .010 .23 .09 .78 .05 .012 .003 .029 .007 .005 .002 .00326494 5160 .57 .78 .029 .012 .22 .07 .77 .05 .011 .002 .026 .006 .003 .001 .00326495 5160 .57 .82 .029 .008 .22 .08 .78 .06 .014 .000 .024 .006 .003 .002 .00326501 1018 .21 .73 .025 .023 .23 .10 .08 .06 .011 .002 .002 .008 .004 .011 .00326503 1018 .19 .85 .018 .007 .25 .11 .06 .05 .013 .002 .003 .007 .005 .010 .00326504 1018 .17 .83 .021 .007 .24 .11 .07 .06 .014 .002 .002 .007 .007 .009 .00326505 1018 .17 .79 .18 .010 .24 .11 .07 .06 .013 .002 .003 .008 .005 .007 .00326507 1018 .18 .85 .024 .015 .19 .11 .14 .06 .023 .003 .004 .008 .006 .010 .00326508 1146 .45 .85 .098 .008 .23 .12 .08 .06 .018 .003 .023 .008 .005 .006 .00326509 1146 .44 .79 .101 .009 .21 .11 .09 .06 .012 .002 .021 .008 .004 .002 .00226510 1090 .87 .95 .025 .014 .46 .11 .11 .06 .014 .023 .003 .008 .007 .001 .00426511 1090 .86 .92 .028 .014 .47 .12 .10 .05 .011 .022 .003 .008 .005 .001 .00426512 1090 .87 .86 .036 .016 .43 .12 .11 .05 .015 .019 .003 .007 .005 .002 .00426514 1080 .83 .74 .022 .010 .21 .08 .08 .04 .009 .001 .021 .005 .003 .003 .00226515 1080 .85 .75 .024 .010 .24 .07 .08 .04 .008 .002 .020 .005 .004 .002 .00226516 1080 .85 .77 .025 .010 .23 .08 .09 .04 .011 .002 .020 .006 .005 .002 .00326519 1080 .86 .75 .027 .013 .23 .09 .09 .04 .010 .003 .020 .006 .003 .001 .00226520 1080 .89 .73 .028 .013 .23 .09 .10 .05 .012 .003 .022 .013 .009 .001 .00326521 1080 .87 .73 .028 .010 .21 .08 .09 .05 .011 .003 .018 .006 .003 .001 .00326535 1541 .40 1.44 .036 .018 .22 .11 .09 .05 .012 .002 .024 .007 .005 .001 .00126538 1050 i4 8 )29 .008 .22 .11 .11 .07 .013 .001 .003 .004 .003 .001 .01626539 1050 .53 .98 .030 .008 .22 .09 .11 .05 .011 .001 .003 .005 .003 .001 .01826540 1050 .52 1.04 .030 .009 .23 .10 .11 .06 .014 .001 .004 .006 .005 .001 .01726541 1050 .51 1.00 .031 .010 .21 .10 .10 .06 .013 .001 .003 .005 .002 .001 .01426555 1045 .46 .76 )28 .015 6 .14 .10 .07 .013 .001 .004 .008 .003 .001 .003[71]Table 4.9 Chemical compositions of the heats monitored at Company C.Heat Grad C Mn S P Si Cu Cr Ni Mo Nb V Sn Pb Zn AlNo. e (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%)D 6122 1045 .48 .72 .018 .007 .24 .16 .05 .05 .006 - .025 .006 - - -D 6123 1045 .49 .69 .018 .008 .24 .15 .05 .05 .006 - .023 .007 - - -C 7653 1045 .47 .73 .017 .014 .26 .11 .05 i4 .006 - .023 .004 - - -C 7654 1045 .46 .82 .018 .014 .29 .12 .08 .05 .005 - .024 .007 - - -C7655 1045 .45 .72 .020 .014 .28 .18 .09 )5 .005 - .022 .007 - - -C 7658 5160 .57 .77 .024 .008 .17 .11 .79 .05 .047 - .015 .006 - - -C 7659 5160 .58 .81 .028 .013 .20 .14 .80 .05 .050 - .016 .007 - - -C 7660 5160 .57 .81 .025 .014 .19 .14 .85 .06 .059 - .018 .009 - - -C 7661 5160 .57 .81 .020 .014 .19 .13 .86 .07 .068 - .018 .008 - - -C 7663 1141 .38 1.42 .029 .020 .24 .11 .11 .03 .004 - 27 .006 - - -C 7664 L325 .21 .88 .024 .010 .28 .39 .89 .70 .011 - .009 .009 - - -A 28184 L325 .24 .88 .019 .019 .34 .46 .87 .62 .009 - .010 .006 - - -D6131 1084 .86 .71 .027 .012 .24 .22 .10 .07 .013 - .023 .008 - - -A 28187 L20 .21 1.10 .026 .021 .21 .15 .07 .05 .007 - .056 .007 - - -A 28188 1045 .46 .69 .020 .021 .27 .13 .08 .04 .004 - .024 .006 - - -D 6135 1045 .46 .69 .020 .010 .24 .16 .05 .04 .004 - .023 .006 - - -A 28191 LI7C .21 .95 .020 .017 .18 .10 .06 .05 .009 - .004 .006 - - -A 28192 LI7C .19 .98 .015 .015 .19 .05 .05 .04 .008 - .004 .003 - - -A 28193 L17C .19 1.00 .023 .016 .25 .27 .09 .06 .012 - .004 .008 - - -D 6143 L17C .21 .98 .019 .008 .19 .14 .02 .05 .005 - .003 .006 - - -[72]Table 4.10 Important casting conditions for the heats monitored at Company B.Heat No. Grade C Oil Type Flow Rate Tundish Superheat Billet(%) (mi/mm) Temp (°C) No.(°C)B24636 1008 .046 Soybean 53 1546 19 636-3-153 1538 11 636-3-253 1530 3 636-3-3B24637 l(XJ8 .040 Soybean 53 1559 41 637-3-3/2-353 1541 23 637-3-8/2-8B24638 1008 .068 Soybean 53 1579 53 638-3-3/2-353 1568 42 638-3-7/2-7B24639 1008 .045 Soybean 53 1572 43 639-3-3/2-353 1549 20 639-3-7/2-7B24640 1008 .035 Soybean 53 1594 64 640-3-253 1566 36 640-3-7/2-7B24641 1008 .041 Soybean 53 1568 39 641-3-3/2-353 1567 38 641-3-7/2-7B24642 1008 .043 Soybean 53 1567 38 642-3-3/2-353 1550 21 642-3-7/2-7B24643 1018 .170 Soybean 53 1572 58 643-3-3/2-353 1582 68 643-3-7/2-7B24644 1018 .180 Soybean 53 1555 42 644-3-3/2-353 1552 39 644-3-7/2-7B24645 1018 .180 Soybean 53 1559 47 645-3-3/2-353 1539 27 645-3-7/2-7B24646 1039 .400 Soybean 53 1548 53 646-3-3/2-353 1534 38 646-3-7/2-7B24647 1039 .420 Soybean 53 1534 39 647-3-7/2-7A23408 1008 .051 Soybean 53 1561 33 408-3-3/2-3B24649 1008 .050 Soybean 53 1564 36 649-3-3/2-340[731A23409 1015 .150 Soybean 40 1546 28 409-3-3/2-330 1538 20 409-3-6/2-620 409-3-8/2-8B24650 1010 .120 Soybean 53---A23412 1010 .090 Mineral_S 53 1553 30 412-3-3/2-344 1553 30 412-3-5/2-534 1541 18 412-3-7/2-7B24653 1010 .094 Mineral_S 34 1546 23 653-3-3/2-324 1561 38 653-3-7/2-7B24654 1012 .120 51-LN 53 1557 32 654-3-3/2-353 1566 45 654-3-7/2-7A23413 1012 .120 51-LN 443424B24655 1010 .110 Soybean 53 1570 48 655-3-3/2-3B24657 4037 .400 Soybean 53 1508 12 655-2-3[741Table 4.11 Important casting conditions for the heats monitored at Company E.Heat No. Grade C Oil Type Flow Rate Tundish Superheat Billet(%) (mi/mm) Temp (°C) No.(°C)26493 5160 .57 Canola 65 1557 79 493-4-1/6-145 1543 65 493-4-2/6-2Manual 1535 57 493-4-3/6-326494 5160 .57 Soybean Manual 1541 63 494-4-1/6-126495 5160 .57 Soybean Manual 1529 51 495-4-1/6-126501 1018 .21 Canola 65 1577 66 501-4-1/6-145 1574 63-25 1577 66 501-4-2/6-226503 1018 .19 Canola 65 1574 61 503-4-1/6-145 1579 66-25 1579 66 503-4-2/6-226504 1018 .17 HEAR 65 1568 54 504-4-1/6-145 1568 52-25 1563 49 504-4-2/6-226505 1018 .17 Soybean 65 1566 51 505-4-1/6-145 1557 42-25 1563 48 505-4-2/6-226507 1019 .17 Mineral_S 65 1574 61 507-4-1/6-145 1566 53-25 1566 53 507-4-2/6-226508 1146 .45 Mineral_S 110 1546 58-Canola 65 1549 61 508-4-1/6-1Canola 45 1549 61 508-4-2/6-2Canola 25 1543 55 508-4-3/6-326509 1146 .44 HEAR 65 1549 60 509-4-1/6-145 1543 54 509-4-2/6-225 1541 52 509-4-3/6-326510 1090 .87 HEAR 65 1527 77 510-4-1/6-145 1527 77 510-4-2/6-225 1524 74 510-4-3/6-3<25 1524 74-26511 1090 .86 Canola 65 1543 92 511-4-1/6-145 1541 90 511-4-2/6-225 1538 87 511-4-3/6-326512 1090 .87 Mineral_S 65 1527 77 512-4-1/6-145 1541 91 512-4-2/6-225 1538 88 512-4-3/6-3[75]26514 1080 .83 Mineral_S 65 1535 78 514-4-1/6-145 1529 72 514-4-2/6-225 1524 67 5 14-4-3/6-326515 1080 .85 Canola 65 1518 63 515-4-1/6-145 1518 63 515-4-2/6-225 1518 63-26516 1080 .85 HEAR 65 1527 72 516-4-1/6-145 1524 69 516-4-2/6-225 1521 66 516-4-3/6-326519 1080 .86 Soybean 65 1541 87 519-4-1/6-145 1535 81 519-4-2/6-225 1529 75 519-4-3/6-326520 1080 .89 Soybean 65* 1535 84 520-4-1/6-145 1527 76 520-4-2/6-225* 1521 70 520-4-3/6-326521 1080 .87 Soybean 100* 1538 85 521-4-1/6-1100 1532 79 521-4-2/6-226535 1541 .40 Soybean 65 1552 62 535-4-1/6-145 1549 59 535-4-2/6-225 1549 59 535-4-3/6-3100 1543 53-26538 1050 .54 Soybean 65 1557 76 538-4-1/6-145 1557 76 538-4-2/6-225 1557 76 538-4-3/6-326539 1050 .53 Canola 65 1552 70 539-4-1/6-145 1549 67 539-4-2/6-225 1538 56 539-4-3/6-326540 1050 .52 HEAR 65 1535 53 540-4-1/6-145 1538 56 540-4-2/6-225 1535 53 540-4-3/6-326541 1050 .51 Soybean 65 1538 55 541-4-1/6-145 1538 55 541-4-2/6-225 1535 52 541-4-3/6-326555 1045 .46 Soybean 65 1549 61 555-4-1/6-1Flux 1535 47 555-6-2* Oscillation Frequency changed to 130 osc/min.[76]Table 4.12 Important casting conditions for the heats monitored at Company C.Heat No. Grade C Oil Type Flow Rate Tundish Superheat Billet(%) (mi/mm) Temp (°C) No.(°C)D6122 1045 0.48 Canola 70 1535 48 122-4/5-1D6123 1045 0.49 Canola 100 1541 54 123-4/5-170 1539 52 123-4/5-225 1537 50 123-4/5-3C 7653 1045 0.47 Mineral_S 100 1536 47 653-4/5-170 1532 44 653-4/5-225 1532 44 653-4/5-3C 7654 1045 0.46 HEAR 100 1540 52 654-4/5-170 1535 49 654-4/5-225 1534 48 654-4/5-3C 7655 1045 0.45 Mineral_O 100 1522 33 655-4/5-170 1526 37 655-4/5-225 1520 31 655-4/5-3C 7658 5160 0.57 Canola 100 1526 47 658-4/5-170 1524 45 658-4/5-225 1525 46 658-4/5-3C 7659 5160 0.58 Mineral_S 100 1529 52 659-4/5-170 1517 40 659-4-225 151 1 34 659-4-3C 7660 5160 0.57 HEAR 100 1525 47 660-4/5-170 1525 47 660-4/5-225 1525 47 660-4/5-3C 7661 5160 0.57 Mineral_O 100 1539 61 661-4/5-170 1530 52 661-4/5-225 1526 48 661-4/5-3C 7663 1141 0.38 Canola 100 1508 17 663-4/5-170 1536 45 663-4/5-270 1535 44 663-4/5-325 1535 44 663-4/5-4[77)C 7664 L325 0.21 Canola 100 1541 36 664-4/5A,5B-170 1540 35 664-4-225 1538 33 664-4-3A 82184 L325 0.24 Mineral_S 100 1525 23 184-4/5-170 1547 45 184-4/5-225 1551 49 184-4/5-3D6131 1084 0.86 Mineral_U 100 1514 61 131-4/5-170 1515 62 131-4/5-225 1510 57 131-4/5-3A 28187 L20 0.21 HEAR 100 1573 64 187-4/5-170 1574 64 187-4/5-225 1574 65 187-4/5-3A 28188 1045 0.46 HEAR 100* 1547 58 1884/51*70* 1546 57 1884/52*25* 1546 57 1884/53*0* 1544 55D6135 1045 0.46 Mineral_U 100* 1543 54 1354/51*70* 1540 51 1354/52*25* 1533 44 1354/53*A 28191 LI7C 0.21 Canola 100 1559 59 191-4/5-170 1571 60 191-4/5-225 1565 54 191-4/5-3A 28192 L17C 0.19 Mineral_S 70 1549 36 192-4/5-1100 1549 36 192-4/5-225 1550 37 192-4/5-3A 28193 L17C 0.19 HEAR 100 1560 50 193-4/5-170 1553 43 193-4/5-225 1540 30 193-4/5-3D 6143 L17C 0.21 Mineral_U 100 1556 45 143-4/5-170 1550 40 143-4/5-225 1550 40 143-4/5-3* Oscillation frequency changed to 96 from 144 osc/rnin.[78]Table 4.13 Summary of the types of oils and the flow rates at which they were used at the threePlantsPLANT BOil Type Flow rate (mi/mm) Steel GradeMineral_S 54, 44, 34 1010Soybean 40, 30, 20 1008, 1010, 1012, 1015,103951-LN 54,44,34,24 1008,1012,1018,4037PLANT EOil Type Flow rate (mi/mm) Steel GradeCanola 65,45,25 1018, 1050, 1080, 1090, 1146, 5160HEAR 65, 45, 25 1018, 1050, 1080, 1090, 1146Mineral_S 65, 45, 25 1018, 1050, 1080, 1541, 5160Soybean 65, 45, 25 1019, 1050, 1080, 1090, 1146PLANT COil Type Flow rate (mi/mm) Steel GradeCanola 100,70,25 1017, 1045, 1141, 5160, L325HEAR 100, 70, 25 1017, 1045, 5160, L20, L325Mineral_S 100, 70, 25 1017, 1045, 5160, L325Mineral_O 100, 70, 25 1017, 1045, 1084, 1141, 5160[791— Copper Wire• - -- Constantan WireRSW Water Outlety Thermocouple•VBulk Water Out Thermocouple•a—__VWater Out “_ __ ___Ddej___ • • -—x—— fWater In --___•-_____Bulk Water In ThermocoupleT MoulJallBaffle I Thermocouples Common Copper Wire forMould Wall All Intrinsic MouldWall ThermocouplesFigure 4.1 Schematic diagram of the set-up for measurement of mould wall and mould coolingwater temperatures.[80]38.1 mmI-.IIIII10.8 mm IIIIIi.__4________ ________14.4 mm 15.8 mmFigure 4.2 Schematic ill u strat ion of the load cell.811T1cD_CI CID0CDfl— 2CD C CD C) CD CD CD CD CD _2C7R)Oscillator TableHousing• Load Cell-Figure 4.4 Schematic diagram showing the placement of LVDTs and load cells.Oscillator TableI I1831Magnetic shieldEEE 2 J Primary [condarv i]Tov2I1 Primary_____Figure 4.5 Schematic diagram ol ihe construction of a LVDT.1841Figure 4.6 Photographer/amplifier system.of MetraByte’ s U riiversal Expansion Interface; expnasion multiplex-Figure 4.7 Photograph of the 8 channel high speed A/D converter and Timer counter interfacefrom MetraB\’te.1851110 v(a.c)From MouldThermocouplesand Water TCsFrom LoadIII 4L__JExcitation voltage18v(d.c)Figure 4.8 Schematic diagram showing the main parts of the data acquisition set-up.110 v(a.c)Ribbon cableI IsolationI I TransformerI II 4I II II I ‘Tmnsj EXP-16I Ii a:c iformerrii;<AllInputO-2Omv>1n__________110v(a.c) AI 15 v (a.c) ‘Excitation Voltgae t‘IHousingFrom Control I 1 0-200 vI0IPanelCasting Speed I (dc) [‘UIIIFrom Control f 1 4-20 ma I))Metal Level_____Panel (d.c)[86][JLLETS FROM PLANT TRIA1*MacroetchVFigure 4.9 Flow chart of the billet inspection procedureExamination ofMacrostructureExamination ofMicrostructure[TOREI 87]Cu Cu+________ ___________+J1z. •1J4 VMCuc’ Cu I+ +V2J2 J3Water Channel DAS BoxEquivalent CircuitVM = voltage measuredJI = Mould Wall Junction V3 = V4 =0 (as Cu-Cu)J2 = Junction in Water ChannelV2 opposes ViCu +/viVMC‘:V2_______J2r v1=vM+V2Figure 4.10 Circuit diagram of the thermocouple connection.[88]MOULD-THERMOCOUPLERESPONSE2-I), UNSTEADY-STATE HEAT-FLOWMATHEMATICAL MODEL OF MOULDMOULD TAPERFigure 4.11 Flow chart showing the models used, and the steps involved, in the analysis ofmould thermocouple data.*FLTERING OF THERMOCOUPLERESPONSEFILTEREDThERMOCOUPLE DATA-ThMPERATURE HEAT-FLUX PROFILE3-I), ELASTO-PLASTICFINITE-ELEMENTMODEL OF MOULD2-I), UNSTEADY-STATEHEAT-FLOWMODEL OF BILLETI[891I YU180170160150140130-CUo no80130 135 140 145 150 155 160 165METAL LEVEL FROM TOP OF MOULD (mm)Figure 4.13 Effect of metal level fluctuation on the temperature recorded by the meniscus thermocouple.pwUICFigure 4.12 A typical unfiltered mould-thermocouple response at Company B.liME (s)ociaC0cicici 0cicbciDo0 ciciDciciciciciciDcici[90]EEI.,-4rC-)-2OFigure 4.14 Various components of a mould oscillation cycle.0.743 /=içative Sthp StanII —..—-----. -------.--------—.UPNegative SUiP STROKE0 0.2 0.3 0.4 0.3 0.6Tw60 DOWN40STROKE.: 7.STROKENegative Sn Start Negal eSthp End-SO0 0.1 02 0.3 0:4 0.5 0.6 0.7TIME[91]Chapter 5: RESULTS OF PLANT TRIALSIn this chapter the raw data collected from the three plant trials is presented without subjectingthe information to any major analysis. Thus the mould temperature and mould-billet friction forcesare presented and a detailed explanation of the observations is, in most cases, deferred until theresults of the various mathematical model calculations become available at the end of chapter 8.5.1 Mould Temperature DataThe dependance of the mould temperature on important operating variables is discussed inthe sections that follows. Typical unfiltered response of selected thermocouples are shown in Figure5.1 while Figure 5.2 is a plot of the standard deviation of temperatures recorded. It needs to be keptin mind that the temperatures shown in this chapter are those recorded by the thermocouples whichare located roughly in the middle of the mould wall. Thus the temperature of the hot face of themould, as calculated by mathematical models, would be significantly higher. It also needs to bepointed out that in graphs in which the thermocouple temperatures have been plotted as a functionof distance below the mould, the measurments were made at discrete points (at each thermocouplelocation), and lines have been drawn through the points.5.1.1 Effect of carbon contentThe effect of carbon content of the billet on the mould wall temperature is best seen in thegraphs for Plant B where data was collected for heats with carbon contents ranging from 0.05% Cto 0.42% C. Figure 5.3 shows the time-averaged axial temperature profile for heats with differentcarbon contents.As was discussed in a previous Chapter, the low mould temperature for a 0.12% and 0.09%carbon heats are as reported by other workers [17,34]. The shrinkage accompanying the phase[92]change from toy, which occurs while the solid shell is very thin for 0.12%- 0.09% C, causes thebillet surface to become rippled thereby increasing the local billet-mould gap. This leads to a dropin heat transfer and, therefore, mould wall temperatures.5.1.2 Effect of oil typeTime-averaged mould wall temperatures are presented in Figures 5.2 through 5.7 from datacollected during trials at Plant C. The mould wall temperatures are given for 4 different oils at twodifferent flow rates for steel grades 1018 (Figures 5.4, 5.5) and 1045 (Figures 5.6, 5.7).It can be seen from these figures that the vegetable based oils (Canola and HEAR) in generalgive somewhat higher mould temperatures than the mineral oils (Mineral_S and Mineral_O). Thedifference in temperature is most discernible at low flow rates and for the 1018 grade.5.1.3 Effect of oil flow rateAt Plant C the flow rates of the oils could be changed in systematic manner for all the fouroils tested. Time-averaged mould wall temperature measurements are presented for the three oilsat three different flow rates in Figures 5.8 through 5.10. In the case of Canola oil the flow wasturned down almost to zero and the effect is shown separately in Figure 5.11. Figure 5.12 showsthe response of selected thernocouples as the flow rate of Canola oil is changed from 0 ml/min to100 mi/mm.Preliminary observations indicate that the increased flow rate of oils leads to an increase inmould heat transfer mainly in the meniscus region. The effect is most pronounced with the mineraloil (Steelskin). It is thought that the enhancement of heat transfer is due to improved thermalconductivity of mould-billet gap on account of its hydrogen content arising from the breakdown ofoil; higher flow rates of oil leading to increased hydrogen content in the gap. A point that needs tobe noted is that the largest enhancement in the mould wall temperature occurs when the oil flow[931rate is changed from 0 mI/mm (no oil) to 25 mI/mm (Figure 5.12). This increase in temperature ismuch higher than the increase that takes place when the flow rate of oil is changed from 25 mi/mmto 100 ml/min.5.1.4 Effect of oscillation frequencyThe oscillation frequency of the mould was changed at Plant C from its normal value of 144cpm to 96 cpm. The corresponding change in oscillation related parameters are summarized inTable 5.1 (note the slight increase in the negative strip period from 0.19 seconds to 0.16 secondswith a decrease in the oscillation frequency). The effect that this change in oscillation frequencyhas on the mould temperature is shown in Figure 5.13. It can be clearly seen from the above figurethat there is an increase in the mould wall temperature with a decrease in the oscillation frequency.This result atfirst glance seems to be at variance with what would be expected; an increase in thenegative strip time (at lower oscillation frequency) should increase the oscillation mark depth ofthe billets thereby increasing the local air gap width leading to a decrease in heat transfer andmould wall temperature.5.1.5 Mould temperature at the three PlantsFigure 5.14 and 5.15 show the time averaged mould wall temperatures collected for the threeplants while casting steel of the same grade. Even allowing for differences in mould wall thicknessand the locatic5n of the thermocouples it can be clearly seen that the mould wall temperatures atCompany C are the highest indicating a significant increase in mould heat transfer over the othertwo plants. This issue, which is ofprime importance, is investigated in detail in subsequent chapters.5.2 Mouid-BiIIet Friction ForcesThe manner in which the mould-billet friction forces are analysed has been discussed in theprevious chapter and are summarised here. The analysis consists of:(a) Observation of the shape of the load cell response curve.[94](b) Computation of the percentage of the negative strip time for which decompression of the loadcell takes place(c) Computation of the load at the start and end of the negative strip period.(d) Calculation of the difference in the maximum load attained during the upstroke and the load atthe start of negative strip period.Such a route of analysis is necessary because, as explained, in the previous chapter, the absoluteloads cannot be compared across trials.5.2.1 Effect of oil type and flow rate on load cell responseThe analysis of the load cell response from Company B has been part of the research workof Brendzy [13]. As mentioned earlier, she had observed that the reduction of oil flow (from 54mi/mm to 24 mI/mm) resulted in increased mould-strand interaction. Furthermore, there is someevidence in her work to suggest that different oils exhibited different degrees of lubricity at thesame flow rate.Load cell data from Comnpany C, however, shows that there is no influence ofeither the typeofoil or itsflow rate on mouldfriction. Figure 5.16 through 5.23 which show the load cell responsefor the four oils at two flow rates for a 1045 grade steel clearly reveal the same general nature ofthe load cell response curve. Table 5.2 shows the other parameters used for comparing load cellresponse and clearly no dependance of oil type or oil flow rate can be discerned. Furthermore,using Brendzy’s analysis /13] of the up stroke part of the cycle it appears that there is a largeamount offriction in this part oft/ic stroke regardless of the rypes of oil or their flow rates.5.2.2 Effect of carbon contentBoth Brendzy [131 and other researchers [25,26,27] have indicated that there is a dependanceof the mould-billet friction forces on the carbon content of the steel being cast. Figure 5.24 - 5.26show the load cell response while casting 1018, 1045 and 5160 steel grade and Table 5.3 summarizes[951the statistics of the load cell response in different periods of the mould displacement. The increasedfriction at lower carbons is clearly seen in the difference between the load at start of negative sthptime and the minimum load recorded during the down stroke. This result is consistent with thatobtained by Stel et al. [27].5.2.3 Effect of mould oscillation frequencyThe effect of mould oscillation frequency on the mould-billet friction is shown in Figures5.27 (144 cpm) and 5.28 (96 cprn) for REAR oil at 25 ml/min and in Figures 5.29 (144 cpm) and5.30 (96 cpm) for HEAR oil at 100 mI/mm for a 1045 grade steel. The corresponding statistics forthe load cell response are presented in Table 5.4. There is a greater decompression of the load cells(more mould-billet interaction) at the lower oscillation frequency of 96 cpm as shown by the difference in load at the start of negative strip time and the minimum load recorded during thedownstroke. The increased mould-billet interaction is also obvious from the duration of thedecompression period which is about 80-85% of the negative strip period for the 96 cpm case against—70 % for the 144 cpm case. (The cross marks on the mould displacement curve for the abovementionedfigures indicate the start and end ofnegative strip time while the open square correspondsto the end of load cell decompression. Thus the position of the open square marker relative to theother two markers is a measure of the percentage of the negative strip time for which the load cellis decompressed).5.2.4 Difference in the load cell signals from the three PlantsFrom the standpoint of mould-billet interaction during the negative period, there are majordifferences in the load cell signals from the three plants. Typical load cell signals for Plant B, Eand C are shown in Figures 5.31, 5.32 and 5.33 and the statistics for the different parts of theoscillation cycle is presented in Table 5.5. A close examination of Figures 5.31, 5.32 and 5.33 showsthat at both Plant B and E the decompression period of the load cell is only 50- 55 % of the negative1.961strip period while at Plant C it is clearly in excess of 70%. This analysis is seen somewhat betterin Figures 5.34, 5.35 and 5.36 where the load cell response is shown forjust one cycle for the threeplants. (As mentioned in the previous section the position of the open square marker relative to thecross markers (Figure 5.3 1-5.33) is a measure of the percentage of the negative strip time for whichthe load cell is decompressed).Clearly then, there is more mould-billet interaction at Plant C than at the other two plants.5.3 Billet Quality EvaluationBillets collected during the trials at the three plants were subjected to an extremely exhaustivequality evaluation details of which have been covered in the previous chapter. A total of 45, 85 and49 billets from Plants B, E and C were cut and macro-etched. Since a description of the results ofthe quality evaluation would be too detailed only the main quality problems observed in the billetsare discussed below.5.3.1 Transverse depressionsAfter lightly shot blasting the billets from Company B to remove scale, surface inspectionfor cracks was carried Out. A characteristic of the grades 1008, 1010 & 1012 was the presence ofsevere transverse depressions present along the billet surface. These depressions appeared on boththe control and test strands. Figure 5.37 is a macro-etch of a longitudinal section of a 1008 gradesteel which contrasts with the relatively smooth surface of a “high carbon” grade billet as shownin Figure 5.38.Observations made by looking into the spray chamber during the casting of the low-carbongrades revealed visible jerking of the billet while exiting the mould. Thus prima-facie it wouldappear that the parabolic taper of the billet mould at Company B is too tight for these low-carbongrades. Severe binding in the mould can lead to the formation of transverse depressions and cracksas explained in a previous chapter.[97]5.3.2 Off-corner internal cracksOff-corner internal cracks were present in billets from Company E & C and absent frombillets from Company B. The depth of cracks beneath the surface at each of the eight off-cornersites (see Figure 5.39) was measured and an average depth was calculated for each steel grade.Furthermore, for each group, the percentage of billets that had cracks at a given site was alsodetermined.For Company E, the most serious quality problem in both the test and control strand billets,is the presence of off-corner internal cracks. A typical macro-etch of a 1090 grade billet showingtypical off-corner, internal cracks is presented in Figure 5.40. Figures 5.41 to 5.45 show the averagedepth of crack at different sites for both the test and control strands for grades 1018, 1050, 1080,1090 and 1146. Longitudinal off-corner depressions were frequently seen on the test and controlstrand billets and Figure 5.46 shows this finding in the form of a bar graph. The principal findingsof the analysis are summarised below.[1] Cracks are present randomly on all eight sites with no apparent selectivity.[2] The cracks are deeper on the test strand billets than on the control strand[3] The crack depths vary from 8-11 mm on the test strand.[4] The severity of cracking is worse in the 1080, 1090 and 1146 grades, with all the billetsexamined containing a crack in at least one off-corner site. Of these three grades the resulfurized steels (1146) had off-corner internal cracks at six of the eight possible sites on boththe test and control strands.[5] Longitudinal off-corner depressions are observed on the straight faces that are notdeformed by the withdrawal rolls.The mechanism by which off-corner, internal cracks form has been discussed earlier. Itinvolves bulging of one of the faces, invariably the wide face, accompanied by rotation of the corner[98][59]. Longitudinal off-corner depressions could form by this mechanism off the corner adjacent tothe bulged face. Cracks form close to the solidification front in the region of low ductility as a resultof the tensile stresses generated at the off-corner by bulging of the wide face and on the adjacentoff-corner due to the rotation of the corner. The depth of the crack beneath the surface is a measureof the shell thickness of the billet at the time the crack formed.The depth of the crack (8-11 mm) is a vital evidence in the analysis of the formation of thecrack. Utilising the shell growth profile from the results of the mathematical model, discussed inChapter 8, it can be shown that, allowing for 20 % variation in shell thickness between the mid-faceand the off-corner (with the latter being lower), the cracks are forming just below the mould. Theoccurrence of cracks in both the test and control strands for all grades suggests that they are morelikely related to the bulging of the wide face below the mould. The greater propensity for crackingin the high carbon and resulphurized grades is because the shell thickness is lower at the mouldexit on account of their longer freezing range. The depth of cracks for these grades is closer onaverage to 8-9 mm in the 1018 grade.An important difference between the test and control strands lies in the average depth ofcracks from the surface. It is clear that the cracks are on average 1-2 mm deeper on the test strand.Since the mechanism of cracking is undoubtedly the same on both strands, this observation suggeststhat the shell thickness at the bottom of the test strand is 1-2 mm greater than on the control strand.This is important as it indicates greater heat transfer in the control strand mould.For Company C, Figure 5.47 is a typical macro-etched transverse section showing the presenceof off-corner, internal cracks. Figures 5.48 - 5.52 show the average depth of cracks at different sitesfor both the test and control strands for grades 1018, 1045, 5160, 1084 and L325 respectively.Results of analysis done on billets from Company C are summarized below.[1] Cracks are present randomly on all eight sites with no apparent selectivity.[991[2] The severity of cracking correlates well with the off-squareness.[3] The crack depths vary from 6-14 mm on the test strand.[4] The cracks are most severe in the 5160, L-325 and 1084 billets in comparison to 1018 and1045 billets.[5] Longitudinal off-corner depressions are observed on all billets.A detailed analysis for the cause of these cracks is left until a later section.5.3.3 Midway cracksIn an analysis somewhat similar to the preceding section, the depth of cracks beneath thesurface at each of the four locations (see Figure 5.39) were measured and an average depth wascalculated for each steel grade. Furthermore, for each group, the percentage of billets that had cracksat a given location was also determined.Figure 5.53 is a macro-etch of a transverse section of a billet from Company E showing typicalmidway cracks. Average depths of these midway cracks for grade 1018, 1050, 1080, 1090 and 1146for Company E are shown in Figures 5.54 to 5.58. These cracks are seen on all grades and predominantly on the wide faces. In some cases the cracks appear to extend to the centreline.A typical macro-etch, for Company C, of a transverse section containing midway cracks isshown in Figure 5.59. Details of the crack depths and percentage of the billets exhibiting midwaycracks are presented in Figures 5.60 to 5.64. Thu sit can be seen that the cracks appear most frequentlyin 1018, L-325 and 1084 grades as compared to 1045 and 5160 billets. There is little difference incrack frequency between the test and control strands and no discernible preference of the cracks toappear adjacent to a given billet face. Comparison of the crack depth to the shell profile (availablefrom Chapter 8) reveals that the midway cracks are forming 2.5 to 3.0 m from the top of the mould.This location corresponds to the bottom of the sprays, as expected, which suggests that the spraysneed to be redesigned to minimize surface reheat of the billets below the secondary cooling zone.[100]The length of the spray chamber (1.74 m), for example, is too short.5.3.4 Surface roughnessA ready measure of the surface roughness of billets is available from the profilometer readingsof the billet surface. These results are put on a firmer footing by combining them with visualobservation of the billet surface.5.3.4.1 Effect of carbonThe effect of carbon on the surface roughness of the billets is best seen at Company B wherebillets were collected for a wide range of carbons. Figures 5.65, 5.66 and 5.67 are photographs ofthe longitudinal section of billets from grades 1008, 1012 and 1039. The wrinkled surface of a 1012grade stands out in direct contrast to the smoother surfaces of the other two grades. This differencein surface roughness is only expected as the phase change from ö to yfor the 1012 grade takes placevery early in the solidification and the solid shell, being thin and thus unable to resist the stressesarising from the contraction accompanying the phase change, buckles giving rise to a “wrink1ed’surface appearance. The phase change for the 1008 grade takes place after it has undergone significant amount of cooling and the solid shell is reasonably thick to withstand phase-change relatedstresses. The higher carbon grade, 1039, when below the solidus is in the y phase and does notundergo ö to yphase transformation.5.3.4.2 Effect of oil typeAt Company B there was no discernible effect of either the oil type or its flow rate on billetquality [13]. This conclusion is based on billet surface evaluation and profilometer measurements.However since changes in the oil type could not be separated from changes in carbon content ofbillets, this finding remains somewhat inconclusive.At Company E, systematic changes were made to oil type and its flow rate and the resultsare discussed below. Figures 5.68 - 5.71 are graphs of oscillation mark depth of four 1018 billets[1011cast with Canola, HEAR, Mineral_S and Soybean oils respectively. Figures 5.72- 5.75 are thecorresponding photographs of the surfaces of these billets. The billet cast with Canola clearly havethe best appearance. The surface of the HEAR oil billet is considerably rougher whilst the billetscast with Mineral_S and Soybean lubricating oils have deeper and more irregular oscillation marks.The profilometer readings for the billets cast with Mineral_S and Soybean oils, further show thatnot only are the oscillation marks more uniform across the face but are also shallower and lesssensitive to oil flow rate than with other oils. The oscillation-mark depths appear to be highlysensitive to flow rate with the Soybean oil, and are very non-uniform for Mineral_S oil. Notwithstanding the inherent superiority of Canola oil, there are occasional problems like sticking andbleeds.The profilometer measurements of oscillation-mark depths of 1080 grade billets fromCompany E for the four oils, Canola, HEAR, Mineral_S and Soybean are shown in Figures 5.76 -5.79 and the corresponding photographs in Figures 5.80 - 5.83. At a flow rate of 45 ml/min theoscillation marks on billets cast with Canola are slightly more uniform while the billets cast withSoybean appear to have bleeds and laps.Profilometer measurement of 1090 grade billet cast at Company E with Canola, HEAR andMineral_S are shown in Figures 5.84- 5.86 and the corresponding photographs are shown in Figures5.87 - 5.89. Neither the type of nor its flow rate appear to have any significant effect on theoscillation-mark depth except with Canola oil where there was a marked deterioration in uniformityat 65 mI/mm.[102]The last set of billets from Company E is the one belonging to the 1146 grade and the profilometer measurement of oscillation mark depths are shown for Canola and HEAR oils in Figures5.90 - 5.91. The depths of these marks are deeper than that observed on billets from grades 1018,1080 or 1090. Furthermore these marks appear quite non-uniform in depth across the face of thebillet. Corresponding photographs of billets are shown in Figures 5.92- 5.93.Figures 5.94- 5.97 shows the oscillation-mark depths for 1018 grade billets from CompanyC cast with Canola, HEAR, Mineral_S and Mineral_O oils at different flow rates, and photographsof the billet surfaces cast at a flow rate of 25 ml/min are shown in Figures 5.98 - 5.101. At low flowrates it is clear that the billets cast with Canola oil have the best appearance with those cast withHEAR oil being relatively similar. The oscillation mark depth support this observation. Withincreasing flow rate the uniformity of the depth of oscillation marks deteriorates for both Canolaand HEAR. This finding is in accordance with the observations made on billet from Company E.The oscillation mark depth for 1045 grade billet from Company C cast with HEAR andMineral_O oils is shown in Figures 5.102- 5. 103 and the corresponding photographs are shownin Figures 5.104 - 5.105. No clear trend is discernible.For the 5160 grade billets from Company C the oscillation mark depths for Canola, HEAR,Mineral_S and Mineral_O oils are presented in Figures 5.106- 5.109. At oil flow rates of 25 mi/mmthe oscillation marks are shallower and more uniform when casting with Canola and Mineral_Swhilst at flow rates of 100 mI/mm Mineral_O performs the best.5.3.4.3 Effect of oscillation frequencyThe effect of oscillation frequency on oscillation mark depth can be seen for Company C bycomparing the oscillation mark depth for 1045 grade billet cast at 96 cpm with HEAR oil (Figure5.110) and Mineral_O oil (Figure 5.111) with Figures 5.102 and 5.103 which give the oscillationmark depth for 1045 grade billet with the same oils at 144 cpm. It is clear that for HEAR oil, at low[103]flow rates, the oscillation marks are more non-uniform at a higher frequency while at high flowrates the differences are not significant. For the billets cast with Mineral_O oil no clear trendsemerge.The oscillation mark depth for 1080 grade billet cast at 130 cpm is given in Figure 5.112.This, when compared to the depths shown in Figure 5.83, reveals that at both high and low flowrates there is a slight decrease in the average depth of oscillation mark. At high flow rates thevariability in oscillation-mark depth is higher.5.3.5 RhomboidityRhomboidity was determined by measuring the difference in the lengths between the twodiagonals of a transverse section of a billet. In almost all cases rhomboidity was less than 1% inCompanies B & E. In case of billets from Company C, the control strand had the most severelyoff-square billets of the 1018, 5160 and L-325 grades as shown in Figure 5.113. It is important tonote that the mould cooling water velocity in the control strand is much lower than that in the teststrand and there is a possibility that the cold face temperature of the control strand mould couldexceed the boiling point of water at the prevailing pressure. This issue is investigated in greaterdetail in a subsequent chapter.5.3.6 Other defectsDetailed examination of the billet surface revealed two defects that can be sources of problemwhile rolling steel billets.5.3.6.1 Craze cracksCraze cracks are a network of cracks usually up to 1 mm deep that can be revealed bymacroetching billet surfaces. Earlier work at the Centre for Metallurgical Processing, UBC, hadshown that increasing copper content of steel resulted in deeper, more interconnected cracks. Thisis consistent with studies done on bloom casting elsewhere [66]. It had also been found that an[104]increase in the Ni/Cu ratio reduced the severity of the cracking.The same phenomenon was found when transverse sections from Company C weremacroetched. A typical example of craze cracking is shown in Figure 5.114. The cracking in itsleast severe form was observed to start at the corners of the inside radius, with the width of coverageincreasing towards the centreline until the entire billet surface was affected. It was also observedthat the other three billet surfaces exhibited similar behaviour, although to a lesser degree for agiven copper content and Ni/Cu ratio.To categorize the severity of cracking, a craze cracking index was formulated [671. The indexconsists of measuring (in mm) the distance across the billet face that was affected by the cracking.It is then multiplied by a reduction factor based on the visual appearance of crack severity. Thisindex is plotted versus copper, nickel, carbon content and nickel/copper ratio in Figures 5.115 to5.118.Importance of the Ni/Cu ratio can be seen when steel grades 1018 and L-325 are compared.The 1018 grade has a copper content ranging from 0.05% to 0.27% with craze cracking indiceswhich vary with copper content. When the L-325 grade is examined, craze cracking indices areobserved to be much lower in spite of the copper content being approximately 0.40%. It appearsthat the Ni/Cu ratio of 0.2-0.8 of the 1018 grade is inferior to the ratio of 1.3-1.8 in grade L-325.5.3.6.2 Zipper marksA new defect, termed “zipper marks” was observed on the surface of billets from CompanyE. This defect is generally associated with a bead of material trapped initially on the mould walland Figures 5.119 and 5.120 show this defect for billets cast with Canola and Soybean oilsrespectively. Beading was generally observed on both the test and control strand billets. A controlstrand grade 1080 billet cast with Soybean oil as a lubricant was sectioned transversely through abead and subjected to metallographic examination. Figure 5.121 is a macro-etch of the transverse[105]section through the bead from which it is evident that the bead has solidified on the surface of thebillet and in this instance is approximately 0.3 mm thick. An EDX analysis of the bead on the SEMindicated that the bead has the same composition as the billet. Thus the bead may be a result ofmicro-bleeding of the solidifying shell. This could be due to poor lubrication which causes tearingand bleeding. The bleeding problem appears to be most severe on high carbon billets which couldbe related to the long freezing range and the susceptibility of the shell to rupture. This defect wasobserved at both high and low flow rates with the four types of oils.[1061Table 5.1 Oscillation characteristics of theoscillator at Plant C for two different oscillationfrequencies.Oscillation Negative Strip Mould LeadFrequency Period (s) (mm)(cpm,Hz)144,2.4 0.16 5.496,1.6 0.19 3.1Table 5.3 Difference in the maximum loadattained during the upstroke and the load at thebeginning of tN while casting different steelgrades at Company C.Steel Grade Load(N)1018 2401045 1965160 196Table 5.2 Difference in the maximum loadattained during the upstroke and the load at thebeginning of tN while casting 1045 grade steelat Company C.Type of oil Load at flow Load at flowrate of oil rate of oil(25 mI/mm) (100(N) mi/mm)(N)Canola 196 187Mineral_S 178 169HEAR 196 182Mineral_O n.a 209Table 5.4 Decompression of the load cellduring the negative strip period and the percentage of negative strip time for which thisdecompression lasts for two oscillationfrequencies at Company C.Oscillatio Load at Load at Period Periodn frequen 25 100 of deco of decocy (cpm, mI/mm mI/mm mpressi mpressiHz) (N) (N) on at 25 on atmI/mm 100(%) mi/mm(%)144 583 627 68 7296 902 920 84 87Table 5.5 Percentage of the negative strip time for which the load cell is decompressed (duringnegative strip period) at Company B, E and C while casting a 1018 grade steel.Company Period of decompression (%)B 50E 52C 78[1071TC at 215 mm200TC at 175 mmi150•.•, ;: :I V; ! . J4 i.j ,.R i;i 1.! / J i ‘‘5’f a •;•) • • / ::!si’./ V\1 UTC at 145 mm (meniscus)50TC at 50 mmI I0 100 200 300 400 500 600Figure 5.1 Unfiltered response of selected thermocouples at Company C while casting a 1045grade billet.[108]101614at-)z 12010>r.i8z6CfJ2C)0 100 200 300 400 500 600 700 800DISTANCE FROM TOP OF MOULD (mm)Figure 5.2 Typical standard deviation values for temperature data collected by thermcouples atCompany C.[109]Figure 5.3 Time-averaged axial temperature profile for different carbon contents at Company B.(Note that the temperatures are those recorded by thermocouples located approximately midwaybetween the cold and hot faces of the mould).UrJ200 300 400DISTANCE FROM TOP OF MOULD (mm)[110]sQUiUiCUiFigure 5.4 Time-averaged axial temperature profile for a 1018 grade of steel billets cast at 25mi/mm of Canola, Mineral_S, HEAR and Mineral_O lubricating oils at Company C. (Note thatthe temperatures are those recorded by thermocouples located approximately midway betweenthe cold and hot faces of the mould).-,cp________________________________________________________________________________________TEEL GRADECanolaM,ncr.d_S50 J HEARL Me- 100 200 300 4(X) 500 600 700800DISTANCE FROM TOP OF MOULD (mm)Figure 5.5 Time—averaged axial temperature profile for a 1018 grade of steel billets cast at 100mI/mm of Canola, Mineral_S. 1-lEAR and Mineral_O lubricating oils at Company C. (Note thatthe temperatures are those i-ecorded by thermocouples located approximately midway bet\veenthe cold and hot faces of the mon Id).P150C- 100Ui5—0011111220200p 180ii 160140c.. 120Li4I- tOO80600 100 200 300 400 500 400DISTANCE FROM TOP OF MOULD (mm)700 800Figure 5.6 Time-averaged axial temperature profile for a 1045 grade of steel billets cast at 25ml/min of Canola, Mineral_S. HEAR and Mineral_O lubricating oils at Company C. (Note thatthe temperatures are those recorded by thermocouples located approximately midway betweenthe cold and hot faces of the mould).200150DFc. 100F50Figure 5.7 Time—averagel axial temperature profile for a 1045 grade of steel billets cast at 100mi/mm of Canola, Mineral S. 1—lEAR and Mineral_O lubricating oils at Company C. (Note thatthe temperatures are those recorded by thermocouples located approxiniatelv midway betweenthe cold and hot faces of the mould ).IEEL GRADE 2045CanolaMineral_SHEARMinrral_O-.I-flCfliSCUsLTEEL GRADE 1045CanolaMineml_SHEARMine.--—---I0 100 200 300 400 500 600 700 800DISTANCE FROM TOP OF MOULD (mm)FNERALS(rnE]——- mcnISCUS100 200 300 400 500 00 700DISTANCE FROM TOP OF MOULD (mm)Figure 5.8 Time-averaged axial temperature profile for a 1018 grade steel billets cast at 25, 70arid 100 nil/miri of Mineral_S oil at Company C. (Note that the temperatures are those recordedby themiocouples located approximately midway between the cold and hot faces of the mould).250Figure 5.9 Time—averaged axial temperature profile for a I 0 8 grade steel billets cast at 25, 70and 100 mI/mm of I—lEAR oil at Company C. (Note that the temperatures are those recorded bythermocouples located approximately midway between the cold and hot faces of the mould).200Pa- 150DI. 100I-50 J800200Pa-ISO. 100F50“0 100 200 3(8) 400 500 600DISTANCE FROM TOP OF MOULD (mm)700 80011131Figure 5.10 Time-averaged axial temperature profile for a 1018 grade steel billets cast at 25,70and 1(X) mi/mm of MineraLO oil at Company C. (Note that the temperatures are those recordedby thermocouples located approximatel’ midway between the cold and hot faces of the mould).250Figure 5. 11 Time—averaged axial temperature profile for a 1018 grade steel billets cast at 0, 25,70 and 100 ml/min of Canola oil ai Company C. (Note that the temperatures are those recordedby thermocouples located approximately midway between the cold and hot faces of the mould).200Ic-)a’ 150DI. 1001-50MINERAL_O (mi/mmJ__nlcmscus0 100 200 3(8) 400 500 600 700DISTANCE FROM TOP OF MOULD (mm)800Pa’ ISODI. 100I50100 200 3(8) 400 500 600 700DISTANCE FROM TOP OF MOULD (mm)80011141L)DICFigure 5.12 Response of selected thermocouples at Company C with change in oil flow rate from0 mi/mm (no oil) to 100 mI/mm of Canola oil for 1018 grade steel billet at Company C. (Notethat the temperatures are those recorded by thermocouples located approximately midwaybetween the cold and hot faces of the mould).NI0050 J•.— menisCus0 I I0 100 200 300 400 500 600 700 800DISTANCE FROM TOP OF MOULD (mm)Figure 5.13 Time—avcra2ed axial temperature profile for a 1045 grade of steel cast at 144 and 96cpm of mould oscillation at Company C. (Note that the temperatures are those recorded by ther—inocouples located approxiniatelv midway between the cold and hot faces of the mould).11151200 300TIME FROM START OF CASTING (s)Figure 5.14 Time-averaged axial temperature profile obtained at Company B, E and C whilecasting a 1018 grade steel billet. (Note that the temperatures are those recorded by thermocouples located approximately midway between the cold and hot faces of the mould).Figure 5. 15 Time—averaged axial teinperature profile obtained at Company 13, E and C whilecasting a 1045 grade steel billet. (Note that the temperatures are those recorded by thermocouples located approximately midway between the cold and hot faces of the mould).200150100500-0 100 200 300 400 500 600DISTANCE FROM TOP OF MOULD (mm)700111610Figure 5.16 Load cell response for a 1045 grade billet cast with Canola oil at 25 mi/mm at Company C.0Figure 5. 17 Load cell response for a 1045Conipany C.grade billet cast with Canola oil at 100 mI/mm atliME (s)1117]WAD CELL - MOULD I1LACEMENT26.7 ‘.. A A2662&5264263026226 12625.9 ... ‘1 -25.S1.5 2 2.5 3 3.5 4 4.5 5liME (s)Figure 5.18 Load cell response for a 1045 grade billet cast with Mineral_S oil at 25 mI/mm atCompany C.26.8— WAD CELL .-. MOULD DISPLACEMENTTIME(s)Figure 5. 19 Load cell response for a 1045 grade billet cast with Mineral_S oil at 100 mI/mm atCompany C.IIISIZZLÔACELL MOULD DISPLACEMENTII52253351(s)Figure 5.20 Load cell response for a 1045 grade billet cast with HEAR oil at 25 mI/mm at Company C.LOAD CELL— MOULD DISPLACEMENT2E \7/A[fñtsc1 1.5 2 2.5 3 3.5 4 4.5TIME(s)Figure 5.21 Load cell response br a 1045 grade billet cast with HEAR oil at 100 mI/mm atCorripany C.11191CFigure 5.22 Load cell response for a 1045 grade billet cast with Mineral_O oil at 25 mI/mm atCompany C.CFigure 5.23 Load cell response for aConpany C.1045 grade billet cast with MineraLO oil at 100 mI/mm at11ME()TIME(s)11201Figure 5.24 Load cell response for a 01 $ grade billet cast with Canola oil at 25 mi/mm at Corn—pan)’ C.Figure 5.25 Load cell response for a 1045 grade billet cast with Canola oil at 25 mI/mm at Corn-pan)’ C.Figure 5.26 Load cell response iou a 5 1 6() grade billet cast with Canola oil at 25 mi/mi n at Company C.7D4E121]Figure 5.27 Load cell response for a 1045 grade billet cast with HEAR oil at 25 mi/mm and at144 cpm of mould oscillation at Company C.Figure 5.28 Load cell response for a 1045 grade billet cast with HEAR oil at 25 mi/mm and at 96cpm of mould oscillation at Company C.11[122]U)AD CEll.-----.--. MOULD nlsplNT1MEFigure 5.29 Load cell response for a 11)45 grade biller cast with HEAR oil at 100 mI/mm and at144 cpm of mould oScillation at Company C.27.8— LOAD CELl....-..--.. MOULD DISPlACEMENT27.6 f f ! ; (1!f VTllvffi(s)Figure 5.30 Load cell response br a 1045 grade billet cast with HEAR oil at 100 mI/mm and at96 cpm of mould oscillation at Company C.123Figure 5.31 A typical load cell response for Company B. (Note: Cross marks indicate start andend of negative strip time; open markers correspond to end of load cell decompression).Figure 5.32 A typical load cell response for Company E. (Note: Cross marks indicate start andend of negative strip time; open markers correspond to end of load cell decompression).Figure 5.33 A typical load cell response for Company C. (Note: Cross marks indicate start andend of negative strip time; open markers correspond to end of load cell decompression).flME()[124]Figure 5.34 A typical load cell response for Company B (enlarged). (Note: Cross marks indicatestart and end of negative strip time; square marker corresponds to end of load cell decompression).Figure 5.35 A typical load cell response for Company E (enlarged). (Note: Cross marks indicatestart and end of negative strip time; square marker corresponds to end of load cell decompression).Figure 5.36 A typical load cell response for Company C (enlarged). (Note: Cross marks indicatestart and end of negative strip time; square marker corresponds to end of load cell decompression).[125]Figure 5.37 Macro-etch of a longitudinal section of a 1008 grade billet from Company B showing transverse depressions and cracks at the base of these depressions (Mag. 0.8 X).[126]i271Macr0t of a iongitUth’ sec0fl of a i039 grade bilet from ComPY B showingFigure 3’ surface Mag. 1.0 X).a ‘smoothBillet Profile OrientationProfilometer Side AIcw18 7 6/DirectionSide BFigure 5.39 Schematic diagram showing the eight off-corner sites of a billet.[128]Figure 5.40 Macro-etch of a transverse section of a 1090 grade billet from Company E showingtypical off-corner, internal cracks (Mag. 0.8 X).[129]Figure 5.41 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1018 grade billet from CompanyE. (Note: Numbers on the top of the barsrepresent percentage of billets with cracksFigure 5.43 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1080 grade billet from CompanyE. (Note: Numbers on the top of the barsrepresent percentage of billets with cracksFigure 5.42 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1050 grade billet from CompanyE. (Note: Numbers on the top of the barsrepresent percentage of billets with cracksFigure 5.44 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1090 grade billet from CompanyE. (Note: Numbers on the top of the barsrepresent percentage of billets with cracksI 2 2 4 fsms2 3 4 6 7 62 4 5 6Il00A00io4—too 10000100100100100II.IIIi4Figure 5.45 Bar graph showing averagedepth of off-corner, internal cracks at eightoff-corner sites on the test and control strandfor a 1146 grade billet from Company E.(Note: Numbers on the top of the bars represent percentage of billets with cracks).Figure 5.46 Bar graph showing longitudinaloff-corner depressions on the test and controlstrand for 1018,1050, 1080, 1090 and 1146grade of billets from Company E.00so01a 7I::.ID1010GRADE[131jFigure 5.47 Macro-etch of a transverse section of a 1045 grade billet from Company C showingtypical off-corner, internal cracks (Mag. 1.0 X).[132]‘ 1G6511I0I6Lilz 0j .0ol0.1[o42oilLi22 2500 110.4101I 2 3 6 7 6Figure 5.48 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1018 grade billet from CompanyC. (Note: Numbers on the top of bars represent percentage of billets with cracks).Ii114ls1.c 1045112[e25iI 03 50 38 250 I8IIrIV/Aol ,Figure 5.49 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1045 grade billet from CompanyC. (Note: Numbers on the top of bars represent percentage of billets with cracks).16700T ST. 11*51 5165Li20 50[ol 2050wI2I1 —I 2 3 6Figure 5.50 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 5160 grade billet from CompanyC. (Note: Numbers on the top of bars represent percentage of billets with cracks).1 2 3 4 5 8Figure 5.51 Bar graph showing averagedepth of off-corner, internal cracks at theeight off-corner sites on the test and controlstrand for a 1084 grade billet from CompanyC. (Note: Numbers on the top of bars represent percentage of billets with cracks).4 016_________________1ThT III 10051[4(XI*TWJ.IS100II8 *00 *0 1005. 100000 if00 1 1.1.00[514 5 6 7 70[133]I,,Figure 5.52 Bar graph showing averagedepth of off-corner, internal cracks at eightoff-corner sites on the test and control strandfor a L-325 grade billet from Company C.(Note: Numbers on the top of bars representpercentage of billets with cracks).l341Figure 5.53 Macro-etch of a transverse section of a 1018 grade billet from Company E showingtypical mid-way cracks (Mag. 0.8 X).[135]Figure 5.54 Bar graph showing averagedepth of mid-way cracks at four locations onthe test and control strand for a 1018 gradebillet from Company E. (Note: Numbers ontop of bars represent percentage of billetswith cracks).Figure 5.56 Bar graph showing averagedepth of mid-way cracks at four locations onthe test and control strand for a 1080 gradebillet from Company E. (Note: Numbers ontop of bars represent percentage of billetswith cracks).Figure 5.55 Bar graph showing averagedepth of mid-way cracks at four locations onthe test and control strand for a 1050 gradebillet from Company E. (Note: Numbers ontop of bars represent percentage of billetswith cracks).Figure 5.57 Bar graph showing averagedepth of mid-way cracks at four locations onthe test and control strand for a 1090 gradebillet from Company E. (Note: Numbers ontop of bars represent percentage of billetswith cracks).CI I\1 36Figure 5.58 Bar graph showing average depth ofmid-way cracks at four locations on the test andcontrol strand for a 1146 grade billet fromCompany E. (Note: Numbers on top of barsrepresent percentage of billets with cracks).1371CCc-)C(_)>i1381i*Figure 5.60 Bar graph showing averagedepth of midway cracks at four locations onthe test and control strand for a 1018 gradebillet from Company C. (Note: Numbers onthe top of the bars represent percentage ofbillets with cracks).Figure 5.61 Bar graph showing averagedepth of midway cracks at four locations onthe test and control strand for a 1045 gradebillet from Company C. (Note: Numbers onthe top of the bars represent percentage ofbillets with cracks).351T I251012 I‘1 I‘ Ii 10010 II5’.III,8 *Figure 5.62 Bar graph showing averagedepth of midway cracks at four locations onthe test and control strand for a 5160 gradebillet from Company C. (Note: Numbers onthe top of the bars represent percentage ofbillets with cracks).Figure 5.63 Bar graph showing averagedepth of midway cracks at four locations onthe test and control strand for a 1084 gradebillet from Company C. (Note: Numbers onthe top of the bars represent percentage ofbillets with cracks)..‘138TI——LW ZW B It I 391Figure 5.64 Bar graph showing averagedepth of mid-way cracks at four locations onthe test and control strand for a L-325 gradebillet from Company C. (Note: Numbers ontop of bars represent percentage of billetswith cracks).I 4OFigure 5.65 Photograph of the surface of a 1008 grade billet from Company B (Mag. 1.0 X).{141jFigure 5.66 Photograph of the surface of a 1012 grade billet from Company B (Mag. 1.0 X).[142]>Cc-)CC(-)C-cCtCCIIAa. a._— AcC. J c!.GgA€ IQIêAllI f f f Lt 11W! RAT! (VunFigure 5.68 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade test strand bifiet cast with Canolaoil at Company E.Figure 5.69 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade test strand billet cast with HEARoil at Company E.aFigure 5.70 Graph showing the influence ofoil flow rate oi-cscill&u mark depth of a1018 grade test strana billet cast with Mineral_S oil at Company E.[144]IAv a. a.v a. ovl6IWJNI AFl!I6Illx -—4.I •1.[ •1Figure 5.71 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade test strand billet cast with Soybean oil at Company E.IA—-a. LT a.°J6-—IllI IiII..tIR I I 1I.. I.1W! RATE (l/mi)161.4L a. a.Ije— Ills I•1’ 1 tI IT46flaw 16Figure 5.72 Photograph of the surface of a 1018 grade billet cast on the test strand with Canolaoil at 65 mi/mm at Company E (Mag. 1.0 X).[145]Figure 5.73 Photograph of the surface of a 1018 grade billet cast on the test strand with HEARoil at 65 mi/mm at Company E (Mag. 1.0 X).[1461Figure 5.74 Photograph of the surface of a 1018 grade billet cast on the test strand with Minera1S oil at 65 mi/mm at Company E (Mag. 1.0 X).[147]Figure 5.75 Photograph of the surface of a 1018 grade billet cast on the test strand with Soybeanoil at 65 mi/mm at Company E (Mag. 1.0 X).[148]Figure 5.76 Graph showing the influence ofoil flow rate on oscillation mark depth of a1080 grade test strand billet cast with Canolaoil at Company E.0.4LI a. LI a. a.T LI a. T0.2502[.tit45FIDI 6612 (muAs) 05Figure 5.78 Graph showing the influence ofoil flow rate on oscillation mark depth of a1080 grade test strand billet cast with Mineral_S oil at Company E.Figure 5.77 Graph showing the influence ofoil flow rate on oscillation mark depth of a1080 grade test strand billet cast with HEARoil at Company E.U a. LI a. OT035jr.1. I03IiI I Ia6602 RATE (ail/ün)aFigure 5.79 Graph showing the influence ofoil flow rate on oscillation mark depth of a1080 grade test strand billet cast with Soybean oil at Company E.LI a. o:wa a. LW a. a0.55-____054LElU0Q16.3[20.45+WW RATE (mVLW a.45 ‘5035r+ •ff456602 RATE frA/mi)25 65II I.G0ALIC I{i It[149]Figure 5.80 Photograph of the surface of a 1080 grade billet cast on the test strand with Canolaoil at 45 mi/mm at Company E (Mag. 1.0 X).[150]Figure 5.81 Photograph of the surface of a 1080 grade billet cast on the test strand with HEARoil at 45 mi/mm at Company E (Mag. 1.0 X).[1511Figure 5.82 Photograph of the surface of a 1080 grade billet cast on the test strand with Mineral_S oil at 45 mi/mm at Company E (Mag. 1.0 X).[152]--Figure 5.83 Photograph of the surface of a 1080 grade billet cast on the test strand with Soybeanoil at 45 mi/mm at Company E (Mag. 1.0 X).[153]Figure 5.84 Graph showing the influence ofoil flow rate on oscillation mark depth of a1090 grade test strand billet cast with Canolaoil at Company E.ILW I. W LW LWt . .f .f f + +451DW RATE (mVm)85Figure 5.85 Graph showing the influence ofoil flow rate on oscillation mark depth of a1090 grade test strand billet cast with HEARoil at Company E.Figure 5.86 Graph showing the influence ofoil flow rate on oscillation mark depth of a1090 grade test strand billet cast with Mmeral_S oil at Company E.FlOW RATE (mI/mAi)ROW RATE (m)/rn)[154]Figure 5.87 Photograph of the surface of a 1090 grade billet cast on the test strand with Canolaoil at 45 mI/mm at Company E (Mag. 1.0 X).-...[1551Figure 5.88 Photograph of the surface of a 1090 grade billet cast on the test strand with HEARoil at 45 mI/mm at Company E (Mag. 1 .0 X).[156]Figure 5.89 Photograph of the surface of a 1090 grade billet cast on the test strand with Mineral_S oil at 45 mi/mm at Company E (Mag. 1.0 X).[157]LI ft LI ft I LI ftI tt4 ItFIDW RATE (mi/mm)Figure 5.90 Graph showing the influence ofoil flow rate on oscillation mark depth of a1146 grade billet cast with Canola oil atCompany E.Figure 5.91 Graph showing the influence ofoil flow rate on oscillation mark depth of a1146 grade billet cast with HEAR oil atCompany E.DI RATE (mi/mm)158]• •; •:: :1Figure 5.92 Photograph of the surface of a 1146 grade test strand billet cast with Canola oil at 65mi/mm at Company E (Mag. 1.0 X).[159jFigure 5.93 Photograph of the surface of a 1146 grade test strand billet cast with HEAR oil at 65mi/mm at Company E (Mag. 1.() X).[160]--.-—-- --PWO RATE (m1/mi.)tooIIT70PlOW RATE (VsoFigure 5.94 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade billet cast with Canola oil atCompany C.Figure 5.95 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade billet cast with HEAR oil atCompany C.•4a. v a. o, N a.SILl t.GRADS 5016IL2E——-H--..*11 --t25 70 500PLOW RATE (mi/mm)Figure 5.96 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade billet cast with Mineral_S oil atCompany C.01-I-Q” -b00.150.1It IFigure 5.97 Graph showing the influence ofoil flow rate on oscillation mark depth of a1018 grade billet cast with Mineral_O oil atCompany C.CRAD0 5058..4N m:w N a. ai N a. a,0.4El a. 6;W a. T El a. Wi1E..Z— .. I0.30CIN1A 25.) mciiii]&30 - —.-..LI0 -— -ILI- —Oil —-‘.1--1-0.4Ia. o El a. or ro a. 0..WE 0.3 )slm. coox 5518 I25 70 100ftOW RATE (rrd/mks)[1611Figure 5.98 Photograph of the surface of a 1018 grade billet cast with Canola oil at 25 ml/min atCompany C (Mag. l.() X).[162]Figure 5.99 Photograph of the surface of a 1018 grade billet cast with HEAR oil at 25 mllmin atCompany C (Mag. 1.0 X).[163]Figure 5.100 Photograph of the surface of a 1018 grade billet cast with Mineral_S oil at 25mi/mill at Company C (Mag. 1.0 X).[164]Figure 5.101 Photograph of the surface of a 1018 grade billet cast with Mineral_O oil at 25mi/mm at Company C (Mag. 1.0 X).[165]Figure 5.102 Graph showing the influence ofoil flow rate on oscillation mark depth of a1045 grade billet cast with HEAR oil atCompany C.LW LWafr 0_Is2flDO R4TE (a1/taoFigure 5.103 Graph showing ihe influence ofoil flow rate on oscilla&o mark depth of a1045 grade billet cast with Mineral_O oil atCompany C.l660.LW a ozW LW (L T•(f1 IS’14 Ie,x -C.I_.t .ftnfl, RL1t (mVm)tooFigure 5.104 Photograph of the surface of a 1045 grade billet cast with HEAR oil at 100 mI/mmat Company C (Mag. 1.0 X).[1671Figure 5.105 Photograph of the surface of a 1045 grade billet cast with MineralO oil at Company C (Mag. 1.0 X).[168]L4Lw a. Lw co a. 0:1025Ic4k’1 S1 02*1)0 oIJr+ H220102 ROTE (mt/min)0102 RATE (*nI/mAi)iào100drH2250102 ROTE (ml/mA)0102 kAlE (rnl/mw)tao(02Figure 5.108 Graph showing the influence ofoil flow rate on oscillation mark depth of a5160 grade billet cast with Mineral_S oil atCompany C.Figure 5.109 Graph showing theinfluierceofoil flow rate on oscillation mark dcpth of a5160 grade billet cast with Mineral_O oil atCompany C.LWFigure 5.106 Graph showing the influence ofoil flow rate on oscillation mark depth of a5160 grade billet cast with Canola oil atCompany C.a a. ILl C! a. Lw025_____S1T11I05C5l60a.[20(5Figure 5.107 Graph showing the influence ofoil flow rate on oscillation mark depth of a5160 grade billet cast with HEAR oil atCompany C.LALV Q. ILl Cl a. Lw000t:i 51G2*D25L60I1.10.0LI a. OCT LI a. 0:1O-iii 1 QV01O45f[2M•PlOt OT (JfmioióoFigure 5.110 Graph showing the oscillationmark depth for a 1045 grade billet cast withHEAR oil and with a mould oscillation of 96cpm at Company C.GALI 0. OCT LI a. a.’wDC 10*1owi RaTE (G/)iioFigure 5.111 Graph showing the oscillationmark depth for a 1045 grade billet cast withMineral_O oil and with a mould oscillationof 96 cpm at Company C.CALI 0. OCT LW 0._____I2[.3bT RAIl (oi/uñn)Figure 5.112. Graph showing the oscillationmark depth oa 1080 grade billet cast withSoybean oil and with a mould oscillation of130 cpm at Company E.I[170]to a. ow to a. woil11 1 U00 1045 IPlOt RAE (ml/m6100Figure 5.110 Graph showing the oscillationmark depth for a 1045 grade billet cast withHEAR oil and with a mould oscillation of 96cpm at Company C.to a. :v to a. TOifST00 1045.f •f {FlOW SAlE (ml/.üu)100Figure 5.111 Graph showing the oscillationmark depth for a 1045 grade billet cast withOlex oil and with a mould oscillation of 96cpm at Company C.0.4to a. W to0.00I 108002520.15: I46 46 46 - 45 40FlOW SAlE (mJ/oAoFigure 5.112 Graph showing the oscillationmark depth for a 1080 grade billet cast withSoybean oil and with a mould oscillation of130 cpm at Company E.1171]Figure 5.114 Macro-etched longitudinal section of a 1018 grade billet showing a typical exampleof craze cracking on the inside radius at Company B (Mag. 1.0 X).[172]300200150U XX100KK KI50I IXX KIL K K, K K!,0 LI 02 0.3 0.4 0.5 0.0 0.7 0.0 09CA1N 14T (2Figure 5.117 Plot ofcraze crack index versuscarbon content of billets collect. at Company C.300250K—I,5 0:4 0ooa aor (2200KX1 K04XX KXXX KXKX*bi K —K100KK K!K KzK KK0 ,K fl.K026 0.1 0.15 02 026 0.2 0.35 0.4 0.46I31P QIN’ID4r (2Figure 5.115 Plot ofcraze crack index versuscopper content of billets collected at Company C.Figure 5.116 Plot of craze crack index versusnickel content of billets collected at Company C.K200XXXXZXzKI100KKK50XX KK0 .K, . U , K,0 04 I 02 1.4 1.0 14N PJ3S3Figure 5.118 Plot of craze crack index versusNi/Cu ratio of billets collected at CompanyC02 0.4 0.0[1731Figure 5.119 Surface of a 1080 grade control strand billet cast with Soybean oil at 65 mi/mm atCompany E showing evidence of beads (Mag. 1.0 X).[174]Figure 5.120 Surface of a 1080 grade test strand billet cast with Canola oil at 65 mI/mm showingsevere “zipper marks” at Company E (Mag. 0.8 X).[175]Figure 5.121 Transverse macroetch through a bead on a 1080 grade billet at Company E (Mag.14 X).[176]Chapter 6: MATHEMATICAL MODELLING OF MOULD BEHAVIOURThe three main mathematical models used in the study are discussed in this and the followingchapter. This chapter discusses the two mathematical models of the mould - the first to obtain theheat flux from the billet to the mould wall and the second to obtain the distortion of the mouldduring service. A detail presentation of the results of the models is deferred till Chapter 8.6.1 Mathematical Model of Heat-Flow in the MouldThe mathematical model of heat flow in the mould is used to obtain the heat flux profile downthe mould length. A pre-existing mathematical model of the mould [18] that simulates the heat flowdown the length of the mould was used. The program models a longitudinal section through themid-face of the mould wall. The input to the model is an assumed heat flux profile which is alterediteratively so as to match the predicted mould temperatures with those recorded by the thermocouples. Owing to the iterative nature of the calculations, the mathematical model was considerablymodified to facilitate it use.A schematic diagram of a longitudinal mid-plane through the mould wall is shown in Figure6.1. The simplifying assumptions of the unsteady-state, two-dimensional, finite-difference, mathematical model are as follows.[1] Transverse heat flow in a direction perpendicular to the plane of interest are negligibly small.This follows from symmetry.[2] Temperature variations due to mould oscillation and metal level fluctuations are ignored.This is permissible as an average value of the temperature distribution in the mould is used.[3] Temperature dependence of the thermal diffusivity has not been considered as its effect onmould temperature is negligible.[177][4] The top and bottom surfaces of the mould are treated as adiabatic surfaces as they havetemperatures that are close to the ambient temperature.[51 The cooling water channel extends to the top and bottom ends of the mould and the coolingwater is in plug flow. The latter assumption is supported by the fact that the water flow isturbulent (Re 70,000) and the channel gap is less than 5 mm. There is no heat transferbetween the cooling water and water jacket.The governing equation for heat conduction in the mould wall in two dimensions is( aT”1 ( (6.1)while a sectional heat balance for the cooling water yields3TpVdC—-—h(z,t)(T(0,z,t)—T(z,t))=0 (6.2)The boundary conditions and the details of the model have been discussed in the originalwork [18]. The mould continuum was discretized (Figure 6.1) and the finite difference equationsfor the system of nodes obtained by transforming the heat flow equations and relevant boundaryconditions for each node into a finite difference form was solved by the alternating direction implicitmethod [68].6.1.1 Characterisation of heat transfer in the water channelIn a procedure described by Rohsenow [69] and discussed in detail by Samarasekera andBrimacombe [18], forced convection boiling curve for water was constructed (Figure 6.2). Thereare three distinct regions of the forced convection boiling curve : Forced Convection (FC), theTransitions region (TR) and the Fully Developed Boiling region (FDB). In the forced convectionregime of the above mentioned curve the heat flux can be calculated by the following expression[18].[178]= h(T Tsat) (6.3)where hJCDH = O.O23p1VD1j8[c]O.4 (6.4)and all values are evaluated at the average temperature of water.The heat flux in the fully developed nucleate boiling region is given byQFD[(Tw_Tsai)x_xjx’ (6.5)fg sJ g0The heat flux in the transition region is given by2 0.5QTR=QI:C[1 J (6.6)The heat flux at the inception of boiling is given by the expressionQFIV 5.28 1 x 103P1.156(1 .8(T T))°°4 (6.7)With a decrease in the number of active nuclei for bubble formation due to degassing of asurface over a long period of time, larger superheats than predicted by Equation 6.7 are requiredto initiate and sustain boiling. The boiling of the water in the water channel can thus cause a hysteresisin the boiling curve leading to thermal cycling- a point explained in detail in the original work[18j. A transient heat flow model was developed to simulate the time-dependent effects of boilingon the thermal field in the mould.The changes made in the programme were done mostly to make the code easier to use.6.2 Mathematical Model of Mould DistortionA three-dimensional, elastic-plastic, finite-element model of the mould wall, described indetail in an earlier publication [62], was used to compute the thermal expansion of the mould duringservice. This model was used as a ‘black box” and a programme was written to generate data for[179]input to the finite-element model. This programme generates the appropriate mesh and uses thetemperature distribution in the mould, available from the previous programme, to associate atemperature with each element in the mould wall. The mesh generated by this programme and usedin the finite-element model is shown in Figure 6.3. It may be mentioned in the passing that themesh-generating programme is semi-general purpose programme in that it can generate the meshgiven the mould dimension and the number of nodes in each direction.The use of a three-dimensional model has been rationalised [62] on the basis of the mouldshape: since the mould is a square or rectangular tube with an inherent rigidity, the corners can beexpected to exert a restraining effect on the movement of the mid-face and thus a three-dimensionalanalysis becomes a necessity.[1801x=o X= XMxZ=0Z=ZFMould JacketZ=ZMCooling WaterFigure 6.1 Mesh used for modelling a longitudinal section of the mould wall [18].[181]EFigure 6.2 Forced convection- heat transfer curve [181.[182)Figure 6.3 Mesh used to model mould distortion [62].Z.w@485 4486t. / 1478/ / 4481/4487/4488447 472’1489 44904473/ (474/449’447/ 1141 •j_—i 4470/J//A/ 39 38 37 / 38 d 39 403.e 29_0 --4. 4 .‘ !L8)/8 /-;,,7/(0/ 772/7Q) Y.v[183]Chapter 7: MATHEMATICAL MODELLING OF BILLET SOLIDIFICATION AND SHRINKAGEThis chapter describes the development of a mathematical model that predicts the progressof solidification and estimates the shrinkage profile of a billet during its passage through the mould.7.1 Mathematical Model of Billet ContractionA mathematical model to describe the heat flow in a continuously cast strand and to computethe shrinkage of the billet as a function of its axial position in the mould was developed. The modelis based on the equation for two-dimensional, unsteady-state heat conduction in one quarter of atransverse slice of the strand, as follows:( T” ( (7.1)The initial and surface boundary conditions (see Figure 7.1 for the mesh used), expressedmathematically, are as followst=O, O<x4, O<y<, T(x,y)=T (7.2)yt>O, x=O, Oy, —k--=q0 (7.3)xt>O, y=O, Ox—, —k--—=q0 (7.4)2Symmetrical heat flow about the center planes is assumed and fomu1ated as follows:x ytO x=—, Oy—. kç=O (7.5)2 2 3xY Xt 0, y = , 0 x , —k9--- = 0 (7.6)[184]Equation 7.1 was solved, subject to the above initial and boundary conditions, using analternating direction implicit, finite-difference method as developed by Peaceman and Rachford[68].Convective heat transfer in the pooi was neglected, an assumption justified by Mizikar [70]and Szekely et al. [711, and the latent heat of solidification was taken into account by consideringequilibrium freezing. The specific heat of the steel was increased in the temperature range betweenthe liquidus and the solidus temperature of the steel to account for the release of latent heat. A linearrelease of specific heat in the mushy zone has been found to be adequate in an earlier work [72].The temperature dependance of the specific heat and thermal conductivity of steel was obtainedfrom the work of Thomas, Samarasekera and Brimacombe [73] and is shown in Table 7.1 and 7.2.To safeguard against nodes that “jump’ over the mushy region in a single time step and thus misstheir latent heat evolution, a post-iterative correction technique was used to readjust the temperatureof these nodes. Calculations were performed with different node sizes and time steps to determinethe optimum magnitude of these variables.7.2 Shrinkage CalculationThe following assumptions were made in the calculation of billet contraction.(i) The different phases present in the billet at any temperature are those given by theFe-C phase diagram. This is in contrast to an earlier work [1] where austenite wastaken to be the only phase present.(ii) The shrinkage of a billet is computed by incorporating the contraction associated withphase change (wherever applicable) with the normal thermal contraction associatedwith cooling.(iii) The effect of ferro-static pressure is neglected.[185](iv) The mechanical behavior of the solidified shell is neglected. Neither the strainsimposed by the stress field nor creep of the solidified shell is included in the model.The initial dimensions of the steel billet were taken as those of the distorted copper mould atthe meniscus. The differential coefficient of linear thermal expansion of steel was calculated bycomputing the fraction of different phases present from the phase diagram by the lever rule andcalculating the shrinkage associated with these phases. This represented a major effort in the presentwork and is examined in some detail in a subsequent section.The calculation of billet contraction was carried out by the following procedure(i) A transverse slice of unit thickness of the billet was allowed to move down the mouldin discrete time steps. At each time step, the temperature distribution in the slice wascalculated.(ii) The calculation of billet contraction was initiated at the instant the first row of nodessolidified. Prior to the initiation of solidification the billet dimensions were taken tobe the dimensions of the distorted mould.(iii) For the shrinkage calculation, at a given time step, the temperature change of eachnode relative to the previous time step was determined.(iv) The length change of each node in the row of solidified nodes, due to the temperaturechange, was calculated relative to the previous time step.(v) The change in the length of the first solidified row was obtained by summing the linearexpansion/contraction of all nodes in that row.(vi) The calculation of the linear expansion/contraction of the second row was started assoon as all nodes in that row had solidified and so on.(vii) As soon as a row solidified, its starting length at that instant was taken as the averagelength of the rows adjacent to it that had already solidified.[186j(viii) The average length of all the solidified rows at any given time step was taken as thebillet dimension at that time step.(ix) This procedure was repeated until the bottom of the mould was reached.The procedure outlined above could be done for rows or columns but the interaction betweenrows and column was not considered. The model does not account for the bulging of the steel shellon account of ferro-static pressure but this approach enables quantification of mould taper necessaryto prevent the outward bulging of the shell.7.3 Transverse Variation of Heat FluxThe drop in heat flux from the mid-face to the corner of the billet is well documented[6,17,32,74]. The drop in heat flux arises from a wider air gap in the corner region compared to thewidth of the gap between the mould wall and the mid-face of the billet. In order to characterize thistransverse heat-flux variation, a trial-and-error approach had been adopted in the past [1] where thecalculated contour of the solid-liquid interface was made to match metallographically observeddark bands in macro-etched transverse billet sections, by varying the heat flux in the transversedirection (Bommaraju et al. [17,58,75], have shown that the dark solidification bands formed nearthe exit of the mould). Additionally, the transverse variation in the heat flux had been allowed tocommence only after an initial solidification time of one second elapsed.Unfortunately the dark solidification bands are extremely difficult to observe in low-carbonbillets and thus a different method was adopted as follows. The heat-transfer coefficient was calculated from the measured heat flux at the mid-face of the billet and the temperature differencebetween the billet surface (calculated) and the cooling water, then was held constant across thebillet face. Subsequently the heat flux at any transverse position on the billet face was obtained bythe product of the heat-transfer coefficient and the difference between the local billet surfacetemperature and the mould cooling water temperature. Toward the corners of the billet, the surface[1871cools more rapidly due to two-dimensional effects and thus the temperature difference, and heatflux decline. This approach to account for the lower heat flux in the corner region of the billet is,at the best, an approximation but appears to give reasonable results.7.4 Calculation of (Carbon- and Temperature-Dependent) Coefficient ofThermal Linear Expansion of Steel.In computing billet shrinkage in the past [11 the coefficient of linear thermal expansion forsteel has been assumed to be a constant, independent of temperature and steel grade. While thisapproach may have been adequate as a first approximation, it is imperative to obtain a more representative value of this coefficient to incorporate the effect of the delta ferrite-gamma phasetransformation during cooling.One of the greatest stumbling blocks in the present work has been the paucity of data on thecoefficient of linear thermal expansion of delta and gamma iron in the temperature range of interest(> 1200 °C). A search of published literature failed to reveal any direct data on the coefficient ofexpansion of gamma and delta phases. The only relevant publication that could be used, to a limitedextent, is the work of Wray [761 in which there is a compilation of the measurements of themechanical, physical and thermal properties of iron and plain carbon steels. However, whenexamined in detail, there is difficulty in using some of the relationships given in the publication.For instance, the equation suggested by Wray for the volume increase of austenite with carbon doesnot appear to show any effect of temperature on the dilation of the gamma iron lattice caused by agiven amount of carbon. This is in contradiction to the work of Ridley and Stuart [77], on whichWray’s equation is based, where the authors clearly show that with increasing temperature theamount of dilation caused by a given amount of carbon, decreases. Furthermore, no source is referred1188]to in Wray’s paper for the equation governing the dilation of delta iron by carbon. However, theapproach suggested in his work was followed in this study to calculate the desired equationsindependently.To be able to predict the thermal expansion-contraction in a two-phase region of the steel, itis necessary to know the temperature and composition dependence (if any) of the density of thetwo phases. Once these are obtained, the phase diagram and the lever rule can be applied to predictthe density and contraction of steel in the two-phase region as indicated earlier. Unfortunately thedensities of the delta and gamma phases, as a function of carbon content and temperature, are notavailable in the temperature range of interest (> 1200 °C). These values, therefore, have to becomputed first from the dimensions of the corresponding unit cell for the pure delta and gammairon and then modified by superimposing the effect of dilation due to the carbon atoms. Thus thelattice parameters of the unit cell are functions of temperature and the carbon content (which dilatesthe cell). Furthermore, the degree ofdilation for a given carbon content may, by itself, be temperaturedependent.7.4.1 Variation of lattice parameter of delta phase with carbon content and temperatureFor the temperature dependence of the specific volume of pure delta iron, Wray’s [76]modification of Lucas’ [78] data was adopted.V = 0.1242 + 8.70 (lOj (T —20) (Lucas) (7.7)V= 0.1234 + 9.38 (lOj(T —20) (Wray) (7.8)where T is the temperature in °C and V is the specific volume in cm3/g of ö iron.Fasiska and Wagenblast, in an earlier work [79], have tabulated the lattice parameter of thealpha phase for different carbon contents at room temperature. Regression analysis on their carefullyobtained experimental data leads to the following equation:[189]a a+8.40(l03)(X) (7.9)where a is the lattice parameter of the phase in A and X is the carbon content of the phase inatomic percentage.The researchers have also noted that the expansion in the lattice of alpha iron (henceforthreferred to as ‘lattice dilation coefficient’), on account of carbon, was not temperature dependent.Based on the last observation this ‘lattice dilation coefficient’ was taken to be representative forthe high-temperature delta iron having a BCC structure as well.7.4.2 Variation of lattice parameter of pure gamma iron with temperatureAs in the case of pure delta iron, the temperature dependance of the specific volume of puregamma iron was also taken from Wray’s [76] modification of Lucas’ data [78].V1= 0.1221 + 9.70 (lOj(T —20) (Lucas) (7.10)V1= 0.1225 + 9.45 (lOj(T —20) (Wray) (7.11)where T is the temperature in °C and V is the specific volume in cm3/g of y iron.7.4.3 Variation of lattice parameter of gamma phase with carbon contentRidley and Stuart, in their work on partial molar volumes of iron-carbon austenite [77], havegraphed the variation with carbon content of the lattice parameters of austenite for a wide range oftemperatures (25 -1200 °C) and given equations for some of the plots. Furthermore, they observedthat the slope of the lines gradually decreased with increasing temperature indicating that the dilationeffect of a specific amount of carbon decreased.As a first step toward obtaining the effect of carbon content and temperature on the dilationcoefficient, the method of least squares was used to fit curves to their raw data. The results aresummarized in Table 7.3 and Figure 7.2. It was then possible to derive, for a given temperature,the effect of carbon content on the dilation coefficient. A separate curve fitting exercise had to be/11901conducted to determine the effect of temperature on the dilation coefficient, the results of whichare shown in Figure 7.3. The parabolic nature of the curve is consistent with the observation ofRidley and Stuart that the slope of the plot of lattice parameter versus carbon content graduallydecreases with temperature.As a result of the curve fitting exercise it was possible to predict the lattice parameters (andtherefore density or specific volume) of the gamma phase for any given carbon content and temperature combination as follows.aT =aT +(0.0317—ll.65 (lOjT—0.05 (107)T2(W) (7.12)Ycwhere T is the temperature in °C, a is the lattice parameter of ‘y phase in A and W is the carboncontent of the phase in weight percent.The variation of the coefficient of linear expansion of steel is shown in Figure 7.4 for threedifferent carbon contents. It should be noted that the plots are of the mean (and not differential)coefficient of linear expansion of steel which is based on a reference length at the solidus temperatureof steel and can only be used to compute contractions in lengthfrom the dimensions at the solidustemperature.The value of the linear expansion coefficient for those grades that do not undergo phasetransformation (C > 0.18%) over the temperature range of interest, is nearly constant with temperature. The plot for steel with a carbon content of 0.15% represents grades (0.10% <C <0.18%)where cooling below the solidus temperature causes shrinkage on account of both thermalcontraction and phase transformation, the latter being much greater in magnitude. In such cases thetotal contraction in the early stages of cooling, is dominated by the effect of phase change in whichthe expansion coefficient has a high initial value reflecting the transformation. With further coolingbelow the transformation-end temperature, the contribution of thermal contraction to the totalcontraction from the solidus becomes progressively higher than the contribution from the phase[19 1]change as can be seen in Figure 7.4. The third plot for steel with a carbon content of 0.05% representsgrades (C <0.10%) that undergo phase change only after the temperature is well below the solidus.The almost constant value of mean expansion coefficient around 1500°C is indicative of contractionon account of thermal cooling of delta phase only. With the onset of phase transformation theexpansion coefficient increases to a relatively large number as explained above and then graduallydecreases with temperature. Because a mean coefficient of linear expansion is being considered,its highest value in the latter case (C = 0.05%) is lower than that attained with the 0.15% carbonsteel.7.5 Model VerificationWhile a detailed discussion on the results from the model is deferred to the next chapter, atabulation of the model predicted shell thickness is given in Table 7.4 along with the measured shellthickness of the billet. As can be seen from the table the predicted shell thickness at the bottom ofthe mould adequately matches those measured from solidification bands in macro-etches of billetsamples.[1921Table 7.1 Enthalpy and specific heat functions used in the heat-flow model [73].Temperature (°C) Enthalpy (kJ/kg) Specific heat (kJ/kg°C)0.Otoll4.3- 499114.3 to 491.4 —23.3 ÷0.456T+ 1.88 x 10T2 0.456 +2x 1.88 x 10T491.4 to 697.1 —13.4 +0.268T +4.18 x 10T2 0.268 +2x4.18 x 1OT697.1 to742.9—595÷1.431T 1.431742.9 to 868.6 —1348.9 +3.849T — 1.883 x 103T2 3.849—2 x 1.883 x 103T868.6 to 1142.9 11.7 +0.648T 0.6481142.9 to T50 228.3 +0.268T + 1.67 x 10T2 0.268+2 x 1.67 x 10TT501 to Tijq H+272T1 CpsoI +T272liq sol liq sal> Tijq 97.5 +0.787T 0.787Table 7.2 Thermal conductivity function for low carbon steel used in the heat flow model [731.Temperature (°C) (kW/mK)0- 800 59.4—0.418T800 - T501 18.4 +0.0094TT01-Tijq K01 + (43 —K01)> Tijq 43[1931Table 7.3 Results of regression analysis on the experimental data of Ridley and Stuart [771.Temperature (°C) Lattice Parameter of gamma iron (A) r225°C 3.5737+0.0316(wt%C) 0.986200°C 3.5878+O.0314(wt%C) 0.989400°C 3.6035+0.0311 (wt % C) 0.993600°C 3.6193+0.0304(wt%C) 0.991800°C 3.6356+0.0296 (wt % C) 0.9941000°C 3.6521 + 0.0291 (wt % C) 0.9861200°C 3.6684+0.0282 (wi’ % C) 0.982Table 7.4 Mid-face shell thickness at the bottom of mould for billets of different grades atCompany B.Carbon (%) Measured Shell Thickness (mm) Computed Shell Thickness (mm)0.05 8.5 8.00.09 7.0 7.10.12 (not observable) 6.00.15 9.5 9.40.42 9.5 9.4[194)______ ______ ______ ______I x=X/2q0Figure 7.1 Mesh used for modelling one quarter of a transverse section of a billet.xy—(1,1) (1,j)0 0 0 00 0 0 0(i,j)0 0 0 00 0 0 0(i,1)----0-- --0-y= Y/2—-—0-— -—0----q--[195]1000°C0 800 C3.68.600C3.66x400°C:3.64 • 200C3.58Symbols: Experimental Dataof Ridley and Smart [77]Les : Best Fit I3.560 I 0.’8 1. 1.6WEIGHT PERCENT CARBONFigure 7.2 Lattice parameter of austenite as a function of carbon content at varioustemperatures F771.[196}C-)CzC-)1J0C-)z0Figure 7.3 Effect of temperature on the dilation coefficient for austenite.TEMPERATURE (°C)[197]50 CARBON (%) /1 0.05 1w 45 0.15—0.42800 900 1000 1100 1200 1300 1400 1500 1600TEMPERATURE (oC)Figure 7.4 Variation of mean-linear coefficient of thermal expansion of steel with temperature.[1981Chapter 8: MODEL PREDICTIONSThis chapter describes the results of the various mathematical models utilized to analysemould thermocouple data. The details of the three models used have been described in the precedingtwo chapters. Most of the model predictions are analysed in this chapter itself while a discussionon the mechanism of mould heat transfer is deferred until the next chapter.8.1 Heat-Flux Profile and Mould Wall TemperaturesAxial heat-flux profiles were obtained by applying the mathematical model of heat flow inthe mould. As explained earlier, the input to the model is an assumed heat-flux profile that is changediteratively until the predicted mould temperature profile matches that measured by mould thermocouples. The results from the model runs are presented below.8.1.1 Typical heat-flux profileFigure 8.1 is a plot of heat-flux profiles as a function of position below the meniscus asobtained while casting different steel grades (C = 0.05% 0.42%), at Company BA typicalfeature of the heat-flux profile is that it starts with a high value at the meniscu.s andthen drops to a lower value after which it remains more or less constant. The shape of the curve isexplained in three parts as follows.[A] The first contact of liquid steel with the copper mould leads to a high value of heat transfer.This corresponds to the peak heat flux.[B] The formation of a solid shell, that pulls away from the mould wall, gives rise to the formationof an air gap, thereby lowering the heat transfer to the mould.[199][C] The almost constant nature of the heat flux in the lower part of the mould is indicative of howwell the parabolic taper of the mould wall, at Company B, follows the shrinking billet profile.(It will be shown in a subsequent section that this taper is excessive for casting low carbonsteels).8.1.2 Effect of carbon content of steelThe effect of carbon content of steel billets on the heat transfer to the mould can be seen inFigure 8.1 for steels with carbon contents between 0.05% and 0.42%.It can also be seen from Figure 8.1 that the lowest heat flux is obtained with a 0.12% C steeland the highest with a 0.42% C steel in the carbon range for which data was collected at this plant.Both these observations are consistent with those reported in literature [17,34]. The low heat fluxin the case of 0.12% C has been attributed to the shrinkage accompanying the 6—7 transformation,which occurs in the solid state closest to the meniscus at this carbon level. Differential shrinkagecauses the surface to become rippled thereby increasing the local billet-mould gap which leads toa drop in heat transfer.It was shown in Table 7.4 that the mathematical model of billet solidification employing theseheat-flux profiles predicts a solid shell thickness at the bottom of the mould in agreement with theexperimentally measured values of the shell thickness (obtained from dark solidification bands).The heat-flux profiles for higher carbon billets cast at Company C are shown in Figure 8.2(C = 0.18%- C = 0.56%). The plot clearly shows there is little dependence of the heat flux on thecarbon content provided the carbon content of the billet is not in the range that undergoes peritecticreaction (C = 0.10%- C = 0.18%).8.1.3 Effect of oil flow rateAs was mentioned in the previous chapter, the flow rates of all four oils tested, could bechanged in a systematic manner at Plant C. Figure 8.3 shows the axial heat-flux profiles for a 1018[200]grade heat cast with Canola oil at 0, 25, 70 and 100 mi/mm. (Note: The axial heat-flux profile with70 mi/mm of Canola oil lies between those obtained with a flow rate of 25 and 100 mi/mm and hasbeen omitted from the graph for clarity). Thus the main effect of the oil flow rate on the heat fluxprofile is seen near the meniscus; to observe the oil effect more clearly an enlarged graph of theheat-flux profile in the meniscus region is shown in Figure 8.4. In addition Figure 8.5 shows, in theform of a bar chart, the amount of heat extracted in the meniscus region of the mould for the fouroil flow rates while Figure 8.6 shows the hot face temperature of the mould in the above cases.It is clear from Figures 8.3 and 8.4 that an increase in the oil flow rate leads to an enhancementin heat transfer at the meniscus. This can be explained by the fact that the temperature of the hotface of the mould at Company C (-280°C) is well above the boiling points of all the oils (Mineral_Soil starts boiling at 170°C while Canola oil begins boiling at 205°C). Consequently any oil that ispushed past the meniscus, during the down stroke of the mould, will boil and the gaseous hydrocarbons thus produced will ‘crack” against the hot steel shell leading to their decomposition predominantly into carbon and hydrogen. Because the thermal conductivity of hydrogen (k = 0.475W/m2°C) is about 7 times that of nitrogen (k = 0.0675 w/m2°C ), the thermal resistance of the airgap (80% Nitrogen) is considerably decreased with the presence of Hydrogen gas, leading to asignificant enhancement in heat transfer. (While a detailed discussion on this subject will bepresented in the next chapter, it needs to pointed out that such enhancements in heat transfer atthe meniscus seem to depend on the temperature ofthe hotface ofthe mould being above the boilingpoint of the lubricating oil). An increase in the flow rate of oil would clearly increase the amountof hydrogen in the gap leading to a further increase in the heat transfer to the mould as shown inFigures 8.3 and 8.4.Figure 8.5 reveals an extremely important feature : an increase in the flow rate of the oil hasa significant effect on the heat transfer at the meniscus only at lower flow rates. With an increase[20Iin oil flow rate from 0 to 25 mi/mm the enhancement in heat transfer is 13% while from 25 to 70mi/mm it is — 2% and from 70- 100 mi/mm it is — 1%. An understanding of the largest effect atlow oil flow rates was obtained from the video tapes of the meniscus region taken during theindustrial trials. Thus it was seen that an increase in the flow rate of oil beyond 25 mi/mm createda discontinuous layer of oil on the liquid surface as excess oil was squirted out of the meniscusarea during the negative strip period. This “excess” oil does not flow past the meniscus but progressively boils on the steel surface itself. Other evidence in support of this is the observation madein an earlier chapter that the flow rate of oil had no influence on mould-billet friction forces asdetermined by the load cell response.An obvious inference from the above discussion is that, provided uniform oil distributionaround the mould periphery is maintained, there is little to be gained, both form the viewpoint ofheat transfer and mould lubrication, in increasing the oil flow rate beyond 25 mI/mm (which isapproximately 0.05 mI/mm per mm of mould perimeter). It will not be out ofplace to mention thatat the three companies at which trials were carried out the standard oil flow rate was 25- 35times the above value.8.1.4 Effect of oil typeHaving established the mechanism by which oil enhances heat transfer, it is now possible tocompare the relative effects of different oils. The axial heat-flux profile for a 1018 grade steel castwith four different oils at Company C is presented in Figure 8.7. To maintain clarity in the figuresonly the two extreme cases are shown viz., a vegetable based oil (Canola) and a mineral oil(Mineral_S). It can be clearly seen that the heat flux for the vegetable based oil is higher than thatfor the mineral based oil.Table 4.3, giving the properties of the oils, indicates that the boiling point of Mineral_S oilis much lower than that of Canola oil (20% of Mineral_S boils off at 230 °C compared to around[202]5% of Canola oil at the same temperature) . Consequently for a given flow rate of lubricant and amould hot face temperature that is sufficient to cause the oils to boil, more of the Canola thanMineral_S oil would make its way below the meniscus leading to a higher heat transfer as explainedin the previous section. This theory is further confirmed by the fact that the heat flux in the meniscusregion is almost the same for the case of Mineral_S oil at 25 mI/mm and the “no oil” situation.The enhancement of heat transfer with the Canola oil in the lower part of the mould is alsoexplained on the basis of an increase in the thermal conductivity of the mould-billet gap arisingfrom the presence of hydrogen. The increase, however, is less than that in the upper part of themould, as expected, because in the lower part of the mould the thermal resistance of the solid shelland the mould-billet gap are comparable. This issue is revisited in the next chapter.8.1.5 Effect of mould-oscillation frequencyThe effect of a change in mould-oscillation frequency is depicted in Figure 8.8 which is aplot from data collected at Company C. As shown in the figure the axial heat-flux profile for the(normal) high frequency of 144 cpm is lower both at the meniscus and toward the bottom of themould. Table 5.1 summarizes the change in oscillation related parameters corresponding to a changein the oscillation frequency of the mould.The explanation for the lower rate of heat extraction at the meniscus is deferred until the nextchapter. The difference in the heat extraction rate in the lower part of the mould can be explainedas follows. An increase in the oscillation frequency of the mould increases the number of oscillationmarks per unit length of the billet which widens the local billet-mould gap thereby lowering theheat transfer to the mould. (It should noted that the depth of the oscillation marks (0.18 mm) is thesame in the two cases). The pitch of the oscillation marks is given by (where V is the castingspeed and f is the oscillation frequency of the mould) so that the number of oscillation marks permeter of billet is 47 and 71 for an oscillation frequency of 96 and 144 cpm respectively. Thus at[203]the high oscillation frequency, the mould leaves 1.5 times as many oscillation marks on the billetas at the low frequency oscillation; thus, on average the local air gap is 1.5 times wider in the 144cpm case. That these differences in the pitch of the oscillation marks can indeed cause the observedenhancement in heat transfer is verified quantitatively in the next chapter.It is not very clear why there is a shift of 20 mm in the position of the heat flux with a changein the oscillation frequency but it may be due to an unintended change in the meniscus level.8.1.6 Difference in the mould heat extraction rate at the three plantsThe axial heat-flux profiles for 1045 grade steel at Companies B, E and C are compared inFigure 8.9 and show that considerably more heat is extracted by the mould at Company C comparedto that at the other two Plants. Figure 8.10 shows a plot of the average heat extracted in the mould(as obtained from the mathematical model) against the heat extracted by the mould cooling water(calculated using the water flow rate and the change in the inlet and outlet water temperature). Theagreement is excellent with two exceptions which corresponded to 1012 and 1010 grades of steelswhere, on account of the low heat flux the increase in the cooling water temperature was around1.5 C°. Since each water thermocouple (measuring inlet and outlet water temperature) has anaccuracy of +1- 1C°, it is suspected that temperature rise of just 1 or 2 C° cannot be very accuratelymeasured. The agreement seen in Figure 8.10 reinforces the validity of the values of heat transferobtained through the use of the mathematical model, and especially the high heat extraction rate atCompany C.The high heat extraction rates at Plant C are also indirectly corroborated by the following.(a) Among the three Plants studied, the tendency for the formation of rhomboid billets was thehighest at Plant C. As explained in an earlier chapter, billet rhomboidity may arise when themould cooling water boils in the cooling channel [1 81. A combination of a high heat extractionrate in the mould and a low flow rate of mould cooling water leads to this condition. The[204]mathematical model of heat transfer in the mould at Plant C has shown that the cold facetemperature of the test mould is 145 °C, very near the boiling point of water at the prevailingpressure. Furthermore, the control mould which had a much lower water flow rate, 9.7 rn/scompared to 12 m/s in the test strand, is more likely to produce rhomboid billets. This indeedwas the case as was shown in Chapter 4.(b) Another piece of evidence, though somewhat subjective, is the observation of the operatingpersonnel at Plant C who complain of low mould life on their caster. The mould material atPlant C is DHP copper which has a lower half softening temperature compared to the recommended Cu-Cr-Zr alloy.Figure 8.11 shows a bar chart of the amount of heat extracted per kilogram of steel cast. Thisapproach eliminates differences in heat extraction at the three plants on account of differences incasting speed and/or section sizes. As can be seen, the moulds at Company C extract around 60%more heat than those at Company B and about 25% more heat than those at Company E. Thiseffectively means that, barring other bottlenecks, theproduction rate at Company C can be increasedsubstantially by increasing casting speed while the production rate at Company B can be increasedsignificantly through mould design. The latter will be evident in the subsequent chapter.8.2 Variables Affecting Distorted Mould ShapeDuring service the billet mould is expected to distort and assume a bulged shape [62]. Thefollowing section examines the influence of the steel composition as well as the pre-existing mouldtaper on the shape the mould assumes when in operation. It needs to be noted that the distortedmould shape is obtained by adding the distortion predicted by the finite-element mathematicalmodel to the mould dimensions prior to it being put in service.(205]8.2.1 Steel compositionSteel composition affects mould distortion by virtue of its affect on heat transfer in the mould(shown in the previous section). The dimension of a new mould and its shape when in service areshown in Figure 8.12 for Company B. In the figure the distorted shape is plotted for a low carbon(C = 0.05%) and a high carbon (C = 0.42%) heat. No significant difference in the distortion of themould is seen for the two cases and either of the two profiles can be used for computation of mouldtaper.8.2.2 Pre-existing mould taperThe following discussion shows that the shape (taper) of the mould prior to being put inservice is a factor that has a profound effect on the shape the mould acquires during service. Thisis only to be expected as, mathematically speaking, the distorted mould profile is obtained by addingthe distortion of the mould to the dimensions of a new mould. This finding, however, has considerable significance for it explains most clearly the differences in the thermocouple and load cellssignals obtained at the three plants as will be shown in the next Chapter.Figures 8.13 and 8.14 show the new and distorted mould profile for Companies E and C.Thus during service the moulds at Company B (figure 8.12) and E do not assume a bulged shapenear the meniscus while the mould at Company C, in contrast, has a substantial “negative taper”.The reasonfor this lies in the measured mould wallprofiles ofthe new mouldsfrom these companies.The steep inward taper ofthe upper region ofthe moulds at Company B(4.9%/m ) and at CompanyE (2.7%/rn) almost negates the outward bulging of the mould wall. The distorted mould wall inthese two cases thus has an almost neutral taper. However, in the case of Company C, the measuredupper taper of the mould is very shallow (0.4%/m) even though the design specification was 1. 8%Im;indeed the as-delivered mould actually had several tapers (0 %/m to 0.4 %/m) as can be inferred[206]from Figure 8.14. Such a mouldduring service would acquire a steep “negative taper” in the meniscuswhich during the negative strip period of the mould oscillation cycle would interact most stronglywith the billet.8.3 Billet Solidification and QualityThe billet mid-face temperature, shell thickness and shrinkage profile were all obtained fromthe mathematical model of billet solidification and shrinkage.8.3.1 Billet mid-face temperatureThe predicted mid-face temperature of a billet passing through the mould of Company B fordifferent steel grades is shown in Figure 8.15. The temperature profiles, as expected, follow theheat-flux profiles shown in Figure 8.1, with the 0.42% carbon steel cooling the most and 0.12%carbon steel cooling the least.8.3.2 Billet shell thicknessThe computed shell thickness for different steel grades at Company B has been tabulated inthe previous chapter (Table 7.4) where it was shown that the predicted shell thicknesses at thebottom of the mould adequately match those measured from the solidification bands in macro-etchesof billet samples. Figures 8.16 - 8.18 show the shell growth profile for different steel grades atCompanies B, E and C respectively.The solid shell profile for a 1080 grade steel, shown in Figure 8.17, clearly indicates that forhigh carbon billets the start of solidification, as determined by achievement of the solidus temperature, is delayed significantly compared to the lower carbon grades. This can be explained interms of the size of the mushy zone. The release of latent heat is simulated in the mathematicalmodel by an increase in the specific heat capacity of the steel while in the mushy zone. Thus underthe same cooling conditions, steels that have a smaller difference between their liquidus and solidustemperature, and therefore a narrower mushy zone, will solidify at a faster rate than steels with a[207]broader mushy zone. The initial slow growth of the solid shell in the case of high carbon steel billetscould potentially have an adverse impact on their surface quality as is explained in a subsequentchapter.8.3.3 Billet shrinkage profileComputed billet shrinkage profiles for medium and low carbon grades are shown togetherwith the distorted mould profile for Company B in Figures 8.19 and 8.20 respectively. Thus the“medium” carbon billets (C > 0.15%) shrink sufficiently to clear the distorted mould while the“low” carbon billet (C <0.15%) shrinks less to cause binding in the mould.The greater shrinkage associated with the “medium carbon” grade is due to the high heat flux(and contraction on account of phase change for 0.15% carbon steel) for these grades. Interestingly,despite the lower heat flux of 0.15% carbon steel (compared to 0.42% carbon steel) the formershrinks at a faster rate initially on account of the large contraction associated with the deltagammaphase change. As discussed in an earlier paragraph, the release of latent heat is simulated in themathematical model by an increase in the specific heat of the steel while in the mushy zone. Thussteels with a narrower mushy zone shrink at a faster rate than steels with a broader mushy zone.Thus, the initiation and subsequent progress of solidification is delayed in the 0.42% carbon steelon account of the large mushy zone (difference between liquidus and solidus temperatures for0.15% carbon = 25 C° and 45 C° for 0.42% carbon steel). After some distance from the meniscus,however, the effect of the higher heat flux of 0.42% carbon steel dominates and the higher rate ofshrinkage for this grade, compared to the 0.05% carbon, can be inferred from the slopes of theirshrinkage profiles.The binding of the billets in the low carbon group in one case (1010 and 1012 grades) stemsfrom the comparatively low heat transfer such that even the high contraction associated with thephase change in these grades is insufficient to cause the billet to shrink enough to clear the mould.[208]The billet dimensions then are larger than those of the mould and this results in binding. In the othercase (1008 grade), though the heat flux is relatively higher compared with the 1010 and 1012 grades,the transformation temperature is low so that before the large transformation shrinkage begins, thebillet has already interacted with the mould.Independent evidence of the binding of the 1008,1010 and 1012 billets in the mould was theappearance of transverse depressions on the billet surfaces and the nature of the load cell responsecurve. The mechanism for the formation of depressions on the surface of billets has been proposedby Samarasekera and Brimacombe in an earlier work [29]. This mechanism has been illustrated inFigure 2.12 which shows a longitudinal section through the shell at an instant when it is stickingto, or binding in, the mould tube. Under these conditions of local mould/shell friction, the shell issubjected to a high tensile stress due to the mechanical pulling of the withdrawal system. Dependingon the magnitude of the stress, the shell can begin to flow plastically and form a neck, much as ina laboratory tensile test. The neck is manifested as a depression on the billet surface. Close to thesolidification front, within about 50 °C of the solidus temperature, however, the steel has virtuallyzero ductility so that, under the influence of tensile strains, a transverse crack forms. Thus bindingin the mould should manifest itself in the form of depressions on the surface of the billet andtransverse cracks either on the surface or in the subsurface.The results of surface roughness measurements of billets carried out by a profilometerdesigned by Bakshi et al. [801 are presented in Figures 8.21 and 8.22. The presence of depressionson the billet surface of a low carbon steel (1008) can be seen in the profilometer trace in Figure8.22 and the photograph of the billet surface Figure 5.37. Furthermore, Brendzy [13] has clearlyshown that the pronounced peaks of the load cell signal during the casting of a 1008 grade steelbillet is an indicator that the billet is binding in the mould.[209jAs the preceding paragraphs have shown, the model predicted binding of the low carbonbillets in the mould has been corroborated by other observations. Clearly then the existing mouldtaper at Company B for the grades 1008, 1010, 1012 is too steep and needs to be relaxed for improvedbillet quality.Figure 8.23 shows the billet shrinkage profile for a 1045 grade billet at Company C. Notethat the shrinkage profile of the billet is well in excess of the narrowing dimensions of the taperedmould such that a gap of about 0.7 mm opens up by the bottom of the mould. Such a gap is expectedto allow the solid shell to bulge under the influence of ferrostatic pressure and to cause a hingingaction at the off-corner sites leading to the generation of internal cracks near the solidification front.This mechanism has been explained in detail in an earlier chapter. The depth of off-corner cracksin billets from this company is between 8 and 12 mm (Chapter 5) indicating that off-corner cracksare forming below 400 mm from the top of the mould, in the zone of second taper. This matcheswell with the billet shrinkage profile (Figure 8.23) which shows a major increase in billet-mouldgap in this zone.The absence of off-corner cracks on billets from Company C where the mould dimensions,particularly in the lower part of the mould, matched the shrinking billet profile reinforces themechanism for the formation of off-corner cracks.[210]IC.‘-I)CARBON SPEED SUPERHEAT(%) (rn/mm) ( °C)0.05 2.4 280.092.3 20—-—- 0.12 2.2 30————V 0.15 2.1 200.42 2.2 30IiEE:y:::::=-..TIME BELOW MENISCUS (s)Figure 8.1 Heat-Flux profiles at Company B for billets with different carbon contents.[211]8000-7000 STEEL GRADEC = 0. 17%6000 C = 0.45%BreakPoint C_0.56%g5000 I1000o ibo 20 3)0 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.2 Heat-Flux profiles at Company C for billets with different carbon contents.[2121ov’J’JSTEELGRADE 10187000________________________NO OIL6000 25 mI/mm100 mI/mm5000-4000-3000-2000Meniscusi000\_______00 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.3 Heat-Flux profiles for 1018 grade steel billets cast with Canola oil at 0, 25, 70 and100 mi/mm at Company C. (Note: plot for 70 mi/mm of oil lies in between those for 25 and 100mi/mm, but has not been shown for clarity).[213]EIFigure 8.4 Heat-Flux profiles for 1018 grade steel billets cast with Canola oil at 0, 25, 70 and100 mI/mm at Company C (enlarged in the meniscus region). (Note: plot for 70 mi/mm of oillies in between those for 25 and 100, but has not been shown for clarity).[214]DISTANCE FROM TOP OF-‘U’)’)39003805. Jo’)’)3750Cd)3600C’)Z35003400-330003200310030000 25 70 100FLOW RATE OF OIL (mi/mm)Figure 8.5 Heat extracted in the meniscus area of the mould for 1018 grade steel billets cast withCanola oil at 0, 25, 70 and 100 mi/mm at Company C.[215j‘+uu-STEEL GRADE 1018350NOOIL300 - 25m1/min100 mI/mm0250‘—S.200,•< / A’-—150Mould cold face0 I I I I I I0 100 200. 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 86 Mould hot face temperature during casting of 1018 grade steel billets with Canola oilat 0, 25 and 100 mI/mm at Company C.[216]7() STEEL GRADE 1018____CANOLA600GMINERAL_S500O400O3O00-.2000Meniscus ii \Jioo1 •0 100 200 300 400 500 600 700 800 900DISTANCE BELOW MOULD (mm)Figure 8.7 Heat-Flux profiles for 1018 grade steel billet cast with Canola, HEAR, Mineral_Oand Mineral_S lubricating oil at 25 mi/mm. (Note the plots for Mineral_O and HEAR oils liebetween those for Canola and Mineral_S oils but have been omitted for clarity).[217]8000STEELGRADE 10457000 96CPMI44CPM6000_____________Meniscus0 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.8 Heat-Flux profiles for 1045 grade steel billets cast at mould oscillation frequencies of96 and 144 cpm at Company C.[2181EcfIFigure 8.9 Heat-Flux profiles for 1045 grade steel billets cast at Companies B, E and C.[219]DISTANCE FROM TOP OF MOULD (mm)ERROR BAR (kW/sqm) V3000 +80 7_____rEV2500V4+200O V++7+, V COMPANY1500 +7’UBiooo V EV_z_ C500 7 Low delta TS.’—.,0- I I0 500 1000 1500 2000 2500 3000 3500CALCULATED MEAN hEAT FLUX (kW/sq.m)Figure 8.10 Graph showing the match between predicted heat extraction rate in the mould andthe heat extracted by the mould cooling water.[220]TjC -t -4 -4 ct C -4 I.SPECIFICHEATEXTRACTIONRATE(kj/kg)0 riiz62.362.2 MOULD DISTORTION__s’NEW MOULDC=O.05%62.162 C=O.42%Meniscus_________________________4.619 4.S--S.5.4. 556180.5.4.61.7 4.4..•%6164.%,615.61.40 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.12 Calculated distortion of the mould during the casting of low and high carbon heats atCompany B.[22290.8STEEL GRADE 101890.6NEW MOULDDISTORThD MOULDMeniscus9 90.2089.6I I I I I0 100 200. 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.13 Computed distortion of the mould during service at Company E.223j72.1STEEL GRADE 104572— NEW MOULD-— DISTORTED MOULD71.9_____________Mensicus71.8z071.771.671.5 I0 100 200. 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.14 Computed distortion of the mould during service at Company C.[224]16O120&1100i0()C0 100 200. 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.15 Surface temperature at the mid-face of the billet for different grades at Company B.[225jC’,C’,zc-)Figure 8.16 Shell thickness at the mid-face of the billets for different steel grades at Company B.DISTANCE FROM TOP OF MOULD (mm)[226]ICl)C.-)zUFigure 8.17 Shell thickness at the mid-face of the billets for different steel grades at Company E.DISTANCE FROM TOP OF MOULD (mm)[22712—STEEL GRADE77 1018C..)<A 1045Meniscus f,I ----- 51602 I F____*1/1,I I I I I I0 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.18 Shell thickness at the mid-face of the billets for different steel grades at Company C.[228]Figure 8.19 Billet shrinkage profile, at Company B, for ‘medium” and ‘high” carbon steels.61.61.DISTANCE FROM TOP OF MOULD (mm)[229]62362.2DISTORTED MOULD-62.l C=O.05%S N.N. C=O.12%F61.9-61.861.7-61.6-61.5-0 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 8.20 Billet shrinkage profile, at Company B, for low carbon steels.[230]ri -0.5--.0 -0.7--0.8--O.9() -—1.1-1.2—1.3-0 20 40 60 80 100 120 140 160DISTANCE ALONG BILLET SURFACE (mm)Figure 8.21 Surface roughness of a “high” carbon steel billet cast at Company B.123110-0.1‘—(1)cf-iLL -0.5zx-.0 -0.7-0.8-0.9Cia)-—1.1-1.2-—1.3— i i a a a a0 40 80 120 160 200 240 280DISTANCE ALONG BILLET SURFACE (mm)Figure 8.22 Surface roughness of a low carbon steel billet cast at Company B.[232]EFigure 8.23 Billet shrinkage profile for 1045 grade steel cast at Company C.[233]600 700DISTANCE FROM TOP OF MOULD (mm)Chapter 9: MECHANISM OF MOULD HEAT TRANSFERA mechanism for heat transfer in the mould is developed in this chapter by linking the resultsobtained from various mathematical models. It will be shown in this chapter that, from the standpoint of heat transfer, the mould can be divided into different zones. As an aid to understanding thevarious zones in the mould, a classical resistance analysis of mould heat transfer is carried out asdiscussed in an earlier work by Samarasekera and Brimacombe [29].Under conditions of steady state, one-dimensional heat flow (which is approximately true incontinuous casting), the heat flux from the solidification front at the solidus temperature T to thecooling water at temperature T can be written as follows:q = —(T—T) (9.1)where RT, the total resistance to heat flow, is given by6 61R7 = (9.2)where the terms refer respectively to in order, to the thermal resistances offered by the shell, thegap width, the mould wall and the mould cooling water.It can be shown that the thermal resistances of the mould wall (— 0.40 cm2 °c/W) and thecooling water (— 0.25 cm2 °CIW) together do not amount to more than 10% of the total thermalresistance. Clearly then the magnitude of the resistance of the mould-billet gap and the solid shelldetermine the heat flow in the mould. The relative importance of the resistance of the two itemscan be seen in Figures 9.1 to 9.3 where the axial profiles of the gap and the shell thermal resistanceshave been plotted as a percentage of the total resistance. In keeping with the results of an earlierwork [29], the gap resistance dominates in the top half of the mould while the shell resistance begins[234]to exert its influence on heat flow towards the bottom of the mould. Heat transfer in the upper halfof the mould is, thus, influenced by factors that influence the gap width and conductivity whilefactors that alter both the width of the gap as well as the thickness of the solid shell determine theheat transfer in the lower part of mould. It will be shown that based on a systematic analysis of themould temperature data, the mould can be divided into three zones from the view point of heattransfer as shown in Figure 9.4.Zone I This zone extends from the meniscus downwards to a distance of approximately 30- 40 mm. Heat transfer in the zone is predominantly affected by the mechanicalinteraction between the mould and the billet.Zone II This zone immediately follows Zone I and extends for 100 - 150 mm. Heat transferin this zone is principally affected by the behaviour of the mould lubricating oil.Zone III This zone extends from the end of Zone II to the bottom of the mould. Heat transferin this zone is primarily influenced by the pitch and depth of oscillation marks.The various pieces of evidence that lead to the division of the mould in the zones describedabove are presented in the sections that follow.9.1 Heat Transfer in Zone IThe extent of this zone is obtained by a consideration of the length over which the mechanicalinteraction between the mould and billet takes place. It is related to the stroke length of the mouldoscillation and the normal metal level fluctuations in any casting operation. This would put thelength of the zone in the range of 30 - 40 mm. The width of the stain marks on the chrome platingon the mould is of the same order as well.9.1.1 The role of mould shape in the heat transfer in Zone IIt has been shown in the previous chapter that whether or not an operating mould acquires anegative taper at the meniscus depends primarily on the upper taper (shallow at Plant C or steep at[235jPlants B and E). This, as explained in the preceding chapter, is because the outward bulging of themould wall, that brings about the negative taper of the upper part of the mould, can be compensatedby a pre-existing (steep) positive taper. As a result, a very steeply (positively) tapered, (2- 3 %/m),mould may acquire, during service, a ‘neutral taper” or even retain some of its original positivetaper.It proposed that a mould with a large negative taper at the meniscus, would interact moststrongly with the solidifying shell during the period of negative strip. It is further proposed thatthis enhanced interaction is responsible for improved billet-mould heat transfer leading to a highvalue of heat flux in the meniscus region. The following paragraphs present evidence for thiscontention.Figures 9.5 and 9.6 are schematic diagrams of moulds that during service have acquired aneutral and a steep negative taper respectively. Thus in the former case, during the negative stripperiod when the mould descends faster than the billet, Figure 9.5, there will be a friction force actingon the billet dragging it down. An equal and opposite reaction force on the mould will act upwards,thereby reducing the compressive load sensed by the load cells. In the case of the steeply negativelytapered mould (Figure 9.6), in addition to the above mentioned friction force, there will be anotherforce arising from the physical “obstruction” or resistance from the billet as the negatively taperedportion of the mould tries to squeeze past the billet. A strongly negatively tapered mould wouldthus squeeze the billet for a greater proportion of the negative strip period than would a mould witha neutral taper. Such a squeezing action is expected to enhance heat transfer. The “squeezing”referred to above arises only on account of the negatively tapered mould wall and is likely to beminimal for a mould with neutral taper.Figures 5.34 to 5.36 are presented as evidence of increased mould-billet interaction. Thesefigures are the load cell signals (one cycle only) together with the mould displacement curve,[236jobtained at the three plants. The position of cross markers refer to beginning and the end of thenegative strip period while the position of the square marker indicates the period for which the loadcells are decompressed. Thus the location of the square marker relative to the cross markers is ameasure of the percentage of the negative strip time for which the load cell is decompressed- alarge percentage would indicate greater mould-billet interaction. (As mentioned in Chapter 4, onaccount of differences in pre-loads of the load cells as well as the distribution of the load betweenthe bolts, 0-rings, springs and the load cells, it is not possible to compare the absolute values ofloads among the three plants). It can be seen that the periodfor which the load cell is decompressed(enhanced mould-billet interaction) is around 50-55% at Plant B and E and in excess of 70% atPlant C.Evidence of increased mould-billet interaction can also be seen in the oscillation mark depthof billets cast with a steeply tapered parabolic mould and a shallow double tapered mould. Figure9.7 shows the oscillation mark depth on the billet cast through parabolic moulds on both the testand the control strands at Company B. Although the test strand operated with a shorter stroke andlower negative strip time (9.5 mm, 0.13 s) than the control strand (12.7 mm, 0. 17s) there is littledifference in the depth of oscillation marks on the billets from the two strands which is contrary toearlier findings [81]. Following the trial at Company B, the mould tubes on the test and controlstrands were replaced with double-tapered tubes which the company conventionally employs; theupper taper of the tubes is designed for 2.75%/rn taper over the first 333 mm followed by a secondtaper of 0.5%/rn. The billets cast on the test strand with a lower negative-strip time (0.13 s) and onthe control strand with higher negative strip time of 0.17 s were also subjected to an evaluation ofsurface roughness. Figure 9.8 shows the results of the surface topography measurements made on0.052% carbon billets casts through the conventional double-taper moulds and a remarkableimprovement in the formation of oscillation marks can be seen on billets cast with lower nega[2371tive-strip time. This clearly indicates that in case of the parabolic mould, there is very little interactionof the mould with the billet and hence the oscillation mark depth are insensitive to differences instroke length and negative strip time. Significant interaction, in the case of the double tapered mould,arising from the negative taper that such a mould acquires during service, causes deeper oscillationmarks on billet cast with longer stroke lengths and negative-strip times.It will now be shown that increased mould-billet interaction leads to a higher rate of heattransfer. The first evidence of the effect can be seen in Figure 9.9 which shows the decompressionload (i.e. the amount of decompression the load cell experiences during the negative strip period -a measure of mould-billet interaction) for 1018 grade steels at Plant C against the heat flux in themeniscus region. (Note : Since all data refers to the same plant differences arising out of loadpartitioning and pre-load of load cells can be ignored and loads can then be analysed as a measureof mould-billet interaction). The figure clearly shows that an increase in the decompression load(more mould-billet interaction) leads to higher rates of heat transfer.A second piece of evidence can be seen in Figure 9.10 which is a plot of the period for whichdecompression of the load cell (an alternate indicator of mould-billet interaction) takes place versusthe heat transfer in the meniscus region for 1045 grade steel cast at Company C. Again the influenceof enhanced mould billet interaction on the heat transfer rate in the meniscus region is obvious -enhanced interaction leads to improved heat transfer.A third piece of evidence is apparent in Figure 9.11 which shows the heat flux in the meniscusregion plotted as a function of the percentage of the negative strip time for which the decompressionof the load cell takes place. The plot corresponds to heat fluxes obtained at two different frequenciesof mould oscillation, 144 cpm and 96 cpm. The lower oscillation frequency has a higher negativestrip time (0.19 seconds) compared to the negative strip time of 0.16 seconds for 144 cpm of mouldoscillation (Table 4.3). As before the higher negative strip time causes a larger heat extraction in12381the meniscus region on account of the increased period of mould-billet interaction. It is intriguing,however, to note that the increase in the negative strip time (with a decrease in the mould oscillationfrequency) leads to an increase in the percentage of the negative strip period, for which thedecompression of the load cell takes place, from 70% to over 85% of the negative strip time. Thiscan only be explained by considering that an increase in the heat transfer, arising from a longernegative strip period, causes the mould to acquire an even greater negative taper. This increasednegative taper in turn leads to a further increase in the period ofmould-billet interaction causingthe decompression time to increase from 70% to about 90% of the negative strip time.Indirect evidence of enhanced heat transfer in the meniscus region arising from increasedmould-billet interaction was also found, as follows:[1] Figure 9.12 shows the heat flux profile on two adjacent walls (ICW and RSW) of the mouldat Plant C. The differences in heat fluxes in the meniscus region on adjacent mould wallshave been observed in earlier plant trials too but have been attributed to the unequal distortionof the adjacent mould wall arising from a two sided mould constraint [29]. As the mould atCompany C has a four-sided constraint (Table 4.2), the two heat fluxes are expected to havethe same value in the meniscus region. The reason why it is not so can be inferred fromFigure 9.13 which shows the (computed) distorted mould profile at the mid-face of the twoadjacent walls. The higher negative taper on the inner-curved wall (that should lead toenhanced mould-billet interaction with that face) is clearly visible. This supports the theorythat the mould shape affects the heat transfer in the meniscus region.[2] As mentioned in an earlier chapter, Singh and Blazek [33], had reported peak heat fluxesof 3000 - 3200 kW/m2,while casting steel billets with 0.40% C through an experimentalmould. This range is considerably lower than the value of7800 kW/m2obtained while castingthe same grade at Company C. The experiments by Singh and Blazek were carried out on[239]a mould which was very different from a conventional mould in particular with regard toits distortion characteristics. The authors point out that their experimental mould was sodesigned as to “prevent” mould distortion. Such an experimental mould would, therefore,not acquire a negative taper and would have a lower heat transfer rate at the meniscus thanan industrial mould such as that used in Company C.[31 In an earlier trial Brimacombe and Samarasekera [29] have reported a meniscus heat fluxof 3000 kW/m2with a double tapered mould (2.6 %/m upper taper) and around 4200 kW/m2with a single tapered mould (0.6%m). The lower heat transfer in the case of double tapermould can be explained by the lack of negative taper (and, therefore, lower mould-billetinteraction), relative to a the single tapered mould.[4] In an industrial trial Samarasekera, Brimacombe and Bommaraju [17], have shown a significant increase in meniscus heat transfer from 4000 kW/m2to 5000 kW/m2with a decreasein the mould cooling water velocity from 7.0 m/s to 5.0 m/s. Such an increase in the meniscusheat transfer can be explained by considering the impact of cooling water velocity on moulddistortion. As the mould becomes hotter with a decrease in mould cooling water velocity,its distortion and thus, its negative taper, increases. Such an increase in negative taper wouldlead to increased mould-billet interaction and therefore heat transfer at the meniscus.[5] Evidence of a decrease in heat transfer with increasing mould taper is also apparent fromthe work of Lorento [82] in which he has shown an improvement in billet rhomboidity withchange in mould taper from single to parabolic. The difference in diagonals of billetsdecreased for a 0.40% C steel from 13 mm (single taper mould), 8 mm (double taper mould),to 3 mm (parabolic taper mould). If this data is reviewed in light of the effect of mould taperon heat fluxes, it clearly points to a decrease in heat transfer as the mould taper changesfrom a single shallow taper to a steep parabolic taper. The decrease in heat transfer arising[240jfrom a progressive increase in the initial positive taper (and a resulting decrease in mouldnegative taper), probably lowers the cold face temperature of the mould below that necessaryto cause boiling of the water in the mould cooling channel. As boiling is suppressed thetendency of the billet to assume a rhomboid shape is also reduced [18].[61 As another indirect evidence of the role of mould shape on heat transfer, the average depthof off-corner cracks on the billets from the control and test strands of Company E can beexamined. On average, the off-corner cracks are deeper on the test strands than on the controlstrand by 1- 2 mm. Since the mechanism of cracking is undoubtedly the same on both strands(viz, bulging in the mould), the above observation suggests that the shell thickness at thebottom of the test strand is 1-2 mm greater than that on the control strand. This impliesgreater heat transfer in the test strand mould. Table 4.2 summarizes the various features ofthe test and control strands and an important difference between the two is the 2.7 %/m taperof the test strand versus the 3.6 %Im taper of the control strand. When in use the steepertaper mould would certainly not have any negative taper and therefore lead to a lower heattransfer in the mould.Finally, Figure 9.14 which is a bar graph of the peak heat flux, as measured in plant trials, inthe meniscus region for moulds with different initial tapers clearly shows that the creation of anegative taper in mould with shallow initial positive taper, causes a greater amount of heat to beextracted by the mould in the meniscus region.9.2 Heat Transfer in Zone IIThe extent of this zone is obtained by examining Figure 8.9 which is a plot of the heat fluxobtained at the three plants. The figure indicates that the heat transfer at all three companies becomes[2411roughly of the same order at around 350 mm from the top of the mould which is approximately 200mm from the meniscus. Taking away the length of zone I (—‘ 40 mm), this leaves about 100-150mm as the length of zone II.The role of oil in mould heat transferThe role of oil in the enhancement of heat transfer has been shown in the previous chapter.Figure 9.15 shows the hot face temperature of the mould for a 1045 grade steel cast at Plants B, Eand C. The strikingly higher hot face temperature of the mould at Plant C is immediately obvious.On account of the higher heat transfer, due to a strongly negatively tapered mould, at Company C,the hot face temperature of 270°C (Figure 8.6, no oil case) is well above the boiling point of thelubricating oils and much below it at Companies B and E. It is thus proposed, that an increase inmould heat transfer beyond that obtained from the mould shape, can be affected by the oil providedthe lwt face temperature of the mould exceeds the boiling point of the lubricating oil, therebyallowing the oil to boil and contribute to the high heat conductivity of the mould-billet gap. Furthermore, with the enhancement in heat transfer the hot face temperature of the mould at CompanyC exceeds 350 °C.That the presence of hydrogen gas in the gap can significantly enhance heat transfer can beseen from the Figure 9.16 where the heat flux between the billet surface and the mould is computedfor different levels of hydrogen contents of the gap. (The thermal conductivity of the mould-billetgap is assumed to be given by the law of mixtures). From Figure 9.19 it is clear that at mould-billetgap width of 0.02 mm, the presence ofjust 25% of hydrogen gas, in the mould-billet gap, is sufficientto increase the heat flux from 3000 kW/m2 to 7000 kW/m2.The fact that the hot face temperature of the mould needs to be above the boiling point of theoil before any enhancement of heat transfer takes place, is supported, indirectly by the followingtwo findings.[2421[A] Despite the high flow rate of oil at Companies B and E (54 mi/mm and 70 mi/mm respectively),the heat transfer in the meniscus region is far lower than that obtained at C with 25 mi/mmof oil. This suggests that oil does not contribute significantly to the heat transfer at CompaniesBandE.[B] Brendzy [13] has shown a dependence of the mould friction force on the oil flow rate. It wasshown in her work that a decrease in the oil flow rate at company B led to an increase in thefriction force, as inferred from the continuously increasing profile of the load cell signalduring the upstroke of the mould. Load cell signals from Plant C, however, are unaffectedby the oilflow rate (ChapterS) and, regardless of the flow rate (25 mi/mm or 100 mi/mm),are of the type categorised as “high friction” response by Brendzy. This would suggest onceagain that on account of the high hot face temperature of the mould at company C, almost alloil vapourises in the meniscus region creating adverse conditions for lubrication. This is thedownside of the high heat transfer rate in the mould.Notwithstanding the “high friction” condition suggested by the load cell response at CompanyC, it remains a fact that the low-carbon billet surfaces themselves did not appear to be affectedadversely as can be seen in Figure 5.94. This observation is consistent with that of Brendzy whofound that a reduction in the oil flow rate resulted in an adverse lubrication condition but did notlead to any measurable deterioration in the billet quality in the range of carbon steels studied (C <0.45%).An implication of the above discussion is that high carbon steel billets (C > 0.60%) which,on account of their long freezing range, form an extremely thin shell in the meniscus region, arethe ones most likely to Jel the impact of adverse lubrication condition. This, in part, may explainthe bleeds and laps associated with casting these high carbon billets at a company with a mouldsimilar to that at Company C /83].[24319.3 Heat Transfer in Zone IIIFigures 9.1 -9.3, show that the effect of the resistance of the solid shell, on the total resistanceto heat flow, begins to become significant in the lower part of the mould (zone III). This should notbe taken to mean, that the resistance of the gap ceases to be important; the first statement onlyemphasizes the fact that in the lower part of the mould, both the shell resistance as well as the gapresistance have a comparable effect on the heat flow.The importance of the oscillation frequency on heat transfer in this region of the mould canbe seen in the heat flux obtained in Company C at 144 and 96 cpm of mould oscillation (Figure8.8). As mentioned in the previous chapter, at a lower frequency there are 47 oscillation marks permeter of the billet at 96 cpm compared to 71 in the other case. It should be noted that the averagedepth of oscillation marks in the two cases was the same, about 0.18 mm. To assess the effect ofthe difference in the pitch of the oscillation marks on the observed heat flux, the resistance to heatflow was plotted for the two cases (Figures 9.17 and 9.18). Thus, the shell resistance is the samein the two cases, but the average gap resistance, in zone III, for billets cast at 144 cpm of mouldoscillation is 2.75 cm2°C/W compared to 1.90 cm2°C/W in the latter case. These resistance are inthe same ratio (—1.5) as the ratio of the pitch of the oscillation marks. This clearly proves that theenhancement in heat transfer on account of a decrease in mould oscillation frequency arises froma decrease in the pitch of the oscillation mark leading to a relatively lower gap resistance comparedto the resistance of the oscillation mark in billets cast at higher mould oscillation frequency.9.4 Interaction of ZonesIt has been shown that the shape of the mould at the meniscus and the interaction of the mouldwith the solidifying shell during the negative strip time is the predominant factor influencing heattransfer in Zone 1. The negative taper of the mould is influenced, among other factors, by the[244]temperature of the mould wall in Zone II. The temperature of the mould wall in Zone II is, in turn,largely influenced by the boiling of oil leading to the presence of Hydrogen gas in the gap. Thusindirectly, oil plays a role in heat transfer in Zone I as well.The presence of oil in the gap between the mould and the billet would also affect the heattransfer in Zone III, by increasing the thermal conductivity of the gap. This effect clearly will bediminished because both the shell resistance and the gap resistance play an equally important rolein heat transfer in Zone III. The presence of hydrogen in the gap in the lower part of the mould isprobably responsible for the higher heat transfer in Zone III for the mould at Company C as comparedto Companies B and E. (Figure 8.9).[2451‘‘flu,_9O80 iz7O IGAP500 100 200. 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 9.1 Axial profiles of gap and shell thermal resistances at Company B.{246j1009080 —‘z —S‘_v70,JN,Cl) GAPCl)6050F4030 SHELLz200 110-0 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 9.2 Axial profiles of gap and shell thermal resistances at Company E.[247]1009080L)z70 Th.‘/Cl) % j\‘vJ” \..--f— 50 ,..J ‘V\GAPA30 SHELL“_2010I I I I I100 200 300 400 500 600 700 800 90()DISTANCE FROM TOP OF MOULD (mm)Figure 9.3 Axial profiles of gap and shell thermal resistances at Company C.[248]HEAT FLUXZONET (—30-4Omm)ZONE II(—100- 150 mm)ZONE III(— 400 mm)Figure 9.4 The three zones of heat transfer in the continuous casting billet mould.Mold-billet interactionHydrogen gasOscillation frequency[2491Positively Tapered Mould WallDynamically distortedmould wall.Molten steelShellGapLack of negative taperreduces mould-billet interactionFigure 9.5 Shape acquired during service by a mould with a steep initial taper.Meniscus7[2501Negatively Tapered Mould WallNegative taper causes significantmould-billet interactionDynamically distortedmould wall.— MeniscusFigure 9.6 Shape acquired during service by a mould with a shallow initial taper.Molten steel— ShellGap[251]E0.0)0CC0CO20)0a)<0•4030201Off— CornerFigure 9.7 Graph showing the effect of negative strip time and stoke length on the oscillationmark depth on 1008 grade billets cast through parabolic taper moulds. (Negative strip time andstroke lengths of the mould at Strand 3 and Strand 4 are 0.13s, 9.5 mm and 0.19 s, 12.7 mmrespectively).ControlStrand2-375m1(est.)0302012-7TestStrand3-3Oil:C54m13-754m10 0ff—Corner[252][öouble Taped O52°/0C 10 Side I07 C Side 2 Strand 206C0440.3.= Strand 30 0’1•I I0ff—Corner 0ff—CornerFigure 9.8 Graph showing the effect of negative strip time on the oscillation mark depth on 1008grade billets cast through double taper mould. (Negative strip time and stroke lengths of themould at Strand 3 and Strand 4 are 0.13s, 9.5 mm and 0.19 s, 12.7 mm respectively).[253]xSTEEL GRADE 10183800Ir=0.911C.., 3600C-)C’)35003400>033003200K3100- I200 400 600 800 1000 1200 1400DECOMPRESSION LOAD (N)Figure 9.9 Effect of ‘decompression’ load on the mould heat transfer during the casting of 1018grade billets at Company C.[254]4400’42004000’____U3800’3600’>03400,3200Figure 9.10 Effect of negative strip time on the mould heat transfer during the casting of 1045grade billets at Company C.[255]STEEL GRADE 1045xx[.96 Ix---------i—----—— I--0.14 0.15 0.16 0.17 0.18 0.19 0.2• PERIOD OF DECOMPRESSION (s)u’J’J440O-Q4200C,) 4000L)38003600>0340o3200I I I-55 60 65 70 75 80 85 90 95PERIOD OF DECOMPRESSION (% of iN)1045xKIr=0.97 IK KjiJuJ—50 100Figure 9.11 Effect of decompression period (measured as a percentage of tN), on the mould heattransfer during the casting of 1045 grade billets.[256]7()()() STEEL GRADE 1018600 ICwRSW5000___________________4O0o100w /0— I I I I I0 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 9.12 Axial heat-flux profiles on two adjacent mould walls at Company C.[257]72.472.3 New RSW72.2 Distorted RSW72.1 /7271.9 \ ‘._._\._New ICW \71.8 Distorted ICW, \%\II.. ‘.—-. \--..71.7:71.6 1/71.5714 I I0 100 200. 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 9.13 Computed mould shape during operation for two adjacent mould walls at Plant C.[258]Figure 9.14 Measured peak heat flux for different initial mould tapers.[259]MOULD UPPER TAPER (%/m)350 STEEL GRADE 1045____COMPANY BL)300 COMPANY ECOMPANY C250__ __ ______________200/ \, —.—.—. ——- I‘S4: 1500•,1500 100 200 300 400 500 600 700 800 900DISTANCE FROM TOP OF MOULD (mm)Figure 9.15 Hot face temperature of the mould wall while casting 1045 grade steel billets atPlants B, E and C.[260]10000--9000- ‘100 %‘ % OF HYDROGEN GAS8000- •%75% •5...__7000‘5_5...6000- \50%.5..5-5...-...5— 5’..5.5>< 5000- S-..\25%5..... 5.......f— 4000- ‘--. 550.02 0.04 0.06 0.08 0.1GAP WIDTH (mm)Figure 9.16 Variation in mould heat transfer on account of different amounts of Hydrogen gas inthe mould-billet gap.[261]IL--—___________________________________________10J)GAPo 100 200 300 400 500 600 700 800DISTANCE FROM TOP OF MOULD (mm)Figure 9.17 Axial profile of gap and shell thermal resistances while casting 1045 grade billets at144 cpm of mould oscillation at Company C.[262]12100C_)I: GAP0 100 200 300 400 500 600 700 800 9(X)DISTANCE FROM TOP OF MOULD (mm)Figure 9.18 Axial profile of gap and shell thermal resistances while casting 1045 grade billets at96 cpm of mould oscillation at Company C.A[263]Chapter 10: DESIGN OF MOULD TAPERSDiscussions in the previous two chapters have clearly revealed the significant effect that thepre-existing mould taper has on the heat flux in the meniscus region. It has also been shown that asteep taper in this region of the mould, designed to compensate for the mould-billet gap and therebyincrease the heat transfer, actually reduces the heat flux, primarily on account of reduced mould-billet interaction. Several findings have been presented to show that enhanced mould-shell interaction leads to higher rates of heat flow in the meniscus region of the mould. A high enough heattransfer in the nieniscus region, that raises the temperature of the hot face of the mould above theboiling point of the lubricating oil, causes the oil to boil, thereby, creating adverse lubricationconditions in the meniscus region. While such adverse lubrication conditions are not a cause ofconcern while casting billets of low carbon grades (C < 0.60%), the same conditions can lead tobleeds and laps on the surface of high-carbon grade billets. This, as has been explained earlier, ison account of the large freezing range of the high carbon billets, which permits only a thin solidshell to form in the meniscus region. This thin shell is likely to develop bleeds or laps in the absenceof a lubricating medium between the shell and the mould. Thus taper in the upper part of the mouldneeds to be designed very carefully and not necessarily from the point of view of extracting thelargest amount of heat in the mould.10.1 Conventional Design of Mould Taper:The walls of the mould are tapered inwardly so as to enhance the rate of mould heat extractionby reducing the air gap between the mould wall and the (shrinking) billet. It is not sufficient,however, to design a mould taper based solely on the shrinkage profile of the billet, because duringservice, the mould wall distorts and moves away from the billet (Figures 8.12- 8.14), opening up2641the air gap further. Thus the design of the taper must take into account the formation of an air gapcaused by both the distortion of the mould while in use and the shrinkage of the billet. Mould tapersare expressed in %Im and are based on the expression given below.(M-M)TaPer_,) (M)(M)X 100 (10.1)where MT and M8 are the dimensions of the mould at the top and bottom respectively and ML is thelength of the mould in meters.The gap between the mould and the billet is, conventionally, higher in the meniscus regionof the mould than in regions below the meniscus. This is on account of the large air gap formed bythe rapid initial shrinkage of the billet (high meniscus heat transfer) and the distortion of the mouldwall away from the billet. To compensate for the air gap, the mould wall normally needs to besteeply tapered (between 3.0 - 5.0 %/m), for the first 100 mm (approximately) from the meniscus.With reference to the shrinkage profile and mould distortion results obtained at Company B, calculation of mould taper based on the conventional philosophy of reducing the mould-billet gap, fora 1008 grade steel, leads to a value of 2.9 %/m for the first 100mm from the meniscus, 1.6%/rn forthe next 423 mm and 0.8%/rn for the remaining part of the mould. It should be pointed out that anunderlying assumption in this kind of calculation is that the heat-flux profile used to design the newtaper is itself not altered, in any significant measure, by the new taper. The present work shows thatthis assumption is grossly in error and that a design aiming to enhance heat transfer in the mouldby increasing the upper taper of the mould, ends up actually reducing heat transfer by decreasingmould-shell interaction.10.2 New Approach to Design of Mould TaperThe discussion in the preceding section has shown that based on the conventional designphilosophy, mould tapers would consist of a steep upper taper of about 3 - 5 %/m followed by a[2651steeper second taper. There are at least two flaws in such a design. Firstly, an implicit assumptionis that the new taper, by itself, does not alter the heat transfer in any significant manner and, secondly,the enhancement of heat transfer in the meniscus region is desirable for all grades of steel. In viewof the findings of the present work, the following new points are considered important for designof mould tapers.[1] Billets with a long freezing range (C> 0.60%) have a very thin shell near the meniscus. Thesurface quality of such billets is strongly affected by the absence of the mould lubricating oilnear the meniscus and it is believed that bleeds and laps are a consequence of adverse frictionconditions in the mould. Such grades of billets, thus, need to be cast through moulds thathave hot face temperatures low enough to prevent the oil from boiling. Thus the taper in suchcases must be designed to reduce heat transfer in the meniscus area.[2] Billets with carbon <0.60 % can be cast through moulds that have tapers, in the meniscusregion, designed to enhance heat transfer. It has been shown that the absence of the mouldlubricating oil and the resulting adverse friction conditions do not affect the quality of thebillet surface to any significant degree.Based on several shrinkage profiles computed during the work, an upper taper of 2.5 %/mand a lower taper between 0.6%/rn to 1.0%/rn is recommended for the high carbon grades (C>0.6%). The break point and the exact value of the second taper needs to be determined from thecasting speed and section size of the billet. The upper taper should extend to the top of the mould.The steep upper taper of the mould would prevent the mould wall from acquiring a negative taperduring operation. The absence of negative taper has been shown to reduce mould-billet interactionand thus the heat transfer. The hot face temperature of such a mould is likely to be around 180 -200°C (calculations based on a mould wall thickness of 12-13 mm and a mould cooling watervelocity of 10-12 mIs) which is below the boiling point of the commonly used vegetable-based[266jlubricating oils.Mould tapers recommended for lower carbon grade billets are an untapered mould wall(0.0%/rn) in the upper region and between 1.0%/rn - 2.0 %/m in the lower region with a break pointof 25-30 mm from the meniscus. The exact value of the lower taper would depend on the castingspeed, section size and carbon content of the steel. The mould wall should be untapered right fromthe top of the mould to 25-30 mm below the meniscus to ensure that even with the normal metallevel fluctuations that arise during casting, the meniscus continues to reside, as per design, in anuntapered region of the mould. Such a mould would, during service, distort and assume a steepnegative taper leading to enhanced mould-billet interaction and, therefore, heat transfer.The enhancement in heat transfer as obtained in a mould that is untapered in the meniscusregion clearly would allow for an increase in casting speed leading to an improvement in productionrates. A measure of the increase can be seen from a very simplistic calculation comparing the castingspeeds for a 0.45% carbon steel at Company C and Company B. Under the conditions of low heattransfer in the parabolic mould at Company B, the casting speed is —P2.0 rn/mm. The heat transferin the mould at Company C, indicative of the high heat extraction rates of a 0.0 %/m upper taperedmould, is roughly 1.5 times the value at Company B. This suggests that the casting speed at CompanyB, can be increased by at least 50%. The calculation assumes that the only constraint to the increasein the casting speed is the attainment of a certain shell thickness of the billet that is sufficientlythick to support the liquid steel at the exit from the mould. It needs to be pointed out that 50%increase in casting speed is a conservative estimate as the heat flux in the properly designed mouldis likely to be higher than that obtained from the mould at Company C. Before the benefit of increasedproduction rates is realised in an operating plant, modifications may have to be carried out on thepositions of the billet-cutting torch and the unbending point to ensure that their locations areappropriate with the higher casting speeds.[267110.3 Other Design Parameters10.3.1 Shaping of mould wallsThe tolerance levels indicated on the engineering drawings of the mould are typically +- 0.254mm. To understand the implications of such a loose tolerance, a simple calculation is performedfor a 120 X 120 mm mould having a design taper of 0.4%/rn . To be correctly manufactured, thedimension of the mould wall 25.4 mm down the length of the mould, should be 119.878 mm. Thisdecrease of 0.122 mm in the mould wall dimensions from its dimension at the top, is less than halfthe tolerance magnitude! The loose tolerance in use currently can lead to a taper in excess of 4- 5%/m in the first few important crns near the top of the mould. Moulds that are explosion formedhave been found to have tolerances that compare favourably with those needed and this method ofmanufacture is strongly recommended. Additionally it may be a good idea to change the measureof taper from %/m to %/cm at least in specifications to the manufacturer to emphasize the importanceof a close tolerance over small distance. It is not unlikely that the use of the unit %/m has shiftedthe focus away from the dimension the mould needs to have over a shorter distance to meet thetaper requirements. Thus, often, moulds have an overall taper that matches the designed one overthe mould length but is grossly inadequate over shorter distances especially in the critical meniscusarea.10.3.2 Mould water pressureWhether or not the mould cooling water boils in the cooling channel depends on the mouldcold face temperature and the pressure in the water channel. A high exit pressure of the mouldcooling water or ‘back pressure as it is commonly referred to, permits the cold face of the mouldto acquire a high temperature without causing the water to boil. In view of the large heat extraction[2681rate now possible with the new mould design of an initial untapered meniscus region, calculationsshow, that the cold face temperature of the mould is likely to reach 150°C. To ensure that waterdoes not boil under this condition the back pressure should be about 20 to 30 psi (138 to 207 KN/m2).10.3.3 Material of the mouldGiven the high heat extraction rates achievable, the mould material should be an alloy ofcopper, chromium and zirconium as this has a higher half softening point than the conventionalDHP copper moulds. This point has been made several times earlier [62], but assumes greatersignificance in light of the higher heat extraction rates being sought.[2691Chapter 11 : SENSOR SIGNALS AND BILLET DEFECTSThis chapter briefly introduce how mould sensors such as Thermocouples, Load Cells andLVDTs can be utilized to detect the creation of adverse casting situations and how corrective steps,if any, can be taken to move away from those conditions.11.1 ThermocouplesThermocouples are typically placed midway between the hot and cold faces of the mouldwall and record temperature that are 100 - 150°C lower than the mould hot face temperature. It hasbeen shown in the course of this work that when sampled at 1 Hz., the thermocouples are capableof detecting several billet defects discussed below.11.1.1 Off-corner CracksThe mechanism by which off-corner cracks form in the mould has been described in an earlierchapter and has been described in some detail by Brimacombe, Samarasekera and co-workers [59].It effectively involves bulging of the mid-face of the billet causing the shell at the mid-face to touchthe mould wall. This bulging results in a hinging action at off-corner locations, and the generationof a tensile strain in the region of low-ductility adjacent to the solidification front which can leadto the generation of off-corner cracks.In view of the fact that the bulging of the shell and its subsequent interaction with the mouldis a pre-requisite for the creation of off-corner cracks, it is expected that the thermocouple(s) in theregion of the mould-billet interaction would register an increase in temperature. This is clarifiedfurther in the Figure 11.1 which is a plot of the profiles of the heat-flux, the billet shrinkage andthe distorted mould wall as a function of distance from the top of the mould. Also marked on thegraph is the location at which the off-corner cracks are estimated to have formed. (The depth of theoff-corner cracks is a measure of the solid shell thickness at the moment of crack generation andthe position of the billet in the mould at that instant can thus be obtained by a consideration of the[270]solid shell profile of the billet). As can be seen in Figure 11.1, the cracks appears to have formedat the position where the heat-flux profile rises and then falls indicating a local increase in heattransfer as may be expected with an interaction of the billet with the mould. It is interesting to notethat the crack appears to have formed at the instant the billet reaches the zone of shallow secondtaper where, on account of the high shrinkage in the upper part of the mould, a fairly large gap(approximately 5 -6 mm) exists between the mould and the billet surface. A gap of this magnitudeis conducive to the bulging of the shell.The response of the thermocouples in the region where the crack appears to have formed canbe seen in Figure 11.2 which is a plot of the mould thermocouple temperature for Company C. Alsoplotted on the same graph is the mould thermocouple response at Company B where off-cornercracks were not observed during the trials. The increase in temperature of the thermocouple atCompany C around 400 mm from the top of the mould is indicative of the bulging of the billet. Asis expected, the thermocouples at Company B do not show a similar trend.The obvious corrective action that can be taken is a judicious decrease in the casting speed.The increase in solid shell thickness, arising from a longer residence time, may give sufficientstrength to the solid shell to prevent bulging of the billet. A decrease in mould heat transfer can beattempted by a change in the metal level of a suitably designed “smart’ mould. This topic is discussedin greater detail in a subsequent section. The ideal way to prevent off-corner cracks is to have amould suitably tapered to compensate the formation of the air gap in the lower part of the mould.11.1.2 RhomboidityUnequal heat transfer on adjacent faces of the billet, among other reasons, can lead to theformation of off-squareness in billets. This unequal heat transfer, that may or may not lead to boilingof water in the cooling water channel, can arise from an imprecisely machined mould. This is onaccount of the significant impact the mould profile, in particular the negative taper of the upper[271]part, has on the heat transfer in the meniscus region.Billet rhomboidity arising from unequal heat transfer on adjacent mould faces can be detectedby thermocouples placed on neighboring mould walls. Figure 11.3 shows the hot face temperatureof the mould on adjacent walls for a case in which billets were cast rhomboid at Company C. Asshown in the figure, the ICW (inner curved wall) has higher mould wall temperature than theadjacent wall. (Examination of rhomboid billet samples has confirmed that the rhomboidity isalways oriented in the same direction). The temperatures recorded by the thermocouples for theabove case is plotted in Figure 11.4 and as can be seen the temperatures of the thermocouples onthe ICW are higher than those recorded by the thermocouples on the RSW. The difference intemperature is affected by the fact that thermocouples are not embedded to the same depth on boththe walls. Raw thermocouple data, thus, need to be analysed carefully.No corrective action, short of changing the mould tube can be taken. In fact, if appropriatequality control measures are in place, such a mould would not be put in service at all.11.1.3 Bleeds and LapsThese defects have predominantly been observed on billets that have carbon contents in excessof 0.60%. This, as has been explained in an earlier chapter, is on account of the adverse lubricationcondition, that may arise in ‘hot” mould, to which the thin initial shell of the high carbon billet isparticularly susceptible.The onset of adverse lubrication condition can be detected both by load cells (as explainedin a subsequent section), as well as by thermocouples. The key factor that controls the sustenanceof the mould lubricating oil on the mould wall is the mould hot face temperature, which forappropriate lubrication should not exceed the boiling point of the oil (>200 °C for vegetable basedoil). Thus any change in the mould thermocouple measurement that amounts to an increase in themould hot face temperature beyond the boiling point of the lubricating oil can be taken to represent[272]adverse lubrication condition for high carbon billet.Corrective action can be taken by increasing the mould cooling water velocity, temporarilystopping the lubricating oil flow or changing metal levels in an appropriately designed “smart”mould. All the actions suggested above will cause a reduction in the rate of mould heat transfer andrestore, with the resumption of oil lubrication, appropriate lubrication condition.11.2 Load CellsIn the study conducted load cells were placed between the mould housing and the oscillatortable and signals were collected at 50 Hz. Arising out of this work it appears that friction signalscan be better analysed if the load cells were located in the oscillator arms and were sampled at 500Hz.11.2.1 Transverse Depressions and CracksThe mechanism for the formation of transverse depressions and cracks put forth by Brimacombe, Samarasekera has been confirmed in this work. The mechanism, discussed in detail in anearlier chapter, suggests that the tensile stresses arising out of the binding of a billet in the mouldcan cause the billet surface to “neck”, much like the specimen used in tensile testing. This “necking”may be accompanied by the formation of transverse cracks if the binding of the billet is severeenough. (The billet may bind in the mould if the taper is too steep).Brendzy [13] has shown that the peaks in the load cell signals during the up stroke of themould, tend to be flat in case of normal operation but undulate when the billet binds in the mould.Figure 2.10, which is a plot of load cell signal obtained during the casting of a billet that is bindingin the mould, illustrates this point.In some cases corrective action can be taken by enhancing heat transfer in the mould byaltering appropriate factors as explained in a subsequent section.[273]11.2.2 Adverse lubrication ConditionsIn a previous work [13] adverse lubrication conditions were shown to exist when the loadcells signals, during the up stroke of the mould tended to rachet up’. Similar results have beenfound in the present work as shown in Figure 11.5 which is a representative load cell signalscorresponding to “no-lubricant” condition. It is possible to quantify the “racheting up” of the loadcell response by comparing the sensor signal with load attained at the end of the downstroke of themould.11.2.3 Mould-Billet Interaction at the meniscusA significant mould-billet interaction at the meniscus has been shown to lead to an increasein heat transfer in the upper part of the mould. The enhanced heat transfer is not desirable whilecasting high-carbon grade steel billets but is needed while casting low-carbon grade billets. Toasses, on line, if sufficient mould-billet interaction is taking place, the load cell , LVDT and castingspeed signals need to be analysed together. As been shown in earlier chapters, if the mould-billetinteraction continues for more than 70% of the negative strip period it leads to high heat transferin the meniscus region. A period of interaction which is less than 50% of the negative strip timewill lead to low heat transfer values.11.3 Towards a “Smart” MouldThe preceding sections have shown how sensors such as load cells and LVDTs can used todetect the onset of adverse casting conditions in the mould. The knowledge that the upper taper ofthe mould influences the heat transfer in the meniscus region is a powerful tool for on-line controlof the process and it is possible to conceive of a ‘smart” mould that is so designed as to permitcasting operation at two three different meniscus levels.[274]Thus, with a mould which is untapered till 50mm below the normal meniscus level (meniscuslevel 1), and is followed by a steep taper to the bottom of the mould, it is possible to temporarilyreduce the heat transfer, if necessary, by dropping the meniscus level to a region of steep taper thatreduces mould-billet interaction. It may also be possible to use the mould design suggested abovefor casting high carbon grades where a low heat transfer in the meniscus region is desired. Themeniscus level for casting high carbon steel would normally be in a zone of steep taper (meniscuslevel 2) which would give rise to a low heat transfer. If a high heat transfer is temporarily desired,as for instance during the binding of the billet in the mould, the meniscus level could be raised tolevel 1. The spray design at the bottom of the mould can be linked to the upper taper of the mouldso as to allow changes in the spray water flux corresponding to changes in the meniscus level.Other corrective actions that can be taken while casting billets through such a mould, to alterheat transfer conditions, are changes in:[1] Mould cooling water velocity.[2] Mould oscillation frequency.[3] Mould lubricating oil flow rate.[4] Billet casting speed.The magnitude of change needed to bring about corrective action will have to be determinedthrough several plant trials and appropriate software will need to be developed to implement thecontrol system.[275]QCI-QIDISTANCE FROM TOP OF MOULD (mm)Figure 11.1 Heat Flux, billet shrinkage and mould wall profile at Company C.276j250 Company CP200Rise in temperature150 ICompany B100100 200 300 400 500 600 700 800DISTANCE FROM TOP OF MOULD (mm)Figure 11.2 Mould thermocouple response while casting billets with off-corner cracks.[277]400_ICwmeniscus0 I I100 120 140 160 180 200 220 240 260DISTANCE FROM TOP OF MOULD (mm)Figure 11.3 Mould hot face temperature indicating unequal heat transfer on adjacent faces.[278]£JU-_______________________________________________________meniscus0 a a a100 120 140 160 180 200 220 240 260DISTANCE FROM TOP OF THE MOULDFigure 11.4 Response of mould thermocouple on adjacent mould walls indicating unequal heattransfer.[279]Ze— z- 0Figure 11.5 Load cell response at Company C obtained while casting billets without mould lubricating oil.TIME (s)[280jChapter 12: SUMMARY, CONCLUSIONS AND RECOMMENDATIONSFOR FUTURE WORKData on mould wall temperature and mould-billet interaction has been collected at three steelcompanies in elaborate plant trials organized for this purpose. Operating billet moulds wereinstrumented with arrays of thermocouples, several load cells and LVDTs. Water temperature atthe inlet and outlet of the mould cooling water channel was measured by thermocouples. The liquidsteel surface in the mould was filmedduring casting to observe the behaviour ofthe mould lubricatingoil at the meniscus. Billets samples were collected at pre-determined intervals and subsequentlyanalysed for surface quality and internal cracks.Thermocouple data were analysed with three mathematical models - two of the mould andone of the billet. Axial mould heat-flux profiles, obtained with a heat transfer model of the mould,were verified by comparing them with the total amount of heat extracted by the mould coolingwater. The distortion of the mould was computed by a mathematical model of mould distortion thatuses the temperature distribution in the mould as obtained form the first model of the mould. Shellthickness and shrinkage profiles of the billet were obtained from a mathematical model of the billetdeveloped during the course of this work. Data on the carbon-and-temperature-dependent coefficient of expansion of steel, necessary to model the shrinkage of steel was not directly available andhad to be computed from the experimental values of the lattice parameter of unit cells of alpha andgamma iron. Several curve fitting exercises had to be carried Out Ofl the experimental data availablein the literature to obtain an expression for the coefficient of expansion that incorporates the effectof temperature and carbon. Model verification was done by comparison of the experimentallymeasured, mid-face shell thickness of the billet with the computed ones.[2811The load cell signals, collected at typically 50 Hz., was analysed through several smallcomputer programs developed in this study. The signals were additionally plotted on 1524 X 1224mm long paper to see differences, if any, in the nature of the signal for different types and flowrates ofmould lubricating oils. Analysis of load cells have very strongly corroborated the mechanismof heat transfer in the mould as developed by the study of thermocouple signals.Detailed analysis of billet quality, both in terms of surface appearance and internal cracks,was carried out. Several mathematical model predictions could be verified by measurements madeon the billet.As a result of the various analyses it has been possible to evolve a consistent theory thatcomprehensively explains the various factors that have been found to affect heat transfer in themould, in particular at the meniscus. The theory has been successfully tested in a separate planttrial that has followed this research work. It has been possible to divide the mould into three differentregions from the point of heat transfer. The other important conclusions from the work are as listedbelow.[1] Heat transfer in the upper half of the mould primarily depends on the mould-billet interactionduring the negative strip period of the mould oscillation cycle. Increased interaction leads tohigher heat transfer in Zone I of the mould (Figure 9.4). The high heat transfer in Zone Iraises the hot face temperature of the mould above the boiling point of the mould lubricatingoil leading to the generation of hydrogen gas in the mould-billet gap. Heat transfer is significantly improved in Zone II by the higher thermal conductivity of the hydrogen gascompared to the conductivity of air. (The enhancement in heat transfer in Zone II leads to afurther outward bulging of the mould wall causing an increase in the interaction of the mouldand the billet, and therefore heat transfer, in Zone I). In Zone III, the presence of hydrogen[2821enhances the heat transfer as does a decrease in the mould-oscillation frequency that leavesfewer oscillation marks on the surface of the billet in comparison to the number of marks leftby a higher frequency of mould oscillation.[2] The design of high upper taper (> 2.5 %Im) in the upper part of the mould, prevents the mouldwall from acquiring a bulged shape (or negative taper), by compensating for the differentialexpansion of the mould wall during operation. The absence of a negative taper results in alower mould-billet interaction and, therefore, heat transfer.[3] Calculations have shown that the thermal resistance of the mould-billet gap accounts for80-85% of the total thermal resistance to heat flow in the mould. An increase in the flow rateof the mould lubricating oil leads to an enhancement in the heat transfer if the hot facetemperature of the mould causes the oil to boil. No significant gain in heat transfer is obtainedby increasing the flow rate of the mould lubricating oil beyond 25 mi/mm as “excess” oilcollects on the surface of the liquid steel and does not flow past the meniscus.[4] The absence of the mould-lubricating oil below the meniscus of a “hot” mould and the adverselubrication condition that arise thereby, have been found to be likely causes of bleeds andlaps in the high carbon grade (C> 0.60%) steel billets. The adverse lubrication condition hasbeen found not to affect the surface quality of medium and low carbon billets (C <0.60%).[5] It has been shown that despite the contraction arising from 6 to yphase change in low-carbongrades, the shrinkage of the billets of these grades is small on account of the very low heattransfer to the mould. Thus if low-carbon billets are cast through a mould designed forhigh-carbon grades, they are likely to bind in the mould leading to transverse depressionsand cracks on the billet surface. The model predicted binding of these grades has beencorroborated by an analysis of load cell signals and by an examination of the billet surface.[283][6] It has been shown that mould sensors (thermocouples, load cells and LVDTs) can be used todetect the formation of some billet defects and factors that have been identified to affect heattransfer in the mould, can be altered to bring about corrective action.[7] New mould tapers have been designed that permit a minimum of 50% increase in castingspeed over the casting speed currently in vogue at several Canadian steel plants.[8] New mould tapers have also been designed to eliminate the formation of bleeds and lapswhile casting high-carbon grade billets.[9] In view of the close dimensional tolerances needed for a properly designed mould, theexplosion forming technique of mould manufacture has been recommended.[10] In view of the high heat transfer now available with an appropriately designed mould, it isimportant to ensure that water velocities and back pressures in the mould cooling waterchannel are sufficient to prevent the boiling of water.RECOMMENDATIONS FOR FUTURE WORKIt is very clear from this research work that the heat transfer in the continuous casting mouldcan be controlled during operation by altering the flow rates of mould cooling water and the mouldlubricating oil. Additionally by changing the level of steel in a correctly designed mould, substantialchanges can be made to the shrinkage rate of the billet being cast. The necessary tools to do so haveall been developed in the research work. These concepts need to be implemented in the form of a“smart” mould that is capable of sensing mould-billet interaction and taking corrective action online.It is possible to work towards the creation of the “smart” mould mentioned above in a seriesof well planned plant trials in which, based on the fundamentals developed so far, conditions arecreated first to adversely affect billet quality and subsequently altered to rectify the anomaly. Thiswould not only test the theories developed so far but would build a database to be used in “pro-[284 Igramming the smart mould. It is in the nature ol such work to take several attempts before the’are tuned to perfection and a period ol five \‘ears W( uld he a minimum necessary to carry out theabove task.With the experience acquired and the mathematical and analytical to1s developed during thestudy ){ mould—sira nd interaction iii oi 1—lubricated billet moulds. it should now he possible to studs’and understand in a relatively short period of time the friction effects in billets cast using mouldpovder as a lubricant.REFERENCES1 R. J. Dippenaar, I. V. Samarasekera and J. K. Brimacombe, ISS Transactions, 1986, Vol.7, pp. 3 1-43.2 S. Watanabe, K. Harada, N. Fujita, Y. Tamura and K. Noro, Tetsu-to-Hagane, 1972, Vol.58, pp. 393-394.3 Y. Aketa and K. Ushijima, Tetsu-to-Hagane Overseas (JISI of Japan), 1962, Vol. 2(4), pp.334-343.4 D. P. Evteev, Stahl in English, Aug. 1969, pp. 708-711.5 A. D. Akimeinko and A. A. Skvortzov, Izov. VUZ Chemaya, Metal!., 1961, (10), pp. 29-36.6 A. Grill, K. Sorimachi and J. K. Brimacombe, Metall. Trans. B 1976, 7B(2), pp. 21 1-216.7 E.G. Zetterland and J. K. Kristiansson, Scand. Journ. of Met., 1983, Vol. 12, pp. 211-216.8 I. V. Samarasekera, Report to Stelco Edmonton Steel Works, unpublished work, 1987.9 E. Herrmann, Handuch des Stranggiessens, Aluminium-Verlag, Dusseldorf, 1958.10 G. J. McManus, Iron Age, Vol. 224, No 4, February 1981, pp. MP 7-9, MP 11.11 M. Wolf, Concast Tech., Vol. 21 (3),1982, pp 7-7.12 M. Wolf, Electr. Furn. Proc., AIME, 1983 (40), pp. 335-346.13 L. Brendzy, M.A.Sc. Thesis, Univ. Of British Columbia, 1990.14 Y. Takamura, S. Mizoguchi, 0. Tsubakihara, T. Kuwabra and M. Saito, Nippon Steel Tech.Report, No 21, pp. 198-201, 1983.15 F. Weinberg, Metall. Trans. B, 1979, Vol lOB, pp. 219-227.16 B. G. Thomas, J. K. Brimacombe and 1. V. Samarasekera, ISS Trans., Vol. 7, 1986, pp.95-105.17 1. V. Samarasekera,J. K. Brimacombe and R. Bommaraju, ISS Trans., Vol. 5, 1984, pp.79-94.18 I. V. Samarasekera and J. K. Brirnacombe, Metal!. Trans., Vol. 13B, March 1982, pp.105-116.19 I. V. Sarnarasekera and J. K. Brimacombe, Ironmaking Steelmaking, 1982, Vol. 9, pp. 1-15286]20 J. K. Brimacombe, E. B. Hawbolt and F. Weinberg, ISS Trans., 1982, Vol. 1, PP. 29-40.21 I. V. Samarasekera, Ph.d Thesis, Univ. Of British Columbia, 1980.22 1. V. Samarasekera and J. K. Brimacombe, Canadian Metall. Quart., Vol. 18, 1979, pp.25 1-266.23 F. P. Bowden and D. Tabor, “The Friction and Lubrication of Solids”, Oxford Univ. Press,1950.24 F. D. Dudley, “Theory and Practice of Lubrication for Engineers”, New York, Wiley, 1956.25 B. Mairy and M. Wolf, Facherichte, Vol. 20, No. r, April 1982, pp. 222-227.26 S. N. Singh and K. E. Blazek, Iron and Steel Maker, October 1987, pp. 36-38.27 J. Stel, J. M. Ranberg, M. C. M. Cornelissen and J. Cijsouw, 1st European Conference onContinuous Casting, Italy, 1991, Pp. 2.377-2.386.28 M. Komatsu,T. Kitagawa, K. Kawakami, 104th ISlJmeeting, September 1982,LectureNo.S927, Trans. ISIJ, 1983, 23, pp. B-86.29 1. V. Samarasekera and J. K. Brimacombe, W. 0. Philbrook Memorial Symposium Conference Proceedings, 1988, PP. 157-171.30 1. V. Samarasekera and J. K. Brimacombe, International Metals Review, 1978, No. 6, pp.286-300.31 A. D. Klipov, A. I. Kolpakov, M. G. Chigrinov and E. R. Ballad, Stalin English, 1971, (2),107-111.32 C. R. Taylor, Metall. Trans., 1975, 6B, PP. 359-375.33 S. N. Singh and K. E. Blazek, Open Hearth Proc., AIME, 1977, 59, pp. 264-28334 S. Deshimaru, S. Orniya, H. Mizota, M. Yao, M. MaedaandT. Imai, Trans. ISIJ, 1984, Vol.24, pp. B-339.35 M. Wolf, Trans. 1SIJ, 1980, Vol. 20, Pp. 7 18-724.36 B. Short, F. Faires and T. Horton, Concast Technology News, Vol. 26, No 3, 1987, Pp. 5-5.37 H. Yarnanaka, M. Ikeda, T. Nishitani and T. Ando, Trans. ISIJ, Vol. 23, No. 10, 1983, pp.F-4.38 M. Wolf, 103rd ISIJ meeting, April 1982, Lecture No. S 149, Trans. ISIJ, 1982, 22, (7), pp.B-204.39 G. Foussal, Rev. Metall., Vol. 4, 1984, pp. 299-307.[287140 M. Komatsu, T. Kitagawa, K. Kawakami, 104th ISIJ meeting, September 1982, Lecture No.S927, Trans. ISIJ, 1983, 23, pp. B-86.41 H. Gloor, Concast Standard News, Vol. 30, No 1, 1991, pp. 4-5.42 H. Mizukami, M. Komatsu, T. Kitagawa and K. Kawakami, 106th ISIJ meeting, October1983, Lecture No. S 1032, Trans. ISIJ, 1984, 21, (6), pp. B-181.43 5. Ohmiya, M. Nakato, Y. Habu, T. Emi, K. Hamagami, H. Bada and Y. Fukuhara, 104thISIJ meeting, September 1982, Lecture No. S926, Trans. ISIJ, 1983, 23, pp. B-85.44 J. Savage and W. H. Pritchard, JISI, Vol. 178, 1954, pp. 269-277.45 I. M. D. Halliday, JISI, Vol 191, 1959, pp. 121-163.46 V. A. Kobelev, Stalin English, 1967, pp. 47 1-473.47 H. Jacobi, G. Komma, K. Wunnenberg, Stahl und Eisen, Vol. 102, 1982, pp. 441-449.48 J. P. Birat, Rev. Metall., Vol 79, 1982, pp. 603-616.49 R. Sato, Proc. 62nd NOH-BOSC, ISS-AIME, 1979, pp. 48-67.50 V. P. Perminov, N. M. Lapotyshkin, V. E. Girskii and A. I. Chizhikov, Stalin English, 1968,Vol. 7, pp. 560-563.51 H. Mori, Testu-to-Hagane, 1972, Vol. 58, pp. 1511-1525.52 W. P. Young and W. T. Whitfield, Open Hearth Proc., AIME, 1968, Vol. 51, pp. 127-132.53 K. Matsunaga, Open Hearth Proc., AIME, 1976, Vol, 59, pp. 228-248.54 K. Ushijima, Continuous Casting of Steel, Iron Steel Inst., London, 1964, pp. 59-7 1.55 Y. Aketa and K. Ushijima, Testu-to-Hagane, 1960, Vol. 46, pp. 1733-1740.56 H. G. Baumann and W. J. Lopmann, Wire World mt., 1974, Vol. 16, pp. 149-155.57 K. Fujinami, Funabashi Steelworks Ltd., Funabashi, Japan, unpublished research, 1977.58 J. E. McConnell, Open Hearth Proc., AIME, 1972, Vol. 55, pp. 56-72.59 R. Bommaraju, J. K. Brimacombe and I. V. Samarasekera, ISS Transactions, Vol. 5, 1984,pp. 95-105.60 J. W. Donaldson, Journal of Metals, December 1965, pp. 1-6.61 D. I. Brown, Journal of Metals, April 1965, pp. 2-8.[288162 I. V. Samarasekera, D. L. Anderson and J. K. Brimacombe, Met. Trans. B, 1982, Vol. 13B,pp. 91-104.63 J. K. Brimacombe, I. V. Samarasekera and R. Bommaraju, Steelmaking Proceedings (ISS),1986, 69, 409-423.64 I. A. Bakshi, A. Perri, J. K. Brimacombe, I. V. Samarasekera and R. P. Smith, U.S. PatentApplication, 1991.65 C. D. Henning and R. Parker, Trans. ASME, May 1967, pp. 146-154.66 M. H. Burden, G. D. Funnel!, A. G. Whitaker and J. M. Young, Solidification and Castingof Metals, The Metals Society London, 1979, pp. 279-286.C.67 I. V. Samarasekera, Report to Sidbec_dosco Steel Plant, unpublished work, June 1991.68 D. W. Peaceman and H. H. Rachford Jr., J. Soc. Indust. App!. Math, 1955, 3, pp 28-41.69 Handbook of Heat Transfer, W. M. Rohsenow and J. P. Hartnett, eds., McGraw Hill BookCo., 1973, pp. 13-1 to 13-75.70 E. A. Mizikar, Trans. TMS-AIME, 1967, 239, pp. 1747-1753.71 J. Szekely and V. Stanek, Metall. Trans., 1970, 1, pp. 119-126.72 R. B. Mahapatra, Ph.D Thesis, Univ. of British Columbia, 1989.73 B. G. Thomas, 1. V. Samarasekera and J. K. Brimacombe, Metall. Trans, 18b, 1987, pp.119-130.74 H. Frederiksson and M.Thegerstrom, Scand. Journ. of Met., 1979, 8, pp. 232-240.75 J. K. Brimacombe, I. V. Samarasekera, N. Walker, 1. Bakshi, R. Bommaraju, F. WeinbergandE. B. Hawbolt, Trans. ISS, 1984,5, pp. 95-105.76 P. J. Wray, Proc. AIME Symp. On Modelling of Casting and Welding Processes, 1980, pp.245-257.77 N. Ridley and H. Stuart, Met. Sci., 1970, 4, pp. 219-222.78 L. D. Lucas, Mern. Sci. Rev. Met., 1964, 2, pp. 1-24.79 E. J. Fasiska and Wagenbiast, Trans. Metal!. Soc. AIME, Nov. 1967, 239, pp. 18 18-1820.80 1. A. Bakshi, E. Osinski, 1. V. Samarasekera and J. K. Brimacombe, Intl. Symp. on DirectRolling and Hot Charging, eds. I. J. Jonas, R. W. Pugh and S. Yue, CIM, 1988, 10, pp.60-69.[289]81 J. K. Brimacombe, I. V. Samarasekera and R. Bommaraju, Proceedings of SteelmakingConference of the Iron and Steel Society of AIME, Washington, D. C., 1986,69, pp.409-423.82 D. P. Lorento, Paper presented to Globetrotters Meetings, 1988.83 1. V. Samarasekera, Report to Stelco Edmonton Steel Works, unpublished work, 1992.[2901

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0078515/manifest

Comment

Related Items