UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Mould thermal response and formation of defects in the continuous casting of steel billets Kumar, Sunil 1996

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

Item Metadata

Download

Media
831-ubc_1996-091104.pdf [ 13.32MB ]
Metadata
JSON: 831-1.0078578.json
JSON-LD: 831-1.0078578-ld.json
RDF/XML (Pretty): 831-1.0078578-rdf.xml
RDF/JSON: 831-1.0078578-rdf.json
Turtle: 831-1.0078578-turtle.txt
N-Triples: 831-1.0078578-rdf-ntriples.txt
Original Record: 831-1.0078578-source.json
Full Text
831-1.0078578-fulltext.txt
Citation
831-1.0078578.ris

Full Text

MOULD THERMAL RESPONSE AND FORMATION OF DEFECTS IN THE CONTINUOUS CASTING OF STEEL BILLETS by S U N I L K U M A R B . T e c h . ( M e t a l l u r g i c a l Engineer ing ) Banaras H i n d u Un ive r s i t y , India , 1987 M . A . S c (Meta ls and Mater ia l s Engineer ing) Un ive r s i t y o f B r i t i s h C o l u m b i a , Canada , 1991 A T H E S I S S U B M I T T E D I N P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F D O C T O R O F P H I L O S O P H Y i n T H E F A C U L T Y O F G R A D U A T E S T U D I E S Department o f Me ta l s and Mater ia l s Eng inee r ing W e accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A January 1996 © S u n i l K u m a r , 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^ e ^ l i ^ - ^ q X e - y i W $ ^ " ^ " o e e r i n The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT W i t h the dr ive to op t imize the design o f bi l le t moulds and establish standard operating practices, s ignificant improvements i n b i l le t quali ty have been achieved i n the last two decades. A s this trend continues, the role o f process upsets (or unsteady-state events) i n the m o u l d , such as metal l e v e l fluctuations and/or poor maintenance, are assuming increasing importance, relative to caster design and operation, i n the generation o f quali ty problems. O f the c o m m o n process upsets encountered i n b i l le t cast ing, r andom metal l eve l fluctuations due to open-stream pour ing , are by far the most significant. The metal l eve l fluctuations originate f rom the roughness o f the metal stream p lung ing f rom the tundish into the m o u l d poo l . Rougher streams entrain more gas i n the m o u l d p o o l creating more bubbles w h i c h upon their eruption f rom the surface generate turbulence and waves. T h e result ing cont inuously shift ing meniscus interacts w i t h the o i l weep ing d o w n the m o u l d w a l l to create variable lubr icat ion and heat-flow condi t ions , both on a g iven face and one face relat ive to another; this generates a number o f qual i ty problems i n b i l le ts . The m a i n focus o f this study was to understand the nature o f metal l eve l fluctuations and their impact on the format ion o f b i l le t defects such as laps, bleeds and rhomboid i ty . A series o f s ix instrumented-mould trials were conducted at four Canad ian m i n i - m i l l s and data on m o u l d w a l l temperature, c o o l i n g water temperature, metal l eve l f luctuations, casting speed, osc i l l a t ion characteristics and mould/strand fr ic t ion, were acquired us ing a computer-based data acquis i t ion system. In addi t ion, the tundish pour ing stream and the meniscus i n the m o u l d were f i l m e d us ing a v ideo camera; and 35 m m photographs were also taken. A number o f bi l le t samples corresponding to the data gathered, were col lec ted for qual i ty evaluat ion. In the more recent trials, a Supervisory C o n t r o l A n d Da ta A c q u i s i t i o n ( S C A D A ) sys tem was instal led and the m o u l d temperatures were moni tored on-l ine as w e l l . T h i s study has uncovered the mechanism o f defect formation i n the meniscus reg ion due to r andom metal l eve l fluctuations. The importance o f peak hot face temperature relative to the b o i l i n g range o f o i l has been h ighl ighted w i t h respect to m o u l d heat transfer and defect format ion. T h e analysis o f m o u l d heat transfer suggests that the greatest enhancement i n heat transfer due to o i l b o i l i n g and/or pyro lys i s , occurs when the peak hot face temperature falls i n the b o i l i n g range o f o i l . A new mechan i sm rooted i n metal l eve l fluctuations, peak hot face temperature and o i l d is t r ibut ion has been proposed to expla in the formation o f laps and bleeds i n b i l le t cast ing. W i t h respect to rhomboid i ty , it was shown that a combina t ion o f h igh m o u l d heat transfer and a short freezing range is important f rom the standpoint o f severity o f the p rob lem, i n addi t ion to the non-un i fo rm heat transfer condi t ions present i n the meniscus region. T h i s mechan i sm explains the higher severity o f rhomboid i ty i n m e d i u m carbon grades observed i n this study and also reported i n the literature. W i t h respect to the development o f the "intell igent m o u l d " , an expert-system-based moni to r ing and control system for b i l le t casting, strategies for the on- l ine , hot inspect ion o f b i l le t defects were out l ined based on a detailed analysis o f measured m o u l d thermal response and b i l le t qual i ty evaluat ion. Further, a number o f interesting simulat ions were undertaken to evaluate the effect o f var ious m o u l d events on the thermal response o f the m o u l d . In the case o f transverse depressions, the m o u l d thermal response was computed us ing the characteristics o f depressions measured on an entire bi l le t ; a good agreement was obtained between the measured and the computed m o u l d thermal response. Thus , the f indings o f this study have a significant role to p lay i n the creat ion o f the "intel l igent m o u l d " and also, i n the development o f a knowledge base for the on- l ine , hot surface inspect ion o f b i l le ts . i i i TABLE OF CONTENTS Abstract i i Table of Contents iv List of Tables v i i i List of Figures x List of Symbols x x i i Acknowledgements x x i v CHAPTER 1 - INTRODUCTION 1 CHAPTER 2 - PREVIOUS WORK 6 2.1 B i l l e t M o u l d and its Thermo-mechan ica l Behav iou r 6 2.2 Qua l i t y Problems, M e n i s c u s and M o u l d / S t r a n d Interaction 7 2.2.1 L a p s and bleeds 8 2.2.2 R h o m b o i d i t y 9 2.2.3 Transverse depressions 12 2.3 Process Upsets , M e t a l L e v e l Fluctuat ions and Qua l i ty Problems 13 2.4 Causes o f M e t a l L e v e l Fluctuat ions i n B i l l e t M o u l d s 15 2.5 Sensors and Instrumentation i n Cont inuous Cas t ing 16 2.5.1 Temperature 16 2.5.2 M o u l d displacement 17 2.5.3 M e t a l l eve l and casting speed 18 2.5.4 Mould / s t r and fr ic t ion 19 2.5.5 Other variables 19 2.6 E x i s t i n g On- l i ne M o n i t o r i n g and C o n t r o l Systems 20 2.7 On- l i ne Detec t ion o f Defects i n Cont inuous Cas t ing 21 i v CHAPTER 3 - SCOPE AND OBJECTIVES 34 CHAPTER 4 - PLANT TRIALS 37 4.1 Instrumentation 37 4.1.1 M o u l d temperature measurement 37 4.1.1.1 Conve r t i ng measured thermocouple voltage into temperature 39 4.1.1.2 Ca lcu l a t i on o f f ina l temperature 4 0 4.1.2 Other measurements 4 0 4.1.3 Da ta acquis i t ion system 41 4.2 Detai ls o f Plant Tr ia l s 42 4.3 A n a l y s i s o f Temperature D a t a 43 4.4 B i l l e t Qua l i t y Eva lua t ion 45 CHAPTER 5 - RESULTS OF PLANT TRIALS 62 5.1 B i l l e t Qua l i t y Eva lua t i on 62 5.1.1 L a p s and bleeds 62 5.1.2 R h o m b o i d i t y 65 5.1.2.1 R h o m b o i d i t y measurement on bi l le t samples 65 5.1.2.2 R h o m b o i d i t y measurement on entire b i l le t length 69 5.1.3 Transverse depressions 7 0 5.2 M e t a l L e v e l Fluctuat ions 72 5.3 Ave rage M o u l d Temperature Da ta 72 5.4 B i l l e t Defects and M o u l d The rma l Response 73 5.4.1 L a p s 73 5.4.2 R h o m b o i d i t y 73 v 5.4.3 Transverse depressions 74 CHAPTER 6 - MODELLING OF MOULD THERMAL RESPONSE 124 6.1 T h e M o u l d Heat Transfer M o d e l 124 6.2 Steady-State M o u l d T h e r m a l Response 128 6.2.1 M o u l d heat f lux and temperature dis tr ibut ion 129 6.2.1.1 Effect o f steel carbon content 129 6.2.1.2 Effect o f osc i l la t ion frequency 130 6.2.1.3 Difference between o i l and m o u l d f lux 130 6.2.2 A n a l y s i s o f thermal resistance to m o u l d heat extraction 131 6.3 Unsteady-State M o u l d T h e r m a l Response 135 6.3.1 T h e r m a l response o f bi l le t moulds 135 6.3.2 T h e r m a l disturbances due to m o u l d osc i l la t ion 135 6.3.3 M e t a l l eve l fluctuations 136 6.3.4 S imu la t i on o f transverse depressions i n the t r ia l D - 2 137 CHAPTER 7 - METAL LEVEL FLUCTUATIONS IN BILLET CASTING 177 7.1 M o n i t o r i n g M e n i s c u s Fluctuat ions w i t h E x i s t i n g Sensors 177 7.2 Sens ing M e n i s c u s Fluctuat ions w i t h Thermocouples 178 7.3 O n - l i n e M o n i t o r i n g o f M e t a l L e v e l Fluctuat ions 180 CHAPTER 8 - LAPS AND BLEEDS IN BILLET CASTING 194 8.1 Detec t ion o f L a p s and Bleeds us ing Thermocouples 194 8.2 M e c h a n i s m o f Fo rma t ion o f L a p s and Bleeds 196 8.3 M i n i m i z i n g L a p s and Bleeds i n B i l l e t Cas t ing 202 v i CHAPTER 9 - RHOMBOIDITY IN BILLET CASTING 220 9.1 Detec t ion o f R h o m b o i d i t y 220 9.2 M e c h a n i s m o f R h o m b o i d i t y 222 9.3 M i n i m i z i n g R h o m b o i d i t y P r o b l e m i n B i l l e t Cas t ing 234 CHAPTER 10 - DEVELOPMENT OF THE INTELLIGENT MOULD 245 10.1 Intell igent M o u l d - A B a c k g r o u n d 245 10.2 On- l i ne Inspection o f B i l l e t s 246 CHAPTER 11 - SUMMARY AND CONCLUSIONS 249 11.1 Genera t ion o f N e w K n o w l e d g e and M a j o r Contr ibut ions 259 11.2 Recommendat ions for Future W o r k 261 REFERENCES 264 APPENDIX A - Data Acquisition Hardware 272 APPENDIX B - Steel Composition 274 APPENDIX C - Billet Quality Index 282 v i i LIST OF TABLES Table 1.1 Operat ion- and machine-related process upsets i n b i l le t cast ing 5 Tab le 2.1 Internal cracks i n b i l le t casting [18] 24 Tab le 2.2 M i d w a y cracks i n b i l le t casting [18] 24 T a b l e 2.3 Surface cracks i n b i l l e t casting [18] 25 Tab le 2.4 R h o m b o i d i t y p rob lem i n bi l le t casting [18] 25 Tab le 2.5 L a p s , bleeds and zipper marks i n bi l le t casting [18] 26 Tab le 2.6 Centre l ine segregation and pinholes /b lowholes i n b i l le t cast ing [18] 26 Tab le 2.7 Breakout p rob lem i n bi l le t casting [18] 27 Tab le 2.8 A general outl ine o f mould-related factors contr ibut ing to adverse mould/strand interaction [18] 27 Tab le 2.9 Research recommended m o u l d characteristics for b i l le t cast ing machines [3] 28 Tab le 4.1 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n tr ial A - 1 46 T a b l e 4 .2 Dep th and ax i a l pos i t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n tr ial A - 2 47 Tab le 4.3 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n tr ial C - l 48 Tab le 4.4 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n trials D - l and D - 2 49 Tab le 4.5 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n t r ia l E - l 50 Tab le 4.6 Coeff icients ao to a 7 for type-T thermocouples used for conver t ing voltage into temperature 51 T a b l e 4.7 M a j o r des ign details and operating condi t ions i n the plant trials 52 Tab le 4.8 Grades o f steels examined i n the plant trials 53 Tab le 4.9 Operat ing parameters that were var ied dur ing the plant trials 53 Tab le 4.10 B o i l i n g range o f lubr icat ing oi ls used i n the plant trials [5] 54 v i i i Table 5.1 R h o m b o i d i t y values ( in m m ) measured on bi l le t samples acquired f rom the test and control strands dur ing t r ia l E - l 75 Tab le 6.1 Properties o f m o u l d f lux employed i n t r ia l A - 2 143 Tab le 6.2 C o m p a r i s o n o f frequency distr ibution o f computed and measured temperature drops at four locations (170, 270, 370 and 465 m m be low the meniscus) for B i l l e t #2 that was cast dur ing T r i a l D - 2 (Heat #541333) 144 Tab le 8.1 Characterist ics o f temperature drops or "val leys" associated w i t h laps (no.6 to 15) at C o m p a n y A determined f rom the response o f m o u l d thermocouples located be low the meniscus 204 Tab le 8.2 M o u l d temperature change (rise and fall) 7 m m above the meniscus corresponding to formation o f laps (no.6 to 15 i n Tab le 8.1) and the inferred speed o f temperature drops or "val leys" d o w n the m o u l d 205 i x LIST OF FIGURES Figure 2.1 Schemat ic d iagram o f a bi l le t casting m o u l d showing its var ious components 29 F igu re 2.2 Schemat ic d iagram showing a bi l le t w i t h non-un i fo rm so l id shel l thickness be ing distorted into r h o m b o i d shape by spray c o o l i n g [22] 30 F igu re 2.3 Effect o f m o u l d taper on rhomboid i ty (off-squareness) for a range o f steel carbon contents [25] 31 F igu re 2.4 M e c h a n i s m o f gas entrainment proposed by S o m m e r v i l l e and M c K e o g h [ 3 5 ] 32 F igu re 2.5 Schemat ic d iagram showing the pr inciples o f metal l eve l detector e m p l o y i n g a radioact ive source 33 F igu re 4.1 Schemat ic d iagram o f the thermocouple design adopted for the plant trials 55 F igu re 4.2 L a y o u t o f m o u l d thermocouples employed for plant t r ia l A - l . (Numbers indicate distance f rom top o f m o u l d i n m m ) 56 F igure 4.3 L a y o u t o f m o u l d thermocouples employed for plant tr ial A - 2 (Numbers indicate distance f rom top o f m o u l d i n m m ) 57 F igure 4.4 L a y o u t o f m o u l d thermocouples employed for plant tr ial C - l (Numbers indicate distance f rom top o f m o u l d i n m m ) 58 F igu re 4.5 L a y o u t o f m o u l d thermocouples employed for plant trials D - l and D - 2 (Numbers indicate distance f rom top o f m o u l d i n m m ) 59 F igure 4.6 L a y o u t o f m o u l d thermocouples employed for plant t r ia l E - l (Numbers indicate distance f rom top o f m o u l d i n m m ) 60 F igu re 4.7 T h e arrangement o f thermocouples for measuring m o u l d w a l l and c o o l i n g water temperature 61 F igu re 5.1 Surface photograph o f a 1.02 pet. carbon steel b i l le t f rom C o m p a n y A showing laps 76 F igure 5.2 Surface photograph o f a 0.84 pet. carbon steel b i l le t f rom C o m p a n y C showing laps 77 F igu re 5.3 Surface photograph o f a 0.45 pet. carbon steel b i l le t surface showing a smooth surface i n the absence o f laps 78 F igure 5.4 Photograph o f the subsurface structure o f the lap shown i n F igure 5.1, reveal ing laps 79 F igu re 5.5 Photograph o f a 0.84 pet. carbon steel b i l le t surface showing a b leed 80 x Figure 5.6 Photograph o f a 0.57 pet. carbon steel bi l le t surface showing bleeds 81 F igure 5.7 Photograph o f the subsurface structure o f the b leed shown i n F igure 5.5 82 F igure 5.8 Effect o f steel carbon content on the severity o f laps and bleeds observed i n the plant t r ia l at C o m p a n y C . Freez ing range o f steels is also shown 83 F igu re 5.9 Effect o f m o u l d taper at the meniscus on the severity o f laps and bleeds observed i n the plant trials at Companies A , C and E 83 F igu re 5.10 M i n i m u m and m a x i m u m values o f rhomboid i ty measured for the various grades examined dur ing tr ial A - 1 84 F igu re 5.11 Frequency distr ibution o f rhomboid i ty values measured for the various grades examined dur ing trial A - 1 84 F igu re 5.12 M a x i m u m values o f rhomboid i ty measured for the various grades examined dur ing tr ial C - l 85 F i g u r e 5.13 Frequency dis tr ibut ion o f rhomboid i ty values measured for the various grades examined dur ing trial C - l 85 F igure 5.14 Range o f rhomboid i ty values measured for the various grades cast on the test strand dur ing tr ial E - l 86 F i g u r e 5.15 R a n g e o f rhomboid i ty values measured for the var ious grades cast o n the control strand dur ing tr ial E - l 86 F igure 5.16 Frequency distr ibution o f rhomboid i ty values measured for the various grades examined dur ing tr ial E - l 87 F igu re 5.17 R h o m b o i d i t y values measured for the various grades examined dur ing tr ial D - l 88 F igure 5.18 R h o m b o i d i t y values measured for the various grades examined dur ing a tr ial at C o m p a n y B 89 F igu re 5.19 M i n i m u m , average and m a x i m u m values o f rhomboid i ty measured for the various grades examined dur ing a tr ial at C o m p a n y B 89 F igu re 5.20 Effect o f m o u l d taper at the meniscus on the severity o f rhomboid i ty observed i n the plant trials at Compan ies A , B , C and E 90 F igu re 5.21 R h o m b o i d i t y measurement a long the length o f B i l l e t #2 for a m e d i u m carbon steel containing boron and t i tanium i n t r ia l D - 2 (Heat # 541333): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurement is 19) 91 x i Figure 5.22 R h o m b o i d i t y measurement a long the length o f B i l l e t #4 for a m e d i u m carbon steel conta ining boron and t i tanium i n tr ial D - 2 (Heat # 541333): (a) measured values and (b) frequency dis tr ibut ion. (Note : To ta l number o f measurement is 19) 92 F igure 5.23 R h o m b o i d i t y measurement a long the length o f B i l l e t #5 for a m e d i u m carbon steel conta ining boron and t i tanium i n tr ial D - 2 (Heat # 541333): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurement is 19) 93 F igu re 5.24 R h o m b o i d i t y measurement a long the length o f B i l l e t #8 for a m e d i u m carbon steel grade containing boron and t i tanium i n t r ia l D - 2 (Heat # 541333): (a) measured values and (b) frequency dis t r ibut ion. (Note: To ta l number o f measurement is 19) 94 F igu re 5.25 R h o m b o i d i t y measurement a long the length o f B i l l e t #10 for a m e d i u m carbon steel grade containing boron and t i tanium i n t r ia l D - 2 (Heat # 541333): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurement is 19) 95 F igure 5.26 R h o m b o i d i t y measurement along the length o f B i l l e t #19 for a m e d i u m carbon steel containing boron and t i tanium i n t r ia l D - 2 (Heat # 541333): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurement is 16) 96 F igure 5.27 Range o f rhomboid i ty values measured along the length o f B i l l e t s #2, #4, #5, #8, #10 and #19 for a m e d i u m carbon steel conta ining boron and t i tanium i n tr ial D - 2 (Heat # 541333) 97 F igu re 5.28 R h o m b o i d i t y measurement a long the length o f B i l l e t #1 for a h igh carbon steel i n tr ial D - 2 (Heat #541351) : (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurement is 20) 98 F igu re 5.29 R h o m b o i d i t y measurement a long the length o f B i l l e t #3 for a h igh carbon steel i n tr ial D - 2 (Heat # 541351): (a) measured values and (b) frequency dis t r ibut ion. (Note : T o t a l number o f measurement is 20) 99 F igu re 5.30 R h o m b o i d i t y measurement a long the length o f B i l l e t #6 for a h igh carbon steel i n t r ia l D - 2 (Heat #541351) : (a) measured values and (b) frequency distr ibution. (Note: To ta l number o f measurement is 20) 100 F igure 5.31 R h o m b o i d i t y measurement a long the length o f B i l l e t #8 for a h igh carbon steel i n t r ia l D - 2 (Heat # 541351): (a) measured values and (b) frequency distr ibution. (Note: To ta l number o f measurement is 20) 101 x i i Figu re 5.32 R h o m b o i d i t y measurement a long the length o f B i l l e t #10 for a h igh carbon steel i n tr ial D - 2 (Heat # 541351): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurement is 20) 102 F igu re 5.33 R h o m b o i d i t y measurement a long the length o f B i l l e t #11 for a h i g h carbon steel i n t r ia l D - 2 (Heat # 541352): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurement is 20) 103 F igu re 5.34 R h o m b o i d i t y measurement a long the length o f B i l l e t #13 for a h igh carbon steel i n tr ial D - 2 (Heat # 541352): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurement is 19) 104 F igu re 5.35 Range o f rhomboid i ty values measured along the length o f B i l l e t s #1, #3, #5, #6, #8, #10, #11 and #13 for a h igh carbon steel i n t r ia l D - 2 (Heats # 541351 and 541352) 105 F igure 5.36 R h o m b o i d i t y measurement a long the length o f B i l l e t #16 for 0.12 pet. carbon steel i n tr ial D - 2 (Heat # 541297): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurement is 13) 106 F igure 5.37 Macro-photograph o f a 0.34 pet. carbon steel (containing boron and t i tanium) bi l le t surface showing transverse depression (Tr i a l D - l ; Heat #531147) 107 F igure 5.3 8 Dep th o f transverse depressions measured at the midface o f B i l le t #2 o f a m e d i u m carbon steel containing boron and t i tanium (Tr i a l D - 2 ; Heat #541333) 108 F igure 5.39 Frequency distr ibution o f the depth o f transverse depressions observed on B i l l e t #2 (Tr ia l D - 2 ; Heat # 541333) 108 F igu re 5.40 L o n g i t u d i n a l w id th ( in casting direction) o f transverse depressions measured at the midface o f B i l l e t #2 o f a m e d i u m carbon steel conta in ing boron and t i tanium (Tr i a l D - 2 ; Heat #541333) 109 F igu re 5.41 Frequency dis tr ibut ion o f the longi tudinal w i d t h o f transverse depressions observed on B i l l e t #2 (Tr i a l D - 2 ; Heat # 541333) 109 F igu re 5.42 Transverse w i d t h (perpendicular to casting direction) o f transverse depressions measured at the midface o f B i l l e t #2 o f a m e d i u m carbon steel containing boron and t i tanium (Tr ia l D - 2 ; Heat #541333) 110 F igure 5.43 Frequency distr ibution o f the transverse w i d t h o f transverse depressions observed on B i l l e t #2 (Tr ia l D - 2 ; Heat # 541333) 110 F igu re 5.44 Dis tance between consecutive transverse depressions observed on B i l l e t #2 (Tr i a l D - 2 ; Heat # 541333) I l l x i i i Figure 5.45 Frequency distr ibution o f distance between consecutive transverse depressions observed on B i l l e t #2 (Tr ia l D - 2 ; Heat # 541333) 111 F igure 5.46 Dis tance between consecutive transverse depressions observed on B i l l e t #4 (Tr ia l D - 2 ; Heat # 541333) , 112 F igure 5.47 Frequency distr ibution o f distance between consecutive transverse depressions observed on B i l l e t #4 (Tr ia l D - 2 ; Heat # 541333) 112 F igure 5.48 Dis tance between consecutive transverse depressions observed o n B i l l e t #5 (Tr i a l D - 2 ; Heat # 541333) 113 F igu re 5.49 Frequency dis tr ibut ion o f distance between consecutive transverse depressions observed on B i l l e t #5 (Tr i a l D - 2 ; Heat # 541333) 113 F igu re 5.50 Dis tance between consecutive transverse depressions observed on B i l l e t #8 (Tr ia l D - 2 ; Heat # 541333) 114 F igure 5.51 Frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t #8 (Tr ia l D - 2 ; Heat # 541333) 114 F igure 5.52 Dis tance between consecutive transverse depressions observed o n B i l l e t #10 (Tr i a l D - 2 ; Heat # 541333) 115 F igure 5.53 Frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t #10 (Tr ia l D - 2 ; Heat # 541333) 115 F igure 5.54 O v e r a l l frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t s #2, #4, #5, #8 and #10 (Tr i a l D - 2 ; Heat #541333) 116 F igure 5.55 E x a m p l e s o f (a) metal l eve l and (b) casting speed signals dur ing tr ial C - l (Heat # C 7 6 6 1 ) . (Note: The nomina l metal l eve l is - 1 3 7 m m f rom top o f m o u l d and the average casting speed is ~33.5 m/s) 117 F igure 5.56 Response o f a thermocouple located above the meniscus corresponding to the metal l eve l s ignal presented i n F igure 5.55 (a) (Tr ia l C - l ; H e a t # C 7 6 6 1 ) 118 F igure 5.57 Average m o u l d temperature profiles employed to compute heat f lux for three steel grades cast dur ing tr ial D - l : 0.12 pet. carbon steel (Heat # 531142); 0.32 pet. carbon steel (Heat # 531146) and 0.84 pet. carbon (Heat #531149) steel 119 F igu re 5.58 M o u l d thermal response for C o m p a n y C corresponding to a 0.84 pet. carbon steel (Heat # D6131) where laps were present, and a 0.46 pet. carbon steel (Heat # D6135) grade where laps were not observed, at two locat ions (a) 310 m m and (b) 620 m m be low the meniscus 120 x i v Figu re 5.59 M o u l d thermal response at a locat ion - 3 0 m m be low the meniscus corresponding to (a) large rhomboid i ty observed i n 0.32 pet. carbon steel (Heat #541333 , B i l l e t #19) and (b) smal l rhomboid i ty observed i n 0.12 pet. carbon steel (Heat # 541297, B i l l e t #16) cast dur ing t r ia l D - 2 121 F igu re 5.60 M o u l d thermal response at a locat ion - 4 5 m m be low the meniscus corresponding to (a) large rhomboid i ty observed i n 0.32 pet. carbon steel (Heat# 541333, B i l l e t#19) and (b) smal l rhomboid i ty observed i n 0.12 pet. carbon steel (Heat # 541297, B i l l e t #16) cast dur ing t r ia l D - 2 122 F igu re 5.61 M o u l d thermal response measured dur ing tr ial D - 2 f rom a heat (# 541333) o f a m e d i u m carbon steel containing boron and t i tan ium corresponding to B i l l e t #2 (wi th 46 depressions) and B i l l e t #19 (no depression), at two locations (a) 270 m m and (b) 565 m m b e l o w the meniscus 123 F igu re 6.1 Schemat ic d iagram o f the longi tudinal section o f the m o u l d w a l l at the midface that was mode l led 145 F igu re 6.2 Out l ine o f the procedure adopted i n the mode l calculat ions o f heat f lux and temperature distr ibution i n the m o u l d w a l l f rom temperature measurements conducted i n the plant trials 146 F igure 6.3 M o u l d heat transfer for 0.12 pet. carbon steel (Heat # 531142) , 0.32 pet. carbon steel (Heat # 531146) and 0.84 pet. carbon steel (Heat # 531149) cast dur ing tr ial D - l : (a) heat f lux profiles and (b) average and peak values 147 F igu re 6.4 A x i a l temperature profi le at (a) hot face and (b) c o l d face for 0.12 pet. carbon steel (Heat # 531142), 0.32 pet. carbon steel (Heat # 531146) and 0.84 pet. carbon steel (Heat # 531149) cast dur ing t r ia l D - l 148 F igu re 6.5 Effect o f osc i l l a t ion frequency on m o u l d heat extraction i n (a) t r ia l C - l (Heat # A 2 8 1 8 8 and C7654) and (b) tr ial A - 2 (Heat # E 3 0 7 4 3 ) 149 F igu re 6.6 Effect o f osc i l la t ion frequency on m o u l d hot and c o l d face temperature profiles i n (a) trial C - l (Heat # A 2 8 1 8 8 and C 7 6 5 4 ) and (b) tr ial A - 2 (Heat # E30743) 150 F igu re 6.7 C o m p a r i s o n o f m o u l d heat extraction for a h igh carbon steel cast w i t h o i l (Heat # E30719) and m o u l d f lux (Heat # E 3 0 7 1 8 ) lubricants i n t r ia l A - 2 151 F igure 6.8 C o m p a r i s o n o f m o u l d temperature for a h igh carbon grade cast w i t h o i l (Heat # E30719) and m o u l d f lux (Heat # E30718) lubricants i n t r ia l A - 2 151 x v Figu re 6.9 Schemat ic d iagram showing the paths o f heat transfer f rom mol ten steel to c o o l i n g water i n the m o u l d and the corresponding series resistance analogue : 152 F igure 6.10 A x i a l profi le o f (a) computed values o f thermal resistance o f mould/s t rand gap and s o l i d shel l and (b) percentage contr ibut ion o f each resistance for the 0.12 pet. carbon steel (Heat # 531142) cast dur ing tr ial D - l . (Note: The average metal l eve l is - 1 8 0 m m f r o m top o f mould) 153 F igu re 6.11 A x i a l profi le o f (a) computed values o f thermal resistance o f mould/strand gap and so l id shel l and (b) percentage contr ibut ion o f each resistance for the 0.32 pet. carbon steel (Heat # 531146) cast dur ing t r ia l D - l . (Note : T h e average metal l eve l i s - 1 8 5 m m f r o m top o f mould) 154 F igure 6.12 A x i a l profi le o f (a) computed values o f thermal resistance o f mould/s t rand gap and so l id shel l and (b) percentage contr ibut ion o f each resistance for the 0.84 pet. carbon steel (Heat # 531149) cast dur ing t r ia l D - l . (Note: The average metal l eve l i s - 1 8 2 m m f rom top o f mould) 155 F igu re 6.13 Effect o f so l id shell thickness on (a) heat f lux and (b) m o u l d temperature for a 0.32 pet. carbon steel (Heat # 531146) cast dur ing tr ia l D - l at 550 m m be low the meniscus 156 F igu re 6.14 Effect o f size o f mould/strand gap on (a) heat f lux and (b) m o u l d temperature for a 0.32 pet. carbon steel (Heat # 531146) cast dur ing tr ia l D - l at 550 m m be low the meniscus 157 F igu re 6.15 A c t u a l values o f percent drop i n heat f lux computed for the 46 transverse depressions present on B i l l e t #2 (Heat # 541333) that was cast dur ing tr ial D - 2 158 F igu re 6.16 Frequency distr ibution o f percent drop i n heat f lux computed for the 46 transverse depressions present on B i l l e t #2 (Heat # 541333) that was cast dur ing trial D - 2 159 F igure 6.17 A n example o f thermal response o f the m o u l d ( -25 m m be low the meniscus) that was computed us ing the mathematical m o d e l o f m o u l d heat transfer and heat f lux profi le f rom C o m p a n y C : (a) hot face; (b) mid- thickness and (c) c o l d face 160 F igu re 6.18 M o u l d osc i l la t ion (displacement) profi le that was emp loyed as an input to the mathematical mode l to compute the transient thermal response o f the m o u l d dur ing m o u l d osc i l l a t ion presented i n F igures 6.19 and 6.20 (Tr i a l D - l , Heat # 531146) 161 F igure 6.19 C o m p u t e d transients i n m o u l d temperature at (a) hot face and (b) mid- thickness thermocouple locat ion due to m o u l d osc i l l a t ion presented i n F igure 6.18 (Tr ia l D - l , Heat # 531146) 162 x v i Figu re 6.20 Effect o f m o u l d osc i l la t ion on disturbances generated i n m o u l d temperatures: (a) m a x i m u m minus m i n i m u m and (b) m a x i m u m minus average temperatures at the hot face and a point m i d w a y between the c o l d face and the hot face (Tr ia l D - l , Heat # 531146) 163 F igu re 6.21 Change i n m o u l d midface temperature at (a) mid- thickness thermocouple loca t ion and (b) hot face due to metal l eve l fluctuations i n 0.12 pet. carbon steel (Heat # 531142) cast dur ing tr ia l D - l 164 F igu re 6.22 Change i n m o u l d midface temperature at (a) mid- thickness thermocouple locat ion and (b) hot face due to metal l eve l fluctuations i n 0.32 pet. carbon steel (Heat # 531146) cast dur ing tr ia l D - l 165 F igu re 6.23 Change i n m o u l d midface temperature at (a) mid- thickness thermocouple locat ion and (b) hot face due to metal l eve l fluctuations i n 0.84 pet. carbon steel (Heat # 531149) cast dur ing tr ia l D - l 166 F igu re 6.24 Sec t ion o f the heat f lux profi le at two locations (a) 270 m m and (b) 465 m m be low the meniscus that was employed as input to the mode l to compute the m o u l d thermal response dur ing the cast ing o f B i l l e t #2 (Tr i a l D - 2 , Heat # 541333) 167 F igu re 6.25 C o m p a r i s o n o f computed and predicted m o u l d w a l l temperatures at two locat ions (a) 270 m m and (b) 465 m m be low the meniscus for B i l l e t #2 (Tr i a l D - 2 , Heat # 541333) 168 F igu re 6.26 A compar i son o f the frequency distributions o f (a) measured and (b) computed temperature drops at four locations 170 m m , 270 m m , 370 m m and 465 m m be low the meniscus: for B i l l e t #2 (Tr i a l D - 2 , Heat #541333) 169 F igu re 6.27 Rela t ionship between wid th o f transverse depressions and the corresponding span o f "val ley" observed i n the computed m o u l d temperature at a locat ion 270 m m be low the meniscus for B i l l e t #2 (Tr i a l D - 2 , Heat # 541333, Average casting speed is 19.9 mm/s) 170 F igu re 6.28 Schemat ic d iagram i l lustrat ing the various components o f a "va l ley" i n (a) heat f lux and (b) m o u l d temperature profi les (span, spacing, depth, t l andt2) 171 F igu re 6.29 Rela t ionship between w i d t h o f transverse depressions and the corresponding value o f t l for "val leys" observed i n the computed m o u l d temperature at a locat ion 270 m m be low the meniscus for B i l l e t #2 (Tr i a l D - 2 , Heat # 541333, Average casting speed is 19.9 mm/s) 172 F igu re 6.30 Rela t ionship between depth o f transverse depressions and the corresponding temperature drop observed i n the computed m o u l d temperature (a) 270 m m and (b) 465 m m be low the meniscus for B i l l e t #2 (Tr ia l D - 2 , Heat # 541333) 173 x v i i Figu re 6.31 Effect o f span o f "val leys" i n heat f lux profi le on the magnitude o f temperature drops at a locat ion 270 m m be low the meniscus at (a) hot face and (b) mid-thickness thermocouple loca t ion for the cases examined for 0.32 pet. carbon steel (T r i a l D - 2 , Heat #541333) 174 F igure 6.32 Effect o f span o f "val leys" i n heat f lux profi le on the magnitude o f temperature drops at a locat ion 465 m m be low the meniscus at (a) hot face and (b) mid-thickness thermocouple locat ion for the cases examined for 0.32 pet. carbon steel (Tr ia l D - 2 , Heat #541333) 175 F igure 6.33 Effect o f spacing between consecutive "val leys" i n heat f lux prof i le on the magnitude o f temperature drops computed at a loca t ion 270 m m be low the meniscus at (a) hot face and (b) mid- th ickness thermocouple locat ion for four different spans for 0.32 pet. carbon steel (Tr ia l D - 2 , Heat # 541333) 176 F igu re 7.1 T w o metal l eve l sensor signals at C o m p a n y C f rom (a) Heat # D 6 1 3 5 showing smal l fluctuations and (b) Heat # A 2 8 1 9 3 showing large fluctuations hav ing standard deviations o f 1.4 m m and 4.8 m m respectively. (Note: The average metal l eve l is - 1 3 8 m m f rom the top o f the m o u l d for both cases) 183 F igure 7.2 Cas t ing speed signals at C o m p a n y C corresponding to the two metal l eve l signals presented i n F igure 1.1(a) and (b). (Note: The average casting speed is - 3 3 . 5 mm/s for both cases) 184 F igure 7.3 Patterns observed i n metal l eve l signals dur ing tr ial D - l for (a) Heat # 531142 - 0.12 pet. carbon steel and (b) Heat # 531146 - 0.32 pet. carbon steel conta ining boron (Note: The average metal l eve l for F igure 7.3(a) is - 1 8 0 m m f rom top o f m o u l d and - 1 6 5 m m f rom the top o f the m o u l d for F igure 7.3(b)) 185 F igure 7.4 Differences i n meniscus fluctuations between opposite faces determined f rom the response o f m o u l d thermocouples located 7 m m above the meniscus at the midface on the four faces for 0.12 pet. carbon steel (Heat #531142) cast dur ing tr ial D - 1 . (Note: T h e average metal l eve l is - 1 8 2 m m f rom the top o f the mould) 186 F igure 7.5 Differences i n meniscus fluctuations between off-corner and midface sites on the east face determined f rom the response o f m o u l d thermocouples located 22 m m above the meniscus at the midface o n the four faces for 0.12 pet. carbon steel (Heat #531142) cast dur ing tr ia l D - l . (Note: The average metal l eve l i s - 1 8 2 m m f r o m the top o f the mould . ) 187 F igure 7.6 Photograph o f the meniscus region showing (a) turbulent and (b) fa i r ly quiet metal l eve l 188 F igu re 7.7 M o u l d thermocouple response at 7 m m above the meniscus corresponding to the two metal l eve l signals f rom C o m p a n y C presented i n F igure 7.1(a) and (b). (Note: The average metal l eve l is - 1 3 8 m m f rom the top o f the m o u l d for both cases) 189 x v i i i Figure 7.8 Es t ima t ion o f metal l eve l f luctuation ( in m m ) at the midface f rom the m o u l d thermocouple response ~7 m m above the meniscus ( in °C) for (a) thermocouple locat ion ~8 m m f rom hot face and (b) hot face. ... 190 F igure 7.9 Frequency dis tr ibut ion o f (a) rise and (b) fa l l i n metal l eve l observed i n the metal l eve l sensor signals f rom trial C - l presented i n F igu re 7.1(a) and (b) 191 F igu re 7.10 Frequency distr ibution o f temperature change measured dur ing metal l eve l rise and fa l l by a thermocouple located 7 m m above the meniscus for the two thermocouple signals f rom C o m p a n y C presented i n F igure 7.7(a) and (b) ; 192 F igu re 7.11 Frequency dis tr ibut ion o f duration o f metal l eve l rise and fa l l measured by a thermocouple located 7 m m above the meniscus for the two thermocouple signals f rom C o m p a n y C presented i n F igu re 7.7(a) and (b) ::: 193 F igu re 8.1 Frequency dis tr ibut ion o f temperature drops registered by four thermocouples located at 162, 212, 302, and 412 m m b e l o w the meniscus dur ing the casting o f (a) Heat # D 6 1 3 5 - 0.46 pet. carbon steel and (b) Heat # D 6 1 3 1 - 0.84 pet. carbon steel cast dur ing t r ia l C - l . (Note: T h e average metal l eve l is - 1 3 8 m m f rom the top o f the mould) 206 F igu re 8.2 Thermocoup le response above and be low the meniscus dur ing the format ion and travel o f a depression associated w i t h a lap for 1.02 pet. carbon steel i n t r ia l A - l (Heat # E25990) . (Note: T h e average metal l eve l is —114 m m f rom the top o f the mould) 207 F igu re 8.3 Frequency dis tr ibut ion o f temperature drops registered dur ing the detection o f laps by four thermocouples located at 8, 113, 217 and 368 m m be low the meniscus for 1.02 pet. carbon steel i n t r ia l A - l (Heat # E25990) . (Note: The average metal l eve l is - 1 1 4 m m f r o m the top o f the mould) 208 F igu re 8.4 Frequency dis tr ibut ion o f interval o f temperature drops registered dur ing the detection o f laps by four thermocouples located at 8, 113, 217 and 368 m m be low the meniscus for 1.02 pet. carbon steel i n t r ia l A - l (Heat # E25990) . (Note: The average metal l eve l is ~114 m m f r o m the top o f the mould) 209 F igure 8.5 M e t a l l eve l rise and fa l l associated w i t h each lap, measured by a thermocouple located 7 m m above the meniscus for 1.02 pet. carbon steel i n tr ial A - l (Heat # E25990) . (Note: T h e average metal l eve l is - 1 1 4 m m f rom the top o f the mould) 210 F igure 8.6 Frequency distr ibution o f temperature change measured dur ing metal l eve l rise and fa l l by a thermocouple located at 7 m m above the x i x meniscus for 1.02 pet. carbon steel i n tr ial A - l (Heat # E25990) . (Note: T h e average metal l eve l is ~ 114 m m f rom the top o f the mould ) 211 F igu re 8.7 Frequency dis tr ibut ion o f duration o f metal l eve l rise and fa l l measured by a thermocouple located 7 m m above the meniscus. No te : T h e average metal l eve l is 114 m m f rom the top o f the mould) 212 F igure 8.8 Effect o f in i t i a l m o u l d taper at the meniscus on (a) measured peak heat f lux and (b) specific m o u l d heat extraction [1] 213 F igu re 8.9 M o u l d hot face temperature profiles relative to the b o i l i n g range o f o i l for (a) hot m o u l d operation at C o m p a n y C and (b) c o l d m o u l d operation at C o m p a n y E 214 F igu re 8.10 Oi l -enhanced heat transfer measured us ing m o u l d thermocouples located i n the meniscus region at C o m p a n y C 215 F igu re 8.11 Effect o f o i l f low rate on the measured heat extraction i n the m o u l d at C o m p a n y C [5] 215 F igure 8.12 Peak heat f lux plotted as a function o f peak m o u l d hot face temperature showing the zone o f oi l -enhanced heat transfer 216 F igure 8.13 Effect o f higher c o o l i n g water ve loc i ty on the peak hot face temperature o f the m o u l d for the casting o f 0.84 pet. carbon steel at C o m p a n y D 217 F igure 8.14 Effect o f negative strip t ime on the measured depth o f osc i l l a t ion marks [23] 217 F igure 8.15 Proposed mechan i sm o f lap formation due to r ise/fal l o f metal l eve l and s t ick ing 218 F igure 8.16 Proposed mechan i sm o f b leed formation due to r ise/fal l o f metal l eve l and s t ick ing 219 F igure 9.1 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m measured m o u l d temperature values for tr ial D - l at (a) 30 m m and (b) 45 m m be low the meniscus 236 F igure 9.2 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m measured m o u l d temperature values for tr ial D - 2 at (a) 30 m m and (b) 45 m m be low the meniscus 237 F igu re 9.3 R h o m b o i d i t y values plotted against measured temperature variat ions (difference between m a x i m u m and m i n i m u m values) at (a) 30 m m and (b) 45 m m be low the meniscus for t r ia l D - l and D - 2 238 F igu re 9.4 R h o m b o i d i t y values plotted against peak hot face temperature for the different steel grades cast i n the various plant trials 239 x x Figu re 9.5 Rela t ionship between measured temperature and corresponding peak hot face temperature for trials D - 1 and D - 2 , 25 to 50 m m be low the meniscus ( in the v i c in i ty o f peak hot face temperature) 240 F i g u r e 9.6 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m values o f m o u l d hot face temperature for tr ial D - l at (a) 30 m m and (b) 45 m m be low the meniscus 241 F igure 9.7 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m values o f m o u l d hot face temperature for tr ial D - 2 at (a) 30 m m and (b) 45 m m be low the meniscus 242 F igure 9.8 C o m p a r i s o n o f so l id shel l thickness at the m o u l d exit for the three steel grades cast i n tr ial D - l for va ry ing heat f lux condi t ions i n the meniscus region 243 F igure 9.9 C o m p a r i s o n o f computed so l id shel l thickness for the three steel grades (0.12 pet., 0.32 pet. and 0.84 pet. carbon steels) for va ry ing heat f luxes, at a locat ion situated (a) 25 m m and (b) 50 m m be low the meniscus 244 F igu re 10.1 T h e major components o f the "intelligent m o u l d " be ing developed for the continuous casting o f steel bi l lets 248 x x i L I S T O F S Y M B O L S Csf E m p i r i c a l constant that depends on the nature o f surface and f l u i d combina t ion Spec i f ic heat o f m o u l d (J kg" 1 °C"') c pw . Speci f ic heat o f water (J kg" 1 "C" 1) C o o l i n g water channel gap (m) f Osc i l l a t i on frequency (Hz) 8 , Acce le ra t ion due to gravity (m s"2) K Heat transfer coefficient on hot face above meniscus ( k W m" 2 "C" 1 ) hFC Heat transfer coefficient - forced convect ion ( k W m"2 °C"') K Heat transfer coefficient on mould/water interface ( k W m ' 2 "C" 1 ) Latent heat o f vapourizat ion (J kg" 1) K T h e r m a l conduct iv i ty ( W m" 1 K " 1 ) km T h e r m a l conduct iv i ty o f m o u l d ( W m" 1 K " 1 ) K T h e r m a l conduct iv i ty o f so l id shel l ( W m"1 K " 1 ) kw T h e r m a l conduct iv i ty o f water ( W m" 1 K " 1 ) P Water pressure (kPa) 4 Heat F l u x ( k W m 2 ) QFC Heat f lux - forced convect ion ( k W m"2) QFD Heat f lux - fu l ly developed nucleate b o i l i n g ( k W m"2) QTR Heat f lux - transition region ( k W m"2) QFN Heat f lux - incept ion o f b o i l i n g ( k W m"2) T h e r m a l resistance o f mould/strand gap ( m 2 °C k W " 1 ) x x i i RT To ta l thermal resistance encountered to heat f l ow i n m o u l d ( m 2 °C k W " 1 ) S Osc i l l a t i on stroke length (mm) t t ime (s) T Temperature (°C) tn Nega t ive strip t ime (s) T, So l idus Temperature (°C) Tsat Saturation temperature o f water (°C) Tw Temperature o f c o o l i n g water (°C) Vc Cas t ing speed (m s"1) Vw M o u l d c o o l i n g water ve loc i ty (m s"1) x, z Coordinate A x i s 8S. Th ickness o f so l id shel l (mm) 5 W Th ickness o f m o u l d w a l l (mm) p V i s c o s i t y o f f l u id (Ns m" 2) n 3.142 p m Dens i ty o f m o u l d (kg m" 3) pw Dens i ty o f water (kg m" 3) p ; Dens i ty o f l i q u i d (kg m"3) p v Dens i ty o f saturated l i q u i d (kg m"3) a Surface tension o f l iqu id /vapour interface (Nm"1) x x i i i ACKNOWLEDGEMENTS I w o u l d l i k e to express m y sincere gratitude to m y research supervisors D r . K e i t h B r i m a c o m b e , Dr . Ind i r a Samarasekera and Dr . John M e e c h for g i v i n g me an opportunity to w o r k w i t h them and also, for p rov id ing excellent guidance, valuable help and encouragement throughout this work , i n spite o f their hectic schedules. I must admit that m y stay at U B C was indeed in te l lec tual ly s t imula t ing ! A n industr ia l study o f this nature s imp ly cannot be been performed without the co-operat ion and support o f members o f the "Intelligent M o u l d G r o u p " at U B C as w e l l as a l l o f the par t ic ipat ing m i n i - m i l l s . I take this opportunity to sincerely thank them a l l . In particular, I w o u l d l i k e to thank N e i l W a l k e r , V l a d i m i r R a k o c e v i c and Prakash A g a r w a l ( U B C ) , B o b P u g h ( A l t a Steel) , J i m Seekings ( M a n i t o b a R o l l i n g M i l l s ) and D o n Loren to ( A c c u m o l d ) for their help and also for several useful discussions at various stages o f this work . I a m grateful to V e s a T o r o l a for his assistance w i t h the literature rev iew on "tundish pour ing stream and metal l eve l f luctuations" that has been inc luded i n this thesis. I a m greatly indebted to The Un ive r s i t y o f B . C , A l t a Steel , M a n i t o b a R o l l i n g M i l l s , H a t c h Associa tes , A c c u m o l d , C o m d a l e Technologies Inc. and the Natura l Sciences and Eng inee r ing Research C o u n c i l o f Canada for f inancia l support o f this study. T h e graduate student years at U B C were quite exc i t ing thanks to the excel lent company o f f e l low students and several friends that I made i n Vancouve r . I a m also grateful to D r . A m i t Chatterjee, Senior Techn i ca l A d v i s o r , Tata Steel , for his cont inued support and encouragement. F i n a l l y , a w a r m thanks to m y fami ly members for their encouragement and also for support ing a l l the major decisions I have made so far; I owe m y success to them ! x x i v CHAPTER 1- INTRODUCTION T h e bi l le t casting process has been extensively studied for the last two decades and numerous l inkages have been established between quali ty problems and design/operat ion o f the cast ing machine [1]; the knowledge has been generated w i t h the help o f measurements made on operating casters at several m i n i - m i l l s i n N o r t h A m e r i c a , and mathemat ical models o f the thermo-mechanical behaviour o f the m o u l d and the design/operation o f spray c o o l i n g . O v e r the years, knowledge gained f rom these studies has been presented through numerous publ icat ions and short courses. M o r e recently, an expert system [2] has been developed to diagnose qual i ty problems i n b i l le t casting. The d r iv ing force for a l l this effort is the important task o f transferring knowledge to people w o r k i n g on the shop floor. U t i l i z i n g valuable inputs f rom research i n the last two decades, the bi l le t casting process has actually evo lved f rom the days o f "tr ial -and-error" w i t h sub-opt imal design, operation and maintenance, to the present state approaching condi t ions that inc lude regular measurements [3]. In this age o f g loba l compet i t ion , the subject o f knowledge transfer w i l l continue to p lay an important role i n the intel l igent operation o f a process and its successful implementat ion w i l l determine the performance o f a company i n the years to come. Knowledgeab l e personnel can help solve operating and qual i ty problems faced by most m i n i m i l l s p roduc ing cast steel bil lets since most processing problems arise due to a l ack o f proper understanding o f the process. The situation today, however , is m u c h i m p r o v e d as compared to the early days o f the process. B u t even w i t h a knowledgeable w o r k force, an important hurdle is the implementa t ion o f knowledge : what happens i f a workforce that is knowledgeable about the process does not implement the knowledge into the day-to-day operat ion ? C l e a r l y , this 1 human element cannot be ignored dur ing technology transfer. Thus , a need arises for a system that can di rect ly implement knowledge i n the process w i t h m i n i m a l human intervent ion - thus the concept o f an "intell igent m o u l d " [1]. T h e "intell igent mou ld" , w h i c h on-l ine w i l l moni tor and op t imize a b i l le t cast ing operat ion w i t h respect to both qual i ty and tonnage, is an expert-system-based too l that w i l l sense events (or upsets) occur r ing i n the process u t i l i z ing sensors instal led on the cast ing machine . T h e signals f r o m these sensors w i l l be interpreted by the "intelligent m o u l d " and correct ive act ion w o u l d be taken on- l ine , s imi la r to a human expert [1]. In designing such a system, it is important at the beg inn ing to understand the c r i t i ca l events (or upsets) occur r ing dur ing the process that can generate qual i ty problems i n the bi l le t . Techniques to detect these events (or upsets) must be developed and f ina l ly strategies to m i n i m i z e and prevent these events (or upsets) f rom occur r ing should be implemented i n the expert system o f the "intell igent m o u l d " . Thus , moni to r ing and cont ro l o f process upsets are at the heart o f the operation o f the "intell igent m o u l d " . T h e developments i n b i l le t casting over the last two decades pose an interesting situation w i t h respect to process upsets. A s bi l le t casters adopt better, op t imal design and operating condi t ions i nc lud ing regular measurements and proper maintenance, the role o f process upsets is beg inn ing to assume increasing importance relative to caster design and operat ion i n the generation o f quali ty problems. In the past, many adverse effects o f process upsets were masked by poor design, operation and maintenance. O f the c o m m o n process upsets encountered i n b i l le t cast ing (Table 1.1), r andom metal l eve l fluctuations due to open-pour tundish streams appear to be the most significant w i t h a direct impact on numerous bi l le t defects that originate i n the meniscus region. Exper ience w i t h bi l le t casters e m p l o y i n g open-stream pour ing reveals that qual i ty problems frequently occur randomly dur ing operation. It is c o m m o n to encounter periods dur ing the cast ing 2 o f a heat when the severity o f a p rob lem drast ically changes w i t h t ime. P rob lems may appear r andomly on a l l strands ( inboard versus outboard), on a l l faces (surface defects) and at a l l orientations ( rhomboidi ty) w i t h no clear preference. There are other instances when problems appear and disappear dur ing the casting o f a heat, or even dur ing sequence cast ing for no obvious reasons. In other words , there is a loss o f control o f the b i l le t cast ing operation and product qual i ty f rom t ime to t ime without any clear trend emerging. The unpredictable behaviour points to a source or an upset that is random, chaotic and unpredictable. M e t a l l e v e l f luctuations, as w i l l be demonstrated later, inherently possess these characteristics and, hence, have become the focus o f attention o f this study. W i t h open-stream pour ing, random metal l eve l fluctuations are caused by rough tundish streams that entrain gas and create bubbles i n the mol ten poo l . W h e n the entrained bubbles erupt f r o m the l i q u i d p o o l , surface turbulence and waves are generated w h i c h cause the meniscus to fluctuate randomly , creating a state o f ferro-dynamic chaos i n the meniscus region. In moulds lubr icated w i t h o i l , the fluctuating meniscus interacts w i t h the o i l weep ing d o w n the w a l l and influences heat transfer and lubr icat ion condit ions around the m o u l d periphery i n a stochastic manner. T h e m o u l d heat transfer, w h i c h is a strong function o f taper at the meniscus [1,4-5] has a s ignif icant effect on the nature o f meniscus /o i l interaction and also on the generation o f b i l le t defects such as transverse depressions, laps, bleeds and even, rhomboid i ty . Notwi ths tanding the excellent progress made i n the past w i t h respect to instrumentation and measurements, l inks between process upsets, especial ly metal l eve l f luctuations, and defect format ion i n b i l le t casting are not fu l ly understood. In addi t ion, real-t ime moni to r ing o f defects and related events i n b i l le t moulds is currently not conducted. T h i s w o r k w h i c h is a part o f a larger on-go ing project [6] that aims to design the "intell igent m o u l d " , addresses t w o major 3 issues i n the development o f the intell igent system - understanding l inks between process upsets and format ion o f qual i ty problems and designing strategies for on- l ine detection o f detects us ing thermocouples instal led i n the m o u l d . T h i s study has focused on metal l eve l fluctuations and has established the effect o f meniscus /o i l interaction on heat transfer and lubr icat ion and f ina l ly , on the format ion o f defects such as laps and bleeds on the surface o f bi l le ts , and rhomboid i ty . A mechan i sm has been proposed to exp la in the formation o f laps and bleeds on the basis o f disturbances to heat transfer and o i l lubr ica t ion i n the meniscus region caused by metal l eve l fluctuations. T h e impact o f metal l eve l fluctuations on the generation o f rhomboid i ty is also examined . T h e effect o f important operating variables on the formation o f these defects is discussed on the basis o f the proposed mechanisms. T h e f indings o f this study are based on m o u l d temperature measurements made i n s ix plant trials under a w ide range o f operating condi t ions. W i t h these temperature measurements, it was not on ly possible to detect the formation o f defects such as laps and rhomboid i ty but also to determine the cause o f the problems w h i c h , i n most cases, c o u l d be l i n k e d to l o c a l fluctuations i n the meniscus as determined f rom the response o f m o u l d thermocouples located above the metal l eve l . W i t h respect to hot inspect ion o f bi l le ts , strategies for the on- l ine detection o f laps and rhomboid i ty i n bi l lets are presented to illustrate how these defects can be detected on- l ine us ing m o u l d thermocouples. T h e mathematical analysis o f the detection o f transverse depressions us ing m o u l d thermocouples is discussed together w i t h measurements made on cast b i l le ts , to h ighl ight some issues related to transient m o u l d thermal response that are c ruc i a l i n the development o f strategies for detection. F i n a l l y , the design o f the "intell igent m o u l d " [1] for b i l le t cast ing is discussed together w i t h the impl ica t ions o f this study on its development . 4 Table 1.1 Operat ion- and machine-related process upsets i n b i l le t cast ing. No. Operating Parameters Upsets Encountered 1 Metal Level Issues * Metal level fluctuations * Change in nominal level 2 Condition of Tundish Stream * Stream shape - ropey, flaring * Stream misalignment * Sparking 3 Condition of Tundish, Nozzles * Blockage of nozzles * Wear of nozzles * Damaged weirs and dams * Flow problems in tundish 4 Casting speed issues * Pinch-roll problems * Sticking, jamming * Variation of negative strip time 5 Liquid Steel Quality * Composition variations * Temperature variations * Cleanliness 6 Mould Cooling Water * Quality * Pressure, flow, temperature * Boiling in cooling water channel 7 Mould Oscillation * Frequency variation * Variation of negative strip time * Mechanical / electrical problems 8 Mould Lubrication * Blockage of oil-slots, flow problems * Boiling / flashing of oil 9 Start-up issues * Dummy bar problems * Problems with opening nozzles * Stream condition 5 CHAPTER 2- PREVIOUS WORK T h i s chapter summarizes the state-of-the-art informat ion on bi l le t cast ing, spec i f ica l ly issues related to cast b i l le t qual i ty , mathematical analysis o f m o u l d thermo-mechanica l behaviour and in-plant measurements. The importance o f meniscus fluctuations i n the generation o f qual i ty problems is discussed together w i t h a review o f the subject o f tundish pour ing stream and metal l eve l fluctuations - an area examined by numerous groups s tudying oxygen p i c k u p and reoxidat ion o f l i q u i d steel. W i t h respect to the development o f the "intell igent m o u l d " , the var ious types o f sensors and measuring instruments employed i n continuous casting and also, the current state o f ex is t ing on- l ine moni tor ing and control systems, are also discussed. 2.1 Billet Mould and its Thermo-mechanical Behaviour T h e bi l le t m o u l d generally consists o f a copper tube, 10-20 m m thick, straight or curved , and he ld concentr ica l ly inside a steel jacket as shown i n F igure 2 .1 . C o o l i n g water f lows upwards through the gap between the m o u l d tube and the steel l iner; steel spacers or screws are used to set a un i fo rm gap between the m o u l d and the jacket dur ing assembly. A back-pressure is appl ied to ensure that the annulus is a lways fu l l o f water. T h e m o u l d is osci l la ted either mechan ica l ly or hydrau l i ca l ly , w i t h the help o f a c a m arrangement to provide the necessary s t r ipping act ion for the newly fo rming so l id shel l . The m o u l d is lubricated w i t h o i l (or more recently, powders) w h i c h is pumped and distributed through a splitter to channels i n an o i l i n g plate f r o m w h i c h it is transmitted un i fo rmly around the periphery o f the m o u l d tube. In the case o f powders , the lubricant is fed either manual ly by operators or by a mechanica l device , and a submerged entry nozz le is emp loyed to del iver the steel to the m o u l d . 6 T h e thermo-mechanical behaviour o f bi l le t moulds and its effect on b i l le t qual i ty has been thoroughly examined by B r i m a c o m b e , Samarasekera and co-workers i n numerous studies u t i l i z i n g operating data col lec ted f rom plant trials coupled w i t h mathematical models to analyse m o u l d heat transfer [7-9], m o u l d tube distort ion [7-9,10-11], b i l le t shrinkage and taper design [5,12-13]. The effect o f various design and operating variables such as m o u l d taper at meniscus [1,4-5], o i l flow-rate [5], c o o l i n g water ve loc i ty [7-9], w a l l thickness [7-9,14-15], and steel grade [5,13,15], on the nature o f m o u l d heat transfer have been established. B a s e d on the predicted temperature distr ibution i n the m o u l d w a l l , events such as b o i l i n g i n the c o o l i n g water channel [7-9], thermal distortion o f the m o u l d w a l l [7-9,10-11] and generation o f negative taper dur ing the casting process [7-9,10-11], have been explained. T h e thermal dis tor t ion o f the m o u l d w a l l is m a x i m u m just be low the meniscus where the m o u l d is the hottest and changes d y n a m i c a l l y dur ing the process i n response to variations i n m o u l d temperature caused by events such as nucleate b o i l i n g i n the c o o l i n g water channel and metal l eve l fluctuations [7]. Further, o p t i m u m m o u l d tapers were designed for various grades o f steels w i t h different so l id i f ica t ion and shrinkage patterns, us ing predictions f rom a thermo-mechanical mode l o f the m o u l d and a b i l le t shrinkage mode l [5,12-13,16]. The thermo-mechanical behaviour o f the m o u l d is important since it influences the nature o f mould/strand interaction [1,16-17], a key event i n the generation o f qual i ty problems i n the m o u l d . 2.2 Quality Problems, Meniscus and Mould/Strand Interaction Cracks , depressions, rhomboidi ty , laps, bleeds, segregation and pinholes are the major qual i ty problems encountered i n cont inuously cast steel bi l lets [2,17-19]. Breakou t o f the s o l i d shel l is another p rob lem that seriously hampers product ivi ty . In addi t ion to be ing related to operator errors, breakouts are c lose ly l i nked to the generation o f l oca l i zed th in/weak regions i n the s o l i d shel l due to the presence o f depressions, laps, bleeds or cracks [2,18]. Tables 2.1 7 to 2.7 summar ize the salient information on quali ty problems i n b i l le t cast ing [2,18]. A s shown i n the tables, the three major causes o f quali ty problems i n b i l le t casting inc lude adverse mould/s t rand interaction, improper spray coo l i ng , and poor l i q u i d steel qual i ty [2,18]. W i t h respect to mould-related defects such as longi tud ina l depressions and cracks, transverse cracks and depressions, laps, bleeds, rhomboid i ty , off-corner internal cracks and breakouts, adverse mould/strand interaction at the meniscus appears to be the most important var iable [1-2,4,12]. The important factors affecting mould/strand interact ion are l is ted i n Table 2.8 [2,18]. The behaviour o f the meniscus, w h i c h is the site o f in i t i a l so l id i f i ca t ion , affects osc i l l a t ion mark formation and also the generation o f numerous qual i ty problems v i a its influence on the nature o f mould/strand interaction i n the meniscus region [1,4,21-22]. A number o f studies have focused on the meniscus region to understand the genesis o f defects i n continuous casting. Surface cracks, both transverse and longi tudina l , observed at the corner and midface locat ions i n bi l lets originate close to the meniscus [2,18-19]. Defects such as off-corner internal cracks and rhomboid i ty have been l i nked to the depth and uni formi ty o f osc i l l a t ion marks [22-23] w h i c h f o r m at the meniscus. L a p s and bleeds i n h igh carbon steels are also thought to be caused b y adverse mould/strand interaction at the meniscus [2,4]. S ince this study focuses on laps, bleeds and rhomboid i ty , the information i n the literature on these defects is discussed i n greater detai l . In addi t ion, the informat ion on transverse depressions obtained f r o m a previous study [21], has been inc luded as background for a mathematical analysis conducted to develop strategies for the on- l ine detection o f transverse depressions us ing m o u l d thermocouples. 2.2.1 Laps and bleeds L a p s and bleeds observed i n h igh carbon grades are l i n k e d to adverse mould/s t rand interaction associated w i t h a sha l low taper at the meniscus [4]. It is be l i eved that a sha l low taper at meniscus gives rise to h igh m o u l d heat transfer because o f enhanced mould/s t rand 8 interaction ar is ing f rom the steep negative taper acquired by the m o u l d dur ing the operation [4]. Thus , the result ing m o u l d hot face temperature w h i c h is greater than the b o i l i n g range o f lubr ica t ing o i l , causes s t ick ing o f so l id shel l i n the m o u l d [4]. T h e poor lubr ica t ing condi t ions together w i t h thin so l id shell i n the h igh carbon grades due to a l o n g freezing range, cause tearing o f so l id shel l and the formation o f bleeds and laps [4]. The l imi t a t ion o f the above mechan i sm is that it does not dis t inguish between the two defects. It w i l l be shown later that the sub-surface structures associated w i t h laps and bleeds are dis t inct ly different w h i c h suggest that the two defects, al though caused by related events, f o r m differently. Thus , there is a need to examine the two problems i n greater detail to uncover the differences. Further, as w i l l be discussed i n a later section, these two defects f o r m randomly dur ing the process. T h e proposed mechan i sm [4] based on poor design (shal low taper at the meniscus and the accompany ing h igh m o u l d heat transfer), does not take into account the role o f process upsets i n the generation o f the two defects. I f poor design was the m a i n issue, the severity o f the p r o b l e m should not drast ical ly change dur ing heats; thus, it is important to focus o n process upsets to exp la in the randomness i nvo lved . It w i l l be shown later that the r andom upsets dur ing the process, par t icular ly metal l eve l fluctuations, can influence the severity o f the two problems dur ing the casting operation. 2.2.2 Rhomboidity In an earlier study on rhomboid i ty i n bi l lets [7,10-11], asynchronous b o i l i n g i n the c o o l i n g water channel , was identif ied as an important contributor. T h e reg ion o f the m o u l d that is most affected by b o i l i n g is the area close to the meniscus. A mechan i sm based o n non-un i fo rm c o o l i n g around the b i l le t periphery was proposed to exp la in the format ion o f these defects i n moulds operating w i t h water veloci t ies be low 8.0 m/s. It was mathemat ica l ly shown that operating w i t h l o w water ve loc i ty results i n intermittent, asynchronous b o i l i n g o n the four 9 c o l d faces. W h e n b o i l i n g occurs, the four faces o f the b i l le t c o o l at unequal rates w h i c h cause non-un i fo rm shrinkage o f the shel l and rhomboid i ty , since colder faces contract more than the hotter faces. T h e above mechan i sm was successful i n exp la in ing a number o f observations o n the severity o f rhomboid i ty . The benefits o f "soft c o o l i n g " (coo l ing w i t h reduced c o o l i n g water f low) i n reducing the severity o f rhomboidi ty , at least on a short term basis, was also expla ined . W h e n the water ve loc i ty is reduced, b o i l i n g on the four c o l d faces becomes more v igorous and less intermittent such that c o o l i n g around the bi l le t periphery becomes more un i fo rm. T h e increase i n severity o f rhomboidi ty w i t h decreasing section size was expla ined through the effect o f w a l l thickness on c o l d face temperature and b o i l i n g . It was argued that smal ler sections (100-130 m m square) are cast through thin moulds w i t h w a l l thickness i n the range o f - 6 . 0 to - 9 . 0 m m as compared to - 1 2 . 7 m m for larger sections. It is w e l l k n o w n that a thinner m o u l d w i l l experience a higher c o l d face temperature and therefore, there is a greater tendency for b o i l i n g to occur i n the c o o l i n g channel . F i n a l l y , the reduct ion i n the severity o f rhomboid i ty by mach in ing hor izontal serrations on the c o l d face o f the m o u l d was also exp la ined on the basis o f the b o i l i n g mechanism; increased roughness o f the c o l d face promotes sustained b o i l i n g and reduces the b o i l i n g hysteresis that causes intermittent b o i l i n g i n the c o o l i n g channel . Ano the r mechan i sm was proposed to expla in the generation o f rhomboid i ty based o n osc i l l a t ion mark formation and non-uni form heat transfer i n the m o u l d and the sprays [1,22]. The p r o b l e m begins w i t h the formation o f deep and non-uni form osc i l l a t ion marks around the b i l le t periphery. In the v i c in i t y o f a deep osc i l la t ion mark, the rate o f heat r emova l is l o w due to a w i d e mould/s t rand gap. O n the other hand, regions o f the b i l le t hav ing sha l low osc i l l a t ion marks experience higher rates o f heat extraction. Thus , the presence o f non-un i fo rm osc i l l a t ion marks on the b i l le t surface gives rise to markedly different heat extraction rates around the 10 bi l le t periphery w h i c h ul t imately leads to a non-uni form so l id shel l . Thus , the b i l le t ex i t ing the m o u l d , a l though reasonably square, has a non-uni form so l id shel l , as shown i n F igure 2.2 w h i c h is a schematic representation o f this concept. In the sprays, the co lder portions o f the strand, hav ing thicker so l id shel l , tend to c o o l faster than the hotter regions because o f the greater thermal path and effects o f unstable b o i l i n g ; the result is non-un i fo rm shrinkage o f the b i l le t and rhomboid i ty . Th i s mechanism is supported by some observations made on bi l le ts dur ing the casting process: (a) the obtuse-angle corners o f rhombo id bi l lets have been found to have the deepest osc i l l a t ion marks [22]. (b) the bi l lets emerging f rom the m o u l d , when observed through a peep-hole, showed that, o f the two corners i n v i e w , one was c o l d (dark) and the other was hot (bright). Subsequent inspect ion o f bil lets on the c o o l i n g bed indicated that the acute-angle corners i n the b i l le t corresponded to the colder corners whereas the hot corners formed the obtuse angle o f the bi l le t [22]. W h i l e the two mechanisms deal ing w i t h non-uni form osc i l l a t ion marks and asynchronous b o i l i n g i n the c o o l i n g water channel may be v a l i d for the cases o f rhomboid i ty examined , it appears that other events on the hot face o f the m o u l d such as metal l eve l fluctuations are more important as far as the in i t i a l so l id i f ica t ion is concerned; metal l eve l fluctuations disturb the dynamic distort ion o f the m o u l d tube and also interact w i t h lubr ica t ing o i l . Further, w i t h most b i l le t casters n o w e m p l o y i n g thicker moulds and higher c o o l i n g water veloci t ies , metal l eve l fluctuations appear to be far more important than asynchronous b o i l i n g i n the c o o l i n g water channel . M e t a l l eve l fluctuations also affect osc i l l a t ion mark format ion, an event c lose ly 11 l i n k e d to rhomboid i ty [22]. Furthermore, since the severity and orientation o f rhomboid i ty change randomly w i t h t ime, it is possible that the p rob lem has its roots i n metal l eve l f luctuations. Ano the r interesting aspect o f rhomboid i ty is the effect o f m o u l d taper and steel grade (carbon content) on its severity as shown i n F igure 2.3 [25-27]. W i t h respect to the severity o f rhomboid i ty , the graph indicates that parabolic-tapered moulds are the best f o l l o w e d by double-tapered moulds and single-tapered moulds are the worst. Further, steel grades w i t h carbon contents i n the range - 0 . 1 7 to - 0 . 4 5 pet. are worse than other grades. T h i s study suggests that parabol ic moulds are a panacea for rhomboid i ty but does not p rov ide any technical explanat ion for the recommendat ion. A l t h o u g h this study does not discuss the effect o f m o u l d taper at the meniscus, the f indings suggest that there may be a l i n k between the m o u l d taper at the meniscus and the severity o f rhomboidi ty . The parabol ic and double tapered moulds generally have steep tapers (around - 2 . 0 pct./m) i n the meniscus region and lower heat transfer w h i l e the s ingle tapered moulds usual ly have sha l low tapers (-0.8 pct. /m) and h igh heat transfer. W h i l e this f ind ing may be true for the l imi t ed cases examined i n the earl ier study [25], rhomboid i ty o f - 1 0 - 1 5 m m was observed i n one o f the plant trials conducted dur ing the present study on parabol ic m o u l d (Company D ) . Furthermore, it is also not clear w h y the m e d i u m carbon grades should experience more problems as compared to other grades. T h e role o f process upsets, par t icular ly metal l eve l fluctuations, w h i c h can influence the severity o f rhomboid i ty , was not considered i n these studies [25-27]. 2.2.3 Transverse depressions Transverse depressions were examined i n previous studies [9,21,28-30] i n l o w carbon ( -0 .09 pet. carbon) and m e d i u m carbon (0.32 pet. carbon) grades conta in ing boron . T w o types o f depressions, namely "nose-type" and "smooth-type" were c o m m o n l y seen i n b i l le t 12 cast ing w i t h o i l [21]. The major difference between the two types o f depressions was the o i l f l q w rate emp loyed dur ing the casting; the o i l f l ow rate was - 4 0 m l / m i n for the "nose-type" a n d - 2 0 m l / m i n for the "smooth-type" [21]. The distr ibution o f "nose-type" and "smooth-type" depressions o n the four faces for a 0.32 pet. carbon steel grade conta ining boron and t i tan ium suggests that the two types o f depressions occur randomly dur ing the process [21]. In this study on bi l le t casting e m p l o y i n g o i l lubr ica t ion and open-pour ing tundish stream, it was found that a metal l eve l rise almost a lways preceded the format ion o f a transverse depression. A mechan i sm based on the interaction o f the fluctuating meniscus and lubr ica t ing o i l behaviour , was proposed to expla in the formation o f transverse depressions [21]. It is thought that when the metal l eve l rises, it traps the lubr icat ing o i l beneath the meniscus between the so l id i fy ing shel l and the m o u l d w a l l [21]. The sudden release o f vapour f rom the trapped o i l generates sufficient pressure to push the shel l so l id i fy ing at the meniscus and create a depression on the strand [21]. Other inf luencing factors that were ident i f ied as important inc lude steel compos i t ion , casting speed, section size and superheat levels [21]. It was also thought that the w iden ing o f the mould/strand gap due to the presence o f a depression on the strand caused a reduct ion i n heat transfer and further aggravated problems o f b i n d i n g and transverse depression [21]. 2.3 Process Upsets, Metal Level Fluctuations and Quality Problems B a s e d on the studies o f qual i ty problems, recommended values were established for important operating and design parameters, to m i n i m i z e formation o f defects i n b i l le t cast ing and are presented i n Tab le 2.9 [3]. W h i l e the previous studies were quite successful i n l i n k i n g qual i ty problems to sub-opt imal caster design and operation, these d i d not adequately address the generation o f defects such as laps and bleeds i n machines e m p l o y i n g the recommended m o u l d design and operating parameters and also, the randomness i n v o l v e d i n the severity o f 13 qual i ty problems i n b i l le t casting e m p l o y i n g open-stream pour ing . Thus , there are other issues that need to be addressed to m i n i m i z e quali ty problems and the associated randomness, i n addi t ion to the various design and operating parameters. T h e source o f randomness associated w i t h quali ty problems and the cause o f occas ional problems i n casters e m p l o y i n g the recommended design and operating parameters, appear to be process upsets or transient events encountered dur ing the cast ing operation. O f the major process upsets l is ted i n Table 1.1, metal l eve l fluctuations appear to be the most s ignif icant since they direct ly impact on the nature o f mould/strand interaction at the meniscus and the format ion o f osc i l la t ion marks. The randomness highl ights the fact that the process upset i n v o l v e d is itself, unpredictable, chaotic and random. M e t a l l eve l fluctuations, as w i l l be demonstrated later i n the thesis, possess these "random" characteristics and hence, have become the focus o f attention o f this study especial ly for laps, bleeds and rhomboid i ty . A l t h o u g h the role o f the meniscus fluctuations i n the generation o f these defects is expected to be quite signif icant , there is very litt le information i n the literature that highl ights this issue. O n l y a few studies have examined the role o f transient metal l eve l fluctuations i n the generation o f defects i nc lud ing transverse depression i n bi l lets [21], longi tudina l depressions i n b looms [31] as w e l l as st icker breakouts i n slab casting [32-34]. L o n g i t u d i n a l depressions and other surface defects i n b l o o m cast ing were found to be caused by the interaction o f the fluctuating meniscus and the m o u l d f lux r i m at the meniscus [31]. T h e instal lat ion o f a new metal l eve l control system s ignif icant ly reduced the inc idence o f these depressions [31]. St icker breakouts i n slab casting were also thought to be caused by the interaction o f metal l eve l fluctuations occurr ing dur ing transient speed operations, w i t h the m o u l d f lux [33]. It was proposed that when the metal l eve l rises, a notch is created i n the shel l due to the interfacial tension between the m o u l d f lux r i m and the s o l i d shel l [33]. W h e n the 14 m o u l d f lux r i m moves downward , it contacts the so l id shel l at a l oca l i zed reg ion above the notch and sticks to the so l id shel l . D u r i n g the upward movement o f the m o u l d , tensile strains on the so l id shel l cause tearing o f the shel l at the notch [33]. 2.4 Causes of Metal Level Fluctuations in Billet Moulds T h e changing meniscus is affected by events occurr ing above it, i.e. i n the tundish and the tundish stream [35], and also those happening be low it, i.e l ower d o w n i n the m o u l d and machine . The pr imary source o f random metal l eve l fluctuations is gas entrainment by the open-pour tundish stream entering the m o u l d [35]. It w i l l be shown later that metal l eve l fluctuations can also originate f rom the lower part o f the m o u l d due to b i n d i n g o f the strand or due to problems w i t h the wi thdrawal system w h i c h cause j e r k i n g o f the strand. G a s entrainment by open-pour tundish streams has been the subject o f numerous invest igat ions, a l l o f w h i c h evaluated the effect o f entrained oxygen on the ox ida t ion o f a l l o y i n g elements i n steel and the generation o f inclus ions [35-42]. T h e f indings o f these studies and others f rom the C h e m i c a l Engineer ing literature [43-45], w h i c h were based on both phys i ca l and mathematical mode l l i ng o f fa l l ing water streams, are useful i n understanding the causes o f meniscus fluctuations. G o k l u and Lange [37] have proposed that gas entrainment occurs i n a f l u id stream when its ve loc i ty exceeds a c r i t i ca l value. F igures 2.4 (a) to (c) schemat ica l ly show that, as the stream becomes more turbulent, its surface becomes increas ingly w a v y and irregular such that the disturbances generated i n the meniscus cont inuous ly increase i n magnitude [35]. Thus , the turbulence l eve l o f the stream exi t ing the tundish becomes the most important cause o f gas entrainment and metal l eve l fluctuations i n the m o u l d . T h e turbulent condit ions present i n the stream at the exit o f the tundish nozz le as w e l l as nozz le design are important parameters that influence the nature o f the stream issu ing f rom the 15 nozz le . T u n d i s h stream turbulence is most inf luenced by turbulence f rom the ladle stream p lung ing into the tundish and, therefore, by the tundish design - shape, size and internal devices l i k e weirs , dams and pour box a l l o f w h i c h serve to isolate the turbulence o f the ladle stream. The major nozz le design variables are shape o f the nozz le inlet, its diameter and aspect ratio, and surface roughness [35,37]. In addit ion, stream diameter and ve loc i ty are factors that inf luence the break-up length and the vo lume o f air entrained by the stream, and eventual ly i n the m o u l d p o o l [35,37]. Furthermore, since this is a c lass ica l instabi l i ty p rob lem, the g rowth o f disturbances or instabili t ies also depends on t ime; thus, the height o f fa l l o f the stream ( tundish- to-mould height) is also important. 2.5 Sensors and Instrumentation in Continuous Casting 2.5.1 Temperature Temperature-based systems have been employed to op t imize the design and operation o f the continuous casting moulds for a number o f years [1,28,46,32-34] as w e l l as to moni tor metal l eve l i n some companies [34]. T w o types o f thermocouples have been used for m o u l d temperature measurements [28,46]. The first is formed in t r ins ica l ly and consists o f a d i s s imi la r metal such as constantan ( C u - 45 pc t .Ni ) i n contact w i t h the copper m o u l d w a l l whereas the second consists o f two thermocouple wires encased i n a sheath o f stainless steel [28,46]. T h e thermocouples have been instal led i n the various systems i n a number o f ways - i n the ver t ical plane as w e l l as around the m o u l d periphery [33]. E a c h configurat ion o f thermocouples ut i l izes its o w n l o g i c to moni tor the thermal state o f the m o u l d and detect problems [33]. Heat - f lux sensors instal led on the c o l d face o f slab casting moulds have also been e m p l o y e d to moni tor m o u l d thermal response [33]. 16 In b i l le t cast ing, temperatures o f operating moulds have been measured by B r i m a c o m b e , Samarasekera and co-workers [1,28] u t i l i z ing single wi re , T y p e - T , copper-constantan int r ins ic thermocouples. The ax ia l temperature distr ibution i n the m o u l d w a l l was measured at the midface o f one or more faces [28,46]. Heat-f luxes were calculated f rom t ime-averaged response o f thermocouples for a w ide range o f casting condit ions [28,46]. C o o l i n g water temperature was measured us ing two-wire , T y p e - T thermocouples instal led at various locat ions i n the water channel i nc lud ing bu lk inlet, bu lk outlet and outlet corresponding to each face [28,46]. M o s t earlier studies focused on the steady-state thermal response o f the m o u l d to understand its thermo-mechanical behaviour and the or igins o f qual i ty problems i n bi l lets [28,33,46]. W i t h respect to real-t ime moni tor ing o f the m o u l d thermal response, there are not many examples i n the literature. The sticker-breakout detection system i n slab cast ing uniquely makes use o f the real-t ime m o u l d thermal response to moni tor breakout problems on- l ine [32-34]. Some recent studies have demonstrated the potential o f detecting surface defects [21,31,47] and irregularities i n shel l thickness [48] us ing thermocouples embedded i n the m o u l d w a l l . T h e informat ion on exis t ing on-l ine systems for continuous cast ing and detection o f defects us ing m o u l d thermocouples is presented i n a later section. 2.5.2 M o u l d Displacement M o u l d displacement dur ing osc i l la t ion have been measured us ing L i n e a r V a r i a b l e Disp lacement Transducers ( L V D T ) . T h i s sensor has been employed for routine check ing o f the m o u l d osc i l l a t ion characteristics. B a k s h i et al.[28] p laced L V D T s on the m o u l d table to measure the osc i l l a t ion characteristics o f the m o u l d - negative strip t ime and m o u l d lead. M o r e recently, the hor izonta l movements o f the m o u l d were moni tored us ing a L V D T - b a s e d sys tem designed at the Un ive r s i t y o f B r i t i s h C o l u m b i a [49]. 17 2.5.3 Metal Level and Casting Speed M e t a l l eve l i n the m o u l d has been measured by a number o f means [34,50]: eddy-current probe, electromagnetic cassette, radioactive source, thermocouples embedded i n m o u l d w a l l , op t ica l sensors, and magnetic f lux sensors. In b i l le t casting, metal l eve l is t radi t ional ly measured us ing a radioactive source coupled to a sensor (Figure 2.5) w h i l e the cast ing speed s ignal is obtained f rom a tachometer attached to the wi thdrawal ro l l s . In most cases, the metal l eve l is mainta ined at a constant value by adjusting the casting speed; this is necessary because there is no p rov i s ion to control metal f l ow f rom the tundish. In the metal l eve l system using an array o f thermocouples instal led i n the m o u l d w a l l i n the meniscus region, the detection o f metal l eve l is based on the m a x i m u m m o u l d temperature recorded by the configurat ion o f thermocouples [50]. It was reported that the thermocouples must be c lose to the hot face o f the m o u l d to achieve a measurable response t ime; however , this affects w o r k i n g l ife o f the m o u l d [50]. Further, it was reported that the thermocouple-based systems have a s low response t ime [50]. D u e to these reasons, thermocouple-based systems have not been w i d e l y used as compared to the eddy-current sys tem or the radioactive-source-based systems [50]. W i t h respect to metal l eve l fluctuations, loca l variat ions are extremely c r i t i ca l f rom the standpoint o f quali ty problems as was demonstrated i n the study on transverse depressions [21]; it w i l l be shown later that other defects such as laps and bleeds can also be tr iggered by l o c a l metal l eve l fluctuations. Observat ion o f the meniscus i n the m o u l d indicates that metal l eve l can vary randomly on the four faces as w e l l as across a g iven face. W h i l e the radioact ive metal l eve l sensor seems adequate for detecting g loba l changes i n metal l eve l w h i c h is the case for main ta in ing a constant l eve l , it appears inadequate i n detecting l o c a l variat ions i n metal l eve l . It w i l l be shown later that thermocouples located above the meniscus and embedded i n 18 the m o u l d w a l l on the four faces at midface and off-corner locat ions, are extremely sensitive to l o c a l changes i n metal l eve l and therefore, w i l l be more useful i n detecting l o c a l metal l eve l fluctuations. 2.5.4 Mould/Strand Friction Mould / s t r and fr ic t ion has been moni tored i n continuous casting moulds w i t h the help o f accelerometers or load cel ls attached direct ly to the m o u l d and also by strain gauges instal led on the osci l la tor shaft to study the lubr icat ion mechanisms i n slab cast ing [33]. Acce lerometers have also been used to assess the osci l la tor condit ions and subsequently develop predict ive maintenance schedules for osc i l la t ion systems [33]. L o a d cel ls have also been used by researchers to measure mould/strand fr ic t ion i n continuous casting moulds . B r e n d z y et al.[29] measured m o u l d fr ic t ion w i t h the help o f load cel ls instal led between the m o u l d hous ing and the osci l la tor table; the measured m a x i m u m load was related to o i l lubr ica t ion issues (o i l type and f l o w rate) and the variations i n the m a x i m u m load i n l o w carbon steels to b i n d i n g o f the strand i n the m o u l d . 2.5.5 Other Variables T h e internal d imensions o f instrumented m o u l d tubes have been measured us ing a L V D T - b a s e d profi lometer [51-55]; the condi t ion o f the l i q u i d metal stream and its pos i t ion w i t h respect to the cross-section o f the m o u l d have been observed dur ing the process us ing v ideo camera; c o o l i n g water f l ow rates have been measured us ing a pitot-tube assembly [56], and o i l flow-rates have been measured manual ly . The bi l lets co l lec ted dur ing the plant trials were inspected for surface defects, internal cracks, segregation and rhomboid i ty . Surface profi le o f bi l lets were also measured w i t h a L V D T - b a s e d profi lometer [57]. 19 2.6 Existing On-line Monitoring and Control Systems T h e exis t ing real-t ime moni tor ing and control systems reported i n the literature are designed for slab casting and are based on mould/strand fr ic t ion and/or thermal response o f moulds . In addi t ion to the c o m m o n l y used automatic m o u l d l eve l contro l , on- l ine systems have also been emp loyed to detect and control st icker breakout [33], to detect slag entering the tundish f rom the ladle [33], to detect surface c rack ing [47], to cont inuously moni tor tundish temperature [50], to moni tor and control tundish l eve l [50], to regulate secondary c o o l i n g water dis t r ibut ion and intensity [50,58], to moni tor and regulate m o u l d powder feeding [50] and to automate start-up o f a casting machine [50]. The M . L . T E K T O R system w h i c h is based on mould/s t rand fr ic t ion, is probably the first sys tem designed for breakout detection i n continuous casting o f slabs [33,59]. D e v e l o p e d by C R M , B e l g i u m , the M . L . T E K T O R system uti l izes accelerometers to measure mould/s t rand f r ic t ion, a parameter related to the generation o f stickers i n the m o u l d [33,59]. A l t h o u g h revolut ionary i n concept, the M . L . T E K T O R system d i d not prove very effective since it generated a large number o f false alarms [33,59]. T h e r m a l moni to r ing systems were found to be more effective i n detecting sticker-related breakouts as compared to systems based on mould/strand fr ic t ion. A number o f temperature mon i to r ing system have been developed for detecting breakouts i n slab casting by leading steel companies i nc lud ing K a w a s a k i Steel , N i p p o n Steel, L T V Steel , and B r i t i s h Steel [33]. Systems based on heat f lux sensors instal led on the c o l d face o f the m o u l d and thermocouples embedded i n the m o u l d w a l l have been employed for breakout detection [33]. T h e details o f the var ious types o f the breakout systems were rev iewed by E m l i n g and D a w s o n [33]. T h e appl ica t ion o f these thermal systems has been extended to include detection o f slag entrapment and surface c rack ing , both o f w h i c h can lead to breakouts [33]. 20 A l a r m s f rom the breakout detection systems are usual ly handled by operators and invo lve manual control . O n numerous occasions, t ime delays associated w i t h manual cont ro l have resulted i n breakouts despite the successful detection o f the p rob lem. T o a v o i d this delay, efforts are n o w directed toward designing automatic control systems. Some companies have designed and are us ing systems for the automatic control o f casting speed and metal l eve l f o l l o w i n g the detection o f st icker breakouts i n slab casting [33]. A n addi t ional d imens ion i n process cont ro l is the development o f "intelligent" systems [1] where emphasis is also on the empowerment o f the process w i t h the knowledge and understanding o f the operation. W h i l e the literature contains informat ion on exis t ing on-l ine moni tor ing and control systems that have been designed for detecting breakouts especial ly sticker breakouts and other related events i n slab cast ing, there are no examples o f such on-l ine moni tor ing / control systems for b i l le t cast ing. 2.7 On-line Detection of Defects in Continuous Casting Qual i ty control for surface defects is based on inspect ion o f c o l d semis p r imar i l y by v i sua l inspect ion on the as-cast surface, after p roof scarfing or after full-face scarfing. Internal qual i ty is evaluated by cutt ing a transverse or a longi tudinal section f rom a strand and then sulphur pr in t ing or macro-etching the cross-section. The severity o f the defect is graded on the basis o f a numer ica l system. W i t h the dr ive to increase the amount o f cont inuously cast semis for direct charging or for direct r o l l i ng , a w ide range o f systems based on opt ica l , ul trasonic, electro-magnetic and thermal processes, have been developed and instal led on product ion plants for the on- l ine hot inspect ion o f as-cast products [50]. The requirements o f these sys tem are quite demanding and include detection o f f laws o f va ry ing sizes at h igh temperatures ( > ~ 9 0 0 ° C ) and at h igh strand speeds, abi l i ty to detect and mark defects, and indicate defect type. Further, the strand must be descaled before the detection process. The comprehensive inspect ion o f a l l types o f defects is quite a formidable task and requires a range o f complex systems w i t h 21 sophist icated software to analyse the sensor data on- l ine . A n d even w i t h the most sophist icated system, defects such as slag spots and pinholes are hard to detect on- l ine [50]. Furthermore, on- l ine sensing and evaluat ion o f the internal qual i ty o f cast strand material is next to imposs ib le w i t h systems based on direct detection [50]. D u e to the diff icult ies associated w i t h the direct detection o f defects w i t h the systems ment ioned above, there is an increasing interest i n on- l ine "intell igent systems" that can cont inuously moni tor the thermal / mechanica l response o f a m o u l d and use the informat ion to detect the format ion o f defects i n the m o u l d . The analysis for assessing the cast strand qual i ty is based on the state o f parameters moni tored both off- l ine (steel chemistry, compos i t ion , cast ing condi t ions) and on- l ine (mould thermal and mechanica l response, casting speed, metal level ) . T h e software and log ic employed i n such systems are based on fundamental and heurist ic knowledge about the operation o f the casting process. The development o f the "intell igent m o u l d " at the Un ive r s i t y o f B r i t i s h C o l u m b i a , is the first attempt at creating an on-l ine, "expert-system-based", process cont ro l tool for b i l le t cast ing [1]. A recent study on bi l le t casting demonstrated the potential o f detecting transverse depressions on the strand us ing thermocouples embedded i n the m o u l d w a l l [21]. T h e transverse depressions, w h i c h loca l ly w i d e n the mould/strand gap, are manifested as a drop i n m o u l d temperature as each depression passes by a g iven thermocouple locat ion [21]. T h i s is the on ly study i n b i l le t casting that examines the detection o f defects us ing thermocouples. A study on b l o o m casting, based on s imi la r pr inciples adopted for the detection o f transverse depressions i n b i l le t casting, showed that longi tudina l depressions and other surface defects can be detected using m o u l d thermocouples [31]. Ano the r interesting w o r k i n slab 22 cast ing demonstrated that approximately 80 pet. o f longi tudina l surface cracks i n peritectic grades c o u l d be detected w i t h the help o f m o u l d temperatures and heat fluxes measured f rom a conf igurat ion o f 72 thermocouples embedded i n the m o u l d w a l l [47]. A l t h o u g h the study on transverse depressions [21] i n b i l le t casting is a good start toward des igning on- l ine systems for detection o f defects, a lot o f w o r k is s t i l l required to translate measured m o u l d temperatures into an equivalent "quali ty index" based o n w h i c h bi l le ts can either be hot charged, set aside for c o l d inspect ion or rejected. F o r example , no mathematical analysis has been done to evaluate the w iden ing o f mould/strand gap associated w i t h transverse depressions and the corresponding transient m o u l d thermal response. In the design o f the "intel l igent m o u l d " , it is important to evaluate the severity o f surface qual i ty problems on the basis o f measured temperature drops. Thus , a mathematical analysis is required to l i n k the magnitude and w i d t h o f measured temperature drops to the severity o f transverse depression on the b i l le t surface. O n l y then w i l l it be possible to develop strategies for the on- l ine hot inspect ion o f bi l le ts . T h e other issue relates to the fact that there is no informat ion i n the literature on the detection o f other defects such as laps, bleeds and rhomboid i ty us ing m o u l d thermocouples. 23 Table 2.1 Internal cracks i n bi l le t cast ing [18] QUALITY PROBLEMS LOCATION CAUSES INFLUENCING FACTORS Off-corner Crack Lower part of mould or very close to mould exit Bulging of solid shell and hinging at off-corners Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Steel composition and superheat Diagonal Crack Spray zone Non-uniform shell generated by the mould Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Asynchronous intermittent boiling in the mould; Deep and non-uniform oscillation marks; steel composition and superheat Asymmetric spray cooling Poor spray design and maintenance; Steel composition and superheat Centreline Crack Near the point of complete solidification Sudden decrease in the centreline temperature at the point of complete solidification Inadequate spray cooling near the point of complete solidification; Steel composition and superheat Pinch-roll Crack Close to the pinch rolls Squeezing on a strand with liquid core Excessive pinch roll pressure; Steel composition and superheat Unbending Cracks Close to the point of unbending Unbending on a strand with liquid centre Excessive bending strains; Steel composition and superheat; High casting speed Table 2.2 M i d w a y cracks i n bi l le t casting [18] LOCATIONS CAUSES INFLUENCING FACTORS Mould exit or in the gap between the mould and the sprays Reheating of the billet surface Mismatch between the mould and the sprays: due to design or maintenance problem; Poor design of cooling jacket near the mould exit; Steel composition and superheat Upper portion of the sprays Reheating of the billet surface Poor spray maintenance : bent or plugged spray nozzles; Steel composition and superheat Lower portion of the sprays or the radiation cooling zone Reheating of the billet surface due to the sprays Sprays : design and maintenance issues; Steel composition and superheat Reheating of dark overcooled patches generated by the mould Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Steel composition and superheat 24 Table 2.3 Surface cracks i n b i l le t casting [18] QUALITY PROBLEMS ORIGINS CAUSES INFLUENCING FACTORS Transverse Crack (and depression) In the mould Pulling action on the strand as a result of binding or sticking in the mould; metal level fluctuations Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Steel composition and superheat Longitudinal Corner Crack In the mould Reheating of the billet corner due to a large mould/shell gap and / or Presence of thin and weak shell at the hotter corners Large corner radius; Presence of corner "key holes"; Mould tube alignment; Asynchronous intermittent boiling in the mould; Deep and non-uniform oscillation marks; Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Steel composition Longitudinal Midface Crack In the mould Excessive reheating of a localized portion of the billet surface; Stream impingement on a face Presence of scratch or gouge marks on the inner surface of the mould wall; Misalignment of metal stream Craze Crack In the sprays Grain boundary embrittlement due to the presence of Cu, Sn or low melting impurities in steel. High level of Cu, Sn or low melting impurity in steel Table 2.4 Rhombo id i t y p rob lem i n bi l le t cast ing [18] ORIGINS CAUSES INFLUENCING FACTORS In the mould and/or the sprays Non-uniform shell generated in the mould Thermo-mechanical behaviour of mould; Adverse mould/shell interaction; Asynchronous intermittent boiling in the mould; Deep and non-uniform oscillation marks; Mould-tube alignment; Steel superheat In the sprays Asymmetric spray cooling Poor spray design and maintenance; Steel superheat 25 Table 2.5 Laps , bleeds and z ipper marks i n b i l le t casting [18] QUALITY PROBLEM ORIGINS CAUSES INFLUENCING FACTORS Laps In the mould Tearing of the solid shell close to the meniscus, as a result of sticking or binding in the mould; problem exacerbated by the presence of thin/weak shell. Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Steel composition and superheat Bleeds In the mould Tearing of the solid shell close to the meniscus, as a result of sticking or binding in the mould; problem exacerbated by the presence of thin/weak shell. Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Steel composition and superheat Zipper marks In the mould Tearing of the solid shell close to the meniscus, caused by a bead of steel sticking on the mould wall. Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Steel composition and superheat, Splashing of liquid steel (metal level variation). Table 2.6 Centrel ine segregation and pinholes /b lowholes i n bi l le t cast ing [18] QUALITY PROBLEM LOCATION CAUSES INFLUENCING FACTORS Centreline Segregation Billet centreline Redistribution by mass flow of solutes rejected at the solidification front. Steel composition, Steel superheat, casting speed, section size and shape. Electro-magnetic stirring. Pinholes and Blowholes Surface/ sub-surface of a billet Reduction in the solubility of C, H, O and N during the solidification and cooling; gases come out of solution and form blowholes or pinholes. Levels of carbon, oxygen, nitrogen and hydrogen in steel. 26 Table 2.7 Breakouts i n bi l le t casting [18] ORIGINS CAUSES INFLUENCING FACTORS Close to obtuse-angle corners of the strand Thin shells generated at the obtuse angle corners due to asymmetrical cooling in the mould Thermo-mechanical behaviour of the mould; Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Mould-tube alignment; Steel superheat Close to transverse depressions or deep oscillation marks on the billet surface Local reduction in the shell thickness due to the presence of depressions or deep oscillation marks Adverse mould/shell interaction; Deep and non-uniform oscillation marks; Thermo-mechanical behaviour of the mould Close to a weak spot in the shell Inadequate shell thickness at the mould exit Insufficient dwell time in the mould; Entrapment of slag or scum between mould and billet Mould overflow Operator error Poor metal control/casting speed control Table 2.8 A general outl ine o f mould-related factors contr ibut ing to adverse mould/s t rand interaction [18] REGION IN MOULD INFLUENCING FACTORS At the meniscus Mould Distortion Magnitude of mould taper at meniscus Cooling water velocity / quality Mould tube misalignment Type of mould constraint Mould wall thickness Position of metal level Type of mould copper alloy Mould design tolerances Mould Lubrication Hot face temperature of mould Oil type - properties Oil flowrate and distribution Oil system cleanliness Cleanliness of meniscus region Mould cooling water temperature Mould Oscillation Negative strip time Mould lead Nature of oscillation marks Metal level fluctuations Away from meniscus Mould Taper Type of taper Value of taper in the lower part 27 Table 2.9 Research recommended m o u l d characteristics for b i l le t cast ing machines [3]. No. Design / Operating Parameters Recommended Values 1 Copper Grade DHP, Silver-bearing copper, Chromium-zirconium copper alloy 2 Taper Double or multiple 3 Wall thickness (minimum value in mm) 13 for smaller billets (100 - 150 mm square); 20 for larger billets ( 200 mm square) 4 Inside corner radius (in mm) 3 to 4 5 Meniscus level from top of copper mould (in mm) 100 to 150 6 Water velocity (in m/s) > 10-11 7 Cooling water channel width (in mm) 3 to 5 8 Mould-tube support (constraint type) Four-sided type near top, top and bottom type 9 Negative strip time (in s) 0.12 to 0.15 10 Mould lead (in mm) 3.0 to 4.0 11 Water quality - total hardness Less than 5 ppm, no deposits on cold face 12 Mould-tube type Explosion-formed 13 Measuring internal dimensions of mould before and after use Mandatory requirement 14 Cleaning of oil slots and region of hot face near meniscus Mandatory requirement 28 1 Mould 2 Steel jacket 3 Housing 4 Support plate 5 Lubricator plate 6 Cover plate 7 Water channel Figure 2.1 Schematic d iagram o f a b i l le t casting m o u l d showing its var ious components [ 1 ]. 29 Off-square billet containing off-corner internal cracks Figure 2.2 Schematic d iagram showing a b i l le t w i t h non-uni form so l id shel l thickness be ing distorted into r h o m b o i d shape by spray c o o l i n g [22]. 30 ? 2 0 E v> Sl5 O U) (0 § 10 <1> O O O c 0) 0> Single — Double Parabolic Q o 0.2 0.4 Carbon Content (%) 0.6 0.8 Figure 2.3 Effect o f m o u l d taper on rhomboid i ty (off-squareness) for a range o f steel carbon contents [25]. 31 (a) Smooth Jet Oscillating annulus • t ; . l. o i • \ • • • - \ -—LAscending bubbles —^-Descending bubbles (b) Rippled Jet Surface Plan view of plunge point Surface Bubbles ^ * Vortex Inward Flow o 0 / • o fi ° °\ " — Ascending bubbles ' ^ • Descending bubbles (1-2 mm dia.) (c) Rough Jet Intense surface roughness^^^ "Boils" (emergent bubbles) :'<,..* Ao".^' f '^Closely packed bubbles »V.* „'• " (2 mm dia.) Figu re 2.4 M e c h a n i s m o f gas entrainment proposed by S o m m e r v i l l e and M c K e o g h [35]. 32 Figure 2.5 A schematic d iagram showing the pr inciples o f metal l eve l detector e m p l o y i n g a radioact ive source. 33 C H A P T E R 3- SCOPE AND OBJECTIVES T h e literature on bi l le t casting strongly indicates that the meniscus is the most c r i t i ca l region i n the m o u l d and its behaviour affects the nature o f mould/strand interaction w h i c h is the root o f most qual i ty problems or ig inat ing i n the m o u l d . Thus , a f luctuating meniscus w i l l adversely inf luence the nature o f mould/strand interaction and s ignif icant ly impact on the generation o f qual i ty problems. Surpr i s ing ly , there is l imi t ed informat ion i n the literature that discusses the characteristics o f metal l eve l fluctuations and their influence on disturbances i n heat f l o w and lubr ica t ion and f ina l ly on quali ty problems such as laps and bleeds i n h igh carbon grades and rhomboid i ty ; al though, some recent studies d i d examine this issue i n b i l le t cast ing [21] and b l o o m cast ing [31] . W i t h the emphasis o f continuous casting operations now shift ing to the use o f on- l ine mon i to r ing and control systems, the detection o f defects us ing m o u l d thermocouples is l i k e l y to become an important source o f informat ion for operators. W h i l e it was shown that a transverse depression can be detected on-l ine us ing m o u l d thermocouples, the magnitude o f the temperature changes, or "val leys" , observed i n m o u l d temperature were not l i n k e d to the s ize o f depressions on the strand and the severity o f the problem. B o t h o f these issues are important i n the design o f automatic, on- l ine qual i ty control systems such as the "intell igent m o u l d " [1]. W i t h respect to the on- l ine detection o f other defects such as laps, bleeds and rhomboid i ty i n b i l le t cast ing, the informat ion i n the literature is insufficient. Furthermore, the potential o f on- l ine detection o f l o c a l fluctuations i n metal l eve l us ing m o u l d thermocouples has not been fu l ly explored . T h e above informat ion cannot be generated f rom laboratory experiments since these s imp ly cannot simulate the industr ial condit ions encountered i n the plant operations where the var ious 34 operating parameters are in ter- l inked i n a fa i r ly complex manner. Thus , plant trials were conducted on operating casters and various variables were measured. In each plant t r ia l , a m o u l d instrumented w i t h thermocouples and mechanica l sensors, was emp loyed and data were co l lec ted dur ing the casting operation for subsequent analysis. B i l l e t samples corresponding to the same t ime slot as the sensor data, were col lected and subjected to qual i ty inspect ion. M o u l d thermal data were analysed to evaluate the potential o f us ing m o u l d thermocouples to detect l o c a l metal l eve l fluctuations and bi l le t defects and to develop strategies i n the "intel l igent m o u l d " for the on-l ine detection o f process upsets and defect formation. A n important element i n the analysis o f m o u l d thermal data is the transient thermal response o f the m o u l d ; thus, mathemat ical m o d e l l i n g o f the transient thermal response o f the m o u l d was also conducted. The strategies developed i n this w o r k to detect defects and other mould-rela ted events, were implemented i n a Supervisory C o n t r o l A n d Da ta A c q u i s i t i o n ( S C A D A ) Sys t em i n a para l le l study [60-62] and tested i n the plant. T h e f o l l o w i n g are the major objectives o f this study: [1] T o carry out industr ial instrumented-mould trials at the various companies to measure m o u l d temperature, m o u l d displacement, metal l eve l , casting speed and acquire b i l le t samples, v ideo and photographs o f the operation. [2] T o study the nature o f metal l eve l fluctuations i n b i l le t moulds us ing the tradit ional radioact ive source sensor as w e l l as m o u l d w a l l thermocouples located above the n o m i n a l meniscus posi t ion. [3] T o translate the temperature change recorded dur ing metal l eve l change ( in °C) to actual metal l eve l change ( in m m ) using a transient heat transfer m o d e l o f the m o u l d . 35 [4] T o relate measured m o u l d thermal response to the format ion o f laps and bleeds i n bi l le ts . [5] T o establish a mechan i sm to expla in the formation o f laps and bleeds i n bi l le ts (especial ly, h igh carbon steel grades) us ing informat ion on m o u l d thermal response, results o f b i l le t qual i ty inspect ion and observations made dur ing the plant trials. [6] T o relate measured m o u l d thermal response to the severity o f rhomboid i ty i n bi l le ts . [7] T o exp la in the generation o f rhomboid i ty i n bi l lets on the basis o f va r iab i l i ty i n heat transfer i n the meniscus region, metal l eve l fluctuations, m o u l d hot face temperature (relative to b o i l i n g range o f lubr icat ing o i l ) and steel compos i t ion (carbon content). [8] T o simulate the effect o f surface defects (transverse depressions) present on the strand, on the transient thermal response o f m o u l d us ing a heat transfer mode l . [9] T o identify strategies for the hot surface inspect ion o f bi l le ts for defects us ing m o u l d thermocouples instal led i n the intell igent m o u l d . 36 CHAPTER 4- PLANT TRIALS In this study, s ix instrumented m o u l d trials were conducted on operating casters at four Compan ies : A , C , D , and E . The m a i n tasks i n v o l v e d i n each o f the trials inc luded acquis i t ion o f data on: m o u l d thermal response, mould/strand fr ict ion and displacement o f the m o u l d dur ing osc i l l a t ion , cast ing speed and metal l eve l . In addi t ion, observations were made at var ious stages o f the t r ia l ; extensive f i l m i n g o f the casting operation was done w i t h a v ideo camera and 35 m m photographs were also taken. Samples o f bil lets cast at the trials were co l lec ted and subjected to a detai led inspect ion at the Un ive r s i t y o f B r i t i s h C o l u m b i a ( U B C ) . T h e details o f the plant trials relevant to this w o r k and the equipment employed for data acquis i t ion are descr ibed i n this chapter. 4.1 Instrumentation 4.1.1 Mould Temperature Measurement The technique successfully implemented by B r i m a c o m b e , Samarasekera and co-workers [28,46] i n earlier plant trials, was appl ied to measure m o u l d temperature. T y p e - T (copper-constantan) s ingle-wire , intr insic thermocouples w h i c h can measure temperature i n the range o f -160 to 3 5 0 ° C [67] were employed to measure m o u l d w a l l temperature [28,46]. F igure 4.1 shows the design o f the thermocouples emp loyed to measure m o u l d w a l l temperatures dur ing the plant trials [28,46]. The water temperature at the inlet and the outlet o f the c o o l i n g channel was measured us ing T y p e - T , two-wi re thermocouples [28,46]. T h e temperature o f the c o l d junc t ion w h i c h was located at the data acquis i t ion system, was measured w i t h a mercury thermometer [28,46]. 37 T h e layouts o f thermocouples adopted for the s ix trials are presented i n F igures 4.2 to 4.6 w h i l e the informat ion on the ax ia l locat ion o f the thermocouples and the respective distances o f the thermocouple tips f rom the c o l d face are tabulated i n Tables 4.1 to 4 .5 . S ince the m a i n focus o f this study was on metal l eve l fluctuations and the generation o f defects i n the meniscus region, a higher concentration o f thermocouples was instal led i n this locat ion. Thermocouples were p laced at the midface on two adjacent faces i n trials A - 1 (where, A represents the company and 1 denotes the t r ia l number) , C - l and E - l wh i l e i n the more recent trials, namely A - 2 , D - l and D - 2 , thermocouples were instal led on the four faces at both the off-corner and midface locat ions. T h e off-corner thermocouples located above the meniscus were u t i l i zed to evaluate l o c a l fluctuations i n the metal l eve l across a g iven face. T h e procedure adopted for ins ta l l ing the thermocouple wires i n the m o u l d was quite elaborate [28,46,51-55]. A bead was first created on the constantan w i r e (diameter = - 0 . 8 m m ) using a T I G w e l d i n g machine and f i led to produce a flat end, approximate ly 0.30-0.40 m m thick. T h e constantan wires were then inserted into heat shr inkable tubes o f diameter 1.60 m m w h i c h were subsequently shrunk onto the constantan wires . Before the thermocouple wires were inserted into the m o u l d w a l l , holes were d r i l l ed i n the baffle tube and the copper m o u l d w a l l . A flat bot tom d r i l l was used to flatten the hole i n the m o u l d and the holes were threaded. T h e thermocouple wires were inserted through the baffle into the holes d r i l l ed i n the m o u l d w a l l and were he ld i n place by copper plugs screwed into the threaded holes. S i l i c o n sealant was emp loyed to prevent any leaks through the baffle. T h e constantan wires were connected i n the c o o l i n g water p l enum to shielded copper wires w h i c h were bunched together and connected to a data acquis i t ion system v i a a hole cut i n the m o u l d hous ing . O n c e again, this hole was securely sealed to ensure a water-tight system. The thermocouple conf igura t ion was tested for e lectr ical continui ty before the f inal implementat ion. 38 Afte r the trials, the thermocouple wires were again tested to identify those that fa i led dur ing the trials. It is clear f rom this study, as w e l l as f rom another one on slab cast ing [63-65], that s ingle-wire thermocouples o f diameter - 0 .8 m m cannot be used for on- l ine systems w h i c h obv ious ly require a more robust wi re . F o r the "intelligent m o u l d " , w o r k is proceeding to design and implement thicker, s ingle-wire thermocouples (diameter =-3.2 m m ) that are not on ly robust but are also easy to instal l i n the m o u l d [66]. H o w e v e r , it must be pointed out that the thicker thermocouple wire w i l l have a s lower response as compared to the wires emp loyed i n the study. 4.1.1.1 Converting measured thermocouple voltage into temperature T h e p o l y n o m i a l employed to convert measured thermocouple voltage (x, V o l t s ) to temperature (T, °C) has the f o l l o w i n g f o r m [67]: T = a0 + a{xl + a2x2 + a3x3 + + a n x n (4-1) where, a0,a{, an are coefficients that are unique to the type o f thermocouple be ing used and n is the order o f the p o l y n o m i a l . The coefficients are obtained f rom O m e g a Temperature Measurement and Handbook and E n c y c l o p e d i a [67] w h i c h reports that accuracy for the T-type thermocouple is +/- 0 .5°C , i f a 7th order p o l y n o m i a l is used. T h e values o f the coefficients, aQ, a{, an, for the T y p e - T thermocouple are l is ted i n Table 4.6. A computer p rogram [68] coded i n Fortran 77 was developed by the B i l l e t Cas t ing G r o u p at U B C to transform measured raw voltage into temperature. In addi t ion to conver t ing voltage to temperature, this program includes a number o f useful features that a l l o w easy manipula t ion o f data. F o r example , data i n binary or asci i format can be transformed f rom 39 a higher frequency to the same or any lower frequency; the voltage o f two or more channels can be manipulated; user-defined cal ibrat ion equations can be set up for different data channels; and user-defined output files can be designed i n a number o f ways for later use. 4.1.1.2 Calculation of final temperature T h e thermocouple connections employed dur ing the plant trials are s h o w n i n F igure 4.7 [51-55]. T h e figure shows that a copper/constantan junc t ion is created i n the c o o l i n g water p lenum. T h e e m f ( V 2 ) generated at this junc t ion , opposes the voltage ( V I ) generated at the copper/constantan junc t ion i n the m o u l d w a l l . Thus , the measured voltage is actually V I minus V 2 . The value o f V 2 is the same as the voltage recorded b y the two-wi re thermocouple for the water temperature. A flange separates the inlet and outlet water f lows as shown i n F igure 4.7. The upper chamber contains the outlet water and therefore, the value o f V 2 is set equal to the voltage recorded for the outlet water temperature for a l l junct ions located i n the upper chamber, and equal to the voltage recorded for the inlet water temperature for a l l junct ions located i n the lower chamber. The voltage corresponding to the reference temperature (measured at the data acquis i t ion computer by a mercury thermometer), was added to V I to obtain the f inal voltage corresponding to the actual temperature o f the m o u l d w a l l . The f ina l voltage was ul t imately transformed to temperature us ing the p o l y n o m i a l reported for the T y p e - T thermocouple, as discussed earlier. 4.1.2 Other measurements The m o u l d displacement was measured w i t h the help o f L V D T s . T h e cast ing speed s ignal was obtained f rom the wi thdrawal r o l l tachometer w h i l e the metal l eve l s ignal was recorded f rom the metal l eve l controller . B o t h metal l eve l and casting speed signals were cal ibrated before the trials. T h e metal l eve l s ignal was cal ibrated by l o w e r i n g a steel b i l le t to 40 different levels i n the m o u l d and measuring the output voltage at each l eve l . T h e cast ing speed was cal ibrated on site by measuring the voltage corresponding to different specif ied speeds. In t r ia l D - 1 , the speed o f the wi thdrawal rol ls was also measured us ing a hand-held tachometer. In trials A - l , C - l and E - l , the o i l f l ow rate was also cal ibrated by measur ing the v o l u m e o f o i l co l lec ted for a specif ied duration and for various o i l f l ow settings, i n a pan p laced be low the m o u l d . A l t h o u g h not analysed i n this work , the mould/strand interaction was also measured i n trials C - 1 , E - 1 , D - l and D - 2 using load cel ls ; the experimental details as w e l l as the analysis o f mechan ica l sensor signals are being examined i n a separate study [69] and w i l l not be discussed here. In a l l the trials, extensive f i l m i n g o f the cast ing operation, i n c l u d i n g those o f the tundish pour ing stream and the meniscus, was carr ied out w i t h a v ideo camera; and 35 m m photographs were taken at various stages dur ing the trials. The internal d imensions o f the m o u l d were measured us ing a L V D T - b a s e d profi lometer before the trials to evaluate the taper. T h e m o u l d dimensions were measured after the trials to check for any permanent dis tort ion o f the m o u l d tube. 4.1.3 D a t a A c q u i s i t i o n Sys tem Metraby te ' s U n i v e r s a l E x p a n s i o n Interface ( M o d e l N u m b e r E X P - 1 6 ) and D A S - 8 board designed for the L B M - P C , were used to acquire sensor data The technical details o f the data acquis i t ion hardware are c o m p i l e d i n A p p e n d i x A . F o r trials A - l , C - l and E - l , Lab tech N o t e b o o k software vers ion for M S - D O S was instal led on a portable T o s h i b a P C - 2 8 6 and e m p l o y e d for data acquis i t ion w h i l e for the recent trials A - 2 , D - l and D - 2 , the M S - W i n d o w s vers ion o f Lab tech No tebook software was used on a 4 8 6 / 6 6 - D X machine . T h e thermal data obtained i n trials A - 2 , D - l and D - 2 c o u l d be acquired at a higher frequency (50-100 H z ) as compared to trials A - l , C - l and E - l (1 H z ) because o f the ava i lab i l i ty o f superior computer hardware and software. The data were stored i n a binary format o n hard d i sk w i t h frequent 41 backups to f loppy disks or power tapes. The frequency o f data acquis i t ion was an important issue i n the real-t ime moni tor ing o f m o u l d thermal response. H o w e v e r , an examina t ion o f data acquis i t ion frequency dur ing the course o f this study suggested that a frequency o f 5 to 10 H z was adequate to moni tor m o u l d temperature us ing thermocouples embedded i n the m o u l d w a l l . In t r ia l D - 2 , a S C A D A system [60-62] instal led i n para l le l w i t h the exis t ing data acquis i t ion system, was used to moni tor on-l ine metal l eve l fluctuations and to detect the format ion o f defects (transverse depressions) i n the m o u l d . Deta i l s o f data acquis i t ion e m p l o y e d for the S C A D A system are described i n detail elsewhere [60-62]. W i t h this on- l ine system, v i sua l observations especial ly those i n v o l v i n g metal l eve l fluctuations and tundish pour ing stream qual i ty cou ld be correlated direct ly w i t h the sensor s ignal . A l t h o u g h this in format ion was quite useful, a more detailed study is required to fu l ly characterize the roughness o f the tundish stream and result ing metal l eve l fluctuations on the basis o f the metal l eve l sensor s ignal and m o u l d thermal response. 4.2 Details of Plant Trials T h e casting practices employed at the s ix plant trials, as l is ted i n Tab le 4.7, were quite diverse. T h e moulds i n use at trials A - l , A - 2 , C - l , D - l , and D - 2 were square w h i l e that at t r ia l E - l was rectangular. F o r the square moulds , the section size var ied f rom 140 to 208 m m w h i l e the cast ing speed ranged f rom 17 to 35 mm/s . The m o u l d taper at the meniscus was about 3.0-4.0 pc t . /m i n trials D - l and D - 2 , 2.7 pct . /m i n tr ial E - l , 0.9 pc t . /m i n t r ia l A - 2 , 0.8 pc t . /m i n t r ia l A - l and about 0.4 pct . /m i n tr ial C - l . The osc i l l a t ion frequency var ied f r o m 160 c p m to 190 c p m w h i l e the osc i l l a t ion stroke ranged f rom 6.0 to 12.7 m m . T h e negative strip t ime 42 i n the s ix trials ranged f rom 0.12 to 0.25 s w h i l e the m o u l d lead var ied f rom 1.4 to 9.7 m m . T h e c o o l i n g water ve loc i ty var ied between 10.0 and 15.0 m/s. T h e U B C o i l dis t r ibut ion sys tem [70] was ins ta l led on a l l the moulds and shrouding was i n effect dur ing the cast ing operations. A w i d e range o f steel grades were cast i n the s ix plant trials as shown i n Tab le 4.8; the steel compos i t ion for the various heats have been c o m p i l e d i n A p p e n d i x B . In trials C - l and E - l , the carbon content ranged f rom 0.18 pet. to 0.84 pet.; i n t r ia l A - l h igh carbon grades conta in ing up to 1.05 pet. carbon were tested along w i t h other grades; i n trials D - l and D - 2 , the carbon grades moni tored were 0.12 pet., 0.84 pet., and 0.32 pet. w i t h and wi thout boron and i n t r ia l A - 2 , on ly certain h igh carbon grades (0.70 to 0.90 pet. carbon) were moni tored . T h e operating condi t ions that were var ied dur ing the trials inc luded water ve loc i ty , osc i l l a t ion frequency and o i l f l ow rate, Table 4.9 . The o i l type was also var ied i n the trials since the phys i ca l properties o f lubr ica t ing o i l , especial ly its b o i l i n g range, differed for the var ious o i l types, Tab le 4.10. 4.3 Analysis of Mould Sensor Data W i t h respect to the m o u l d thermal data, two types o f analysis were carr ied out. The first analysis treated the steady-state response o f the m o u l d i n w h i c h the average m o u l d heat f lux prof i le , and temperature distr ibution i n the m o u l d w a l l were computed for various operating condi t ions us ing an exis t ing m o u l d heat transfer mode l [5,7-9]. The procedure for ca lcu la t ing these parameters was s imi la r to that adopted i n earlier studies [5,7-9]. A technique discussed i n earlier publ icat ions [5,46], to filter the effect o f metal l eve l fluctuations f rom the m o u l d thermal response, was adopted i n this study. In this method, the "meniscus thermocouple" located just above the nomina l meniscus posi t ion was first identif ied as a reference point . Temperature data for a l l thermocouples were extracted for those t ime periods w h e n the temperature o f the "meniscus thermocouple" was w i t h i n 5 ° C o f its mean value; it is expected 43 that the extracted data corresponded to a "constant" metal l eve l . A n addi t ional fi l ter based on cast ing speed s ignal was appl ied to further refine this process and data were extracted for periods w h e n the cast ing speed was wi th in - 2 mm/s o f the mean value. T h e extracted data were f ina l ly t ime-averaged and used as input to the m o u l d heat transfer mode l . The second type o f analysis focused on the transient response o f the m o u l d to evaluate detection o f metal l eve l fluctuations and also, the generation o f qual i ty problems such as laps, bleeds and rhomboid i ty . The output f rom the data convers ion program [68] was impor ted into the Quat t ro-Pro for W i n d o w s spreadsheet for further analysis. Time-temperature graphs were plotted and v i sua l ly examined to detect obvious trends i n the data. T h e patterns i n the signals were manua l ly evaluated and strategies to detect surface defects were developed based on this analysis . Fast-fourier analysis was also attempted for the transient analysis o f metal l eve l and temperature signals. The characteristic frequencies for these signals were found to be i n the region o f - 0 . 2 to 0.4 H z ; i n compar ison, the frequency o f m o u l d osc i l l a t ion was about - 2 H z . H o w e v e r , due to the l imi t ed scope o f the current study, this analysis was not pursued further i n detai l . A Computa t iona l Intel l igence (CI) software, details o f w h i c h have been discussed elsewhere [60-62], was developed for the S C A D A system embedded i n the "intel l igent m o u l d " , to detect patterns i n m o u l d temperature, on- l ine . The software was tested us ing measurements made on the time-temperature graphs plotted. Af te r successful testing, the software was e m p l o y e d for automatic detection o f patterns i n the temperature data for a w i d e range o f condi t ions . 44 4.4 Billet Quality Evaluation B i l l e t samples, 300 to 400 m m long and corresponding to the m o u l d thermal data, were acquired 2 to 3 times dur ing a heat for quali ty inspect ion. B i l l e t evaluat ion inc luded surface inspect ion, internal inspect ion and d imens iona l checks. T h e details o f the procedures adopted for the inspect ion are described i n detail elsewhere [5,7,46,51-55]. The m a i n inspect ion tasks carr ied out were as fo l lows : * D i m e n s i o n a l checks - difference between two diagonals ( rhomboidi ty) * Internal inspect ion - cast structure, cracks, segregation, porosi ty * Surface inspect ion - depressions, laps, bleeds, osc i l l a t ion marks, cracks * Sub-surface inspect ion - structure adjacent to laps and bleeds F o r this study, the important issues were the inspect ion for laps, bleeds, transverse depressions and rhomboid i ty . The inspect ion procedures adopted at trials A - l , A - 2 , C - l , D - l and E - l were the same. In tr ial D - 2 , since the focus was on the on- l ine detection o f defects i n the m o u l d , the who le b i l le t corresponding to the thermal data was acquired and its entire surface was inspected for defects. In the case o f transverse depressions, the depth, w i d t h and distance between consecutive depressions were measured on the east face o f the b i l le t . These data were e m p l o y e d to validate the mathematical s imulations that were carr ied out to detect transverse depressions us ing m o u l d thermocouples. W i t h respect to rhomboid i ty , the length o f the two diagonals was measured every - 3 0 5 m m (1 foot) a long the b i l le t length. T h i s was necessary because i n the preceding tr ial D - l , the rhomboidi ty changed dur ing the heats and twis t ing o f bi l le ts i n the c o o l i n g beds was evident. 45 4.1 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n T r i a l A - l . T/C Number - Distance from top of mould (mm) - Hole Depth (mm) NORTH EAST SOUTH WEST 01-077-8.60 01-077-9.20 01-077-9.10 01-076-8.80 02-092-8.50 02-091-9.20 02-091-9.05 02-093-8.85 03-108-8.55 03-107-9.10 03-106-8.95 03-106-8.80 04-123-8.60 04-121-9.10 04-121-9.00 04-121-8.95 05-138-8.50 05-137-9.10 05-136-9.00 05-135-8.90 06-152-9.20 06-152-9.20 06-152-9.10 06-150-8.50 07-167-9.20 07-167-9.15 07-167-9.00 07-166-8.80 08-183-9.15 08-182-9.30 08-181-9.10 08-180-8.60 09-198-9.30 09-197-9.20 09-197-9.10 09-196-8.70 10-213-9.20 10-211-9.25 10-211-9.05 10-211-8.60 11-227-9.20 11-227-9.25 11-227-9.10 11-226-8.70 12-242-9.20 12-242-9.30 12-241-8.80 12-241-8.70 13-257-9.20 13-257-9.30 13-256-8.90 13-256-8.80 14-272-9.10 14-271-9.30 14-272-8.95 14-271-8.75 15-287-9.05 15-287-9.30 16-302-9.05 16-302-9.30 17-332-9.10 17-331-9.30 18-362-9.00 18-361-9.20 19-391-9.00 19-393-9.10 20-483-9.10 20-483-9.15 21-533-9.05 21-535-9.15 22-583-9.05 22-584-9.10 23-627-9.05 23-634-9.20 46 Table 4.2 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n T r i a l A - 2 . T/C Number / Distance from top of mould (mm) / Hole Depth (mm) SOUTH NORTH EAST WEST 01-054-12.02 02-096-11.99 01-095-11.33 01-095-11.71 01-095-12.56 03-112-12.15 02-109-11.71 02-109-12.10 02-109-12.28 04-125-11.45 03-124-11.87 03-122-11.93 03-122-12.01 05-140-12.21 04-139-11.99 04-138-11.60 04-139-12.11 06-140-11.82 05-153-11.97 05-138-11.79 05-138-11.97 07-140-12.07 06-155-12.01 06-138-11.97 06-138-12.35 08-156-11.44 07-154-12.13 07-151-11.85 07-156-12.21 09-156-11.63 08-169-11.95 08-153-12.02 08 not working 10-156-11.77 09-169-11.69 09-152-12.30 09-154-12.33 11-168-11.57 10-170-12.40 10-170-11.23 10-168-12.18 12-170-11.81 11-184-12.26 11-184-11.70 11 not working 13-170-12.26 14-186-12.23 15-210-12.42 16-235-13.35 17-260-13.36 18-291-12.74 19-319-12.69 20-321-12.25 21-321-12.92 22-349-12.98 23-391-11.95 24-391-11.73 25-392-13.29 26-421-12.26 27-503-12.94 28-533-11.95 29-533-12.88 30-531-12.72 31-571-12.28 32-571-12.79 33-570-12.90 34-588-12.29 35-601-12.01 36-601-12.32 37-601-13.19 38-615-12.72 39-632-12.44 47 Table 4.3 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n T r i a l C - 1 . T/C Number - Distance from top of mould (mm) - Hole Depth (mm) ICW - Centreline RSW - Centreline 01-067-8.00 01-065-7.75 02-085-7.90 02-080-7.80 03-101-7.70 03-100-7.80 04-116-7.90 04-116-7.80 05-131-7.90 05-130-7.80 06-146-7.80 06-144-7.80 07-162-7.90 07-160-7.70 08-177-7.90 08-174-7.80 09-191-7.90 09-189-7.75 10-216-7.85 10-214-7.75 11-240-7.90 — 12-300-7.90 11-299-7.80 13-320-7.90 12-318-7.85 14-350-7.80 13-349-7.80 15-381-7.80 14-378-7.55 16-411-7.90 15-409-7.65 17-441-7.80 16-438-7.80 18-471-7.90 17-468-7.90 19-501-7.80 18-499-7.65 20-551-7.90 19-551-7.65 21-602-7.80 20-609-7.70 22-652-7.90 21-649-7.75 23-702-8.10 22-699-7.80 24-750-8.10 23-749-7.80 Note: Thermocouples were installed only on two faces of the mould - RSW and ICW; RSW refers to the right straight wall and ICW stands for inner curved wall. 48 Tab le 4.4 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n Tr ia l s D - l and D - 2 . T/C Number - Distance from top of mould (mm) - Hole Depth (mm) EAST NORTH SOUTH WEST 1-095-7.81 2-131-7.79 1-130-7.85 1-130-8.02 1-131-7.94 3-145-7.77 2-145-7.78 2-145-7.80 2-144-8.04 4-160-7.78 3-161-8.95 3-161-7.79 3-158-7.64 5-161-7.74 4-160-7.91 4-160-7.96 4-160-7.95 6-162-7.74 5-162-7.58 5-160-7.96 5-160-7.82 6-175-7.63 6-173-7.63 6-173-7.92 6-175-8.19 8-191-7.69 7-189-8.29 7-191-7.85 7-190-7.73 9-190-7.70 8-190-7.64 8-190-7.95 8-190-8.38 10-191-7.85 9-190-7.69 9-191-8.14 9-190-7.84 11-205-7.77 10-205-8.71 10-205-8.39 10-205-7.88 12-205-7.86 11-205-7.57 11-204-7.72 11-205-8.02 13-206-7.71 12-206-7.40 12-205-8.34 12-206-8.01 14-220-7.76 13-220-7.57 13-220-8.06 13-221-8.29 15-235-7.76 14-234-7.63 14-235-7.79 14-235-8.43 16-290-7.48 17-315-7.66 18-335-7.78 19-385-7.69 20-433-7.68 21-484-7.64 22-535-7.60 23-630-7.62 24-729-7.56 15-730-7.63 15-731-7.60 15-729-7.78 25-729-7.56 16-729-7.33 16-731-7.83 16-730-8.13 17-730-7.48 17-731-8.36 17-730-8.30 49 Tab le 4.5 Dep th and ax ia l posi t ion o f thermocouples used to moni tor m o u l d w a l l temperature i n T r i a l E - l . T / C Number - Distance from top of mould (mm) - Hole Depth (mm) I C W - Centreline R S W - Centreline 01-085-09.90 01-084-10.00 02-099-09.85 02-099-10.15 03-114-10.00 03-115-10.10 04-129-10.00 04-130-10.00 05-144-09.85 05-145-10.00 6A-159-09.90 6A-160-09.80 6B-159-09.65 6B-160-09.90 6C-158-09.90 6C-159-09.80 6D-158-09.80 6D-158-09.90 6E-159-09.75 6E-158-10.25 07-188-09.80 07-190-09.60 08-218-09.70 08-218-09.50 09-248-09.75 09-250-09.35 10-277-09.70 10-277-09.45 11-327-10.10 12-358-09.95 12-358-09.90 13-388-09.85 13-390-09.20 14-418-09.90 14-419-08.80 15-448-09.60 15-448-08.75 16-479-10.20 16-478-09.80 17-509-10.20 17-509-09.65 18-569-10.00 18-569-09.65 19-629-10.15 19-629-09.65 20-690-09.95 20-690-09.85 21-750-09.45 21-750-09.90 Note: Thermocouples were installed only on two faces of the mould - RSW and ICW; RSW refers to the right straight wall and ICW stands for inner curved wall. RSW has only 24 thermocouples; T / C #11 could not be installed. 50 Table 4.6 Coeff icients SLQ to a 7 for T y p e - T thermocouples used for conver t ing voltage into temperature. COEFFICIENTS VALUE a<> + 0.100860910 a. + 25727.94369 a2 767345.8295 a3 + 78025595.81 a4 9247486589 a5 + 6.97688 X 10" a« 2.66192 X 1013 a7 + 3.94078 X 1014 51 Table 4.7 M a j o r design details and operating condit ions i n the plant trials. Operating Parameters Trial A-l Trial A-2 Trial C - l Trial D-l Trial D-2 Trial E-l Machine type Straight Straight Curved Curved Curved Curved Mould Material DHP DHP Cu-Cr-Zr DHP DHP Cu-Cr-Zr Section size (mm X mm) 152 X 152 203 X 203 140 X 140 209 X 209 209 X 209 127 X 178 Mould Length (mm) 734.0 734.0 835.0 812.8 812.8 812.8 Metal Level (mm) 135 152 130 165-185 165-185 115 Mould Taper Single Single Multiple Parabolic Parabolic Double Mould Taper at meniscus (pct/m) .0.8 0.9 0.4 3.0-4.0 3.0-4.0 2.7 Wall Thickness (mm) 16.0 19.8 16.0 15.6 15.6 19.8 Channel Gap (mm) 3.175 3.560 3.175 4.99 4.99 3.175 Corner Radius (mm) 3.175 3.175 3.175 6.35 3.175 3.175 Water velocity (m/s) 14.1 10.4 12.4 10.4 10.4 18.0 Constraint Type 4-sided 4-sided 4-sided 4-sided 4-sided 4-sided Oil Flow (ml/min) 125 26 25,70 & 100 55 55 25,45 & 65 Casting speed (mm/s) 31.3 16.7 31.7 19.1 19.1 27.5 Oscillation Stroke (mm) 12.7 12.7 11.2 7.0 6.0 6.4 Oscillation Frequency (cpm) 190 160 & 100 144 & 96 110 140 170 & 130 Negative-strip time (s) 0.13 0.17 & 0.25 0.16&0.19 0.19 0.15 0.12 & 0.13 Mould Lead (mm) 8.2 9.7 & 8.1 5.4 & 3.1 2.6 2.4 2.3 & 1.4 52 Table 4.8 Grades o f steels examined i n the plant trials (*). TRIALS GRADES EXAMINED A-l 1020, 1035, 1040, 1080, 1090, 1095 A-2 1070, 1080, 1090 C-l 1017, 1020, L325, 1141, 5160, 1045, 1084 D-l 1008, 1035 (with and without Boron), 1080 D-2 1008, 1035 (with Boron), 1080 E-l 1018, 1146, 1541, 5160, 1050, 1080, 1090 Table 4.9 Operat ing parameters that were var ied dur ing the plant trials. TRIALS OPERATING PARAMETER CHANGED A-l Oil flowrate: 50. 100 ml/min Oscillation frequency: 160. 190 cpm A-2 Oscillation frequency: 100. 160 cpm C-l Oil tvpe: Canola. Hear. Mineral. Soybean Oil flowrate: 0. 25. 75. 100 ml/min Oscillation frequency: 96. 144 cpm D-l Water velocity: 6.66. 10.44 m/s Metal level: 165. 180 mm D-2 Water velocity: 10.44. 5.67 m/s E-l Oil tvpe: Canola. Hear. Mineral. Sovbean Oil flowrate: 25. 45. 65 ml/min (*) The steel compositions of the various heats monitored are listed in Appendix B. 53 Table 4.10 B o i l i n g range o f lubr icat ing oi ls used i n the plant trials [5]. Oil Type Boiling Point (°C) Start 20 pet. 50 pet. 90 pet. Canola 205 280 315 335 Hear 215 280 320 335 Mineral-S 170 230 300 330 Mineral-O 205 270 315 335 Soybean 180 275 320 335 51-LN 205 300 335 350 Note: pet. denotes the fraction of oil (in percentage) that vapourizes at a given temperature. 54 M o u l d W a l l C o p p e r P l u g B r a s s Insert O - R i n g B r a s s M a l e Insert T h e r m o c o u p l e W i r e ( C o n s t a n t a n Wi re ) T h e r m o c o u p l e T ip Baf f le C o o l i n g C h a n n e l Figure 4.1 Schematic d iagram o f the thermocouple design adopted at the plant trials. 55 North and East South and West • 1 75 • 1 75 • 2 90 • 2 90 • 3 105 • 3 105 • 4 120 • 4 120 • 5 135 • 5 135 • 6 150 • 6 150 • 7 165 • 7 165 • 8 180 • 8 180 * 9 195 ' 9 195 • 10 210 • 10 210 • 11 225 • 11 225 • 12 240 • 12 240 ' 13 255 • 13 255 • 14 270 • 14 270 • 15 285 • 16 300 • 17 330 • 18 360 • 19 390 • 20 480 • 21 530 • 22 580 • 23 630 F igu re 4.2 L a y o u t o f m o u l d thermocouples employed for tr ial A - l . (Numbers indicate distance f rom top o f m o u l d i n m m ) . 56 South North, East and West • 1 50 • 2 100 • 3 110 • 4 125 5 ' • 6 • 7 140 8* • 9 •10 155 11 • • 12 •13 170 • 14 185 " 15 210 • 16 235 • 17 260 • 18 290 19- • 20 •21 320 . 22 350 23- • 24 •25 390 • 26 420 • 27 500 28- • 29 •30 530 31 • • 32 •33 570 • 34 585 35 ' • 36 •37 600 • 38 615 • 39 630 Figure 4.3 L a y o u t o f m o u l d thermocouples employed for t r ia l A - 2 . (Numbers indicate distance f rom top o f m o u l d i n m m ) . 57 ICW RSW 1 50 1 50 • 2 85 • 2 85 3 100 • 3 100 4 115 • 4 115 5 130 • 5 130 6 145 • 6 145 7 160 • 7 160 • 8 175 • 8 175 • 9 190 • 9 190 • 10 215 • 10 215 • 11 240 • 12 300 • 11 300 • 13 320 • 12 320 • 14 350 • 13 350 • 15 380 • 14 380 • 16 410 • 15 410 • 17 440 • 16 440 • 18 470 • 17 470 • 19 500 • 18 500 • 20 550 • 19 550 • 21 600 • 20 600 • 22 650 • 21 650 • 23 700 • 22 700 • 24 750 23 750 F igure 4.4 L a y o u t o f m o u l d thermocouples employed for t r ia l C - l . (Numbers indicate distance f rom top o f m o u l d i n m m ) . 58 East North, South and West • 1 95 • 2 130 • 1 130 • 3 145 • 2 145 4- • 5 • 6 160 3 • - 4 • 5 160 • 7 175 • 6 175 8- • 9 •10 190 7 • - 8 • 9 190 11" •12 •13 205 10- -11 •12 205 •14 220 •13 220 "15 235 "14 235 •16 290 •17 315 •18 335 •19 385 •20 435 •21 485 •22 535 •23 630 24 • •25 730 15- «16 •17 730 Figure 4.5 L a y o u t o f m o u l d thermocouples employed for trials D - l and D - 2 . (Numbers indicate distance f rom top o f m o u l d i n m m ) . 59 Side A - 1 2 9 mm Side B-181 mm • 1 85 • 1 85 • 2 100 • 2 100 • 3 115 • 3 115 • 4 130 • 4 130 • 5 145 • 5 145 • • • 6 " ' 160 • 6 • 160 • 7 190 • 7 190 • 8 220 • 8 220 • 9 250 • 9 250 • 10 280 • 10 280 • 11 330 • 11 330 ' 12 360 • 12 360 • 13 390 • 13 390 ' 14 420 • 14 420 • 15 450 • 15 450 • 16 480 • 16 480 ' 17 510 • 17 510 • 18 570 • 18 570 • 19 630 • 19 630 • 20 690 • 20 690 • 21 750 • 21 750 F i g u r e 4 . 6 L a y o u t o f m o u l d thermocouples employed for t r ia l E - l . (Numbers indicate distance f rom top o f m o u l d i n mm) . 60 Water out ma Mould Wall Baffle Divider • Thermocouple Tip * Junction In water plenum * 1 Wall Outlet Water T/C #2 Bulk Outlet Water T/C #3 Bulk Inlet Water T/C Constantan wire Copper wire "* Direction of Water Flow Figure 4.7 T h e arrangement o f thermocouples for measuring m o u l d w a l l and c o o l i n g water temperature. 61 C H A P T E R 5 - RESULTS OF PLANT TRIALS T h e pre l iminary results o f the plant trials are presented i n this chapter. These inc lude b i l le t qual i ty evaluat ion, moni to r ing metal l eve l fluctuations and m o u l d thermal response. 5.1 Billet Quality Evaluation B i l l e t samples ( -300 to 400 m m long) col lected dur ing the plant trials were subjected to qual i ty evaluat ion, as described earlier [5,28,46,51-55]. B i l l e t s were inspected for var ious qual i ty problems such as laps, bleeds, rhomboid i ty , and transverse depressions. T h e results o f the qual i ty evaluat ion are discussed i n this section: 5.1.1 Surface Laps and Bleeds A typ ica l surface o f a h igh carbon (1.02 pet. carbon) steel b i l le t f rom C o m p a n y A (trial A - l ) w i t h severe laps is shown i n F igure 5.1. The surface o f a h igh carbon (0.84 pet. carbon) steel b i l le t f rom C o m p a n y C is shown i n F igure 5.2. In both cases, the pattern o f laps is quite i rregular and accompanying osc i l la t ion marks are distorted. In compar i son , the surface o f a b i l le t o f a 1045 grade f rom C o m p a n y C where the laps were not observed, F igu re 5.3, appears quite smooth. A longi tudina l section was cut through the lap shown i n F igure 5.1 and the sub-surface was po l i shed and etched w i t h p ic r ic ac id ; a photograph o f the section is shown i n F igu re 5.4. T w o dist inct so l id i f ica t ion laps (or hooks) are evident. The first lap is c lear ly indica t ive o f part ial freezing o f the meniscus because the dendrites are perpendicular to a whi te l ine demarcat ing the meniscus. T h i s suggests h igh heat extraction i n this region poss ib ly due to the shel l s t i ck ing to the m o u l d . S t i ck ing is also evident f rom the nature o f the osc i l l a t ion marks i n the v i c in i t y o f the laps; the p i tch o f the osc i l la t ion marks l oca l l y decreased o n b i l l e t samples 62 that exhib i ted laps. A depression o f up to 1 m m is evident i n the figure and appears to be related to the over f low o f l i q u i d steel. The importance o f the depression w i t h regard to the detection o f laps us ing m o u l d thermocouples w i l l be h ighl ighted i n Chapter 8. Ano the r surface defect, namely bleeds, that are s imi la r to laps were observed o n the bi l le t samples examined i n this study, al though the frequency and size o f bleeds were m u c h smaller than those o f laps. The photographs o f bi l le t surfaces exh ib i t ing bleeds i n 0.84 pet. carbon steel and 0.57 pet. carbon steel are shown i n Figures 5.5 and 5.6 respect ively. W h e n compared to the laps observed i n F igure 5.1 and 5.2, the bleeds are general ly smal ler i n size and do not extend across the surface o f the bi l le t . Here again, the pattern is quite irregular and the accompanying osc i l l a t ion marks are distorted. S t i c k i n g at the meniscus is evident i n the case o f bleeds as w e l l , since the osc i l la t ion marks associated w i t h them are irregular. T o examine the o r ig in o f bleeds i n b i l le t casting, a longi tudina l section was cut through the b leed shown i n F igure 5.5 and the sub-surface was pol i shed and etched w i t h p i c r i c ac id . A photograph o f the section shown i n F igure 5.7 reveals tearing o f the so l id shel l and in t rus ion o f l i q u i d steel. Further, as i n the case o f laps, the bleeds were also often found to be associated w i t h a sma l l "depression" on the b i l le t surface. W i t h regard to the occurrence o f the two defects, different patterns were observed. F o r example , i n C o m p a n y A , bleeds were rarely seen on h igh carbon bi l le t samples that exhib i ted severe laps; bleeds were most ly seen on bi l lets where the p rob lem o f laps was less severe. T h i s was also evident i n C o m p a n y C where h igh carbon bi l lets w i t h severe laps d i d not exhibi t many bleeds. Here again, bleeds were less frequent as compared to the laps. In C o m p a n y E , where the p r o b l e m o f laps was not as severe as i n C o m p a n y A and C o m p a n y C , laps and bleeds were randomly observed on bil lets . A s mentioned earlier, on a l l o f the bi l le ts samples that exhib i ted the two defects, the osc i l la t ion marks were irregular, suggesting s t i ck ing o f the 63 strand i n the meniscus region. The distorted osc i l la t ion marks associated w i t h both laps and bleeds suggest that the two defects are caused by s imi la r events occur r ing i n the meniscus reg ion and also, that the p rob lem is worse i n the case o f laps. Thus , the two defects were grouped together to evaluate the effect o f variables such as steel carbon content and m o u l d taper i n the meniscus region as w e l l as o i l flow-rate and o i l types. In trials A - l and C - l , laps and bleeds were randomly seen on bi l le ts at h igh and l o w o i l f l o w rates and w i t h different oi l- types, w i t h no clear trends emerging. Fur thermore, laps and bleeds were not observed on any specific corner or face, and the severity and size o f the two defects var ied dur ing the trials. The observed randomness as w e l l as the i r regular pattern o f osc i l l a t ion marks observed on the bil lets w i t h laps and bleeds suggest that metal l eve l fluctuations were i n v o l v e d i n the generation o f these defects. W i t h respect to steel grade, the p rob lem o f laps and bleeds was par t icular ly severe i n h i g h carbon heats, as indicated i n F igure 5.8, w h i c h also shows the freezing range o f the steel grades examined i n t r ia l C - l . Thus , the h igh carbon steels w i t h 0.8 pet. carbon or greater, exhibi t the greatest incidence o f laps and bleeds w h i l e hav ing a so l id i f ica t ion range o f ~ 1 0 0 ° C as compared to ~ 3 0 - 5 0 ° C for steels w i t h 0.45 pet. carbon or lower . It is expected that the h igh carbon grades consequently sol id i fy a weaker shel l that is h igh ly susceptible to the format ion o f laps and bleeds i n the meniscus region. Ano the r interesting f inding w i t h respect to laps and bleeds i n bi l lets is the role o f m o u l d taper at the meniscus. A s shown i n F igure 5.9, the percentage o f h i g h carbon bi l le ts exh ib i t ing laps and bleeds was considerably lower for C o m p a n y E as compared to companies A and C . A s shown i n Tab le 4.7, the taper at the meniscus for tr ial E - l was 2.7 pc t . /m as compared to on ly 0.8 and 0.4 pc t . /m for trials tr ial A - l and C - l respectively. 64 A s was the case i n C o m p a n y E , laps and bleeds were occas iona l ly observed i n h igh carbon bi l le ts at C o m p a n y D where a steep m o u l d taper o f -3 .5 -4 .0 pc t . /m was e m p l o y e d at the meniscus together w i t h a c o o l i n g water ve loc i ty o f 10.5 m/s. It is also interesting to note that the p r o b l e m o f laps and bleeds i n h igh carbon bi l lets at C o m p a n y A (trial A - 2 ) w h i c h uses a m o u l d w i t h a sha l low taper o f 0.8 pct . /m at the meniscus, c o u l d be reduced considerably by l o w e r i n g the osc i l l a t ion frequency f rom 160 to 100 c p m . Thus , i n addi t ion to m o u l d taper, there are other design/operating variables - c o o l i n g water ve loc i ty and osc i l l a t ion frequency -that inf luence the generation o f laps and bleeds i n bi l le ts . T h e adverse effect o f process upsets and maintenance practice on laps and bleeds was observed at Compan ies D and E . D u r i n g the trials at C o m p a n y E , the lubr ica t ing o i l d is t r ibut ion system had serious problems and on a number o f occasions manual o i l feeding was necessary. It is possible that poor lubr icat ion condit ions at C o m p a n y E contr ibuted to the laps and bleeds observed. Ano the r example o f a process upset causing defects i n bi l lets was found i n a recent plant tr ial at C o m p a n y D where laps and bleeds i n h igh carbon grades (normal ly not a serious problem) suddenly started forming dur ing the casting process [71]. A n examina t ion o f the operating condi t ions revealed a b reakdown i n the c o o l i n g water treatment sys tem w h i c h caused excessive scale deposits to fo rm on the c o l d face o f the m o u l d [71]. Thus , c o o l i n g water qual i ty is another important factor for laps and bleeds. 5.1.2 Rhomboidity 5.1.2.1 Rhomboidity measurement on billet samples In trials A - l , C - l , E - l and D - l , 2 to 3 b i l le t samples ( -300 to - 4 0 0 m m long) corresponding to the thermal data, were col lected dur ing every heat and rhomboid i ty (difference between the two diagonals) was measured on the samples. T h e absolute value o f 65 the difference is a measure o f the severity o f the rhomboid i ty p rob lem w h i l e the s ign (posi t ive or negative) associated w i t h the value is indicat ive o f the orientation o f rhomboid i ty . A l t h o u g h the number o f b i l le t samples considered for this analysis is sma l l , the results presented i n this section do highl ight the fact that the severity and orientation o f rhomboid i ty change w i t h t ime dur ing the casting operation, w i t h no clear trend emerging. T h e m i n i m u m and m a x i m u m rhomboid i ty values measured for s ix steel grades cast dur ing t r ia l A - l are presented i n F igure 5.10. It is clear f rom the figure that the m a x i m u m value for most grades is be low - 6 . 0 m m , except for the 0.80 pet. carbon steel where the value is - 8 . 0 m m . The 0.90 and 1.00 pet. carbon bi l le t samples have the lowest rhomboid i ty values whereas the 0.20 pet. and 0.40 pet. carbon grades have values i n the range o f - 4 . 0 to - 6 . 0 m m . T h e frequency dis tr ibut ion o f rhomboid i ty for the s ix grades examined i n t r ia l A - l is shown i n F igure 5.11. It is interesting to note that the absolute rhomboid i ty i n the 0.80 pet. carbon bi l le ts is - 1 . 0 to - 8 . 0 m m ; c lear ly , the severity o f rhomboid i ty p r o b l e m var ied considerably dur ing the trials. The rhomboid i ty values for 0.40 pet. carbon bi l le ts var ied between - 3 . 0 and - 6 . 0 m m w h i l e a l l o f the values for the 0.15 pet. and 0.20 pet. carbon bi l lets were less than - 4 . 0 m m . T h e rhomboid i ty measurements for tr ial C - l are presented i n Figures 5.12 and 5.13. T h e m a x i m u m rhomboid i ty (absolute values) for the 1045 and 1084 are the lowest ( -2 .0 to - 3 . 0 m m ) , whereas it exceeds - 4 . 0 m m for the other grades. T h e 1020 grade and the 5160 grades are the worst w i t h m a x i m u m values exceeding - 8 . 0 m m , w h i l e the 1018 grade has a m a x i m u m rhomboid i ty o f about 6.0 m m . The frequency dis t r ibut ion o f rhomboid i ty values shows that for both the 1020 and 5160 grades, the range o f rhomboid i ty values encountered is large, i.e., - 1 . 0 to - 1 0 . 0 m m . The 1018 grade has values i n the range o f - 1 . 0 to - 6 . 0 m m . T h e 1045 and 1084 bi l lets had the lowest rhomboid i ty values i n the range o f - 0 . 0 to - 3 . 0 66 m m . T h e rhomboid i ty measurements for tr ial E -1 presented i n Tab le 5.1 for both the test and cont ro l strands, indicate that the severity and orientation o f rhomboid i ty not o n l y changes dur ing a sequence but also w i t h i n a single heat. The range o f values encountered is presented i n Figures 5.14 and 5.15 for the test and control strands respect ively. T h e rhomboid i ty values were sma l l (less than or equal to - 4 . 0 m m ) for most grades except the 1050 grade where values up to - 6 . 0 m m were observed. O n the control strand, the situation was s l ight ly different for the 1080 and 1090 where large values o f rhomboid i ty up to - 7 . 0 m m were observed. T h e frequency dis tr ibut ion o f rhomboid i ty for the four grades examined i n t r ia l E - l is shown i n F igu re 5.16. The rhomboid i ty values for most grades are generally smal l , around 1.0 to 4.0 m m ; and on ly a sma l l percentage o f the 1018 and 1050 grades had rhomboid i ty values i n the range o f - 4 . 0 to - 6 . 0 m m . T h e rhomboid i ty measurements for tr ial D - l are presented i n F igu re 5.17. The rhomboid i ty i n the 0.12 pet. carbon steel is almost negl ig ib le w h i l e that i n the 0.84 pet. carbon steel is about 4.0 m m . The rhomboid i ty values for 0.32 pet. carbon steel grades, bo th w i t h and wi thout boron, are quite interesting. The 0.32 pet. grade without boron has the worst rhomboid i ty ; the first b i l le t (B i l l e t #2) that corresponds to the in i t i a l stage o f the heat, hav ing a h igh rhomboid i ty value o f - 1 0 . 0 m m . The severity o f the p rob lem reduces w i t h t ime and the rhomboid i ty value for the second sample is on ly - 4 . 0 m m . T h e two boron grades have very different rhomboid i ty values; bi l lets f rom the first heat conta in ing boron have l o w rhomboid i ty values o f 1.0 to 2.0 m m whereas those f rom the second heat have higher values o f about 4.0 to 6.0 m m . T h e rhomboid i ty was also measured i n an earlier plant t r ia l at C o m p a n y B where a parabol ic m o u l d w i t h a taper o f about 4.9 pct . /m at the meniscus was e m p l o y e d [72]. A s 67 shown i n F igure 5.18, the rhomboid i ty values for a l l the grades except 1040 are sma l l . B i l l e t s o f the 1040 grade have a large range o f rhomboid i ty values between - 2 . 0 and - 8 . 0 m m . A grade-wise b reakdown o f the m i n i m u m , average and m a x i m u m values o f rhomboid i ty is presented i n F igure 5.19. A g a i n , bi l lets o f a l l grades have smal l rhomboid i ty except the 1008 and 1040 grades where the values exceed - 4 . 0 m m . T h e measured values o f rhomboid i ty for the various grades examined i n the plant trials at Compan ie s A , B , C , and E are plotted against m o u l d taper at the meniscus i n F igure 5.20. T h e values o f rhomboid i ty for a taper o f 0.4 pct . /m employed i n t r ia l C - l , fa l l i n a range o f - 0 . 0 to - 1 0 . 0 m m . The values o f rhomboid i ty for tr ial A - l , where the taper was 0.80 pct . /m, also have a large range o f - 0 . 0 to - 8 . 0 m m . M o s t values o f rhomboid i ty i n tr ial E - l where the taper was 2.7 pct . /m, are w i t h i n - 4 . 0 m m . F o r C o m p a n y B , where the taper i n the meniscus region was - 4 . 9 pct . /m, rhomboidi ty values for over 80 pet. o f the bi l lets are w i t h i n - 2 . 0 m m and on ly a few bi l le t samples f rom a 1040 grade have large values o f up to - 7 . 0 m m . T h e above results suggest that the rhomboid i ty is less o f a p r o b l e m i n operations e m p l o y i n g steep m o u l d tapers i n the meniscus region. It is quite clear f rom the measurements made on the 300 to 400 m m long b i l le t samples that the severity and orientation o f rhomboidi ty changes dur ing the course o f a heat and that it does not f o l l o w any clear pattern. Howeve r , it must be pointed out that the results may not be statistically significant since the number o f bi l le t samples evaluated was sma l l i n most cases. Nonetheless , the analysis may point to two important points w i t h regard to the rhomboid i ty p rob lem: (i) m o u l d taper at the meniscus influences the severity o f the rhomboid i ty p rob lem and (ii) the severity and orientation o f rhomboid i ty change w i t h t ime dur ing the casting operation. These issues w i l l be further examined i n Chapter 9. 68 5.1.2.2 Rhomboidity measurement on entire billet length It is clear f rom measurements made i n t r ia l E - l that the severity and orientation o f rhomboid i ty change w i t h t ime i n the heat (and sequence). T h i s was also observed dur ing tr ia l D - l where twis t ing o f the bi l le t on the c o o l i n g bed was evident. S ince rhomboid i ty changes dur ing the heats, the difference between the two bi l le t diagonals was measured every - 3 0 0 m m (1 foot) a long the entire length o f bi l lets (each about 6 m long) dur ing the t r ia l D - 2 . It must be h ighl ighted that this was the first t ime that rhomboid i ty measurements were made a long the entire b i l le t length. S i x bi l lets ( N o s . 2 , 4 , 5, 8 , 1 0 , 1 9 ) f rom a heat o f m e d i u m carbon (0.32 pet. carbon) grade containing boron, f ive bi l lets ( N o s . l , 3, 6, 8, 10) and two bi l le ts (Nos . 11, 13) f rom two consecutive heats o f a sequence o f a h i g h carbon (0.84 pet. carbon) grade, and one bi l le t o f a l o w carbon (0.12 pet. carbon) steel were examined . F o r the m e d i u m carbon grade containing boron , the orientation and magni tude o f rhomboid i ty var ied w i t h t ime as shown i n F igure 5.21 to 5.26 w h i l e the range o f values measured for each bi l le t is summarized i n F igure 5.27. T h e rhomboid i ty for B i l l e t #2 is w i t h i n - 5 . 0 m m for most o f the length except for a meter or so when values as h igh as - 7 . 0 m m are observed. T h e rhomboid i ty fluctuates between —1-6.0 and —5.0 m m for B i l l e t #4, between -+2.0 and —6.0 m m for B i l l e t #5 and between -+3.0 and —5.0 m m for B i l l e t #8. T h e rhomboid i ty for B i l l e t #10 varies w i t h i n - 5 . 0 to - 6 . 0 m m for the first - 4 . 0 meter or so, after w h i c h it suddenly increases to —10.0 m m for about 0.5 meter after w h i c h i t comes back to —5.0 m m . T h e rhomboid i ty for the in i t i a l 1.5 meter o f B i l l e t #19 is around —5.0 m m but changes drast ical ly to about —1-8.0 m m and then to —15.0 m m i n a space o f about - 3 . 0 meters. 69 In the h igh carbon grades, the situation is quite different as shown i n F igures 5.28 to 5.35. T h e orientation o f rhomboid i ty does not change w i t h t ime (posit ive values are observed throughout) and the magnitude is about 2.0 to 7.0 m m for most o f the t ime. T h e rhomboid i ty for the l o w carbon bi l le t is very smal l and magnitude does not exceed ~2.0 m m , F igu re 5.36. T h e above measurements indicate that the orientation and magnitude o f rhomboid i ty change dur ing the course o f a heat. Thus , any conclusions d rawn about the severity o f the rhomboid i ty p rob lem f rom measurements made (as was the case i n the trials A - l , C - l , E - l , D - l and at C o m p a n y B ) on 300-400 m m long bi l le t samples, must be evaluated w i t h extreme caut ion. T h i s point w i l l be examined again i n Chapter 9. 5.1.3 Transverse Depressions Figu re 5.37 shows a photograph o f a transverse depression seen i n a 0.32 pet. carbon steel conta ining boron and t i tanium, that was cast dur ing tr ial D - l (Heat N o : 531147) . T h e informat ion on depth and wid th o f transverse depressions together w i t h distance between consecut ive depressions, is quite important i n the development o f strategies for the "intell igent m o u l d " for on-l ine detection us ing m o u l d thermocouples. A l t h o u g h there is informat ion i n literature on the types o f transverse depression (nose- and smooth-type depressions) [21], there is no informat ion i n the literature on the size and dis tr ibut ion o f depressions present on the b i l le t surface. A detailed analysis o f transverse depressions was therefore conducted. In tr ial D - 2 , transverse depressions on 0.32 pet. carbon steel bi l le ts conta in ing boron and t i tanium, were found to be more severe i n the early part o f the heat as compared to the later part. F i v e bi l le ts (#2, #4, #5, #8, #10) f rom the test strand that corresponded approximately to the first - 3 0 minutes o f a heat were kept aside for detailed inspect ion. Ano the r B i l l e t (#19), for w h i c h the S C A D A system detected no depressions, was col lec ted later i n the heat ( -40-50 70 minutes after the start o f the heat). A total o f 46 depressions were seen on the east face o f B i l l e t #2, 34 on B i l l e t #4, 14 on B i l l e t #5, 19 on B i l l e t #8, and 24 on B i l l e t #10. A s predicted by the S C A D A system, no depression was observed on the east face o f B i l l e t #19. It must be pointed out that the t ime taken by the d u m m y bar to clear the p i n c h rol ls was about 300-400 s; this impl i e s that B i l l e t #2 and part o f B i l l e t #4 were cast dur ing a per iod w h e n the d u m m y bar was s t i l l engaged by the pinch-rol ls . B i l l e t #2 was subjected to a detailed analysis for depth, w i d t h and spacing o f transverse depressions. T h e depth (d) o f each transverse depression was measured together w i t h its w id th , w t and w,, at the midface i n the transverse and longi tudinal directions respect ively. T h e depth o f the depressions is presented i n F igure 5.38 w h i l e a frequency dis t r ibut ion o f the same is shown i n F igure 5.39; it is clear f rom the two figures that the depth o f about 70 pet. o f the depressions is between 0.5 and 2.0 m m . W i t h regard to the w i d t h (w,) o f the depressions i n the long i tud ina l (casting) direct ion, it is clear f rom Figures 5.40 and 5.41 that the w i d t h o f about 70 pet. o f the depressions is roughly between 45 and 75 m m . In compar i son , as evident f rom Figures 5.42 and 5.43, around 75 pet. o f the depressions stretch 100 to 175 m m across the b i l le t w h i c h is approximately 75 pet. o f the bi l le t section size. T h e distance between consecutive depressions was measured for a l l the f ive bi l le ts hav ing depressions. The results are shown i n Figures 5.44, 5.46, 5.48, 5.50 and 5.52 w h i l e the corresponding statistical distributions are presented i n Figures 5.45, 5 .47 ,5 .49 , 5.51 and 5.53. T h e distance between consecutive depressions on B i l l e t #2 l ies between 50 and 200 m m for about 95 pet. o f the cases observed; thus, the depressions occur quite frequently, every 5-10 seconds or so. T h e percentage o f values l y i n g i n the range o f 50 to 200 m m is about 75 for B i l l e t #4, about 50 for B i l l e t #5, about 55 for B i l l e t #8 and about 70 for B i l l e t #10. T h e percentage o f values exceeding 300 m m is 0 for B i l l e t #2, less than about 10 for B i l l e t #4 and 71 about 30 for B i l l e t s #5, #8 and #10. The overa l l frequency dis t r ibut ion for the f ive bi l le ts is presented i n F igure 5.54. It is clear that about 70 pct .of values are between 50 and 200 m m ; this translates to about 5 to 10 s for a casting speed o f - 1 9 mm/s . Thus , these measurements suggest that there is some randomness i n the generation o f depressions dur ing the course o f a s ingle heat. 5.2 Metal Level Fluctuations In this study, signals f rom the metal l eve l control ler were acquired together w i t h casting speed signals. T h e metal l eve l s ignal obtained f rom a heat i n t r ia l C-1 is shown i n F igu re 5.55(a) together w i t h the corresponding casting speed profi le , F igure 5.55(b). T h e temperature recorded by a thermocouple located above the meniscus was also employed to sense metal l eve l fluctuations. A n example o f the response o f a m o u l d thermocouple located about 7 m m above the meniscus, corresponding to the metal l eve l s ignal plotted i n F igure 5.55(a) is shown i n F igu re 5.56. Thermocouples were also used to examine the difference i n the metal l eve l fluctuations on the four faces as w e l l as across a g iven face. T h e character o f metal l eve l fluctuations based on the response o f various sensors employed dur ing the trials is discussed i n Chapter 7. 5.3 Average Mould Temperature Data T o calculate the average m o u l d heat f lux profi le and the temperature dis t r ibut ion i n the m o u l d w a l l , m o u l d temperature data obtained f rom the plant trials were first f i l tered and t ime-averaged as expla ined i n the previous chapter. A n example o f the average measured m o u l d temperature for three grades o f steel examined i n tr ial D - l is presented i n F igu re 5.57. The measured temperature profiles are s imi la r to those reported i n earlier publ icat ions [13,14]. T h e 72 average temperature values were employed as inputs to a mathematical mode l for steady-state heat transfer calculat ions. The descript ion o f the mathematical m o d e l and the results o f mode l predict ions are presented i n Chapter 6. 5.4 Billet Defects and Mould Transient Thermal Response In addi t ion to moni to r ing metal l eve l fluctuations, the transient thermal response o f the m o u l d was also employed to detect bi l le t defects such as laps, bleeds, rhomboid i ty , and transverse depression. Some examples o f m o u l d temperature response i n heats exh ib i t ing these problems are presented i n this section w h i l e a detailed analysis and d iscuss ion o f the format ion and detection o f these defects are presented i n Chapters 8 and 9. 5.4.1 Laps In t r ia l C - l , laps were observed p r imar i ly i n the h igh carbon grades w h i l e bi l le ts f rom 1045 grades d i d not exhibi t this problem. Figures 5.58(a) and (b) compare the m o u l d thermocouple signals f rom a 0.84 pet. carbon steel w i th another conta in ing 0.46 pet. carbon at two locat ions (310 and 620 m m be low the meniscus) respectively i n the m o u l d . It is clear that the var iab i l i ty i n the thermocouple response for the 0.84 pet. carbon steel is quite significant as compared to that o f the 0.46 pet. steel. T h i s suggests that the drops or "val leys" i n m o u l d temperature observed i n the 0.84 pet. carbon steel are associated w i t h laps. T h e nature o f m o u l d thermocouple response w i t h respect to the formation o f laps is discussed i n Chapter 8. 5.4.2 Rhomboidity In tr ial D - 2 , rhomboid i ty o f up to 15.0 m m was measured on B i l l e t #19 for the 0.32 pet. carbon steel containing boron and t i tanium. In compar ison , the rhomboid i ty for B i l l e t #16 for the 0.12 pet. carbon steel was extremely smal l (less than - 2 . 0 m m ) . F igures 5.59 and 5.60 73 compare the average m o u l d temperatures for the two cases at two locat ions (30 m m and 45 m m ) b e l o w the meniscus. The temperatures plotted i n the figures were obtained by averaging the measured temperature every 25 s. It is clear f rom the two figures that the b i l le t w i t h h igh rhomboid i ty is associated w i t h a large difference between the m a x i m u m and m i n i m u m m o u l d temperatures recorded on the four faces. F o r the 0.12 pet. carbon bi l le t , the temperatures are not on ly lower , but the difference between the m a x i m u m and m i n i m u m temperatures is m u c h smaller . It w i l l be shown later i n Chapter 9 that the average difference between the m a x i m u m and m i n i m u m m o u l d temperatures on the four faces is an important cr i te r ion that can be e m p l o y e d i n the knowledge base o f the "intell igent m o u l d " to detect rhomboid i ty i n bi l le ts . 5.4.3 Transverse depressions In t r ia l D - 2 , a higher frequency o f transverse depressions was observed i n the in i t i a l stage o f the heat (B i l l e t #2 had 46 depressions on the east face) and a lower frequency i n the latter part o f the heat (billet #19 had no depressions on the east face). F igures 5.61(a) and (b) compare the response o f two m o u l d thermocouples instal led on the east face at 260 and 560 m m be low the nomina l meniscus l eve l , for the two bil lets . T h e two figures indicate that the presence and absence o f transverse depressions can be detected w i t h m o u l d thermocouples. T h i s observat ion is consistent w i th the f indings o f an earlier study on transverse depression [21]. T h e strategy for the detection o f transverse depressions i n the m o u l d , based on bi l le t evaluat ion, measured m o u l d thermal response, and mode l predict ions is discussed i n Chapter 6 and 10. 74 Tab le 5.1 R h o m b o i d i t y values ( in m m ) measured on bi l le t samples acquired f r o m the test and control strands during tr ial E - l ( * ) . Grade Heat No. Test Strand Control Strand Sample 1 Sample 2 Sample 3 Sample 1 Sample 2 Sample 3 1018 26503 -1 -4 +4 -2 26504 +1 +2 +1 -3 26505 +1 0 0 0 26507 +3 +2 +1 +3 1146 26508 +1 +1 +1 -4 +4 +2 26509 +1 0 +1 +4 +3 2 1050 26539 0 +2 -4 +3 +6 26540 +6 +1 +2 +5 26541 +1 +2 +4 +4 +5 1080 26514 +3 0 +1 +1 -4 +3 26515 +1 +1 0 0 26516 +1 +1 0 +1 +2 -1 26519 +1 +1 -2 +5 -1 +5 26520 +1 -1 +2 +3 +3 -3 1090 26510 0 -2 +2 -1 +2 26511 +1 +1 +1 -3 -4 +4 26512 0 -1 +1 +2 +7 (*) F o r the f ive grades moni tored, b i l le t samples were col lec ted f rom selected heats dur ing a sequence for both the test and the control strand. 75 Figure 5.1 Surface photograph o f a 1.02 pet. carbon steel b i l le t f rom C o m p a n y A showing laps. 76 Figure 5.2 Surface photograph o f a 0.84 pet. carbon steel b i l le t f rom C o m p a n y C showing laps. 77 Figure 5 .3 Surface photograph o f a 0.45 pet. carbon steel b i l le t surface s h o w i n g a smooth surface i n the absence o f laps. 78 79 80 Figure 5.6 Photograph o f a 0.57 pet. carbon steel bi l le t surface showing bleeds. 81 Casting Direction Figure 5.7 Photograph o f the subsurface structure o f the bleed shown i n F igure 5.5. 82 Figure 5.8 Effect o f steel carbon content on the severity o f laps and bleeds observed i n the plant tr ial at C o m p a n y C . Freez ing range o f the steels is also shown. F igure 5.9 Effect o f m o u l d taper at the meniscus on the severity o f laps and bleeds observed i n the trials at Companies A , C and E -83 12 Steel Grades Figure 5.10 M i n i m u m and m a x i m u m values o f rhomboid i ty measured for the var ious grades examined dur ing trial A - l . 100 • 0.15 C g 0.20 C £ 3 0.40 C BJJ 0.80 C g § 0.90 C • 1.00 C 0-2 2-4 4-6 Difference between Diagonals (mm) Figure 5.11 Frequency dis tr ibut ion o f rhomboid i ty values measured for the var ious grades examined dur ing tr ial A - 1 . 84 12 ? E 1018 1020 1045 5160 1084 Steel Grades Figu re 5.12 M a x i m u m values o f rhomboidi ty measured for the various grades examined dur ing tr ial C - l . 100 >. u g 50 3 cr 0) 25 7\ / / / / 2-4 4-6 6-8 8-10 Difference between Diagonals (mm) Figure 5.13 Frequency distr ibution o f rhomboid i ty values measured for the var ious grades examined dur ing tr ial C - l . 85 10 + 6 mm - +3 mm +1 mm I + 3 i mm + 1 mm 0 mm 0 mm 1 -1 mm ' - 2 • mm -10 4 mm + -+-1018 1146 1050 1080 Steel Grades 1090 Figure 5.14 Range o f rhomboid i ty values measured for the various grades cast on test strand dur ing t r ia l E - l . 10 ? E <n I  5 o D ) CJ § 0 QJ 0J g - 5 QJ 3= -10 +4 mm + 4 • i + 6 mm i i mm + 5 + 7 mm i mm i -3 • mm • -4 mm - 4 . . . . • i mm - 4 i • • mm - A i 1 mm t 1018 1146 1050 Steel Grades 1080 1090 Figure 5.15 Range o f rhomboid i ty values measured for the various grades cast on control strand dur ing tr ial E - l . 86 100 • 1018 S 1146 ^ 1050 H11080 • 1090 0 0-2 2-4 4-6 6-8 Difference between Diagonals (mm) Figure 5.16 Frequency distr ibution o f rhomboid i ty values measured for the var ious grades examined dur ing t r ia l E - l . 87 0.12 C 0.32C 0.32C(B) 0.32C (B) 0.84 C Steel Grades Figure 5.17 R h o m b o i d i t y values measured for the various grades examined dur ing t r ia l D - 1 . 88 100 0-2 2-4 4-6 6-7 Difference between Diagonals (mm) Figure 5.18 R h o m b o i d i t y values measured for the various grades examined dur ing tr ial at C o m p a n y B . 1008 1010 1012 1015 Steel Grade 1018 1040 Figure 5.19 M i n i m u m , average and m a x i m u m values o f rhomboid i ty measured for the var ious grades examined dur ing tr ial at C o m p a n y B . 89 Difference between Diagonals (mm) Figure 5.20 Effect o f m o u l d taper at the meniscus on the severity o f rhomboid i ty observed i n trials at Compan ies A , B , C and E . 90 (a) (b) F igure 5.21 R h o m b o i d i t y measurement a long the length o f B i l l e t #2 for a m e d i u m carbon steel conta ining boron and t i tanium i n t r ia l D - 2 (Heat # 541333): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurements is 19). 91 (a) F igure 5.22 R h o m b o i d i t y measurement a long the length o f B i l l e t #4 for a m e d i u m carbon steel conta ining boron and t i tanium i n t r ia l D - 2 (Heat #541333) : (a) measured values and (b) frequency distr ibution. (Note: T o t a l number o f measurements is 19). 92 (a) § -10 0) 3= 5 -15 -I 1 1 1 1 1 1 1 1 0 1 2 3 4 Distance along Billet Length (m) (b) F igure 5.23 R h o m b o i d i t y measurement a long the length o f B i l l e t #5 for a m e d i u m carbon steel conta ining boron and t i tanium i n t r ia l D - 2 (Heat #541333) : (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurements is 19). 93 (a) (b) F igure 5.24 R h o m b o i d i t y measurement a long the length o f B i l l e t #8 for a m e d i u m carbon steel conta ining boron and t i tanium i n t r ia l D - 2 (Heat # 541333): (a) measured values and (b) frequency distr ibution. (Note: To ta l number o f measurements is 19). 94 (b) F igure 5.25 R h o m b o i d i t y measurement a long the length o f B i l l e t #10 for a m e d i u m carbon steel conta ining boron and t i tanium i n tr ial D - 2 (Heat # 541333): (a) measured values and (b) frequency distr ibut ion. (Note: T o t a l number o f measurements is 19). 95 (a) (b) 0-2 2-4 4-6 6-8 8-10 10-12 12-14 14-16 Difference between Diagonals (mm) Figure 5.26 R h o m b o i d i t y measurement a long the length o f B i l l e t #19 for a m e d i u m carbon steel conta ining boron and t i tanium i n trial D - 2 (Heat # 541333): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurements is 16). 96 15 £ — 10 w re c O) 5 re § 0 J2 -5 V O c 0) 3= + 1 mm + 6 mm + 7 mm „ +3 mm + 2 mm + 2 mm -15 • 7 mm - 5 mm • 6 mm - 5 mm •• 10 mm -15 mm #1 #4 #5 #8 Billet # #10 #19 Figure 5.27 Range o f rhomboid i ty values measured along the length o f B i l l e t s #2, #4, #5, #8, #10 and #19 for a m e d i u m carbon steel conta ining boron and t i tan ium i n tr ial D - 2 (Heat #541333) . 97 (a) Distance along Billet Length (m) Figu re 5.28 R h o m b o i d i t y measurement a long the length o f B i l l e t #1 for a h igh carbon steel i n t r ia l D - 2 (Heat # 541351): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurements is 20). 98 (a) (b) F igure 5.29 R h o m b o i d i t y measurement a long the length o f B i l l e t #3 for a h igh carbon steel i n t r ia l D - 2 (Heat # 541351): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurements is 20). 99 (a) F igu re 5.30 R h o m b o i d i t y measurement a long the length o f B i l l e t #6 for a h igh carbon steel i n tr ial D - 2 (Heat # 541351): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurements is 20). 100 0-2 2-4 4-6 6-8 8-10 10-12 12-14 14-16 Difference between Diagonals (mm) Figure 5.31 R h o m b o i d i t y measurement a long the length o f B i l l e t #8 for a h igh carbon steel i n tr ial D - 2 (Heat # 541351): (a) measured values and (b) frequency dis t r ibut ion. (Note: To ta l number o f measurements is 20). 101 (a) 0 1 2 3 4 5 6 7 Distance along Billet Length (m) Figure 5.32 R h o m b o i d i t y measurement a long the length o f B i l l e t #10 for a h igh carbon steel i n t r ia l D - 2 (Heat # 541351): (a) measured values and (b) frequency dis tr ibut ion. (Note: To ta l number o f measurements is 20). 102 (a) Figure 5.33 R h o m b o i d i t y measurement a long the length o f b i l le t #11 for a h igh carbon steel i n t r ia l D - 2 (Heat # 541352): (a) measured values and (b) frequency dis t r ibut ion. (Note: To ta l number o f measurements is 20). 103 (a) S -10 V 3= 5 -15 -I 1 (- 1 1 ' 1 1 1 0 1 2 3 4 Distance along Billet Length (m) Figure 5.34 R h o m b o i d i t y measurement a long the length o f B i l l e t #13 for a h igh carbon steel i n t r ia l D - 2 (Heat # 541352): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurements is 19). 104 15 ? JL -m J- + 8 m m w +7 mm e , + 7 mm +7 mm « i- +6 mm +6 mm ¥ i 4 r T i i I I I * • • + 3 mm * * + 3 r r S o T -« 1 + 2 mm + 2 mm + 2 mm + 2 mm + 2 mm 0) XJ -5 V u c 4) JS-10 i t -15 -1 1 1 1 1 h #1 #3 #6 #8 #10 #11 #13 Billet Number Figure 5.35 Range o f rhomboid i ty values measured along the length o f B i l l e t s #1, #3, #5, #6, #8, #10, #11 and #13 for a h igh carbon steel i n t r ia l D - 2 ((Heats # 541351 and 541352). 105 M a c r o photograph o f a longi tudinal section o f a 0.34 pet. carbon steel (containing boron and titanium) bi l let showing transverse depression (Tr i a l D - l ; Heat # 531147). 107 (a) ? 15 E, «T 10 5 + -5 + -10 -15 Billet #16 1 2 3 Distance along Billet Length (m) Figure 5.36 R h o m b o i d i t y measurement a long the length o f B i l l e t #16 for a 0.12 pet. carbon steel i n tr ial D - 2 (Heat # 541297): (a) measured values and (b) frequency distr ibut ion. (Note: To ta l number o f measurements is 13). 106 19 25 Depression # Figure 5.38 Dep th o f transverse depressions measured at the midface o f B i l l e t #2 o f a m e d i u m carbon steel containing boron and t i tanium (Tr i a l D - 2 ; Heat # 541333) . 40 30 +• 0.5-1.0 1.0-1.5 1.5-2.0 2.0-2.5 2.5-3.0 3.0-3.5 Depth of Depression (mm) Figure 5.39 Frequency distr ibution o f the depth o f transverse depressions observed on B i l l e t #2 (Tr ia l D - 2 ; Heat # 541333). 108 90 O 12U "35 tn a> k. a a> Q o | (TJ C '•5 "5> c o S 60 30 13 19 25 31 Depression # 37 43 Figure 5.40 L o n g i t u d i n a l w id th ( in casting direction) o f transverse depressions measured at the midface o f B i l l e t #2 o f a m e d i u m carbon steel conta ining boron and t i tanium (Tr i a l D - 2 ; Heat # 541333). 45-55 55-65 65-75 75-85 85-95 95-105 105-115 115+ Longitudinal Width of Depression (mm) Figure 5.41 Frequency distr ibution o f longi tudina l w id th o f transverse depressions observed on B i l l e t #2 (Tr ia l D - 2 ; Heat # 541333). 109 200 150 E E, c o w </) 0) I— a a> a o 100 v £ 50 > c ra 13 19 25 Depression # 31 37 43 Figu re 5.42 Transverse w i d t h (perpendicular to casting direction) o f transverse depressions measured at the midface o f B i l l e t #2 o f a m e d i u m carbon steel conta in ing boron and t i tanium (Tr ia l D - 2 ; Heat # 541333). 50- 50-75 75-100 100-125 125-150 150-175 Transverse Width of Depression (mm) 175-200 Figure 5.43 Frequency dis tr ibut ion o f transverse w i d t h o f transverse depressions observed on B i l l e t #2 (Tr i a l D - 2 ; Heat # 541333). 110 Figure 5.44 Dis tance between consecutive transverse depressions observed on B i l l e t #2 (Tr ia l D - 2 ; Heat #541333) . F igure 5.45 Frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t #2 (Tr i a l D - 2 ; Heat # 541333) . I l l Figure 5.46 Dis tance between consecutive transverse depressions observed on B i l l e t #4 (Tr ia l D - 2 ; Heat #541333) . F igure 5.47 Frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t #4 (Tr i a l D - 2 ; Heat # 541333) . 112 2400 1 3 5 7 9 11 13 15 Depression # Figure 5.48 Dis tance between consecutive transverse depressions observed on B i l l e t #5 (Tr i a l D - 2 ; Heat #541333) . F igure 5.49 Frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t #5 (Tr ia l D - 2 ; Heat # 541333). 113 Figure 5.50 Dis tance between consecutive transverse depressions observed on B i l l e t #8 (Tr ia l D - 2 ; Heat #541333) . F igu re 5.51 Frequency distr ibution o f distance between consecutive transverse depressions observed on B i l l e t #8 (Tr i a l D - 2 ; Heat # 541333) . 114 1600 9 11 13 15 Depression # Figure 5.52 Dis tance between consecutive transverse depressions observed on B i l l e t #10 (Tr ia l D - 2 ; Heat # 541333). 50 40 ^ 3 0 + o c 3 O20 + 10 + Billet #10- 23 depressions + 50- 50-100 100-150 150-200 200-250 250-300 Distance between Depressions (mm) 300+ Figure 5.53 Frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t #10 (Tr ia l D - 2 ; Heat # 541333) . 115 50 Billets #2, #4, #5, #8 & #10 - Total 132 depressions Figure 5.54 O v e r a l l frequency distr ibution o f distance between consecut ive transverse depressions observed on B i l l e t s #2, #4, #5, #8 and #10 (T r i a l D - 2 ; Hea t#541333) . 116 Figu re 5.55 Examples o f (a) metal l eve l and (b) casting speed signals observed dur ing tr ial C - l (Heat # C 7 6 6 1 ) . (Note: T h e nomina l metal l eve l is - 1 3 7 m m f rom the top o f the m o u l d and the average casting speed is - 3 3 . 5 mm/s) . 117 (a) Inner curved w a l l 110 o 100 60 ICW • 7 mm above meniscus H 1 H 1 1 (-100 200 300 400 Time (s) 500 H H 600 (b) R i g h t straight w a l l 110 o 1 0 0 t 1 so OJ a E £ 80 3 O 2 70 + 60 RSW - 7 mm above meniscus 100 200 300 Time (s) 400 500 H 1 1 1 1 1 1 1 H 600 Figu re 5.56 Response o f a thermocouple located above the meniscus corresponding to the metal l eve l s ignal presented i n F igure F igure 5.55(a) (Tr ia l C - l ; Heat # C 7 6 6 1 ) . 118 200 0 -| 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ' — I 0 100 200 300 400 500 600 700 800 Distance from Top of Mould (mm) Figu re 5.57 Ave rage m o u l d temperature profiles employed to compute heat f lux for three steel grades cast dur ing tr ial D-l: 0.12 pet. carbon steel (Heat # 531142) , 0.32 pet. carbon steel (Heat #531146) and 0.84 pet. carbon steel (Heat # 531149) . 119 (a) 170 O 160 120 - 0.84 pet. C — 0.46 pet. C 310 mm below meniscus 200 250 300 350 400 450 Time (s) 500 550 600 (b) 160 O 150 0.84 pet. C — 0.46 pet. C 620 mm below meniscus 110 -4-200 250 300 350 400 450 Time (s) 500 550 600 Figure 5.58 M o u l d thermal response at C o m p a n y C corresponding to a 0.84 pet. carbon steel (Heat # D6131) where laps were present and a 0.46 pet. carbon steel (Heat # D 6 1 3 5 ) where laps were not observed, at two locat ions (a) 310 m m and (b) 620 m m be low the meniscus. 120 Figure 5.59 M o u l d thermal response at a locat ion - 3 0 m m be low the meniscus corresponding to (a) large rhomboid i ty observed i n 0.32 pet. carbon steel (Heat # 541333 , B i l l e t #19) and (b) sma l l rhomboid i ty observed i n 0.12 pet. carbon steel (Heat # 541297, B i l l e t #16) cast dur ing tr ial D-2. 121 (a) 125 100 200 Time (s) 300 400 (b) 200 O 175 0 1 150 a E « 125 3 o 100 75 - e — B 100 North East West 45 mm below meniscus 200 Time (s) 300 South 400 Figu re 5.60 M o u l d thermal response at a locat ion - 4 5 m m be low the meniscus corresponding to (a) large rhomboid i ty observed i n 0.32 pet. carbon steel (Heat # 541333 , B i l l e t #19) and (b) smal l rhomboid i ty observed i n 0.12 pet. carbon steel (Heat # 541297, B i l l e t #16) cast dur ing tr ial D-2. 122 (a) 175 95 -Bi l let #2 — B i l l e t #19 270 mm below meniscus + 4-200 210 220 230 240 250 260 Time (s) 270 280 290 300 (b) 120 60 •Bil let #2 — B i l l e t #19 565 mm below meniscus + 200 210 220 230 —I 1 1 1 1— 240 250 260 Time (s) -t-270 280 290 300 Figu re 5.61 M o u l d thermal response measured dur ing tr ial D - 2 f rom a heat (# 541333) o f a m e d i u m carbon steel containing boron and t i tanium corresponding to B i l l e t #2 w i t h 46 depressions and B i l l e t #19 w i t h no depressions, at two locat ions (a) 270 m m and (b) 565 m m be low the meniscus. 123 C H A P T E R 6 - M A T H E M A T I C A L M O D E L L I N G O F M O U L D T H E R M A L RESPONSE A n exis t ing mathematical mode l was employed to compute the ax ia l prof i le o f heat f lux and temperature dis tr ibut ion i n the m o u l d w a l l i n the same way as reported i n earl ier publ icat ions [7-10]. T h e m o d e l was also adapted to study the thermal response o f the m o u l d to events such as metal l eve l fluctuations, m o u l d osc i l la t ion , and the presence o f transverse depressions on the b i l le t surface that l oca l l y w i d e n the mould/strand gap and cause a reduct ion i n heat f lux . The details o f the mode l are described i n this chapter together w i t h some pre l iminary results and analysis o f mode l predictions. 6.1 The Mould Heat Transfer Model T h e m o u l d heat transfer mode l simulates heat f low i n a longi tud ina l section through the midface o f the m o u l d w a l l [7-10]. F igure 6.1 shows a schematic d iagram o f the long i tud ina l section o f the m o u l d w a l l at the midface that was mode l l ed [7-10]. T h i s m o d e l is based on the equation for two-dimens iona l , unsteady-state heat transfer that was so lved us ing a finite-difference formulat ion [7-10]. The unsteady-state formulat ion was adopted to a l l ow for c y c l i c changes i n m o u l d temperatures due to nucleate b o i l i n g i n the c o o l i n g water channel [7-10]. T h e m o d e l w h i c h is coded i n Fortran 77, computes the ax ia l profi le o f m o u l d heat f lux and the temperature dis tr ibut ion i n the m o u l d w a l l [7-10]. The f o l l o w i n g assumptions were adopted i n the formula t ion o f the mathematical mode l [7-10]: (i) Heat f l ow i n the direct ion perpendicular to the longi tudina l midface plane (i.e. transverse direction) is negl ig ib le . 124 (i i) Heat transfer between the c o o l i n g water channel and the water jacket is negl ig ib le . ( i i i ) Var ia t ions i n heat f lux due to m o u l d osc i l la t ion and metal l eve l fluctuations are ignored. ( iv) The thermal conduct iv i ty o f copper is independent o f temperature; it was shown i n the or ig ina l study [7-10], w i t h the help o f a sensit ivi ty analysis, that the effect o f temperature-dependent conduct iv i ty on the calculated temperature dis t r ibut ion i n the m o u l d is negl ig ib le . (v) The c o o l i n g water channel extends f rom the top o f the m o u l d to its bo t tom and the water i n the c o o l i n g channel is i n p lug- f low and the f low is turbulent. (vi) T h e top and bot tom surfaces o f the longi tudinal midface plane are assumed to be adiabatic. T h e n the differential equation for heat transfer i n the two dimensions (x and z directions) is as fo l lows : _3_ dx a + * f 3 T A F o r p l u g f l ow , the heat transfer i n the water channel is expressed as fo l lows [7-10] : ?wVwdwcpw— - hw(z,t){T(0,z,t)-Tw(z,t)} = 0 (6.2) T h e boundary condit ions employed i n the mode l are summar ized as fo l lows [7-10]: (i) C o l d face o f m o u l d : x = 0, 0 < z < Zm, t>0 ~kmlt = K(z,t){T(0,z,t)-Tw(z,t)} (6.3) 125 ( i i ) T o p o f m o u l d w a l l : 0 < x < X m , z=0, t>0 ~k-Yx • 0 <6-4> ( i i i ) B o t t o m o f the m o u l d w a l l : 0 < j t < X m , z=Zm, t>0 - * - f " 0 <6-5> (iv) H o t face o f the m o u l d be low the meniscus: x=Xm, Zf<z<Zm, t>0 -km^- = qz(z) (6.6) (v) H o t face o f the m o u l d above the meniscus: JC = Xm, 0 < z < Zf, t > 0 ~kmlt = ha(z,t){T(Xm,z,t)-Ta} (6.7) (vi) Inlet water temperature: z=Zm Tw = Tt (6.8) (v i i ) In i t ia l temperature distr ibution i n mou ld : 0<x<Xm, 0<z<Zm, t = 0 T = T0 (6.9) 126 T h e heat transfer between the copper m o u l d w a l l and the c o o l i n g water occurs by forced convec t ion o w i n g to the h igh c o o l i n g water ve loc i ty w h i c h is i n the range o f 6.0 to 20.0 ms" 1. T h e heat f lux i n the water channel for the three regions o f the forced convec t ion b o i l i n g curve has been discussed i n detail i n earlier publicat ions [7-10]. The heat transfer coefficient hFC for the forced convec t ion regime is determined at the average temperature o f water us ing the f o l l o w i n g correlat ion [7-10]: nFCUh V K f J = 0.023 v -4 V K f J .(6.10) T h e heat f lux QFC for forced convect ion is calculated as fo l lows : QFC = h F C ( T w - T S J .(6.11) T h e heat f lux for the fu l ly-developed nucleate b o i l i n g region QFD is g iven by the f o l l o w i n g equation: Q FD (Tw - T s a t ) K J (P/-Pv)g o-go .(6.12) T h e heat f lux i n the transit ion region QTR is g iven by the f o l l o w i n g equation: QTR = QFC\1 + Q FD Q FC Q V l 0 . 5 FN V QFD J .(6.13) T h e heat f lux at the incept ion o f bo i l i ng QFN is g iven by the f o l l o w i n g equation: 127 QFN = 5.28lxW-'PlA56(l.S0(Tw-Tsat)f J (6.14) T h e m o u l d w a l l was discret ized i n the x and z directions as shown i n F igure 6.1 and finite difference equations were set up for the configurat ion o f nodes us ing the govern ing equation o f heat transfer and the appropriate boundary condit ions (Equations 6.1 to 6.14) [7-10]. The Al te rna t ing D i r e c t i o n Impl ic i t ( A D I ) technique [73] was adopted to solve the resul t ing simultaneous equations to obtain the f inal temperature dis tr ibut ion i n the m o u l d w a l l . 6.2 Steady-state Mould Thermal Response T h e m o u l d temperatures measured i n the plant trials were emp loyed to calculate the steady state thermal response o f the m o u l d (heat f lux profi le and temperature dis t r ibut ion i n the m o u l d w a l l ) . The important steps i n v o l v e d i n the mode l calculat ions are out l ined i n F igure 6.2. A s shown i n the figure, the first step is the calcula t ion o f "fil tered" average values for the measured m o u l d temperature us ing appropriate filters for "meniscus thermocouple", metal l eve l and cast ing speed [28,46]. A n assumed heat f lux profi le is then appl ied as a boundary cond i t ion on the hot face o f the m o u l d and the temperature distr ibution i n the m o u l d w a l l is computed and compared w i t h the "filtered" temperature . I f the two temperature profi les do not match, the input heat f lux profi le is altered and the temperature dis tr ibut ion is re-calculated for the new heat f lux prof i le . T h i s procedure is repeated unt i l the computed temperatures match w i t h the measured values w i t h i n about 1°C. The heat f lux profi le and temperature dis t r ibut ion were calcula ted for a range o f operating condit ions - steel grade, osc i l l a t ion frequency and also, the type o f m o u l d lubricant (o i l and m o u l d f lux) . 128 6.2.1 Mould heat flux and temperature distribution 6.2.1.1 Effect of steel carbon content Three grades o f steel cast dur ing tr ial D - l were considered for this analysis; these inc lude: 0.12 pet. carbon steel (Heat # 531142), 0.32 pet. carbon steel (Heat # 531146) and 0.84 pet. carbon steel (Heat #531149) . The computed ax ia l m o u l d heat f lux prof i le for three grades are presented i n F igure 6.3(a) w h i l e the average and peak values o f the heat f lux are plot ted i n F igure 6.3(b). The ax ia l temperature profi le for the hot face and the c o l d face o f the m o u l d for the three grades examined are presented i n F igure 6.4(a) and (b) respect ively. F igure 6.3(a) shows that the m a x i m u m value o f the heat f lux is attained at the meniscus be low w h i c h the heat f lux drops to a lower value. T h i s trend is s imi la r to those obtained i n previous studies [5, 7-10, 14]. The peak heat f lux value corresponds to the site o f in i t i a l so l id i f ica t ion where the l i q u i d steel contacts the copper m o u l d [ 5 , 7 - 1 0 , 1 4 ] . W h e n the l i q u i d steel sol idi f ies , it pu l l s away f rom the m o u l d w a l l creating a gap w h i c h causes a reduct ion i n the heat f lux [5 ,7 -10 , 14]. The peak heat f lux for various grades evaluated at C o m p a n y D is highest for the 0.84 pet. carbon steel ( -6500 k W / m 2 ) f o l l owed by 0.32 pet. carbon steel ( - 4 5 0 0 k W / m 2 ) and is the lowest for 0.12 pet. carbon steel ( -2500 k W / m 2 ) . T h e peak hot face temperature fo l lows a s imi la r trend; it is highest for the 0.84 pet. carbon grades ( ~ 2 9 0 ° C ) and lowest for the 0.12 pet. carbon grade ( ~ 1 5 0 ° C ) ; the temperature for the 0.32 pet. carbon steel is about 2 5 0 ° C . It w i l l be shown later that the magnitude o f the peak hot face temperature relat ive to the b o i l i n g range o f o i l , is an important variable w i t h respect to the generation o f qual i ty problems such as laps/bleeds and rhomboid i ty . 129 6.2.1.2 Effect of oscillation frequency T h e cases examined to evaluate the effect o f osc i l la t ion frequency on m o u l d heat transfer inc luded 144 c p m (t n=~0.16 s) and 96 c p m (t n=~0.20 s) for a 0.46 pet. carbon steel (Heats # A 2 8 1 8 8 and C 7 6 5 4 ) i n tr ial C - l as reported i n a previous study [5], and 160 c p m (t n=~0.18 s) and 100 c p m (t n=~0.25 s) for a 0.93 pet. carbon steel (Heat # E 3 0 7 4 3 ) i n tr ia l A - 2 . The results o f heat f lux calculat ions for the two cases are shown i n Figures 6.5(a) and (b) w h i l e the corresponding m o u l d temperatures are presented i n Figures 6.6(a) and (b). In tr ia l C - l , as the osc i l l a t ion frequency was reduced f rom 144 c p m to 96 c p m , an enhancement i n m o u l d heat extraction occurred w h i c h caused the peak hot face temperature to increase by about 3 5 ° C f rom - 3 7 5 to ~ 4 1 0 ° C . T h i s behaviour was attributed to an increased mould/s t rand interaction due to a longer negative strip t ime at the lower osc i l l a t ion frequency [5]. In tr ial A - 2 , however , as the frequency was decreased f rom 160 c p m to 100 c p m , the m o u l d heat transfer decreased and caused the peak hot face temperature to reduce by about 6 0 ° C f rom ~ 3 2 5 ° C to ~ 2 6 5 ° C . Thus , the two results are not consistent and it appears that there are other factors i n v o l v e d , part icularly the depth o f osc i l la t ion marks w h i c h increases w i t h increasing negative strip t ime [23]. A more detailed discuss ion o f the above f indings w i l l be presented i n Chapter 8 together w i t h its impl ica t ions on the formation o f laps and bleeds i n bi l le ts . 6.2.1.3 Difference between oil and mould flux In tr ial A - 2 , a h igh carbon grade was cast w i t h both m o u l d f lux (Heat # E 3 0 7 1 8 ) and o i l (Heat # E30719) for the same operating condi t ions; the properties o f the m o u l d f lux e m p l o y e d for the tr ia l are l is ted i n Table 6.1. The results o f heat f lux calculat ions for the two cases are shown i n F igure 6.7 w h i l e the corresponding m o u l d temperature profi les are presented i n F igure 6.8. These calculat ions suggest that the m o u l d heat transfer for the case 130 e m p l o y i n g m o u l d f lux is about 15 pet. l ower than for the heat cast w i t h o i l lubr ica t ion w h i l e the corresponding peak hot face temperature is lower by ~ 5 0 ° C for the heat e m p l o y i n g m o u l d f lux as lubricant, as shown i n F igure 6.8. 6.2.2 Analysis of thermal resistance to mould heat extraction In addi t ion to the steady state mode l calculat ions, a s imple analysis was also conducted to evaluate the various resistances encountered dur ing heat f l ow f rom the l i q u i d steel i n the m o u l d to the c o o l i n g water. It is well-establ ished for a continuous cast ing m o u l d that heat f rom l i q u i d steel (latent heat and superheat) released at the so l id i f ica t ion front, is transferred to c o o l i n g water by several paths i n series as shown i n F igure 6.9: (a) conduct ion through the so l id i fy ing shel l (b) conduct ion (and radiation) across the s teel /mould gap (c) conduct ion through the copper m o u l d w a l l (d) convec t ion to the m o u l d c o o l i n g water A s s u m i n g steady state and one-dimensional heat f l ow i n the b i l le t m o u l d for the sake o f s imp l i c i t y , the heat f lux f rom the sol id i f ica t ion front at the solidus temperature Ts to the c o o l i n g water at a temperature Tw can be represented by the f o l l o w i n g relat ionship: where RT is the total thermal resistance encountered to heat f l ow and is g iven as fo l lows : (6.15) K (6.16) 131 The heat flow i n the m o u l d described by Equat ions (6.15) and (6.16) is s imi l a r to the current flow through a series o f resistances i n an electr ical c i rcui t as shown i n F igure 6.9 [14]. T h e thermal resistance o f the copper m o u l d w a l l and the c o o l i n g water are m u c h smal ler than the thermal resistances o f the so l id shel l and the mould/strand gap [14]; these resistances were therefore neglected i n the calculat ions. The relative importance o f the so l id steel shell and mould/strand gap resistances was assessed us ing heat fluxes measured at the plant trials. Three steel grades (0.12 pet., 0.32 pet. and 0.84 pet. carbon steels) that were cast dur ing tr ial D - l were analysed. F o r each grade, the total thermal resistance was first calculated us ing Equa t ion (6.15). T h e s o l i d shel l thickness was computed us ing a finite-difference mode l for bi l le t so l id i f ica t ion [5,10,20,76], that employs the measured heat flux as the surface boundary condi t ion . T h e thermal resistance o f the s o l i d shel l at a g iven locat ion i n the m o u l d was calculated by s imp ly d i v i d i n g the shel l thickness by the thermal conduct iv i ty o f steel. The mould/strand gap resistance was calculated us ing the f o l l o w i n g equation: R = R T -y K kw h c j .(6.17) Figures 6.10 to 6.12 show the axia l profiles o f the gap and shel l resistances calculated for the three grades and the relative contr ibut ion o f each to the total thermal resistance. T h e thermal resistance o f the mould/strand gap is m u c h greater than that o f the so l id shel l and therefore, the m o u l d w i l l respond more strongly to changes i n the mould/s t rand gap than to variat ions i n so l id shel l thickness. T o estimate the effect o f changes i n so l id shel l thickness on the steady state temperature response o f the m o u l d , some s imple calculat ions were performed for the 0.32 pet. carbon steel 132 cast at t r ia l D - 1 (Heat #531146) . The so l id shel l thickness for this grade at a midface loca t ion ( - 5 5 0 m m be low the meniscus) where the heat f lux value is i n the range o f - 1 3 0 0 - 1 4 0 0 k W / m 2 , was computed us ing the b i l le t so l id i f ica t ion mode l [5,10,20,76] to be - 1 1 . 2 m m . F o r this ca lcula t ion , the mould/strand gap was kept constant at - 0 . 2 m m , a value corresponding to a shel l thickness o f - 1 1 . 2 m m . F igure 6.13(a) shows that the heat f lux decreases w i t h increasing s o l i d shel l thickness. F o r the above condi t ions, the heat f lux changes on ly about - 5 0 k W / m 2 per m m change i n so l id shel l thickness. A t the mid-thickness loca t ion i n the m o u l d w a l l ( roughly at the thermocouple tip locat ion) , the change i n m o u l d temperature is also quite sma l l ~ 2 ° C per m m change i n so l id shel l thickness, F igure 6.13(b). Thus , it may not be possible to di rect ly measure changes i n so l id shel l thickness us ing thermocouples embedded i n the m o u l d w a l l . T h e above f indings are quite important f rom the standpoint o f detection o f defects us ing m o u l d thermocouples instal led i n an on-l ine system such as the "intell igent m o u l d " . W i t h respect to the detection o f defects such as transverse depressions, m o u l d thermocouples can be used quite effect ively since these defects cause a w iden ing o f the mould/s t rand gap. W i t h respect to the rhomboid i ty problem, where non-uni form so l id shel l thickness is a key parameter [22], m o u l d thermocouples w i l l be less effective i n the direct measurement o f changes i n so l id shel l thickness. It w i l l be shown later that non-uni form c o o l i n g that actually generates the var iable shel l thickness i n the meniscus region, can be employed as an indirect means to infer the presence o f rhomboid i ty , on- l ine . T h i s point and other related issues w i l l be discussed further i n Chapter 9. T h e heat f lux is a function o f the total thermal resistance. Thus , the presence o f a defect on the strand surface that widens the mould/strand gap w i l l cause a reduct ion i n heat f lux . Ca lcu la t ions were also performed to estimate the effect o f w i d e n i n g o f the mould/s t rand gap 133 on the steady state temperature response o f the m o u l d . F o r the 0.32 pet. carbon steel (Heat # 531146) cast at tr ial D - l , the mould/strand gap at a midface locat ion ( -550 m m b e l o w the meniscus) where the heat f lux value is - 1 3 0 0 to - 1 4 0 0 k W / m 2 , was estimated to be about - 0 . 2 m m . T h i s estimate w h i c h is based on the predictions o f m o u l d distort ion and bi l le t shrinkage [5], is consistent w i t h previous calculat ions based on thermal resistances. T h e temperature o f the m o u l d at the hot and c o l d faces was estimated to be ~ 1 1 0 ° C and ~ 6 0 ° C respect ively, for an assumed gap thermal conduct iv i ty o f about 0.25 W / m ° C [7]. It was shown earlier i n Figures 5.38 and 5.39 that the depth o f about 70 pet. o f the transverse depressions present on B i l l e t #2 i n tr ial D - 2 (Heat # 541333) l ies between - 0 . 5 and - 2 . 0 m m w h i l e the m a x i m u m depth was measured to be - 3 . 3 m m . Thus , i n the presence o f transverse depressions, large gaps can open up between the m o u l d and the strand w h i c h can lead to significant reduct ion i n heat transfer. F igure 6.14(a) shows the effect o f w i d e n i n g o f mould/s t rand gap o f up to 4.0 m m , on the heat f lux profi le w h i l e F igure 6.14(b) shows the temperature at the hot face, the c o l d face and at mid- thickness pos i t ion corresponding to the loca t ion o f the thermocouple tip. A s expected, the heat f lux decreases f r o m a value o f - 1 4 0 0 to - 2 0 0 k W / m 2 when the w i d t h o f the mould/strand gap is increased f rom 0.2 to 4.0 m m . Furthermore, there is l i t t le change i n heat transfer once the w i d t h o f the mould/s t rand gap exceeds about 2.0 m m . The m o u l d w a l l temperature at the hot face fo l lows the pattern i n heat f lux and decreases f rom a value o f - 1 1 0 ° C to ~ 4 0 ° C . T h e above calculat ions were repeated for the 46 depressions observed on B i l l e t #2. The percent drop i n heat f lux for each depression is plotted i n F igure 6.15 w h i l e a statistical dis t r ibut ion o f the drops i n heat f lux is presented i n F igure 6.16. A s shown i n the two figures, the drop i n heat f lux for a l l o f the 46 depressions is greater than - 5 0 pet., and most o f the drops l ie i n the range o f - 6 0 to - 9 0 pet.. These results have been employed i n Sec t ion 6.3.4 134 to calculate the transient m o u l d thermal response for B i l l e t #2 cast dur ing tr ial D - 2 . 6.3 Unsteady-State Mould Thermal Response T h e m o u l d heat transfer mode l was adapted to simulate the effect o f transient events such as m o u l d osc i l l a t ion , metal l eve l fluctuations and the presence o f defects such as transverse depressions on the m o u l d thermal response o f the m o u l d . 6.3.1 Thermal response of billet moulds The response t ime o f the m o u l d , part icularly at the loca t ion o f the t ip o f the thermocouples, to an event occurr ing at the hot face is important w i t h respect to the detection o f variat ions observed i n m o u l d temperature dur ing the cast ing operation. W h e n a heat f lux prof i le is appl ied to the hot face o f a m o u l d set at an in i t i a l temperature o f 2 5 ° C , the temperature o f the m o u l d increases and reaches steady state i n about 10 seconds as shown i n F igu re 6.17. T h e t ime to attain steady state temperatures at the c o l d face, hot face and a loca t ion m i d w a y between the hot and the c o l d face (about 8 m m f rom the hot face that is roughly the loca t ion o f the thermocouple tip) is about 10 seconds. D u r i n g the operation, however , the in i t i a l temperature o f the m o u l d is not 2 5 ° C but i n the region o f ~ 1 0 0 - 3 0 0 ° C . Therefore, the t ime to reach steady state after a loca l change i n heat f lux w i l l be shorter than 10 s. 6.3.2 Thermal disturbances due to mould oscillation T h e m o u l d is osci l la ted dur ing the casting operation at a frequency o f about ~2 H z . T h i s must cause disturbances i n m o u l d temperature. In this study, it was necessary to determine whether the temperature fluctuations generated f rom m o u l d osc i l l a t ion affected the thermocouples instal led approximately m i d w a y i n the m o u l d w a l l between the hot and the c o l d 135 faces. In addi t ion, it was also necessary to f ind out whether the thermal disturbances were signif icant compared to those generated by metal l eve l fluctuations, especia l ly i n the meniscus region. T h e osc i l l a t ion o f the m o u l d was s imulated by s imp ly translating the heat f lux profi le ax ia l ly up and d o w n (amplitude was set equal to the osc i l la t ion stroke length) the m o u l d f rom a f i xed metal l eve l pos i t ion and at a certain osc i l la t ion frequency. M o u l d osc i l l a t ion s imula t ion was conducted for the 0.32 pet. carbon steel cast at trial D - l ; the osc i l l a t ion frequency was f ixed at 100 c p m and the stroke length was 7.0 m m . W h e n m o u l d osc i l l a t ion starts after an in i t i a l steady state per iod (Figure 6.18), the temperature o f the m o u l d w a l l declines and reaches a new steady state as shown i n Figures 6.19(a) and (b). It can be noted f rom the figure that the t ime taken by the m o u l d to reach this new steady state is about 5 to 10 s. T h e m a x i m u m , average and m i n i m u m temperatures were calculated for locat ions above and b e l o w the meniscus, after the new steady state was achieved. A sample result o f the calculat ions performed is presented i n Figures 6.20(a) and (b) w h i c h show that the thermal disturbance is greatest at the meniscus and drast ically drops to very l o w values for other locat ions i n the m o u l d . The disturbance at the hot face is quite s ignif icant ( ~ 4 0 - 5 0 ° C ) as compared to that at the midpoin t ( location o f thermocouple tip) where the difference at the meniscus is on ly ~ 5 ° C . Thus , w i t h respect to detection o f these disturbances us ing m o u l d thermocouples instal led roughly m i d w a y between the hot and the c o l d faces, these are quite sma l l compared to those generated by metal l eve l fluctuations. 6.3.3 Metal level fluctuations T o be able to moni tor metal l eve l fluctuations us ing thermocouples located above the meniscus , it is necessary to estimate the metal l eve l ( in m m ) f rom temperature ( in °C) recorded 136 by a thermocouple. T h e metal l eve l change (rise and fall) was s imulated i n the mathematical mode l by s imp ly m o v i n g the average measured heat f lux profi le ax ia l ly up and d o w n the m o u l d for a desired duration. The effect o f metal l eve l change on the temperature var ia t ion at the hot face as w e l l as at mid-thickness thermocouple locat ion was computed at a loca t ion ~8 m m above the meniscus, for the three grades (0.12 pet., 0.32 pet. and 0.84 pet. carbon steels) examined i n tr ial D - l . The results are presented i n Figures 6.21 to 6.23. T h e temperature change calculated for the 0.12 pet. carbon is the lowest f o l l owed by 0.32 pet. carbon steel and then by the 0.84 pet. carbon. T h i s trend exact ly fo l lows the values o f peak heat f lux for the three grades shown i n the F igure 6.3. T h e relat ionship between measured temperature change and corresponding metal l eve l change depends on m o u l d hot face temperature, distance o f the thermocouple f r o m the average meniscus pos i t ion and also, distance o f the thermocouple tip f rom the hot face o f the m o u l d . A s shown i n Figures 6.21 to 6.23, the t ime duration o f the change i n metal l eve l is another var iable that also influences this relationship since there is a delay i n the response o f the m o u l d to thermal disturbances generated on the hot face. These issues w i l l be further examined i n Chapter 7. 6.3.4 Simulation of transverse depressions observed in trial D-2 A s ment ioned earlier, thermocouples w i l l be employed to detect transverse depressions i n the "intell igent m o u l d " . T o be useful for the on- l ine detection o f depressions, the "intel l igent m o u l d " should be able to estimate the characteristics (depth and width) o f depressions f rom the measured m o u l d thermal response. Based on these estimates, it w i l l be possible to make decis ions on- l ine to assess b i l le t qual i ty. T o examine the above issues, the mathemat ical mode l was emp loyed to predict the m o u l d thermal response i n the presence o f transverse depressions 137 observed on the east face o f B i l l e t #2 cast dur ing tr ial D - 2 (Heat # 541333) . F o r these calcula t ions , the measured data on transverse depression (depth, w i d t h and spacing) presented i n F igures 5.38 to 5.41 for B i l l e t #2 were incorporated as inputs to the m o d e l . In the calculat ions that were carr ied out, it was assumed that each transverse depression creates a "va l ley" i n the heat f lux and also, that the "va l ley" moves d o w n the m o u l d at a rate equal to the casting speed, the average value o f w h i c h i n t r ia l D - 2 was - 1 9 . 9 mm/s . Further, the depth o f a "va l ley" i n the heat f lux represents the depth o f a g iven depression w h i l e the w i d t h o f the "va l ley" i n the heat f lux was assumed to be equal to that o f the depression i n the cast ing direct ion. The wid th o f the depression i n the casting di rec t ion was converted into an equivalent " d w e l l t ime" ( in seconds) by s imply d i v i d i n g the w i d t h ( in m m ) by the cast ing speed (19.9 mm/s) . In this analysis, the depth o f the depression was incorporated into the mode l input as a percent reduction i n heat f lux us ing the relationship presented i n F igu re 6.14(a). T h e transient condi t ion was s imulated by m o v i n g the "val ley" i n the heat f lux d o w n the m o u l d at a rate equal to the casting speed. A s an example , F igure 6.24 shows a section o f the resul t ing heat f lux profi le (at two locations i n the mould) that was incorporated as a boundary cond i t ion i n the m o d e l calculat ions. T h e m o u l d thermal response was then computed us ing the mathematical m o d e l and compared w i t h the measured m o u l d w a l l temperature corresponding to B i l l e t #2. T h e results are presented i n F igure 6.25 for two thermocouple locations ( - 2 7 0 and - 4 6 5 m m b e l o w the meniscus) . A s shown i n F igure 6.25, the mode l predicted m o u l d thermal response matches quite c lose ly w i t h the measured response and indicates that m o u l d temperature can s ignif icant ly drop i n the presence o f transverse depressions. In both cases, most o f the temperature drops were between - 1 0 and ~ 5 0 ° C . W h i l e the magnitude o f temperature drops are quite comparable , 138 the two temperatures sometimes are out o f phase f rom each other. T h i s is expected since a constant casting speed was assumed i n the mode l calculat ions, whereas i n the actual operation, the cast ing speed is not constant and changes i n response to variations i n metal l eve l . T h e S C A D A software was employed to compute the magnitude o f temperature drops for each "va l ley" seen i n both measured and computed m o u l d temperature. F igure 6.26 shows a compar i son between the computed and the measured temperature response o f the m o u l d ; most o f the temperature drops observed for both cases l i e between - 1 0 and ~ 4 0 ° C . O n c e again, the agreement between the two results is quite reasonable as evident f rom Tab le 6.2. F r o m the above analysis, it is apparent that a transverse depression is manifested as a "va l ley" or drop i n m o u l d temperature as it moves past a g iven thermocouple loca t ion . T h e span o f the "va l ley" is a function o f the w id th o f the depression i n the cast ing d i rec t ion w h i l e the magnitude o f temperature drop is direct ly related to the depth o f the depression. Thus , by measur ing the depth and span o f a temperature drop, it should be possible to predict the depth and w i d t h o f transverse depressions generated on the bi l le t surface. T o examine the above issue, the span o f "val leys" i n the m o u l d thermal response was computed and compared w i t h the actual w i d t h o f transverse depressions measured i n the cast ing di rect ion. T h e results presented i n F igure 6.27 indicate a poor correla t ion between the span o f the "valleys!' i n m o u l d temperature and the actual w i d t h o f the depression i n the casting di rect ion. Thus , the span o f "val leys" observed i n m o u l d thermal response cannot be e m p l o y e d to predict the w i d t h o f depressions generated on the bi l le t . T o address the above issue further, the total span o f the "va l ley" i n a temperature prof i le was d i v i d e d into two parts as shown i n F igure 6.28, t l and t2 (total span = t l +12) where, the t ime t l is the per iod i n the "val ley" dur ing w h i c h the temperature falls unt i l a m i n i m u m value 139 is reached, w h i l e t2 is the per iod dur ing w h i c h the temperature rises to a higher value f o l l o w i n g the attainment o f the m i n i m u m value. The computed values o f t l was plotted against the w i d t h o f transverse depressions observed on B i l l e t #2. F igure 6.29 shows that a fa i r ly good correlat ion exists between the computed values o f t l and the wid th o f depressions measured on the bi l le t surface. Thus , t l can be employed i n the "intelligent m o u l d " to estimate the w i d t h o f transverse depressions. Furthermore, the average value o f t l for a l l o f the transverse depressions present on B i l l e t #2 was found to be roughly - 0 . 7 5 times that o f the actual span o f depressions measured o n the bi l le t surface for a casting speed o f 19.9 mm/s . T h e temperature drops computed at two locations ( -8 m m f rom the hot face) i n the m o u l d are plot ted against the depth o f depressions i n Figures 6.30. A l t h o u g h there is some scatter i n the data, it is apparent that the magnitude o f temperature drops increases w i t h increas ing depth o f transverse depressions. The scatter i n the data obv ious ly indicates that other parameters are i n v o l v e d , par t icular ly since the m o u l d temperature does not reach steady-state for each "va l ley" present i n the heat f lux. Because o f the transients i n v o l v e d , the magnitude o f temperature drop observed for a g iven transverse depression must be inf luenced by factors such as the w i d t h o f the depression as w e l l as its distance f rom depressions preceding it and also, those f o l l o w i n g it (i.e., distance between consecutive depressions). Thus , calculat ions were carr ied out to examine the effect o f these variables on the magnitude o f temperature drops. It was shown earlier i n F igure 6.15 that the m o u l d heat f lux can drop s ignif icant ly ( -50 to - 9 0 pet.) due to the widen ing o f the mould/strand gap i n the presence o f transverse depressions observed on B i l l e t #2 i n tr ial D - 2 w h i l e the span o f depressions were between - 2 . 0 and - 8 . 0 s. Calcu la t ions were carr ied out for a number o f cases w i t h va ry ing percent reductions i n heat f lux (50 to 90 pet.) and spans (2.0 to 10.0 s). In this ca lcula t ion , o n l y one "va l ley" was considered to isolate the effect o f the other variable, i.e., distance between consecut ive 140 depressions. T h e magnitude o f the predicted temperature drops for the various cases examined are c o m p i l e d i n Figures 6.31 and 6.32 for two locations (430 and 730 m m ) i n the m o u l d . Thus , the m o u l d does not reach steady state for the range o f spans encountered. Ca lcu la t ions were also conducted for various spacings (distance between consecut ive depressions) between depressions for a series o f 10 "val leys" m o v i n g d o w n the m o u l d . T h e percentage drop i n heat f lux for each va l ley was assumed to be 70 pet. for this ca lcula t ion . T h e range o f span was again 2.0 to 10.0 s w h i l e the distance between consecut ive depressions was between 0.0 and 20.0 s. The results presented i n F igure 6.33 h ighl ight the importance o f the distance between consecutive depressions on the magnitude o f temperature drops. F o r a g iven size (width , depth and spacing) o f depressions, the effect o f cast ing speed is quite important since a higher casting speed w i l l lead to a smaller d w e l l t ime i n the m o u l d . F o r example , a depression 20 m m wide w i l l have a span o f 1.0 s for a cast ing speed o f 20 mm/s and on ly 0.5 s at a speed o f 40 mm/s . Thus , the thermal response o f the m o u l d w i l l be different for the same depression size. In addit ion, the t ime between consecut ive depressions w i l l also be affected by casting speed due to the same reasons. Thus , the sensi t ivi ty o f the "intel l igent m o u l d " to defects on the strand w i l l vary w i t h casting speed and it is important to account for the above parameters w h i l e designing strategies i n the "intell igent m o u l d " to detect defect and to assign some sort o f a "quali ty index" to bi l le ts ; a p re l iminary d iscuss ion on "bil let qual i ty index" for the "intelligent m o u l d " is presented i n A p p e n d i x C . T h e issue o f cast ing speed is quite important as w e l l since a number o f b i l le t section sizes (and therefore, casting speeds), are encountered i n the same plant. T h e m a i n objective o f the above exercise was to develop a better understanding o f the patterns observed i n m o u l d thermocouple signals and ut i l ize this knowledge to develop strategies for detecting defects i n the m o u l d . A t the start o f the development o f the "intel l igent 141 m o u l d " , it was not fu l ly clear, whether or not, each "va l ley" i n m o u l d temperature represented a transverse depression on the strand surface. T h i s analysis p rov ided a mathematical and fundamental basis to the approach adopted for detection o f defects and showed that each "va l ley" ( inc lud ing incomplete ones) i n the temperature was indeed a transverse depression. O n the basis o f this analysis, a nove l technique was developed i n a para l le l study [62], for the on- l ine detection o f surface defects (transverse depressions) i n the m o u l d . 142 Table 6.1: Properties o f m o u l d f lux employed i n T r i a l A - 2 [52]. C h e m i c a l P r o p e r t i e s P h y s i c a l P r o p e r t i e s B a s i c i t y -> 0.75 - 0.85 Softening Point (°C) -> 1 1 4 5 - 1175 S i 0 2 -> 25.0 - 28.0 M e l t i n g Poin t ( °C) -> 1 1 8 5 - 1215 C a O -> 19.8 - 22.8 F l o w i n g Poin t ( °C) -> 1220 - 1250 M g O -> 2.0 max A 1 2 0 3 -> 1 2 . 7 - 1 4 . 7 . V i s c o s i t y (un i t s ) T i 0 2 -> 1.0 max a t l 3 0 0 ° C -> 12.0 - 12.4 F e 2 0 3 -> 3.0 - 4.0 at 1 4 0 0 ° C -> 05.2 - 05.6 M n O -> 0.5 max N a 2 0 -> 3.3 - 4.3 Dens i ty (kg/m 3 ) -> 600 to 800 K 2 0 -> 2.0 max F -> 4.0 - 5.0 C - T o t a l -> 22.5 - 24.5 C - Free -> 2.0 max B 2 0 3 -> 0.5 max H 2 0 at 6 0 0 ° C -> 1.0 max H 2 Q at 105°C -> 0.5 max 143 Table 6.2 C o m p a r i s o n o f frequency dis tr ibut ion o f measured and computed m o u l d temperature drops at four locations (170, 270, 370 and 465 m m b e l o w the meniscus) for B i l l e t #2 that was cast dur ing T r i a l D - 2 (Heat # 541333) . Temperature Drops (°C) Frequency Distribution (pet.) Computed and (Measured) 170 mm 270 mm 370 mm 465 mm Below 10 0 ( 3 ) 0 ( 5 ) 0 ( 5 ) 2 ( 1 4 ) 10 to 20 11(23) 17 (38) 41 (50) 57 (48) 20 to 30 39 (42) 3 9 ( 3 5 ) 43 (27) 41 (35) 30 to 40 3 5 ( 2 3 ) 3 5 ( 1 8 ) 15 (15) 0 ( 3 ) 40 to 50 15(9) 9 ( 3 ) 0 ( 3 ) 0 ( 0 ) Above 50 0 ( 0 ) 0 ( 2 ) 0 ( 0 ) 0 ( 0 ) Note: The table corresponds to values presented in Figure 6.26 for measured and computed temperature drops. 144 Figure 6.1 Schematic d iagram o f the longi tudinal section o f the m o u l d w a l l at the midface that was mode l led [7]. 145 Operating Parameters Material Properties Model Parameters Measured Temperature in Plant Trials I Filter and Average HEAT T R A N S F E R MODEL I Calculated Temperature Profile Input Heat Flux Profile Convergence Is Calculated Temperature = Measured Temperature ? NO Modify Input Heat Flux Profile YES Print Final Results Figure 6.2 Out l ine o f the procedure adopted i n the mode l calculat ions o f heat f lux and temperature distr ibution i n the m o u l d w a l l f rom temperature measurements conducted i n the plant trials. 146 (a) Figu re 6.3 M o u l d heat transfer for 0.12 pet. C (Heat # 531142) , 0.32 pet. C (Heat # 531146) and 0.84 pet. C (Heat # 531149) steels that were cast dur ing tr ia l D - l : (a) heat f lux profiles and (b) average and peak values. 147 (a) 300 0 -I 1 1 1 1 1 1 1 1 1 1 0 200 400 600 800 1000 Distance from Top of Mould (mm) (b) 150 O 3 100 0) Q. E i2 u 50 RJ LL O o 0.12 pet. C 0.32 pet. C — 0.84 pet. C + + + 200 400 600 800 Distance from Top of Mould (mm) 1000 Figure 6.4 A x i a l temperature profiles at (a) hot face and (b) c o l d face for 0.12 pet. C (Heat # 531142), 0.32 pet. C (Heat # 531146) and 0.84 pet. C (Heat # 531149) steels that were cast dur ing tr ia l D-l. 148 (a) 100 200 300 400 500 600 700 Distance from Top of Mould (mm) 800 900 (b) 100 200 300 400 500 600 Distance from Top of Mould (mm) 700 800 Figure 6.5 Effect o f osc i l l a t ion frequency o n m o u l d heat extraction i n (a) t r ia l C - l (Heats # A 2 8 1 8 8 and C 7 6 5 4 ) and (b) t r ial A - 2 (Heat # E30743) . 149 Figure 6.6 Effect o f osc i l la t ion frequency on m o u l d hot and c o l d face temperatures i n (a) t r ia l C - l (Heats # A 2 8 1 8 8 and C 7 6 5 4 ) and (b) tr ial A - 2 (Heat # E 3 0 7 4 3 ) . 150 Figure 6.7 C o m p a r i s o n o f m o u l d heat extraction for a h igh carbon steel cast w i t h o i l (Heat # E 3 0 7 1 9 ) and m o u l d f lux (Heat # E30718) lubricants i n tr ial A - 2 . F igure 6.8 C o m p a r i s o n o f m o u l d temperature for a h igh carbon steel cast w i t h o i l (Heat # E30719) and m o u l d f lux (Heat # E30718) lubricants i n tr ial A - 2 . 151 Figure 6.9 Schematic d iagram showing the paths o f heat transfer f rom mol ten steel to c o o l i n g water i n the m o u l d and the corresponding series resistance analogue [14]. 152 (a) 2.00 Distance from Top of Mould (mm) (b) Figure 6.10 A x i a l profi le o f (a) computed values o f thermal resistance o f mould/s t rand gap and s o l i d shel l and (b) percentage contr ibut ion o f each resistance for the 0.12 pet. carbon steel (Heat # 531142) cast at t r ia l D - l . (Note: T h e average metal l eve l is - 1 8 0 m m f rom top o f mould) . 153 (a) F igure 6.11 A x i a l profi le o f (a) computed values o f thermal resistance o f mould/s t rand gap and s o l i d shel l and (b) percentage contr ibut ion o f each resistance for the 0.32 pet. carbon steel (Heat # 531146) cast at tr ial D - l . (Note: T h e average metal l eve l is - 1 8 5 m m f rom top o f mould) . 154 (a) (b) Distance from Top of Mould (mm) Figure 6.12 A x i a l profi le o f (a) computed values o f thermal resistance o f mould/s t rand gap and s o l i d shel l and (b) percentage contr ibut ion o f each resistance for the 0.84 pet. carbon steel (Heat # 531149) cast at tr ial D-l. (Note: T h e average meta l l eve l is - 1 8 6 m m f rom top o f mould) . 155 (a) 1.6 CM 1.2 X JI 0.8 re OJ X T J 3 0.4 o Figure 6.13 Effect o f so l id shel l thickness on (a) heat f lux and (b) m o u l d temperature for a 0.32 pet. carbon steel (Heat # 531146) cast dur ing tr ia l D - 1 at 550 m m b e l o w the meniscus. 156 (a) 1.6 CM 3 o 0.0 1.0 2.0 3.0 4.0 Size of Mould/Strand Gap (mm) Figure 6.14 Effect o f size o f mould/strand gap on (a) heat f lux and (b) m o u l d temperature for a 0.32 pet. carbon steel (Heat # 531146) cast dur ing tr ia l D - l at 550 m m be low the meniscus. 157 90 ^80 3 .£ 70 Q. O c <U e n O 60 d> CL 50 -4"t"l"l"l»l"l"l wl wl w lM l"l wl*Fl"l*l"l"l*1 wlM l"l"l"l"l"l"l*l"l"l"l"l"l"l*l"l"l"l*l"l"l"l"l"l ' ' l 1 6 11 16 21 26 31 36 41 46 Depression # Figure 6.15 A c t u a l values o f percent drop i n heat f lux computed for the 46 transverse depressions present on B i l l e t #2 (Heat # 541333) that was cast dur ing t r ia l D - 2 . 158 Figure 6.16 Frequency distr ibution o f percent drop i n heat f lux computed for the 46 transverse depressions present on B i l l e t #2 (Heat # 541333) that was cast dur ing tr ia l D - 2 . 159 Figure 6.17 A n example o f thermal response o f the m o u l d w a l l ( -25 m m b e l o w the meniscus) that was computed us ing the mathematical mode l o f m o u l d heat transfer and heat f lux profi le f rom C o m p a n y C : (a) hot face, (b) mid- thickness and (c) c o l d face. 160 0 2 4 6 8 10 12 14 16 18 20 Time (s) Figure 6.18 M o u l d osc i l l a t ion (displacement) profi le that was emp loyed as an input to the mathematical mode l to compute the transient thermal response o f the m o u l d dur ing m o u l d osc i l la t ion (Figure 6.19 and 6.20) (Tr i a l D-l, Heat # 531146) . 161 (a) (b) 175 G-165 + O 0 k_ 1 155 &_ Oi Q. E ^ 145 3 O 135 + 125 At Hot Face ~5 mm above meniscus Old Steady state : A Start of Mould Ul A . New Steady state Oscillation VI  A i i i i i i i i 8 10 12 Time (s) 14 16 18 20 135 O 1 3 0 O 0) 0) a E ,2 120 o S 115 4-110 At Thermocouple Location ~5 mm above meniscus Old Steady state Start of Mould Oscillation 8 10 12 Time (s) 14 16 18 20 Figure 6.19 C o m p u t e d transients i n m o u l d temperature at (a) hot face and (b) mid- th ickness thermocouple locat ion due to m o u l d osc i l la t ion presented i n F igure 6.18 (Tr i a l D-l, Heat #531146) . 162 (a) | -50 -25 0 25 50 75 100 125 Distance From Meniscus (mm) Figure 6.20 Effect o f m o u l d osc i l la t ion on disturbances generated i n m o u l d temperatures: (a) m a x i m u m minus m i n i m u m and (b) m a x i m u m minus average temperatures at the hot face and mid-thickness locat ion. (T r i a l D-l, Heat #531146) 163 (a) ^40 o o <D IS 30 8 mm above meniscus At Thermocouple Location Change in Metal Level (mm) (b) ^.80 O o V 3 « 60 8 mm above meniscus At Hot Face 10 .Os 10 15 20 Change in Metal Level (mm) 25 Figure 6.21 Change i n m o u l d midface temperature at (a) mid- thickness thermocouple loca t ion and (b) hot face due to metal l eve l fluctuations i n 0.12 pet. carbon steel (Heat # 531142) cast dur ing tr ial D-l. 164 (a) (b) -t-10 15 20 Change in Metal Level (mm) Figure 6.22 Change i n m o u l d midface temperature at (a) mid- thickness thermocouple loca t ion and (b) hot face due to metal l eve l fluctuations i n 0.32 pet. carbon steel (Heat # 531146) cast dur ing tr ial D-l. 165 (a) (b) F igure 6.23 Change i n m o u l d midface temperature at (a) mid- thickness thermocouple loca t ion and (b) hot face due to metal l eve l fluctuations i n 0.84 pet. carbon steel (Heat # 531149) cast dur ing tr ial D-l. 166 15 25 35 45 55 65 75 Time (s) 85 95 105 115 125 Figure 6.24 Sect ion o f the heat f lux profi le at two locat ions 270 m m and 465 m m b e l o w the meniscus that was employed as an input to compute the m o u l d thermal response dur ing the casting o f B i l l e t #2 (Heat # 541333) i n t r ia l D - 2 . 167 (a) 165 270 mm below meniscus Predicted — Measured 15 25 35 45 55 65 75 Time (s) 115 (b) 115 •105 465 mm below meniscus Predicted — Measured 55 25 -i h-35 H 1 H H 1 h- H h—+" 45 55 65 75 85 Time (s) 95 105 115 125 Figure 6.25 C o m p a r i s o n o f computed and predicted m o u l d w a l l temperature at two locat ions (a) 270 m m and (b) 465 m m be low the meniscus for B i l l e t #2 (T r i a l D - 2 ; Heat # 541333) having 46 transverse depressions. 168 (a) 10- 10-20 20-30 30-40 40-50 Temperature Drop (°C) 50+ (b) 40-50 Temperature Drop (°C) Figu re 6.26 C o m p a r i s o n o f the frequency distr ibution o f (a) measured and (b) computed temperature drops for B i l l e t #2 (Tr i a l D - 2 ; Heat # 541333) at four locat ions 170 m m , 270 m m , 370 m m and 465 m m be low the meniscus. 169 159.8 139.8 § 119.8 '55 V) 0) a a> O 99.8 O 79.8 •«—< 5 59.8 a • • • • • a a a • n o • • • • • a • • • • a a • a • a 39.8 * 1 h I i • • • • H 1 1 1 8.0 + 7.0 _ c 6.0 o </> (0 5-0 | Q • H 1 h 4.0 3.0 c as a in 2.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 Span of Valley in Temperature (s) Figu re 6.27 Rela t ionship between the w id th o f transverse depressions ( in the cast ing direct ion) and the corresponding span o f "va l ley" observed i n the computed m o u l d temperature at a locat ion 270 m m be low the meniscus for B i l l e t #2 (T r i a l D - 2 ; Heat # 541333, Average casting speed is 19.9 mm/s) . 170 (a) 1.2 1.0 o ~ 0.8 ca 0.6 -f 13 0 4 a) z 0.2 0.0 fh Span -T> : \<h- Spacing 4-7J Depth SL 4-14 16 18 20 22 24 26 Time (s) 28 30 32 34 (b) 160 O S-140 a> 3 +J CO k. 0) a. E a \-120 100 + 80 kr- t1 - | > k r - t2 H^ L < } k O — S p a c i n g — f j ^ 4-14 16 18 20 22 24 26 Time (s) 28 30 32 34 Figure 6.28 Schematic d iagram i l lustrat ing the various components o f a "va l ley" i n (a) heat f lux and (b) m o u l d temperature profiles and the associated terms (span, spacing depth, t l and t2) . h F 5 171 159.8 8.0 • Predicted Fitted Value of t1 (s) Figure 6.29 Rela t ionship between wid th o f transverse depressions ( in the cast ing direction) and the corresponding values o f t l for the "val leys" observed i n the computed m o u l d temperature at a locat ion 270 m m be low the meniscus for B i l l e t #2 (Tr i a l D - 2 ; Heat #541333 , Average casting speed is 19.9 mm/s) . 172 (a) 10 15 20 25 30 35 Temperature Drop (°C) 40 (b) 4.0 0.0 465 mm below meniscus E E 3.0 -ion Depressi to o • a of i .1 .0 -a o a _ a a • a • • • a • • • a • • D a a a • 10 15 20 Temperature Drop (°C) 25 o c a • 30 Figure 6.30 Rela t ionship between depth o f transverse depressions and magnitude o f corresponding temperature drops observed i n the computed m o u l d temperature at (a) 270 m m and (b) 465 m m be low the meniscus for B i l l e t #2 (T r i a l D - 2 ; Heat # 541333, Average casting speed is 19.9 mm/s) . 173 0 4 - 1 1 1 1 1 1 1 1 2 4 6 . 8 10 Span of Valley In Heat Flux (s) (b) F igure 6.31 ( Effect o f span o f "val leys" i n heat f lux profi le o n the magnitude o f temperature drops at 270 m m be low the meniscus at (a) hot face and (b) mid- thickness thermocouple locat ion for the cases examined for 0.32 pet. carbon steel (T r i a l D - 2 ; Heat #541333) . 174 (a) F igu re 6.32 Effect o f span o f "val leys" i n heat f lux profi le on the magnitude o f temperature drops at 465 m m be low the meniscus at (a) hot face and (b) mid- thickness thermocouple locat ion for the cases examined for 0.32 pet. carbon steel (Tr ia l D - 2 ; Heat #541333) . 175 (a) F igu re 6.33 Effect o f spacing between consecutive "val leys" i n heat f lux prof i le on the magnitude o f temperature drops computed at 270 m m b e l o w the meniscus at (a) hot face and (b) mid-thickness thermocouple loca t ion for four different spans for 0.32 pet. carbon steel (Tr ia l D - 2 ; Heat # 541333) . 176 C H A P T E R 7 - M E T A L L E V E L FLUCTUATIONS IN B I L L E T CASTING 7.1 Monitoring Meniscus Fluctuations with Existing Sensors In b i l le t cast ing, the metal l eve l is t radi t ionally moni tored w i t h a radioact ive source coup led to a sensor. In this study, analysis o f these metal l eve l signals revealed several interesting patterns that p rov ided useful insights about the state o f the process. T h e most s t r ik ing features were the numerous fluctuations i n metal l eve l s ignal and their apparent randomness; it must be pointed out that a detailed mathematical analysis was not attempted i n this study to characterize the randomness i n metal l eve l . Figures 7.1(a) and (b) present sensor signals f rom two different heats (# D 6 1 3 5 and A 2 8 1 9 3 ) i n tr ial C - l . It is clear that the metal l eve l r andomly rose and fe l l w i t h the rise almost a lways fo l lowed by a drop be low the average l eve l , p robably induced by a change i n casting speed. The casting speed profi le shown i n F igure 7.2(a) and (b) f o l l o w e d the metal l eve l signals shown i n F igure 7.1(a) and (b), quite c lose ly w i t h a sma l l t ime lag , as expected. T h e second feature o f the metal l eve l variations related to their severity w h i c h appeared to change w i t h t ime, again without a clear trend emerging. T h e standard devia t ion ( S D ) o f the metal l eve l s ignal was employed as a means to quantify the state o f metal l eve l fluctuations dur ing a certain per iod i n a heat. M e t a l l eve l signals plotted i n Figures 7.1(a) and (b) compare two very different cases, one showing a large S D o f 4.8 m m and the other hav ing a sma l l S D o f 1.4 m m . T h e th i rd feature o f the metal l eve l variations is quite interesting since it can be used to infer the event causing the fluctuations. Figures 7.3(a) and (b) show two very different patterns i n metal l eve l signals observed over 400 s i n two heats f rom tr ia l D - l hav ing different carbon 177 content. F igure 7.3(a), w h i c h is a plot o f metal l eve l s ignal for a 0.12 pet. carbon steel grade (Heat # 531142) , exhibits random variations o f about 5-10 m m whereas F igure 7.3(b), w h i c h is the s ignal for a 0.32 pet. carbon steel w i t h boron additions (Heat # 531147) , shows metal l eve l variat ions o f about 20 m m ; interestingly, the latter fluctuations appear to be per iodic , occur r ing almost every 15-20 s. W h e n observed c lose ly , however , the larger w a v e - l i k e pattern has smal ler fluctuations o f 5-10 m m superimposed on it. In grades exh ib i t ing the wave- fo rm seen i n F igure 7.3(b), transverse depressions, tr iggered by metal l eve l fluctuations, were observed on the surface o f the bi l lets [21]. T h e transverse depressions reduced m o u l d heat transfer due to the enlarged mould/s t rand gap. T h i s resulted i n l ower shel l shrinkage and caused the strand to b i n d and hang momentar i ly i n the m o u l d w h i c h l ed to a sudden rise i n metal l eve l . In turn, this event tr iggered an increase i n the casting speed and released the strand i n a j e rk ing mot ion . 7.2 Sensing Meniscus Fluctuations with Thermocouples In the present study, thermocouples embedded i n the m o u l d w a l l above the meniscus were used to measure metal l eve l fluctuations indirect ly around the meniscus per iphery for Heat # 531142 i n tr ial D - l . In F igure 7.4 temperature fluctuations on opposite faces are compared w h i l e temperature responses across a g iven face are shown i n F igure 7.5. S ince the average metal l eve l pos i t ion was 182 m m f rom the top o f the m o u l d , the responses o f the m o u l d thermocouples located at the midface o f the four wa l l s at 175 m m were used to compare metal l eve l f luctuation on the four faces. A s shown i n F igure 7.4, the fluctuations on the opposite faces differ f rom one another and the magnitude o f the difference changes w i t h t ime. A signif icant difference can be noted when the temperature is r i s ing o n one face and fa l l ing on the other; and when s imi la r trends are seen, the rates o f r ise/fal l on the two faces are different or d isp laced f rom one another by a certain t ime interval . Signals f rom a set o f two off-corner 178 and one midface thermocouples located at 160 m m on the east face were used to compare the differences i n metal l eve l fluctuations across the east face again for Heat # 531142 i n tr ial D - l , F igu re 7.5. The loca l metal l eve l fluctuations at the two off-corner locat ions and the midface differ f rom one another, again i n a random manner. A n example o f var iable metal l eve l fluctuations across a g iven face o f the m o u l d can be seen i n the photograph o f the meniscus shown i n F igu re 7.6(a); i n compar ison , the photograph presented i n F igu re 7.6(b) w h i c h shows a fa i r ly quiet meniscus across the face. The basis for us ing m o u l d temperature as an indirect measure o f metal l eve l var ia t ion is quite l o g i c a l . F o r a g iven metal l eve l posi t ion, the m o u l d w a l l assumes a certain temperature prof i le w i t h the m a x i m u m value located roughly 25 m m be low the meniscus. W h e n the meniscus moves closer to a thermocouple located above the meniscus, the temperature o f the m o u l d i n the v i c i n i t y o f the thermocouple increases; on the other hand, when the metal l eve l fal ls , the l o c a l temperature decreases. A s expected, the magnitude o f the temperature change is propor t ional to the extent o f metal l eve l movement , as was seen i n Chapter 6. T h e thermal response o f the m o u l d 7 m m above the meniscus, corresponding to the metal l eve l s ignal plotted i n F igure 7.1(a) and (b), is shown i n F igure 7.7(a) and (b). T h e metal l eve l signals are superimposed on the thermal response to h ighl ight the fact that m o u l d thermal response provides a good indica t ion o f the state o f meniscus fluctuations. Ca lcu la t ions were performed us ing the m o u l d heat transfer mode l [7-10] to relate the measured metal l eve l ( in m m ) to temperature change (in °C) recorded by the thermocouple. A s described i n Chapter 6, metal l eve l change was s imulated i n the mode l by s imp ly m o v i n g the average measured heat f lux prof i le ax ia l ly up and d o w n the m o u l d over a desired t ime interval . T h e computed relat ionship between metal l eve l change (mm) and corresponding temperature change (°C) is plot ted i n F igure 7.8 for three different t ime intervals. A c c o r d i n g to this ca lcu la t ion , for the 179 observed temperature changes o f 10 -30°C seen i n F igure 7.7(a) and (b), the corresponding metal l eve l f luctuation is i n the range o f 5-10 m m . It is interesting to note that the var ia t ion i n temperature is greater i f the rise i n metal l eve l occurs over a longer t ime per iod since more t ime is avai lable for steady state to be achieved. 7.3 On-line Monitoring of Metal Level Fluctuations In a recent plant tr ial at C o m p a n y A [74], meniscus fluctuations were observed v i sua l ly dur ing the casting operation and were found to be l i nked to the shape o f the tundish stream, i.e., ropey or smooth. La rge variations i n metal l eve l were associated w i t h extremely rough tundish streams whereas the smal l deviations were indicat ive o f a smooth stream [74]. Fur thermore, at t r ia l D - 2 where meniscus fluctuations were moni tored on-l ine (using m o u l d thermocouples located above the metal level) by the S C A D A system [60 -62] , the l i n k between the roughness o f the tundish stream and meniscus fluctuations was also readi ly apparent. A l t h o u g h this observat ion is quite important, a more detailed study is needed to characterize the turbulence o f the tundish stream on the basis o f fluctuations i n m o u l d thermal response or metal l eve l sensor s ignal . T h e metal l eve l sensor s ignal and m o u l d thermocouple response above the meniscus for the two cases shown i n Figures 7.1 and 7.7, suggest that the standard devia t ion o f the s ignal can be a good index to characterize the nature o f metal l eve l fluctuations. H o w e v e r , it must be pointed out that the standard deviat ion method, w h i c h employs a s ingle number to characterize meta l l eve l fluctuations may not be adequate. It is interesting to note f rom Figures 7.1 and 7.7, that the metal l eve l fluctuations are indeed made up o f a number o f rises and falls i n the leve l . These fluctuations are quite important and it w i l l be demonstrated i n Chapter 9 that laps and bleeds can be triggered by the rise and 180 fa l l i n metal l eve l . Thus , i n cases where the number o f rise and/or fa l l events is important, it m a y be useful to moni tor the number and magnitude o f the events dur ing a certain per iod i n the heat, to characterize metal l eve l fluctuations, i n addi t ion to the standard deviat ion. O n the basis o f the above analysis, a technique to moni tor rise and fa l l i n metal l eve l was implemented i n the S C A D A system [62] to detect metal l eve l fluctuations, on- l ine . The methodology adopted is quite s imple . The first derivat ive o f the temperature (or metal level) data is first computed and the slopes are examined to detect locat ions o f zero values i n the slope. O n c e these locat ions are detected, the magnitude and duration o f the rise and fa l l i n metal l eve l are calculated and the informat ion is transferred to the expert system embedded i n the "intel l igent m o u l d " for further analysis. The details o f the method and its performance relat ive to some other techniques examined for pattern recogni t ion, were pursued i n a para l le l study [60-62] and w i l l not be discussed here. T h e above technique was appl ied to the metal l eve l signals presented earl ier i n F igu re 7.1 (a) and (b), to detect the number o f rises and falls i n metal l eve l ; the results are presented i n F igu re 7.9 (a) and (b). A s expected, the case w i t h smaller S D has a smal ler number o f rises and falls as compared to the case w i t h a higher S D . The number o f rises and falls i n the 5 to 10 m m range for the case wi th higher S D is almost double that o f the l ower S D . T h e number o f rises and falls observed i n the range 10-15 m m , is comparable for the two cases, a l though the number for the higher S D case is s l ight ly higher. Furthermore, a sma l l fraction o f metal l e v e l rises for the higher S D case is greater than 15 m m ; interestingly, the magnitude o f metal l eve l falls is a lways smaller than 15 m m for the case w i t h lower S D . T h e above analysis was repeated for the two thermocouple signals presented i n Figures 7.7(a) and (b). The results are presented i n F igure 7.10 (a) and (b). F o r the case w i t h sma l l metal l eve l fluctuations, a total o f 89 rises and falls were seen. O f these 89 fluctuations, about 181 68 ( -75 pet.) were less than 5 ° C and therefore, smal l . The number o f s ignif icant rises and fal ls , i.e. above 10°C were on ly 6 ( -7 pet.) and 5 ( -5 pet.) respectively. O n the other hand, i n the case w i t h severe metal l eve l fluctuations, a total o f 63 fluctuations were seen. T h e number o f s ignif icant rises and falls were 37 (59 pet.) and 33 (52 pet.) respect ively w h i l e the number o f rises and falls that were less than 5 ° C were on ly 19 (30 pet.) and 20 (32 pet.) respect ively. The durat ion o f metal l eve l rise and fa l l was also computed for the two cases. T h e results are shown i n F igure 7.11(a) and (b). In the case w i t h smal l metal l eve l fluctuations about 75 pet. o f the rises and falls happened w i t h i n 5 s and the remain ing 25 pet. occurred w i t h i n 5-10 s. In the case w i t h severe metal l eve l fluctuations, the number o f rises/falls that occurred between 0 to 5 s and 5 to 10 s were about 50 pet. each. 182 (a) 120 150 100 200 300 400 Time (s) 500 600 (b) 120 Standard Deviation 4.8 mm Figure 7.1 Signals f rom metal l eve l sensor at C o m p a n y C f rom (a) Heat # D 6 1 3 5 showing sma l l fluctuations and (b) Heat # A 2 8 1 9 3 showing large fluctuations hav ing standard deviations o f 1.4 m m and 4.8 m m respectively. (Note: T h e average metal l eve l is - 1 3 8 m m f rom the top o f the mould) . 183 (a) 39 0 100 200 300 400 500 600 Time (s) Figure 7.2 Cas t ing speed signals at C o m p a n y C corresponding to the two metal l eve l signals presented i n F igure 7.1(a) and (b). (Note: T h e average cast ing speed is - 3 3 . 5 mm/s) . 184 (a) 165 150 Time (s) 300 (b) 150 Time (s) 300 Figure 7.3 Patterns observed i n metal l eve l signals dur ing tr ial D - l for (a) Heat # 531142 -0.12 pet. carbon steel and (b) Heat # 531147 - 0.32 pet. carbon steel conta ining boron. (Note: The average metal l eve l for F igure 7.3(a) is - 1 8 0 m m f rom top o f m o u l d and - 1 6 5 m m f rom top o f m o u l d for F igure 7.3(b)). 185 (a) 200 250 300 Time (s) 350 400 (b) 200 250 300 Time (s) 350 400 Figure 7.4 Differences i n meniscus fluctuations between opposite faces determined f rom the response o f m o u l d thermocouples located 7 m m above the meniscus at the midface on the four faces for 0.12 pet. carbon steel (Heat # 531142) cast dur ing t r ia l D - 1 . (Note: T h e average metal l eve l is - 1 8 2 m m f rom the top o f the mould . ) 186 (a) 65 Left Off-corner Midface 35 200 250 300 Time (s) 350 400 (b) 65 Left Off-corner Right Off-corner O 35 200 250 1 300 Time (s) 350 400 Figu re 7.5 Differences i n meniscus fluctuations between off-corner and midface sites on the east face determined f rom the response o f m o u l d thermocouples located 22 m m above the meniscus at the midface on the four faces for 0.12 pet. carbon steel (Heat # 531142) cast dur ing tr ial D - l . (Note: The average metal l e v e l is - 1 8 2 m m f rom the top o f the mould. ) 187 (a) (b) F igure 7.6 Photograph o f the meniscus region showing (a) turbulent and (b) fa i r ly quiet metal leve l . 188 (a) 150 ° i l 2 5 <D k. 3 •*—< re 0) Q.100 E Oi H O 75 50 Temperature — Metal Level Standard Deviation 2°C 120 130 150 100 200 300 Time (s) 400 500 600 (b) 150 — Temperature 100 200 300 Time (s) 400 500 120 600 Figu re 7.7 M o u l d thermocouple response at 7 m m above the meniscus corresponding to the two metal l eve l signals f rom C o m p a n y C presented i n F igu re 7.1 (a) and (b). (Note: T h e average metal l eve l is - 1 3 8 m m f rom the top o f the mould) . 189 (a) 150 Change in Mould Temperature (°C) (b) 300 Change in Mould Temperature (°C) Figure 7.8 Es t ima t ion o f metal l eve l f luctuation ( in m m ) at the midface f r o m the m o u l d thermocouple response 7 m m above the meniscus ( in °C) for (a) thermocouple locat ion ~8 m m f rom hot face and (b) hot face corresponding to F i g u r e 7.7. 190 (a) (b) F igu re 7.9 Frequency distr ibution o f (a) rise and (b) fa l l i n metal l eve l observed i n the metal l eve l sensor signals f rom trial C - l presented i n F igu re 7.1(a) and (b). 191 (a) F igure 7.10 Frequency distr ibution o f temperature change measured dur ing metal l eve l rise and fa l l by a thermocouple located 7 m m above the meniscus for the two thermocouple signals f rom trial C - l presented i n F igure 7.7(a) and (b). 192 (a) 100 (b) 5- 5-10 10-15 Duration of Temperature Change (s) 100 5- 5-10 10-15 Duration of Temperature Change (s) Figu re 7.11 Frequency distr ibution o f duration o f metal l eve l rise and fa l l measured by a thermocouple located 7 m m above the meniscus for the two thermocouple signals f rom trial C - l presented i n F igure 7.7(a) and (b). 193 C H A P T E R 8 - LAPS AND BLEEDS IN B I L L E T CASTING 8.1 Detection of Laps and Bleeds using Thermocouples T h e real-t ime thermal response o f the m o u l d was employed i n previous studies to detect surface defects such as transverse depressions i n bi l lets [21] and long i tud ina l depressions i n b looms [31]. A s indicated earlier, the presence o f surface depressions causes the mould/s t rand gap to w i d e n and loca l l y reduce the heat f lux [21,31]. The result is a "va l ley" i n the m o u l d temperature response w h i c h moves d o w n the m o u l d wi th the depression at the wi thd rawa l speed [21,31]. L a p s , w i t h the sub-surface structure shown i n F igure 5.4, also cause a "depression-l ike" w i d e n i n g o f the mould/strand gap. S ince m o u l d heat transfer i n the upper reg ion o f the m o u l d is sensit ive to changes i n the mould/strand gap, the same technique can therefore be appl ied to detect the format ion o f laps i n the m o u l d . The size o f the gap is smaller for laps than for transverse depressions; thus, the temperature drop is expected to be smal ler but can be detected. T h e temperature response o f the two heats f rom C o m p a n y C , a 0.84 pet. carbon grade exh ib i t ing laps and a 0.46 pet. carbon grade where neither laps nor bleeds were a p rob lem, was analysed to compare the patterns i n the two thermal histories. T h e bi l le t surfaces for the two grades were presented earlier i n F igure 5.2 and 5.3 a long w i t h the corresponding m o u l d thermocouple responses i n Figures 5.58(a) and (b). A s shown i n these figures, the m o u l d thermal response for the 0.84 pet. carbon heat exhibi ted a larger number o f "val leys" as compared to that for the 0.46 pet. carbon steel. The temperature drops were analysed for the two heats at four locat ions i n the m o u l d ; the results are presented i n F igure 8.1(a) and (b). A s s h o w n i n the f igure, over 90 pet. o f the temperature drops for the 0.46 pet. carbon heat where laps and bleeds were not a p rob lem, are i n the range 5 -10°C and the remain ing values are less than 15°C. F o r 194 the 0.84 pet. carbon steel w i t h laps, the temperature drops are larger. F o r example , on ly 65 pet. o f the temperature drops recorded by the thermocouple located at - 2 1 2 m m be low the meniscus , are i n the 5 - 1 0 ° C range w h i l e over 10 pet. o f the drops are greater than 15°C. F r o m the above compar i son , it appears that the presence o f laps (and the associated depression) on the surface o f the strand can generate fluctuations i n m o u l d temperatures that can be detected w i t h m o u l d thermocouples. A n in-depth analysis o f the formation and propagation o f laps (and depressions associated w i t h laps) at C o m p a n y A (trial A - l ) dur ing the casting o f 1.02 pet. carbon steel was undertaken by examin ing the response o f thermocouples located at 8 m m , 113 m m , 217 m m and 368 m m b e l o w the meniscus. L o c a l meniscus fluctuations on the same face were moni to red w i t h a thermocouple located 7 m m above the average metal l eve l . A l l these thermocouples were instal led at the midface o f the m o u l d w a l l at a distance o f about - 7 m m f rom the hot face. A total o f 17 laps manifested as "val leys" i n the temperature data, were detected dur ing a 600 s per iod for this heat. F igure 8.2 shows the thermocouple response corresponding to the format ion o f a lap (#8 i n Tables 8.1 and 8.2) and its propagation d o w n the m o u l d . T h e metal l eve l falls at around 232 s after a rise, as indicated by the response o f the thermocouple located above the meniscus , and a "va l l ey" o f about 3 5 ° C appears i n the m o u l d temperature recorded 8 m m b e l o w the meniscus. T h i s "va l ley" i n the temperature data propagates d o w n the m o u l d w a l l and is detected, after a t ime delay, by subsequent thermocouples situated at 113 m m , 217 m m and 368 m m b e l o w the meniscus. The speed o f the downward movement o f the "val leys" was calculated to be roughly equal to the casting speed ( -25 to - 3 2 mm/s) as shown i n Table 8.2. T h e frequency distr ibution o f temperature drops registered by the four thermocouples located be low the meniscus is shown i n F igure 8.3 for the 17 laps (or "va l leys" i n m o u l d 195 temperature) detected; most o f the temperature drops are between - 1 0 and ~ 3 5 ° C . T h e t ime interval o f temperature drop w h i c h is a measure o f the w id th o f depressions (associated w i t h the laps) on the b i l le t surface, lies between - 1 0 and - 2 5 s for most laps, F igu re 8.4. F igu re 8.5 shows the temperature changes measured by the thermocouple located - 7 m m above the meniscus, just pr ior to the formation o f laps and reveals that each event is associated w i t h a rise and fa l l i n metal l eve l . Thus , l oca l metal l eve l fluctuations are c r i t i ca l i n the generation o f laps on the b i l le t surface. A s shown i n F igure 8.6, the temperature change associated w i t h most metal l eve l fluctuations is - 5 to ~ 2 5 ° C whereas, as indicated i n F igure 8.7, the duration o f the rise and fa l l is - 5 to - 2 5 s. W i t h respect to bleeds, a s imi la r strategy can be employed because bleeds are also often associated w i t h a "surface depression" w h i c h lead to w iden ing o f the mould/s t rand gap. H o w e v e r , due to the sma l l size and distr ibution o f bleeds across a g iven face, F igure 5.6, m o u l d thermocouples located at the midface may not be adequate for the detection. Thus , it may be preferable s imp ly to moni tor metal l eve l fluctuations and m o u l d w a l l temperatures and mainta in values w i t h i n a desired range where bleed formation is m i n i m a l . 8.2 Mechanism of Formation of Laps and Bleeds It is apparent f rom Figure 8.2 that laps originate very close to the meniscus since the thermocouple located 8 m m be low the meniscus shows a drop i n m o u l d temperature at the t ime o f lap formation. A s mentioned earlier, the drop i n m o u l d temperature occurs because the w i d t h o f the mould / she l l gap increases due to the depression associated w i t h the lap shown i n F igure 5.4. A s the lap (and the associated depression) moves d o w n the m o u l d at a speed equal to the wi thd rawa l rate, the successive thermocouples register a "va l ley" i n m o u l d temperature shown i n F igu re 8.2. W i t h respect to the behaviour o f the meniscus at the t ime o f lap format ion , it is 196 clear that a fa l l i n metal l eve l triggers the lap seen i n F igure 8.2. T h e importance o f lubr ica t ion i n the generation o f laps and bleeds is evident f rom C o m p a n y E where poor maintenance o f the o i l dis tr ibution system and manual feeding o f o i l created condi t ions where the hot face o f the m o u l d was starved o f the lubricant . T h e p r o b l e m o f poor lubr ica t ion and s t ick ing is evident f rom the photographs o f laps and bleeds, shown i n Figures 5.1 and 5.6 respectively, w h i c h also exhibi t distorted osc i l l a t ion marks. T h e situation o f poor lubr icat ion can also arise when the o i l weep ing d o w n the m o u l d w a l l vaporises before it reaches the meniscus region because the m o u l d hot face temperature is higher than the b o i l i n g range o f the lubr icat ing o i l ( -205 to ~ 3 3 5 ° C ) . The sudden increase i n the frequency o f laps and bleeds observed at C o m p a n y D [71], due to b reakdown o f the c o o l i n g water treatment system, is obv ious ly related to an excess ively h igh hot face temperature generated as a result o f scale deposi t ion on the c o l d face o f the m o u l d [71]. Thus , the hot face temperature o f the m o u l d is c r i t ica l f rom the standpoint o f lubr ica t ion and s t i ck ing . T h e influence o f m o u l d taper on the severity o f laps and bleeds i n b i l le ts , as shown i n F igu re 5.9, is l i n k e d to its strong effect on m o u l d heat extraction [1,4-5], F igu re 8.8, and therefore, to the hot face temperature o f the m o u l d . The m o u l d hot face temperature was calculated f rom measured temperature data i n plant trials at companies C and E us ing a m o u l d heat transfer mode l [7-10]. A s a result, two types o f m o u l d operation can be ident i f ied as important i n the formation o f laps and bleeds - denoted as the hot m o u l d and c o l d m o u l d i n F igure 8.9(a) and (b). A t C o m p a n y C , where laps and bleeds were a p rob lem, a "hot m o u l d " operat ion was i n effect since the peak hot face temperature o f the m o u l d was greater than the b o i l i n g range o f the lubr icat ing o i l ( -205 to 3 3 5 ° C ) . O n the other hand, at C o m p a n y E where the laps and bleeds were considerably less severe, a " co ld m o u l d " was i n operat ion since the peak hot face temperature o f the m o u l d w a l l was lower than the b o i l i n g range o f the o i l . The 197 difference between the moulds at the two companies can be attributed to the d i s s imi la r tapers at the meniscus , Table 4.7, w i t h C o m p a n y C having a sha l low taper, 0.4 pc t . /m [1,4-5] and the highest heat extraction, F igure 8.8. The reason g iven for the s ignif icant increase i n heat transfer i n moulds w i t h sha l low taper i n the meniscus region, is the enhanced mechanica l interact ion dur ing the operation between the negatively tapered m o u l d and the so l id fy ing shel l dur ing the negative-strip per iod o f the osc i l la t ion cyc le [1,4-5]. In addi t ion to m o u l d taper, lubr icat ing o i l also influences m o u l d heat transfer and hot face temperature as can be seen i n F igure 8.10. Thus , an enhancement i n m o u l d heat extract ion at C o m p a n y C o f about 20 percent was observed when the o i l supply was increased f rom a near-zero value to about 100 m l / m i n . F igure 8.11 shows the effect o f o i l f low-rate on the enhancement i n heat transfer i n the meniscus region [5]. A t C o m p a n y C , where the m o u l d is "hot", the lubr ica t ing o i l pyrolyses more readi ly producing hydrogen-r ich vapour w h i c h enhances the thermal conduct iv i ty o f gases i n the gap [1,4-5]. S ince the mould / she l l gap constitutes the largest resistance to heat f l ow i n the meniscus region [1,4-5], the l o c a l heat transfer rate is enhanced. The extent to w h i c h oi l -enhanced heat transfer is a funct ion o f peak hot face temperature o f the m o u l d is shown i n F igure 8.12 where the m o u l d peak heat f lux is plotted as a function o f peak hot face temperature for various m o u l d c o o l i n g systems under different operating condi t ions; the o i l b o i l i n g range is superimposed on the plot to h ighl ight the zone o f oi l -enhanced heat transfer. The figure suggests that the greatest enhancement i n heat transfer occurs i n the b o i l i n g region o f the lubr icat ing o i l . A l t h o u g h the effect o f m o u l d taper at the meniscus on the hot face temperature o f the m o u l d is strong, other operating parameters are also important. In C o m p a n y D where a steep m o u l d taper o f 3.5 pct . /m was employed and a "co ld" m o u l d operation is expected, the peak hot face temperature dur ing the casting o f h igh carbon steels was calculated to be about 2 9 0 ° C 198 w i t h a c o o l i n g water ve loc i ty o f 10.5 m/s and a w a l l thickness o f 15.6 m m . Thus , the m o u l d was "moderately hot" and s t ick ing problems and laps/bleeds were occas iona l ly seen under these condi t ions . Calcu la t ions were performed to simulate the operation for an increased c o o l i n g water ve loc i ty o f 15 m/s. A s shown i n F igure 8.13, the hot face temperatures c o u l d be lowered f rom 2 9 0 ° C to be low the b o i l i n g range o f o i l ; and thus, lubr ica t ion and s t i ck ing problems c o u l d be suppressed. A s is evident f rom Figure 8.12, a reduction i n hot face temperature due to the increased c o o l i n g water ve loc i ty suppresses any enhancement i n heat extract ion due to o i l b o i l i n g and pyro lys i s w h i c h lowers the hot face temperature even further. Thus , by a s imple increase i n the c o o l i n g water ve loc i ty , the m o u l d operation can be transformed f r o m "moderately hot" to "co ld" . T h e va l id i ty o f this ca lcula t ion can be seen i n F igure 8.13 w h i c h shows that at C o m p a n y E where a steeply-tapered m o u l d was employed together w i t h a h igh c o o l i n g water ve loc i ty o f 18 m/s, the peak hot face temperature was on ly ~ 2 1 0 ° C . T h e role o f osc i l l a t ion frequency on the severity o f laps and bleeds at C o m p a n y A is also related to its effect on the hot face temperature o f the m o u l d . Osc i l l a t i on frequency influences the peak hot face temperature v i a its influence on negative strip t ime, mould / she l l interaction at the meniscus , and the nature o f osc i l la t ion marks (depth and pi tch). The cases examined were 144 c p m ( t„=0.16) and 100 c p m (t n=0.20) i n tr ial C - l , and 160 c p m ( t„=0 .18 s) and 100 c p m (t n=0.25s) i n tr ial A - 2 . In tr ial C - l , as the frequency was reduced f rom 144 c p m to 100 c p m , an increase i n m o u l d heat extraction was observed w h i c h l ed to a higher peak hot face temperature [5]; this increase i n m o u l d heat transfer was attributed to enhanced m o u l d shel l interaction as a result o f the longer negative strip t ime [1,4-5]. In t r ia l A - 2 , however , as the frequency was decreased f rom 160 c p m to 100 c p m , the m o u l d heat transfer decreased by about 15 pet. and the peak hot face temperature dec l ined about 5 0 ° C f rom ~ 3 2 5 ° C to ~ 2 7 5 ° C . Thus , the m o u l d temperature was transformed f rom "hot" to "moderately hot". T h i s reduct ion i n heat transfer may be due to an increase i n the depth o f osc i l l a t ion marks , and therefore, a wider 199 mould/s t rand air gap. F r o m the above f indings, it appears that there are two oppos ing factors at p lay namely mould / she l l interaction at the meniscus and depth o f osc i l l a t ion marks. W h i l e it is k n o w n that increased mould /she l l interaction at the meniscus enhances l o c a l heat transfer i n the m o u l d , it is also true that increased mould / she l l interaction results i n deeper osc i l l a t ion marks (Figure 8.14) [23] w h i c h w i l l cause a reduction i n heat transfer especia l ly i n the upper part o f the m o u l d where the w id th o f the mould/strand gap governs heat transfer. It is very l i k e l y that the depth o f osc i l la t ion marks at C o m p a n y A becomes more s ignif icant f r o m the standpoint o f heat transfer, as compared to increased mould / she l l interaction, when t n is changed f r o m 0.18 s to 0.25s; unfortunately, osc i l la t ion mark depth c o u l d not be measured on the b i l le t samples at C o m p a n y A since the strand surface was ro l l ed by foot-rolls present c lose to the m o u l d exit . O n the other hand, at C o m p a n y C , where the range o f t n is 0.16 - 0.20 s, the effect o f mould / she l l interaction on heat transfer is greater than that o f depth o f osc i l l a t ion marks . T h e carbon content o f the steel also influences heat extraction i n the m o u l d [15,22]. A s s h o w n i n F igure 6.3, the peak heat f lux for various grades evaluated at C o m p a n y D is the highest for the 0.84 pet. carbon steel ( -6500 k W / m 2 ) f o l l owed by 0.32 pet. carbon steel ( - 4 5 0 0 k W / m 2 ) and is the lowest for 0.12 pet. carbon steel ( -2500 k W / m 2 ) . The peak hot face temperature fo l lows a s imi la r trend; it is the highest for the 0.84 pet. carbon grades ( ~ 2 9 0 ° C ) and the lowest for the 0.12 pet. carbon grade ( ~ 1 5 0 ° C ) ; the temperature for the 0.32 pet. carbon steel is about 2 5 0 ° C . Thus , the m o u l d operates "co ld" for the 0.12 pet. carbon steel w h i l e it is "hot" for the 0.32 pet. and 0.84 pet. carbon steels. The severity o f laps and bleeds i n h igh carbon steels may be partly related, then, to the higher heat transfer as compared to other grades. It is apparent f rom the above f indings that the formation o f laps is related to the behaviour o f the meniscus and the hot face temperature o f the m o u l d relat ive to the b o i l i n g range o f lubr ica t ing o i l . Thus , a mechanism rooted i n metal l eve l fluctuations, and their impact on 200 lubr ica t ion and m o u l d heat transfer i n the meniscus region, is proposed to exp la in the formation o f laps. A schematic d iagram o f the proposed mechan i sm for lap format ion is shown i n F igure 8.15. In operating "hot" moulds , where the peak hot face temperature exceeds the b o i l i n g range o f the lubr ica t ing o i l , a sudden rise i n metal l eve l causes the lubr ica t ing o i l to vapourise and leave a dry, unlubricated region on the m o u l d w a l l beneath the meniscus. Subsequent ly, when the metal l eve l drops, the meniscus contacts the unlubricated (or par t ia l ly lubricated) region and sticks on the m o u l d face because the lubr icat ing o i l descends more s l o w l y than the d ropping meniscus. T h i s phenomenon was evident i n videos o f the hot face o f the m o u l d i n the meniscus reg ion w h i c h shows that whenever the metal l eve l falls, it leaves the o i l f i l m behind. The s t i ck ing is accompanied by h igh heat transfer w h i c h gives rise to a part ial so l id i f ica t ion o f the meniscus observed i n F igure 5.4 and formation o f a lap on the b i l le t surface upon over f low o f l i q u i d steel. T h e mechan i sm proposed for the formation o f bleeds as shown schemat ica l ly i n F igure 8.16, is related to the same events that trigger the formation o f laps on the b i l le t surface. H o w e v e r , the bleeds f o r m be low the meniscus due to tearing o f the so l id shel l w h i l e the laps f o r m at the meniscus due to over f low o f l i q u i d steel. The so l id shel l can tear/rupture because o f the p u l l i n g force exerted on the strand by the wi thdrawal rol ls dur ing the per iod when the she l l is stuck to the m o u l d . T h e shel l rupture causes the l i q u i d steel to f l ow out and so l id i fy on the m o u l d surface as a b leed w h i c h is evident i n the photograph shown i n F igure 5.7. T h e effect o f variables such as m o u l d taper at the meniscus, c o o l i n g water ve loc i ty and qual i ty , osc i l l a t ion frequency and maintenance o f o i l dis t r ibut ion system on the severity o f laps and bleeds observed i n the plant trials at the various companies is consistent w i t h the proposed mechanism. T h e effect o f steel carbon content expla ined on the basis o f the longer freezing range and the result ing weak so l id shel l at the meniscus, is also consistent w i t h the above 201 mechanism. 8.3 Minimizing Laps and Bleeds in Billet Casting B a s e d o n the d iscuss ion presented i n the preceding section, f o l l o w i n g are some o f the major recommendat ions suggested to m i n i m i z e the formation o f laps and bleeds i n bi l le ts : (i) S ince random metal l eve l fluctuations trigger the format ion o f laps and bleeds, it is important to m i n i m i z e the turbulence i n the ex i t ing tundish streams by ensur ing proper f l ow i n the tundish and isolat ing turbulence introduced by the ladle stream us ing f low cont ro l devices [75]. The tundish -to- m o u l d height must be as sma l l as possible to m i n i m i z e growth o f instabili t ies on the open-pour stream. T h e use o f submerged entry nozzles , especia l ly for larger section sizes, w i l l help m i n i m i z e frequent metal l eve l fluctuations. T u n d i s h nozz le c logg ing , due to the presence o f s o l i d inc lus ions i n the steel also may create disturbances on the tundish stream [75]. (i i) A "co ld m o u l d " practice, where the peak hot face temperature o f the m o u l d is lower than the b o i l i n g range o f lubr icat ing o i l , must be adopted. The m a i n requirements to achieve this condi t ion include a steep taper at the meniscus o f about 2.0 pc t . /m together w i t h a c o o l i n g water ve loc i ty greater than - 1 4 - 1 5 m/s and excel lent c o o l i n g water qual i ty . ( iv) T h e o i l lubr ica t ion system must be designed and maintained to de l iver a un i fo rm f low o f o i l around the m o u l d periphery [16,28]. (v) T o achieve adequate str ipping action and m i n i m i z e s t ick ing o f strand i n the m o u l d , a negative strip t ime that is maintained between 0.12 and 0.16 s is r ecommended [16,23]. 202 (vi) A M n : S ratio lower than about 25:1 leads to poor h igh temperature strength o f steel [76] w h i c h aggravates problems o f tearing o f the so l id shel l . Thus , a M n : S ratio o f steel greater than about 25:1 may be beneficial to m i n i m i z e bleed format ion. In addi t ion to a l l the above issues, the design o f metal l eve l control system c o m m o n l y emp loyed i n b i l le t cast ing is also cr i t ica l . The current design w h i c h maintains metal l eve l at a desired pos i t ion by altering the casting speed, is not desirable since changes i n cast ing speed inf luence osc i l l a t ion characteristics (negative strip t ime and m o u l d lead) and also, disturb the metal l eve l . In a recent plant tr ial at C o m p a n y D where negative strip t ime was moni tored on- l ine , it was found that the negative strip t ime var ied s ignif icant ly f rom the set average value o f about 0.14 s as a result o f this design [62]. Furthermore, i n some m i n i - m i l l s , the cast ing speed is also l i n k e d to osc i l l a t ion frequency; i n this case the osc i l l a t ion frequency also becomes dependent on metal l eve l fluctuations. The system that is desired includes a constant metal l eve l and a constant casting speed that is isolated f rom the osc i l la t ion frequency. W i t h such a system, it w i l l be possible to maintain a constant negative strip t ime and m o u l d lead dur ing the cast ing operation. 203 Tab le 8.1 Characterist ics o f temperature drops or "val leys" associated w i t h laps ( N o . 6 to 15) at C o m p a n y A determined f rom the response o f m o u l d thermocouples located be low the meniscus. L a p # D i s t a n c e o f T h e r m o c o u p l e s b e l o w M e n i s c u s ( m m ) 8 113 217 368 6 Temperature D r o p (°C) 7 11 12 9 Transi t o f D r o p (s) 20 12 20 20 7 Temperature D r o p (°C) 15 19 23 20 Transi t o f D r o p (s) 14 24 26 24 8 Temperature D r o p (°C) 34 24 27 19 Transi t o f D r o p (s) 34 19 24 24 9 Temperature D r o p (°C) 16 14 17 14 Transi t o f D r o p (s) 22 18 20 22 10 Temperature D r o p (°C) 15 13 23 15 Transi t o f D r o p (s) 11 12 12 12 11 Temperature D r o p (°C) 34 15 17 14 Transi t o f D r o p (s) 12 16 20 16 12 Temperature D r o p (°C) 28 34 34 24 Transi t o f D r o p (s) 20 32 30 20 13 Temperature D r o p (°C) 22 18 14 11 Transi t o f D r o p (s) 18 24 22 20 14 Temperature D r o p (°C) 15 13 16 12 Transi t o f D r o p (s) 14 14 16 14 15 Temperature D r o p (°C) 31 19 19 13 Transi t o f D r o p (s) 14 20 16 14 204 Tab le 8.2 M o u l d temperature change (rise and fall) 7 m m above the meniscus corresponding to format ion o f laps (nos. 6 to 15 i n Table 8.1) and the inferred speed o f temperature drops or "val leys" d o w n the m o u l d . Lap # Temperature Rise CC) Temperature Fall CO Time of Rise (s) Time of Fall (s) Casting Speed (mm/s) 6 4 4 5 5 26 7 2 10 5 10 32 8 6 19 10 20 26 9 6 21 6 30 32 10 4 2 5 5 25 11 4 16 7 17 26 12 8 14 10 10 26 13 13 13 18 16 26 14 7 5 15 8 32 15 4 21 8 15 28 205 (a) 100 75 u g 50 3 cr 25 +• 5-10 + 162 mm E 3 + 212 mm |§1 + 3 0 2 ™ D + 4 1 2 m m _l 1 1 1 10-15 15-20 20-25 25-30 30-35 Temperature Drop (°C) 35+ (b) 100 + 162 m m S t 212 m m | + 302 mm [ | + 412 mm 5-10 10-15 15-20 20-25 25-30 Temperature Drop (°C) 30-35 Figure 8.1 Frequency dis tr ibut ion o f temperature drops registered by four thermocouples located at 162, 212, 302, and 412 m m be low the meniscus dur ing the casting o f (a) H e a t — D 6 1 3 5 - 0.46 pet. carbon steel and (b) Heat - D 6 1 3 1 - 0.84 pet. carbon steel dur ing t r ia l C - l . (Note: T h e average metal l eve l is - 1 3 8 m m f rom the top o f the mould) . 206 120 I I 200 210 220 230 240 250 260 270 200 Z » 3O0 Hmt(i) Figure 8.2 Thermocouple response above and be low the meniscus dur ing the format ion and travel o f a depression associated w i t h a lap for 1.02 pet carbon steel i n tr ial A - l (Heat # E25990) . (Note: T h e average metal l eve l is - 1 1 4 m m f rom the top o f the mould) . 207 80 Temperature Drop (°C) Figure 8.3 Frequency distr ibution o f temperature drops registered dur ing the detection o f laps by four thermocouples located at 8, 113 ,217 and 368 m m b e l o w the meniscus for 1.02 pet carbon steel i n tr ial A - l (Heat # E25990) . (Note: The average metal l eve l is ~114 m m f rom the top o f the mould) . 208 80 Interval of Temperature Drop (s) Figure 8.4 Frequency distr ibution o f interval o f temperature drops registered dur ing the detection o f laps by four thermocouples located at 8, 1 1 3 , 2 1 7 and 368 m m be low the meniscus for 1.02 pet carbon steel i n tr ial A - l (Heat # E25990) . (Note: T h e average metal l eve l is ~114 m m f rom the top o f the mould) . 209 Figure 8.5 M e t a l l eve l rise and fa l l associated w i t h each lap, measured by a thermocouple located 7 m m above the meniscus for 1.02 pet carbon steel i n tr ial A - l (Heat # E25990) . (Note: T h e average metal l eve l is ~114 m m f rom the top o f the mould) . 210 60 5- 5-10 10-15 15-20 20-25 25-30 30+ Temperature Change (°C) Figu re 8.6 Frequency distr ibution o f temperature change measured dur ing metal l eve l rise and fa l l by a thermocouple located at 7 m m above the meniscus for 1.02 pet carbon steel i n tr ial A - l (Heat # E25990) . (Note: T h e average metal l eve l is 114 m m f rom the top o f the mould) . 211 5-10 10-15 15-20 20-25 25-30 Duration of Meniscus Rise and Fall (s) 30+ Figure 8.7 Frequency dis tr ibut ion o f duration o f metal l eve l r ise and fa l l measured by a thermocouple located 7 m m above the meniscus for 1,02 pet carbon steel i n tr ial A - l (Heat # E25990) . (Note: The average metal l eve l is 114 m m f rom the top o f the mould) . 212 (a) Mould Taper at Meniscus (pct./m) Figure 8.8 Effect o f in i t i a l m o u l d taper at the meniscus on (a) measured peak heat f lux and (b) specif ic m o u l d heat extraction [1]. 213 (a) 500 120 130 140 150 160 170 180 190 Distance from Top of Mould (mm) (b) 500 O 400 0J t_ 3 2 300 8. E OJ 200 0) u CJ li. O 100 X Oil Boiling 90 pct.(-335°C) Oil Boiling Start (~205°C) Metal Level -115 mm 4-100 110 120 130 140 150 Distance from Top of Mould (mm) H 1-160 170 Figure 8.9 M o u l d hot face temperature profiles relative to the b o i l i n g range o f o i l for (a) hot m o u l d operation at C o m p a n y C and (b) c o l d m o u l d operation at C o m p a n y E . 214 275 H 1 1 1 1 1 1 h 100 150 200 250 300 350 400 450 500 550 600 Time (s) Figu re 8.10 Oi l -enhanced heat transfer measured us ing m o u l d thermocouples located i n the meniscus region at C o m p a n y C 25 70 Oil Flowrate (ml/min) 100 Figure 8.11 Effect o f o i l f l ow rate on the measured heat extract ion i n the m o u l d at C o m p a n y C [ 5 ] . 215 0 td m cd PL. Oil Liquid Boiling Range of Oil Oil Vapour 1 0 0 0 0 6 0 0 0 7 0 0 0 4 0 0 0 2 0 0 0 1 0 0 0 Zone of o i l - e n h a n c e d heat transfer 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 o Company A V Company B • Company C A Company D o Company E Fitted 3 5 0 4 0 0 4 5 0 5 0 0 P e a k H o t F a c e T e m p e r a t u r e , C Figure 8.12 Peak heat f lux plotted as a function o f peak m o u l d hot face temperature showing the zone o f oi l -enhanced heat transfer. 216 400 O 0 d> 5 300 Q. E o H o u co 200 o X ^ co o 0. 100 Canola Oil - Boiling 90 pet. 290°C Reduction in hot face temperature due to increased water velocity Company E - 200-205°C BB Canola Oil - Boiling Start 4-175°C 4 - 4 -10 12 14 16 18 Mould Cooling Water Velocity (m/s) 20 Figure 8.13 Effect o f higher c o o l i n g water ve loc i ty on the peak hot face temperature o f the m o u l d for the casting o f 0.84 pet. carbon steel at C o m p a n y D . 0.20 E (fl (0 0.15 c o '•ZJ ra o O 0.10 o JZ +-* a o a 0.05 • Midface • Off-corner O Off-corner ra . p . ra d5 .CD. ra CD H 1 1 1 1 h H h -+—4-0.12 0.14 0.16 0.18 0.20 0.22 0.24 0.26 0.28 0.30 Negative Strip Time (s) Figure 8.14 Effect o f negative strip t ime on the measured depth o f osc i l l a t ion marks [23]. 217 T 3 2 2 O © *j « P c *» a « ? © w © <e a> a) — « c5 w 2 S U T <" 3 <•> 3 "3 "o © <« •?c S 3 ii S «i •£ cu 13 O e cs cs C «H „ g M S 1 .5 £ « a w V !t 2 w A a. a 6X) e "S J3 S J2 60 © "5 2 "w>«3 © cu w 3 d V > 3 13 a o = -0 35  2 5 - a * IU r> -« iu A 2-a *3 © + J "E "3 "3 3 cs 8 .a „ © CS _ © cu - © U 2 3 3 3 <U a o T3 C c3 > 'ci 1 o •c O •4—» <D 3 TJ e o CL, c3 X5 O u CO O 2 CL, IT) oo CJ U l 3 218 O 3 ~ . cu JJ S v O et e r s «T2 o O 3 t-i to <-» to «i i s U u «> <l x J > X S «> » SJ U " u 1-cS »- es— CU es "2 C J C U C -o es 3 * a s ! « D o u a — S3.32 8 ts a u «j u « S J = S * — a cj « a) 3 ^£ 2 *d o *3 & T3 T3 C es u 2 •S38"-. S o u — o-c 2=5 •§ 2 3 3 CO es. > T3 a > 4) OJ OJ •c OJ 3 T3 C .2 OJ OJ o fi o OJ n . OJ ofl C/> (-1 o .5 o o U H C A 1-H OO OJ u, 3 00 2 219 C H A P T E R 9 - RHOMBOIDITY IN B I L L E T CASTING 9.1 Detection of Rhomboidity A s pointed out earlier, rhomboid i ty usual ly begins i n the m o u l d due to non-un i fo rm c o o l i n g as a result o f intermittent nucleate b o i l i n g i n the c o o l i n g water channel [7-10] and other reasons i nc lud ing non-uni form m o u l d distort ion and m o u l d tube misa l ignment . W h e n non-un i fo rm c o o l i n g condit ions exist i n the m o u l d , the bi l le t ex i t ing the m o u l d has a non-un i fo rm s o l i d shel l [22]. T h e bi l le t at this point is reasonably square since it must c o n f o r m to the shape o f the m o u l d [22]. In the sprays, the bi l le t undergoes variable c o o l i n g due to the non-un i fo rm shel l thickness [22]. The c o l d corners have a thicker so l id shel l and experience a more rapid c o o l i n g rate as compared to the hot corners that have a thinner so l id shel l [22]. T h i s impl ies that rhomboid i ty can be generated even w i t h a perfectly designed spray c o o l i n g system. Thus , a non-un i fo rm s o l i d shel l thickness appears to be an important parameter w i t h respect to the rhomboid i ty p rob lem i n b i l le t casting. It is evident f rom the calculat ions performed i n Chapter 6 on thermal resistances to heat f l o w i n b i l le t moulds , that the mould/strand gap resistance dominates i n the upper part o f the m o u l d and the resistance o f the so l id shel l begins to exert some influence l o w e r d o w n i n the m o u l d (Figures 6.10 to 6.12). These calculat ions indicate that thermocouples embedded i n the m o u l d cannot be effect ively used to detect variations i n shel l thickness since the thermal resistance offered by the so l id shel l is m u c h smaller than the gap resistance. T h e m o u l d parameter that can be moni tored is the heat transfer i n the meniscus region that actual ly generates a non-uni form so l id shel l . It is evident f rom F igure 6.3 that m o u l d heat f lux is the highest i n the v i c in i t y o f the meniscus but drops to a lower , stable value i n the regions 220 b e l o w the meniscus. It is be l ieved that the non-uni form s o l i d shel l generated i n the m o u l d has its o r ig in c lose to the meniscus, where the heat extraction is the highest. Thus , m o u l d temperatures measured dur ing trials D - l and D - 2 , were examined to assess i f any correlat ion exis ted between the nature (uniformity) o f heat extraction i n the meniscus region and the rhomboid i ty values measured on bi l le t samples. T h e measured temperatures on the four faces o f the m o u l d at about 25 to 50 m m b e l o w the meniscus ( locat ion o f peak hot face temperature) were plotted against rhomboid i ty values measured on bi l le t samples col lected dur ing trials D-1 and D - 2 . The temperature data considered for this analysis corresponded to the t ime when the part icular b i l le t sample was i n the m o u l d . T h e data acquired at 100 H z at tr ial D - l and 60 H z at t r ia l D - 2 , were first reduced to 5 H z for ease o f handl ing us ing the computer program [68] discussed i n Chapter 4. A s ment ioned earlier, this frequency was found to be adequate for analysing thermocouple signals. T h e measured temperature on the four faces o f the m o u l d was analysed and two values ( m a x i m u m and m i n i m u m ) were extracted f rom the data, every 0.2 s i n the 5 H z data. The two values were averaged over a per iod o f 300-400 s and u t i l i zed i n the analysis o f rhomboid i ty . T h e results are presented i n F igure 9.1 and 9.2 for trials D - l and D - 2 respectively. F o r t r ia l D - 2 , since rhomboid i ty values were measured a long the entire b i l le t length as shown i n Figures 5.21 to 5.36, the m a x i m u m rhomboid i ty values were considered. F igure 9.2 shows the rhomboid i ty values plotted against the two temperature values ( m a x i m u m and m i n i m u m ) measured 30 and 45 m m be low the meniscus. It is evident f rom this figure that for the 0.32 pet. carbon steel, rhomboidi ty values o f - 1 5 m m and - 1 0 m m are associated w i t h a large temperature range (difference between m a x i m u m and m i n i m u m values o f around ~ 6 0 ° C ) . O n the other hand, rhomboid i ty values that are less than ~6 m m for this grade, are associated w i t h a smaller temperature range (wi th differences o f - 3 0 to 4 0 ° C ) . The 0.12 pet. carbon steel 221 does not have a rhomboid i ty p rob lem and the temperature difference is on ly ~ 2 5 ° C w h i l e the h igh carbon steel bi l lets have rhomboid i ty values i n the range o f - 4 . 0 to 6.0 m m and temperature differences o f ~ 5 0 ° C . In t r ia l D - l , a trend s imi la r to those observed i n t r ia l D - 2 is seen, F igure 9.1. The rhomboid i ty value o f - 1 0 m m for a 0.32 pet. carbon steel is associated w i t h a large difference between m a x i m u m and m i n i m u m temperatures o f ~ 6 0 ° C . The bi l lets w i t h smal ler rhomboid i ty values were associated w i t h a temperature difference o f - 2 5 to 3 0 ° C for the 0.32 pet. carbon (wi th and wi thout boron) and 0.84 pet. carbon steels. Here again, the 0.12 pet. carbon steel b i l le t sample had a l o w rhomboid i ty value together w i t h a sma l l value for the temperature difference. F r o m the above analysis, it appears that the observed rhomboid i ty is l i n k e d to the temperature var ia t ion (difference between m a x i m u m and m i n i m u m temperatures on the four faces) measured be low the meniscus. Thus , rhomboid i ty measurements were plot ted against the corresponding temperature var ia t ion for trials D - l and D - 2 i n F igure 9.3. In spite o f some scatter i n the data, it appears that cases w i t h severe rhomboid i ty are associated w i t h a large difference between m a x i m u m and m i n i m u m temperatures measured on the four faces i n the reg ion - 2 5 to 50 m m be low the meniscus. Thus , it is possible to detect the non-uni formi ty i n heat transfer i n the meniscus region, that causes the rhomboid i ty i n b i l le t cast ing. B a s e d on the above f indings, it is apparent that the average temperature var ia t ion can be e m p l o y e d i n the "intell igent m o u l d " to detect the severity o f rhomboid i ty i n the m o u l d . 9.2 Mechanism of Rhomboidity It appears f rom the above discuss ion that non-uni form m o u l d heat extract ion i n the meniscus region triggers rhomboid i ty i n b i l le t casting. T o evaluate the mechan i sm generating 222 non-un i fo rm heat transfer (and the rhomboidi ty problem), measurements made o n bi l le t samples co l lec ted at the plant trials were analysed on the basis o f m o u l d heat transfer and peak hot face temperature calculated us ing the m o u l d heat transfer mode l discussed i n Chapter 6. The importance o f m o u l d heat transfer i n the meniscus region is evident f rom F igu re 5.20 w h i c h shows the effect o f m o u l d taper i n the meniscus region on the difference between diagonals measured on cast bi l le ts . It is apparent f rom the figure that rhomboid i ty is less o f a p r o b l e m i n operations e m p l o y i n g steep m o u l d tapers (2.7 and 4.9 pct./m) i n the meniscus region as compared to those e m p l o y i n g sha l low tapers (0.4 and 0.8 pct. /m). A s imi la r result was also reported i n an earlier study conducted by Lorento [25] i n a number o f m i n i - m i l l s across N o r t h A m e r i c a . T h e influence o f m o u l d taper on the severity o f rhomboid i ty is l i n k e d to its strong effect on m o u l d heat extraction [1, 4-5] , F igure 8.8, and therefore, to the peak hot face temperature. T h e analysis o f hot face temperature o f the m o u l d indicates that the m o u l d was " c o l d " for the cases w i t h steep tapers (2.7 and 4.9 pct./m) w h i l e for the operations e m p l o y i n g sha l low tapers (0.4 and 0.8 pct. /m), the m o u l d was quite "hot". It must be pointed out that rhomboid i ty was occas ional ly observed at C o m p a n y D (trial D - l and D - 2 ) where a parabolic-tapered m o u l d w i t h a taper at the meniscus o f - 3 . 5 pct . /m, was employed . Thus , this observation does not fit w e l l w i t h the results presented i n F igu re 5.20. T h i s anomaly can be expla ined on the basis o f m o u l d hot face temperature relat ive to the b o i l i n g range o f o i l . A s ment ioned earlier i n Chapter 8, al though the m o u l d taper at the meniscus on the hot face temperature o f the m o u l d is quite strong, other operating factors are also important. A t C o m p a n y D where a steep taper o f - 3 .5 pct . /m was emp loyed and a " co ld m o u l d " operat ion is expected, the peak hot face temperature was calculated to be i n the range o f - 2 5 0 to 2 9 0 ° C , w i t h a c o o l i n g water ve loc i ty o f 10.5 m/s and a w a l l thickness o f 15.6 m m . Thus , the m o u l d was "moderately hot" and rhomboid i ty was observed. W i t h a higher c o o l i n g water ve loc i ty , it 223 is possible to lower the hot face temperature f rom ~ 2 9 0 ° C to be low the b o i l i n g range o f o i l and achieve a " c o l d m o u l d " condi t ion s imi la r to that observed at C o m p a n y E (Figure 8.13) where the rhomboid i ty p rob lem was m i n i m a l . Thus , the hot face temperature o f the m o u l d relative to the b o i l i n g range o f o i l , may be a significant parameter w i t h respect to rhomboid i ty . F igure 9.4 shows the rhomboid i ty values as a function o f peak hot face temperature for a range o f carbon steels cast at the various plant trials. The peak hot face temperatures o f m o u l d corresponding to bi l le t samples w i t h rhomboidi ty values greater than 8.0 m m , l i e i n the b o i l i n g range o f o i l or, are close to it. F o r example , out o f the 14 cases w i t h rhomboid i ty values equal to, or greater than, 4.0 m m , there was on ly one instance where the peak temperature was less than ~ 2 0 5 ° C and two cases, where the peak temperature exceeded ~ 3 3 5 ° C . A l t h o u g h the number o f data points evaluated for the above analysis is l imi ted , F igure 9.4 suggests that the hot face temperature (relative to the b o i l i n g range o f o i l ) may be important w i t h respect to the severity o f rhomboid i ty . It can be argued that since the greatest enhancement i n heat transfer occurs when the peak hot face temperature falls i n the b o i l i n g range o f o i l (Figure 8.12), sma l l differences i n the hot face temperature on the four faces can generate s ignif icant var iab i l i ty i n m o u l d heat extraction. The above arguments are quite interesting but their va l id i ty must be tested i n future studies i n v o l v i n g a larger database where the effect o f other variables have been adequately isolated. The other important issue, as mentioned before, is that rhomboid i ty must be measured along the entire bi l le t length instead o f a long sma l l 300 to 400 m m l o n g samples because the severity and orientation o f rhomboid i ty can change s ignif icant ly i n a s ingle bi l le t . A l t h o u g h the above observations provide useful insights about the importance o f hot face temperature relative to the b o i l i n g range o f o i l , these do not exp la in the cause o f non-un i fo rm c o o l i n g condi t ions required to generate rhomboid i ty . Thus , hot face temperatures were computed for the measured ( m i n i m u m and m a x i m u m ) m o u l d temperature values, presented 224 earl ier i n Figures 9.1 and 9.2, i n the region be low the meniscus. T h e translation o f measured values into hot face temperature was done us ing a s imple method. A n equation was der ived f rom a regression analysis conducted on a range o f measured temperature data (at a mid- thickness thermocouple locat ion) and the corresponding hot face temperature values predicted us ing the m o d e l [68] discussed i n Chapter 6. F igure 9.5 shows the calculated hot face temperature against the measured temperature (at thermocouple locat ion); the regression l ine (r 2 = 0.96) obtained i n the analysis is also shown. Th i s equation was employed to compute the hot face temperatures corresponding to the measured m o u l d temperatures i n the meniscus region shown i n Figures 9.1 and 9.2. W i t h respect to the intell igent m o u l d , this technique can be quite attractive for on- l ine moni to r ing since it can be adopted to q u i c k l y estimate the hot face temperatures i n the meniscus region f rom the measured temperatures without actually execut ing the finite-difference mathemat ical mode l . T h e rhomboid i ty values are plotted i n F igure 9.6 and 9.7, as a funct ion o f the estimated hot face temperature for trials D - 1 and D - 2 at the two locations 30 and 45 m m b e l o w the meniscus. A l t h o u g h the trend observed for the range o f temperature is very s imi la r to that seen for the measured values, several interesting points about the nature o f m o u l d heat transfer can be noted f rom the two figures. F igure 9.6 shows that for the case w i t h rhomboid i ty o f - 1 0 . 0 m m i n t r ia l D - 1 , the m i n i m u m temperature (~ 1 5 5 - 1 6 0 ° C ) is less than the b o i l i n g point o f o i l (a " co ld m o u l d " operation) w h i l e the m a x i m u m value ( ~ 2 7 0 ° C ) l ies i n the b o i l i n g range o f o i l (a "hot m o u l d " operation). In this situation, the heat transfer for the two extreme cases is expected to be different as shown i n F igure 9.6. T h e face having a temperature o f ~ 2 7 0 ° C w i l l experience enhanced heat transfer due to o i l vapourisat ion w h i l e the face hav ing the lowest temperature is u n l i k e l y to experience any enhancement. The peak heat f lux value for the case w i t h higher temperature 225 is expected to be -2 .0-2 .5 times higher, F igure 8.12. The above trend is also observed i n tr ial D - 2 for the case w i t h rhomboid i ty - 1 0 . 0 m m , al though the situation is less severe since the m i n i m u m temperature ( ~ 1 9 0 ° C ) is higher than i n the previous case. F o r the b i l le t w i t h a large rhomboidi ty value o f - 1 5 m m i n tr ial D - 2 , a l though both m a x i m u m and m i n i m u m temperatures l ie i n the b o i l i n g range o f o i l (a hot m o u l d operation), the m i n i m u m value ( ~ 2 2 0 ° C ) is extremely close to the start o f b o i l i n g w h i l e the m a x i m u m value (-315°C) is w i th in ~ 2 0 ° C o f the temperature at w h i c h - 9 0 pet. o f b o i l i n g is complete . A l t h o u g h the two temperatures fa l l w i th in the b o i l i n g range o f o i l , the heat f lux values for the two cases are quite different; the heat f lux corresponding to the m a x i m u m temperature is again -2 .0-2 .5 t imes higher than that for the m i n i m u m value. Th i s situation cou ld have been even worse i f the m a x i m u m value was greater than the b o i l i n g range o f o i l and the m i n i m u m value, less than the b o i l i n g range o f o i l . F o r the l o w carbon steel, the m a x i m u m and m i n i m u m values o f hot face temperature o f the m o u l d are less than the b o i l i n g point o f o i l for both trials D - l and D - 2 ; thus a l l the faces are " co ld" . Thus , the heat transfer for this grade is expected to be fai r ly un i fo rm on the four faces. It is not surpris ing that rhomboid i ty is not a p rob lem for this grade. T h e interaction o f the meniscus w i t h the lubr icat ing o i l w i l l depend on the peak hot face temperature o f the m o u l d relative to the bo i l i ng range o f o i l . In a "hot m o u l d " operation, where the peak hot face temperature o f the m o u l d lies i n the v i c in i t y o f the b o i l i n g range o f the o i l , the region be low the meniscus is fa i r ly dry (less l i q u i d o i l ) . Thus , when the meniscus rises, the peak temperature o f the m o u l d moves along w i t h it, f rom a region o f less, or no, o i l to another where l i q u i d o i l is present. W h e n this happens, the o i l is trapped and vaporises; this can result i n an enhancement i n heat transfer o f about 20 pet. (Figures 8.10 and 8.11). O n the other hand, when the metal l eve l falls, the peak temperature o f the m o u l d moves into a region 226 w i t h less or no o i l . N o enhancement is expected due to o i l i n this situation since the region i n the v i c i n i t y o f the nomina l locat ion o f the peak hot face temperature is dry. H o w e v e r , there might be some enhancement i n heat transfer i n loca l i zed regions because o f s t i ck ing w h i c h w i l l l i k e l y occur due to l ack o f lubricant on the m o u l d w a l l . Thus , when the metal l eve l is r i s ing at one locat ion i n the m o u l d and fa l l ing at another, significant variations i n heat transfer can occur and the s o l i d shel l thickness generated w i l l be different. In a " co ld m o u l d " operation, since the peak temperature o f the m o u l d is less that the b o i l i n g range o f o i l , metal l eve l fluctuations w i l l not cause changes i n heat transfer since l i q u i d o i l is un i fo rmly present i n the meniscus region and also, i n the area adjacent to the peak hot face temperature. D u r i n g metal l eve l fluctuations, the meniscus and the peak temperature w i l l m o v e f rom a region w i t h o i l , to another hav ing o i l as w e l l . Thus , variat ions i n heat transfer due to metal l eve l fluctuations are expected to be m i n i m a l . It is not surpr is ing that the rhomboid i ty p r o b l e m is almost nonexistent for the cases o f "co ld m o u l d " operation examined i n this study. In the past "soft c o o l i n g practice" that invo lves operating w i t h reduced c o o l i n g water ve loc i ty , was emp loyed to temporari ly alleviate the p rob lem o f rhomboid i ty i n moulds w i t h sha l low taper, al though there is no measured data to back the explanat ion [3]. T h i s phenomenon was expla ined earlier on the basis o f nucleate b o i l i n g i n the c o o l i n g water channel [3]. T h e mechan i sm based on the interaction o f metal l eve l fluctuations and lubr ica t ing o i l , can also be used to exp la in the effect o f "soft coo l ing" . T h e reduction i n c o o l i n g water ve loc i ty s ignif icant ly increases the hot face temperature o f the m o u l d w h i c h causes the lubr ica t ing o i l i n the region close to the meniscus to vaporize. In this case the peak hot face temperature o f the m o u l d can reach temperatures m u c h greater than ~ 3 5 0 ° C . It must be emphasized here that "soft c o o l i n g " represents an extreme case o f "hot m o u l d operation" where the hot face temperature is m u c h 227 higher than the "unstable" bo i l i ng range o f o i l . Thus , the area i n the v i c in i t y o f the meniscus, w i t h no or less o i l expands, mak ing the heat transfer less sensitive to metal l eve l fluctuations, at least on a short term when "soft c o o l i n g " practice is i n effect. S ince "soft c o o l i n g " was occas ional ly adopted to m i n i m i z e rhomboid i ty , the p rob lem appears to be random i n nature. It is quite possible that metal l eve l fluctuations were i n v o l v e d since this is by far the most c o m m o n , random variable encountered i n b i l le t casting e m p l o y i n g open-stream pour ing . I f nucleate b o i l i n g i n the coo l ing channel was the cause, the rhomboid i ty p r o b l e m w o u l d exist at a l l times i n the absence o f soft c o o l i n g ; it is very u n l i k e l y that the severity o f nucleate b o i l i n g w o u l d change so drast ically dur ing a heat such that the rhomboid i ty p r o b l e m became serious, f rom time to t ime. W i t h metal l eve l fluctuations, since randomness is inherent, the severity o f non-uni form heat transfer can drast ical ly change dur ing a s ingle heat or i n a sequence. It is also possible to encounter problems w i t h non-uni form o i l dis t r ibut ion on the four faces dur ing the casting operation. Th i s can also generate variable heat transfer i n the meniscus. In addi t ion to the peak hot face temperature, the other important point evident i n F igure 9.4 w i t h regard to the severity o f rhomboidi ty is the effect o f steel grade, par t icular ly the carbon content. Thus , once again, it is apparent that steel grades c o n t a i n i n g - 0 . 1 5 t o - 0 . 4 5 pet. carbon are more sensitive to rhomboid i ty compared to others; - 6 0 pet. o f the cases examined i n this carbon range have rhomboid i ty values greater than, or equal to, - 4 . 0 m m . F o r the h igh carbon steels w i t h carbon levels greater than - 0 . 4 5 pet., on ly one case o f large rhomboid i ty ( -8 .0 m m ) was observed i n the 10 cases examined. A n d consistent w i t h earlier observations, a l l o f the b i l le t samples conta in ing less than -0 .15 pet. carbon had rhomboid i ty values less than - 2 . 0 m m . T h e above pattern agrees quite w e l l w i t h the trends reported i n several independent studies [25-27], that m e d i u m carbon steels containing - 0 . 2 0 to 0.40 pet. carbon are the worst w i t h 228 respect to rhomboid i ty . W h i l e the effect o f steel carbon content appears to be an important one, none o f these studies expla ined w h y steels containing ~0.15 to 0.45 pet. carbon were sensitive to rhomboid i ty . T o investigate further the effect o f steel carbon content on the severity o f rhomboid i ty , s o l i d shel l thickness profi les obtained us ing the b i l le t so l id i f ica t ion mode l [5,18,76], were evaluated for the three grades (0.12 pet., 0.32 pet. and 0.84 pet. carbon steels). It must be pointed out that the " so l id" shel l is defined by the locus o f the solidus temperature for the three grades. T h e heat f lux prof i le employed i n these calculat ions were presented earlier i n F igure 6.3. T h e effect o f variat ions i n heat f lux on the shel l thickness i n the meniscus region was also evaluated; this was achieved by ar t i f ic ia l ly setting the heat f lux values i n the upper part o f the m o u l d at 75 pet., 125 pet. and 150 pet. o f the nomina l heat f lux. A s shown i n F igure 9.8, the so l id shel l at the bot tom o f the m o u l d is thickest ( - 12 m m ) for the 0.32 pet. carbon steel, f o l l owed by - 1 0 m m for the 0.84 pet. carbon steel and is the lowest ( -8 m m ) for the 0.12 pet. carbon steel. A l t h o u g h the 0.84 pet. carbon steel has a higher heat f lux , compared to the 0.32 pet. carbon steel, the so l id shel l calculated at the m o u l d exit is actual ly thinner. Thus , m o u l d heat f lux is not the on ly parameter con t ro l l ing so l id shel l thickness. F o r a g iven casting speed and superheat l eve l , the freezing range o f the steel (difference between l iquidus and sol idus temperature) i n question, is the other parameter that appears to be c r i t i ca l . The 0.32 pet. carbon steel w h i c h has a thicker shel l at the m o u l d exit , has a freezing range o f ~ 5 0 ° C as compared to the 0.84 pet. carbon steel where the freezing range is m u c h longer ( ~ 1 0 0 ° C ) . T h i s point becomes more clear when values o f so l id shel l thickness are compared for the three grades and for the four heat fluxes i n the meniscus region. T h e values o f s o l i d shel l thickness computed at the midface o f the strand at a locat ion - 2 5 m m b e l o w the 229 meniscus , are presented i n F igure 9.9. A t the nomina l heat flux value, so l id i f ica t ion has not c o m m e n c e d at the 25 m m locat ion for the 0.12 pet. and 0.84 pet. carbon grades, w h i l e the so l id she l l thickness for the 0.32 pet. carbon grade is - 0 . 5 m m . F o r the reduced heat flux, so l id i f ica t ion is de layed for a l l three grades. W h e n the heat flux is set at 125 pet. o f the n o m i n a l value, the she l l thickness is - 1 . 0 m m for the 0.32 pet. carbon steel, - 0 . 4 m m for the 0.84 pet. carbon steel and on ly - 0 . 2 m m for the 0.12 pet. carbon steel. T h e above calculat ions show that i f the heat f lux dropped by 25 pet. o n one face but remained at the nomina l value on another, the difference between the so l id shel l thickness at the two locat ions w i l l be -0 .5 m m for the 0.32 pet. carbon steel. In the other two grades, since so l id i f ica t ion does not commence at this locat ion , there w i l l be no influence on the progress o f so l id i f ica t ion . A n even worse situation arises when the heat f lux drops by 25 pet. on one face and increases by 25 pet. on another. In this case, the difference between the s o l i d shel l thickness at the two locat ions is - 1 . 0 m m for the 0.32 pet. carbon grade, - 0 . 5 m m for the 0.84 pet. carbon grade and on ly - 0 . 2 m m for the 0.12 pet. carbon grade. Thus , the impact o f non-un i fo rm heat transfer i n the meniscus region i n generating a non-uni form so l id shel l is greater for the 0.32 pet. carbon steel than for the other two grades. It is apparent f rom the above discuss ion that the or igins o f rhomboid i ty are related to the non-un i fo rm c o o l i n g condit ions exis t ing i n the meniscus region, w i t h i n about 50 m m be low the meniscus . A s expla ined before, the interaction o f metal l eve l fluctuations and lubr ica t ing o i l i n moulds w i t h different hot face temperatures, generates the non-un i fo rm heat transfer i n the meniscus region. The importance o f non-uni form c o o l i n g is evident i n F igu re 9.3 w h i c h shows that bi l lets w i t h severe rhomboid i ty are associated w i t h a large difference between the m a x i m u m and m i n i m u m temperatures. T h e variable heat transfer o n the four faces i n the meniscus region, generates a non-uni form so l id shel l around the b i l le t periphery. Ca lcu la t ions 230 indicate that the impact o f non-uni form c o o l i n g is the greatest for grades o f steel (0.32 pet. carbon steel i n the above discussion) for w h i c h the m o u l d heat extract ion is reasonably h igh and at the same t ime, the freezing range is short. O n the other hand, for the other two grades, this combina t ion is not that favourable; for example , the 0.12 carbon steel has a l o w heat transfer due to the rough surface ar is ing f rom the peritectic reaction, w h i l e the 0.84 pet. carbon grade has a l ong freezing range. The above arguments strongly suggest that rhomboid i ty depends both on the magnitude o f m o u l d heat transfer and the freezing range o f the steel. T h i s new mechan i sm based on metal l eve l fluctuations, m o u l d hot face temperature (relative to the b o i l i n g range o f the o i l ) and the freezing range o f steel, is quite significant because it takes into account the properties o f steel as w e l l as m o u l d heat transfer. W h e n compared to o i l casting, rhomboid i ty is m u c h smaller or almost non-existent i n b i l le t casting w i t h m o u l d f lux lubr icat ion. T h i s can be expla ined on the basis o f metal l eve l fluctuations and m o u l d heat transfer. The metal l eve l fluctuations i n operations w i t h m o u l d f lux lubr ica t ion are l i k e l y to be smaller since a submerged-entry nozz le is used to de l iver mol ten metal f rom the tundish to the m o u l d , and consequently less gas is entrained. In addi t ion, w i t h the use o f m o u l d f lux , the m o u l d heat transfer is lower and therefore, the s o l i d shel l generated at the meniscus is less susceptible to non-uniformit ies. Furthermore, the non-uni formi ty i n heat transfer due to o i l b o i l i n g and pyro lys i s , is complete ly e l iminated. In addi t ion to metal l eve l fluctuations w h i c h appear to be a dominant var iable , a misa l igned tundish stream can also result i n non-uni form heat transfer on the four faces. T h e non-un i fo rm dis t r ibut ion o f lubr ica t ing o i l on the different faces is another factor that can also contribute to variable heat transfer i n the meniscus region. In such situations, however , s t i ck ing and related problems w i l l be evident on the faces starved o f lubr icat ion. 231 A nothe r reason for the variable m o u l d heat transfer is the non-un i fo rm dis t r ibut ion o f defects such as transverse depression on the b i l le t surface; the faces w i t h transverse depressions w i l l have a reduced heat transfer as compared to the ones where transverse depressions are absent. S i m i l a r l y , the presence o f non-uni form osc i l la t ion marks on the b i l le t surface w i l l result i n var iable mould/strand gap and non-uni form heat transfer. It is evident f rom F igure 8.14 that the osc i l l a t ion marks tend to become deeper and more non-uni form at longer negative strip t ime (greater than about 0.17 s) [23]. Thus , negative strip t ime can also influence the severity o f rhomboid i ty . T h e presence o f dark and bright patches on the strand at the m o u l d exi t can also contribute to rhomboid i ty . H o w e v e r , i n this case, the sprays must also be i n v o l v e d . T h e dark, overcooled patches c o o l m u c h more rapidly i n the sprays due to transition b o i l i n g as compared to the brighter, hotter regions on the strand where f i l m b o i l i n g prevai ls . The result ing non-un i fo rm c o o l i n g o f the strand can lead to distort ion o f the b i l le t and rhomboid i ty . T h e dark and bright patches are thought to be generated i n more than one way . It is be l i eved that l oca l s t ick ing on a face leads to h igh heat transfer w h i c h creates a dark, overcooled patch on the strand. W h e n this happens, the steel shrinks and causes a gap to open up on the other face(s) generating bright patches due to reduced heat transfer. In addi t ion to s t i ck ing , the dark and bright patches can also be generated dur ing adverse metal l eve l f luctuations. W h e n metal l eve l fluctuations occur, certain por t ion o f the meniscus can experience h igh heat transfer and at the same t ime, other regions may be subjected to l ower heat transfer rates. F o r example , w h e n the metal l eve l l oca l ly rises, it can trap the lubr ica t ing o i l beneath the meniscus between the so l id i fy ing shel l and the m o u l d w a l l . The sudden release o f vapour can s igni f icant ly enhance the heat transfer i n the loca l i zed region. Furthermore, the metal l eve l fluctuations can also contribute to s t ick ing i n the meniscus region i n "hot m o u l d " operations, as postulated i n Chapter 232 8 deal ing w i t h laps and bleeds. Thus , the or igins o f dark and bright patches can be l i n k e d to metal l eve l fluctuations, hot face temperature o f the m o u l d and poor lubr ica t ion . Unfortunately , a direct correlat ion cou ld not be obtained between the presence o f dark/bright patches and the severity o f rhomboid i ty because it was extremely diff icul t to observe the strand ex i t ing the m o u l d . T h e adverse effect o f poor c o o l i n g water qual i ty on rhomboid i ty i n b i l le t cast ing e m p l o y i n g o i l has been h ighl ighted i n earlier studies [7-11]. Poo r c o o l i n g water qual i ty is a c r i t i ca l var iable that s i m p l y cuts across a l l the other operating and design parameters i n b i l le t cast ing w i t h o i l lubr ica t ion . It is quite clear that operating w i t h poor c o o l i n g water w i l l lead to scale deposits on the c o l d face o f the m o u l d w h i c h w i l l cause excess ively h igh m o u l d w a l l temperature and numerous operating problems inc lud ing abnormal m o u l d distort ion, asynchronous b o i l i n g o f water on the c o l d face and vapourisat ion o f lubr icat ing o i l on the hot face. Thus , it is imposs ib le to achieve a "co ld m o u l d " operation when operating w i t h poor water qual i ty and hence, it is mandatory to have excellent c o o l i n g water quali ty to m i n i m i z e rhomboid i ty . B a s e d on the arguments presented i n this section, it can be conc luded that any design and operating variables that generate a "hot m o u l d " operation as w e l l as var iable hot face temperature o n the four faces, w i l l adversely affect the severity o f rhomboid i ty . B u t the p rob lem, whether r andom or not, w i l l depend on the nature o f the dominant variables - for example , metal l eve l f luctuations, tundish stream alignment (and o i l distribution) versus non-un i fo rm channel gap (misal ignment) and variable w a l l thickness. Furthermore, as pointed out earlier, the sprays are also i n v o l v e d , al though indirect ly , i n the generation o f rhomboid i ty , i n addi t ion to mould-re la ted problems [22]. 233 9.3 Minimizing Rhomboidity in Billet Casting B a s e d on the discuss ion presented i n the preceding section, f o l l o w i n g are the major recommendat ions proposed to m i n i m i z e the severity o f rhomboid i ty i n bi l le ts : (i) A "co ld m o u l d " practice must be adopted such that the peak hot face temperature o f the m o u l d for a l l o f the four faces, is l ower than the b o i l i n g range o f lubr ica t ing o i l . The m a i n requirements to achieve this condi t ion inc lude a steep taper at the meniscus o f about 2.0 pct . /m together w i t h a c o o l i n g water ve loc i ty greater than - 1 4 - 1 5 m/s and excellent c o o l i n g water quali ty. It must be emphasized that "soft c o o l i n g practice" should not be employed at any t ime because it causes excessive m o u l d dis tor t ion and serious lubr ica t ion problems. (i i ) S ince random metal l eve l fluctuations generate variable c o o l i n g condi t ions i n the meniscus region, it is important to m i n i m i z e the turbulence i n the ex i t ing tundish streams by ensuring proper f l ow i n the tundish and isolat ing turbulence in t roduced by the ladle stream using f low control devices [75]. The tundish -to- m o u l d height must be as sma l l as possible to m i n i m i z e growth o f instabili t ies on the open-poured stream. T h e use o f submerged entry nozzles , especial ly for larger section sizes, w i l l help m i n i m i z e frequent metal l eve l fluctuations. T u n d i s h nozz le c logg ing due to the presence o f so l id inc lus ions i n the steel also may create disturbances on the tundish stream [75]. ( i i i ) T h e o i l lubr ica t ion system must be designed and mainta ined to de l iver a u n i f o r m f low o f o i l around the m o u l d periphery [16,28]. (iv) A negative strip t ime that is maintained between 0.12 and 0.16 s is r ecommended [16,23] to achieve un i fo rm osc i l la t ion mark depth across a g iven face. T h i s w i l l also m i n i m i z e s t i ck ing and generation o f dark and bright patches. 234 In addi t ion to the mould-related issues, spray c o o l i n g system must be designed and operated such that the heat transfer on the four faces is un i fo rm [77]. 235 75 100 125 150 Measured Temperature (°C) 175 200 75 100 125 150 175 Measured Temperature (°C) 200 ure 9.1 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m measured m o u l d temperature values for tr ial D - l at (a) 30 m m and (b) 45 m m b e l o w the meniscus. 236 (a) 16 12 § 8 5 -EB-0.12 pet. C M 0.32 pet. C 0.80 pet. C + 75 100 125 150 Measured Temperature (°C) 175 200 (b) 16 w jS 12 O O) n 5 c 8 -EB-0.12 pet. C 0.32 pet. C 0.80 pet. C EB-+ + 75 100 125 150 175 Measured Temperature (°C) 200 225 Figure 9.2 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m measured m o u l d temperature values for t r ia l D - 2 at (a) 30 m m and (b) 45 m m b e l o w the meniscus. 237 (a) 0 10 20 30 40 50 60 Measured Temperature Range (°C) Figu re 9.3 R h o m b o i d i t y values plotted against measured temperature var ia t ion (difference between m a x i m u m and m i n i m u m values) at (a) 30 m m and (b) 45 m m b e l o w the meniscus for trials D - l and D - 2 . 238 E 1 6 tn Sl2 O O) 03 Q § 8 B o m a> u e <D mm CJ 4 + -• 0 100 -H=-<0.15 pet. C • 0.15-0.45 pet. C V > 0.45 pet. C <]— Boiling Range of Oil —1> -V--v-V V V +• + 150 200 250 300 350 Peak Hot Face Temperature (°C) 400 Figure 9.4 R h o m b o i d i t y values plotted against peak hot face temperature for the different steel grades cast i n the various plant trials. 239 Figu re 9.5 Rela t ionship between measured temperature and corresponding peak hot face temperature for trials D-l and D-2, 25 to 50 m m be low meniscus ( in the v i c in i t y o f peak hot face temperature). 240 (a) 100 150 200 250 Hot Face Temperature (°C) 300 350 (b) 100 150 200 250 Hot Face Temperature (°C) 300 350 Figure 9.6 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m values o f m o u l d hot face temperature for tr ial D - l at (a) 30 m m and (b) 45 m m b e l o w the meniscus. 241 (a) 16 12 4 + -EB-0.12 pet. C 0.32 pet. C -£s -0.80 pet. C EH—-m + Boiling Range of Oil 4 -4 > too 150 200 250 300 Peak Hot Face Temperature (°C) 350 (b) 16 12 4 + -EB-0.12 pet. C 0.32 pet. C 0.80 pet. C EEH -EB -A Boiling Range of Oil 1 1 1 h -150 200 250 300 Peak Hot Face Temperature (°C) - o 100 350 Figure 9.7 R h o m b o i d i t y values plotted against m i n i m u m and m a x i m u m values o f m o u l d hot face temperature for tr ial D - 2 at (a) 30 m m and (b) 45 m m be low the meniscus. 242 0.12 pet. C 0.32 pet. C Steel Grades 0.84 pet. C Figure 9.8 C o m p a r i s o n o f s o l i d shel l thickness at the m o u l d exi t for the three steel grades cast at t r ial D - l for va ry ing heat f lux condit ions i n the meniscus region. 243 (a) 2.0 oo (/> 0) 5 1.0 u 0) JC 0.5 CO 0.0 70 0.12 pet. C -E3- 0.32 pet. C 3.84 pet. C -25 mm below meniscus 80 90 100 110 120 130 140 Percentage of Original Heat Flux (%) 150 160 (b) 2.5 ? 2 . 0 co 1.5 o c O 1.0 + 0) W 0 . 5 0.0 0.12 pet. C -E3- 0.32 pet. C - A - 0.84 pet. C -50 mm below meniscus 70 80 90 100 110 120 130 140 Percentage of Original Heat Flux (%) 150 160 Figure 9.9 C o m p a r i s o n o f computed s o l i d shel l thickness for the three steel grades (0.12 pet., 0.32 pet. and 0.84 pet. carbon steels) for va ry ing heat f luxes, at two locat ions situated (a) 25 m m and (b) 50 m m be low the meniscus. 244 C H A P T E R 10 - D E V E L O P M E N T OF T H E INTELLIGENT M O U L D 10.1 Intelligent Mould - A Background T h i s w o r k is a part o f an overa l l project to design an intel l igent m o u l d [1,6]. T h e major components o f the system are shown i n F igure 10.1. The system, w h i c h is currently under development includes sensors (thermal and mechanical) , data acquis i t ion hardware, a computa t ional intel l igence (CI) based software interface [60-62] and an expert sys tem consis t ing o f fundamental and heuristic knowledge o f the b i l le t casting process. The C I modu le performs two tasks. It provides an interface between the data acquis i t ion hardware and the expert system and also processes the raw sensor data to detect various trends ident i f ied as useful for subsequent analysis by the expert system. The f o l l o w i n g are the m a i n tasks performed by the system: (i) to detect bi l le t defects such as transverse depressions [21], laps/bleeds and rhomboid i ty on- l ine - a means for on- l ine hot inspect ion o f bi l le ts . ( i i ) to moni tor events i n the m o u l d such as metal l eve l fluctuations, non-un i fo rm heat transfer around the m o u l d periphery and var ia t ion o f negative strip t ime dur ing the operation. ( i i i ) to sound alarms when the m o u l d operation strays f rom the specif ied operating l imi t s . ( iv) to take preventive action by automatical ly cont ro l l ing operating parameters or adv is ing the operators to modi fy current settings. A c r i t i ca l component o f the system is the knowledge base contained i n the expert system. T h e knowledge base includes an in-depth analysis o f the operating condi t ions such as steel 245 superheat and compos i t ion , an estimation o f parameters such as the hot face temperatures f rom a mathematical mode l o f the m o u l d embedded i n the system as w e l l as strategies for ana lyz ing the m o u l d temperature on-l ine . 10.2 On-line Inspection of Billets On- l i ne inspect ion o f bil lets i n the m o u l d is probably the most interesting feature o f immedia te interest to operators i n the m i n i - m i l l s . T h e current w o r k on laps/bleeds and rhomboid i ty as w e l l as the earlier work on transverse depressions [21] s trongly suggest that thermocouples can be used to detect these defects on-l ine i n the m o u l d . T h i s f ind ing is the basis o f on- l ine inspect ion o f bil lets i n the "intell igent mou ld" . Transverse depression. laps and bleeds - A s far as the procedure for on- l ine detection o f defects such as laps, bleeds and depressions is concerned, the C l - m o d u l e i n the "intell igent m o u l d " tracks and computes drops (or "val leys") i n m o u l d temperature and the span o f the va l leys at 4 to 5 locations i n the m o u l d together w i t h the behaviour o f the meniscus . T h i s in format ion is transferred to the expert system w h i c h evaluates the data and establishes the seriousness o f the p rob lem using a reasoning based on fuzzy log ic . S ince transverse depressions, laps and bleeds are associated w i t h w i d e n i n g o f the mould/s t rand gap and y i e l d a s imi la r pattern i n the m o u l d temperature, the expert system distinguishes between these defects us ing informat ion contained i n its knowledge base about peak hot face temperature, freezing range o f steel, and the magnitude o f the observed temperature drop. T h e dec i s ion m a k i n g strategy is based o n the knowledge about the format ion o f laps, bleeds and transverse depressions. The peak hot face temperature o f the m o u l d is first evaluated to assess the nature o f the m o u l d operation. A s was discussed earlier, transverse depressions were more frequent i n "moderately hot" m o u l d operations at C o m p a n y D [21 ] w h i l e 246 laps and bleeds were more severe i n "hot" m o u l d operations at Compan ies A and C . T h e freezing range o f steel i n question is evaluated next to check the grade be ing cast ( long versus short freezing range). S ince laps and bleeds are associated w i t h a smal ler temperature drop ( or "val ley") as compared to transverse depressions, the magnitude o f drop i n m o u l d temperature relat ive to the base temperature is assessed to evaluate the w i d t h o f the air gap created i n the m o u l d by the surface defect. A s pointed out earlier, m o u l d thermocouples located at the midface may not be adequate for the detection o f bleeds because o f their sma l l size and dis t r ibut ion across a g iven face. The "intell igent m o u l d " w i l l handle such situations by s i m p l y mon i to r ing metal l eve l fluctuations and m o u l d w a l l temperatures and main ta in ing values w i t h i n a desired range where bleed formation is m i n i m a l . Rhomboidity - W i t h respect to rhomboid i ty , the "intell igent m o u l d " w i l l moni tor the non-un i fo rm c o o l i n g condit ions i n the meniscus region that generate rhomboid i ty i n bi l le ts . Thus , temperature o f the four faces o f the m o u l d (25-50 m m be low the meniscus) w i l l be e m p l o y e d to detect any non-uniformity i n c o o l i n g condi t ions. Ma thema t i ca l m o d e l results w i l l be e m p l o y e d to estimate hot face temperature f rom measured temperatures us ing prev ious ly der ived s imple empi r i ca l relationships. Neu ra l networks can also be developed to achieve the same. The expert system w i l l make decisions based on informat ion on non-un i fo rm heat transfer and also the freezing range w h i c h determines the sensit ivity o f a steel grade to rhomboid i ty . 247 PROCESS VISION USER INTERFACE COMDALE/C EXPERT SYSTEM * Smart Monitoring * Smart Alarming * Smart Control CI MODULE AND DATA ACQUISITION SYSTEM T T T MOULD SENSOR SIGNALS A * Thermocouples L * Strain Gauges i * LVDT's * Casting Speed * Metal Level t CONTINUOUS CASTING MOULD Figu re 10.1 The major components o f the "intell igent m o u l d " be ing deve loped for the continuous casting o f steel bi l le ts . 248 CHAPTER 11 - SUMMARY AND CONCLUSIONS A comprehensive study was undertaken to evaluate the impact o f metal l eve l fluctuations on m o u l d heat transfer and lubr icat ion i n the meniscus region as w e l l as generation o f defects such as laps, bleeds and rhomboid i ty , to develop strategies to detect process upsets and bi l le t defects on- l ine , to identify mechanisms o f formation o f defects (such as laps, bleeds and rhomboid i ty ) , and to provide recommendations to m i n i m i z e or e l iminate the p rob lem. D a t a on m o u l d w a l l temperature, metal l eve l , casting speed and b i l le t evaluat ion were acquired at s ix plant trials at four Canadian mini-s teel m i l l s . In the plant trials that were conducted, operat ing bi l le t moulds were instrumented w i t h an array o f thermocouples together w i t h L V D T s , load cel ls and/or strain gauges. The water temperature was also measured at the inlet and outlet o f the c o o l i n g channel . M e t a l l eve l and casting speed signals were obtained f rom plant instrumentation. B i l l e t samples were col lected at regular intervals and subjected to a detailed qual i ty evaluat ion. D u r i n g these trials, tundish stream and meniscus i n the m o u l d were f i lmed w i t h a v ideo camera and 35 m m photographs were also taken. The thermocouple data were subjected to two types o f analysis . A s a part o f the first analysis , steady state heat f lux and m o u l d w a l l temperature dis tr ibut ion were computed us ing a m o u l d heat transfer mode l [7-10]. In the second analysis, raw data were evaluated to f ind trends i n the m o u l d temperature. Based on the various findings and results o f b i l le t qual i ty evaluat ion, strategies were developed to detect defects, on-l ine. The m o u l d heat transfer m o d e l was adapted to evaluate transient thermal response o f the m o u l d to events such as m o u l d osc i l l a t ion , metal l eve l fluctuations and presence o f defects (such as transverse depressions) on the strand. 249 A s a result o f the various analysis conducted on m o u l d temperature data and b i l le t qual i ty evaluat ion, it is n o w possible to expla in the formation o f laps, bleeds and rhomboid i ty on the basis o f interaction o f metal l eve l fluctuations, m o u l d peak hot face temperature and lubr ica t ion i n the meniscus region. In addit ion, it is also possible to detect these defects on- l ine us ing thermocouples embedded i n the m o u l d w a l l . T h e f o l l o w i n g is a summary o f the salient points o f the thesis: [ 1 ] Instrumented m o u l d trials were carr ied out to measure m o u l d temperature, metal l eve l and cast ing speed; b i l le t samples were acquired and video/photographs o f the cast ing operation were taken. [2] T h e character o f metal l eve l fluctuations was studied us ing signals f r o m metal l eve l sensor and m o u l d thermocouples located above the meniscus. [3] T h e format ion o f laps and bleeds has been expla ined o n the basis o f a mechan i sm rooted i n metal l eve l fluctuations, m o u l d thermal response, lubr ica t ion and steel carbon content. [4] R h o m b o i d i t y i n b i l le t originates f rom variable heat transfer condi t ions ex is t ing i n the meniscus region that generate non-uni form so l id shel l thickness around b i l le t periphery. It was found that the impact o f non-uni form heat transfer i n the meniscus reg ion is the greatest on m e d i u m carbon steels. [5] Transient events i n the m o u l d and detection o f transverse depressions were s imula ted us ing an adapted vers ion o f an exis t ing m o u l d thermal mode l [7-9] and measurements made on cast bi l le ts . [6] P re l imina ry strategies for the detection o f defects i n the "intel l igent m o u l d " have been formulated. 250 The major conclus ions o f this w o r k are presented as fo l lows under seven major headings: A. Metal Level Fluctuations [1] M e t a l l eve l f luctuation i n b i l le t moulds has been identif ied as an important process upset that can create variable heat transfer and lubr icat ion condit ions i n the meniscus reg ion , and trigger qual i ty problems such as laps, bleeds, and rhomboid i ty i n bi l le ts . T h e impact o f metal l eve l fluctuations on lubr icat ion and heat transfer i n the meniscus reg ion depends on the magnitude o f peak hot face temperature o f the m o u l d , relative to b o i l i n g range o f lubr ica t ing o i l . [2] T h e most s t r ik ing feature o f metal l eve l fluctuations is its inherent randomness. T h e fluctuations (rise and fa l l i n level) differ not on ly on the four faces, but also across a g iven face. Furthermore, its severity changes w i t h t ime, w i t h no clear trend emerging . M e t a l l eve l fluctuations have been l i nked to the shape o f the tundish stream (whether smooth or rough); large variations are associated w i t h rough streams whereas the sma l l deviat ions are indica t ive o f a smooth stream. In addit ion, metal l eve l fluctuations are also affected by b ind ing o f the strand lower d o w n i n the m o u l d . B. Mould Heat Transfer and Peak Hot Face Temperature [1] T h e magnitude o f peak hot face temperature o f the m o u l d relative to the b o i l i n g range o f lubr ica t ing o i l , is a c r i t i ca l parameter w i t h respect to defect format ion and process control i n b i l le t cast ing. Based on the peak hot face temperature o f the m o u l d , two types o f m o u l d operations were elucidated: c o l d m o u l d and hot m o u l d . In the " c o l d m o u l d " operation, the peak hot face temperature o f the m o u l d is lower than the b o i l i n g range o f lubr ica t ing o i l ( be low ~ 2 0 5 ° C ) w h i l e i n the "hot m o u l d " operation, the peak hot face temperature is 251 greater that the b o i l i n g range o f lubr icat ing o i l ( ~ 2 0 5 - 3 3 5 ° C or higher) . [2] Enhancement i n heat transfer due to o i l vapourizat ion is a strong funct ion o f the magnitude o f peak hot face temperature and occurs when the peak hot face temperature is greater than ~ 2 0 5 ° C . In addi t ion, it was found that the greatest enhancement i n heat transfer due to o i l vapour iza t ion occurs when the peak hot face temperature falls i n the b o i l i n g range o f lubr ica t ing o i l . [3] In addi t ion to m o u l d taper at the meniscus, the peak hot face temperature depends on c o o l i n g water qual i ty , c o o l i n g water ve loc i ty , osc i l l a t ion characteristics and steel carbon content. Increasing the c o o l i n g water ve loc i ty f rom 10.5 m/s to 15.0 m/s results i n a significant reduction i n the hot face temperature. The effect o f negative strip t ime on m o u l d heat transfer, and therefore, peak hot face temperature, invo lves two major issues: extent o f mould/s t rand interaction at the meniscus and the depth o f osc i l l a t ion marks generated. W i t h respect to carbon content o f steel, it was found that the peak hot face temperature o f the m o u l d is highest for the h igh carbon (0.84 pet. C ) fo l l owed by m e d i u m carbon (0.32 pet. C ) and then the l o w carbon (0.12 pet. C ) . The presence o f defects such as transverse depressions that w i d e n the mould/strand gap, reduces the heat extract ion capabi l i ty o f moulds w h i c h leads to a lower peak hot face temperature. F o r "hot m o u l d " operations, a reduct ion i n the hot face temperature results i n suppression o f enhancement i n heat extract ion due to o i l b o i l i n g and pyro lys i s , w h i c h lowers the peak hot face temperature even further. 252 C. Laps and Bleeds in Billet Casting [1] L a p s and bleeds are more severe i n h igh carbon grades hav ing long freezing range (around ~ 1 0 0 ° C ) and i n operations, where the peakho t face temperature is greater than the b o i l i n g range o f o i l . [2] T h e observed pattern o f laps and bleeds on the b i l le t surface is quite i r regular and the accompany ing osc i l la t ion marks are distorted. The severity o f laps and bleeds changes r andomly dur ing heats, w i t h no clear trend emerging. [3] T h e longi tud ina l section through the laps shows two dist inct sub-surface so l id i f ica t ion hooks . The first is indicat ive o f partial freezing o f the meniscus because the dendrites are perpendicular to the white l ine demarcating the meniscus. T h i s indicates a pe r iod o f h igh heat extract ion due to s t i ck ing to the m o u l d . S t i ck ing is also evident f r o m the nature o f osc i l l a t ion marks i n the v i c in i t y o f the laps; the p i tch o f osc i l l a t ion marks l o c a l l y decreased on b i l le t samples that exhibi ted laps. T h e laps are associated w i t h a sma l l surface depression (up to 1 m m deep). [4] T h e m o u l d temperature response above the meniscus indicates that a fa l l i n metal l eve l triggers the formation o f laps i n h igh carbon grades; the fa l l i n metal l eve l is a lways preceded by a rise i n metal l eve l . A mechan i sm rooted i n metal l eve l fluctuations and its impact on lubr ica t ion and heat transfer i n the meniscus reg ion is proposed to exp la in the format ion o f laps i n "hot m o u l d " operations. A rise i n metal l eve l causes the lubr ica t ing o i l to vapourise and create a dry, unlubricated region on the m o u l d w a l l beneath the meniscus. Subsequently, when the l eve l drops, the meniscus contacts the unlubricated reg ion and sticks on the m o u l d face. The s t ick ing is accompanied by h igh heat transfer w h i c h gives rise to partial so l id i f ica t ion and formation o f lap upon over f low o f l i q u i d steel. 253 [5] W h e n compared to laps, bleeds are smaller i n size and do not extend across the surface o f the bi l le t . T h e accompanying osc i l la t ion marks are irregular and distorted indica t ing s t i ck ing to the m o u l d . The longi tudinal section through a b leed shows tearing o f so l id shel l . B leeds are also associated w i t h a smal l depression on the b i l le t surface. [6] B leeds are caused by tearing o f so l id shel l because o f p u l l i n g force exerted o n the strand by the wi thdrawa l rol ls dur ing the per iod when the shel l is stuck to the m o u l d . T h e mechan i sm o f metal l eve l fluctuations and s t ick ing i n "hot m o u l d " operat ion is the same as that proposed for laps. D. Rhomboidity in Billet Casting [ 1 ] R h o m b o i d i t y is more severe i n the m e d i u m carbon grades (~0.15 to - 0 . 4 5 pet. carbon) and also, i n "hot m o u l d " operation where the peak hot face temperature i n the v i c i n i t y o f the b o i l i n g range o f o i l . T h e orientation and magnitude o f rhomboid i ty i n b i l le t cast ing was found to change randomly dur ing heats. [2] R h o m b o i d i t y is l i nked to variable m o u l d heat extraction that exists i n the meniscus region and gives rise to a non-uni form so l id shel l . It was found that i n severe cases rhomboid i ty were associated w i t h a large average difference between the m a x i m u m and m i n i m u m temperature measured on the four midfaces, about 25-50 m m be low the meniscus . [3] In addi t ion to non-uni form heat transfer i n the meniscus region, the severity o f rhomboid i ty also depends on the magnitude o f m o u l d heat transfer and freezing range o f steel. T h e new understanding based on metal l eve l fluctuations, m o u l d peak hot face temperature (relative to b o i l i n g range o f o i l ) and the freezing range o f steel, is quite significant because it takes into account the so l id i f ica t ion o f steel as w e l l as m o u l d heat transfer. 254 [4] W h e n m o u l d heat transfer drops by 25 pet. at One loca t ion and increases by 25 pet. at another, the difference i n so l id shel l thickness generated at a locat ion , 25 m m be low the meniscus , is —1.0 m m for the 0.32 pet. carbon steel, - 0 . 4 m m for the 0.84 pet. carbon steel and on ly - 0 . 2 pet. carbon steel. Thus , the impact o f non-uni form c o o l i n g i n generating a non-un i fo rm so l id shel l , is the greatest for the 0.32 pet. carbon steel for w h i c h the m o u l d heat extraction is reasonably h igh ( -4500 k W / m 2 ) and at the same t ime, the freezing range is short ( ~ 5 0 ° C ) . O n the other hand, this combina t ion is not that favourable for the other two grades. F o r example , the 0.12 carbon steel has a l o w heat transfer ( - 2 5 0 0 k W / m 2 ) due to the rough surface ar is ing f rom the peritectic reaction, w h i l e the 0.84 pet. carbon grade has a l ong freezing range ( ~ 1 0 0 ° C ) . [5] T h e variable heat transfer condit ions i n the meniscus region that generates rhomboid i ty can be created by metal l eve l fluctuations. In a "hot m o u l d " operations, metal l eve l fluctuations can cause the meniscus , to move f rom regions w i t h less or no o i l to another, where l i q u i d o i l is present. N o n - u n i f o r m meniscus fluctuations on the four faces and also across a g iven face, can cause portions o f the meniscus to move into regions where o i l is l i q u i d and at the same t ime, force some other por t ion into region that may be starved o f o i l . Thus , large variations can be generated i n the meniscus region i n a r andom manner. O n the other hand, i n a "co ld m o u l d " operation, variations i n heat transfer due to metal l eve l fluctuations are expected to be smal l since l i q u i d o i l is present on the m o u l d w a l l , bo th above and be low the meniscus. [6] T h e presence o f dark and bright patches on the strand at the m o u l d exi t can also contribute to rhomboid i ty ; however , i n this case, the sprays must also be i n v o l v e d . It is be l i eved that the dark and bright patches are caused by loca l s t i ck ing o f the strand and/or adverse metal l eve l fluctuations. 255 [7] T h e other sources o f variable heat transfer i n the meniscus region that can contribute to rhomboid i ty include (a) non-uni form distr ibution o f o i l on the four faces and also across a g iven face and (b) non-uni form distr ibution o f defects such as transverse depressions on the b i l le t surface. In addi t ion to these factors, some design variables such as (a) m o u l d tube misal ignment , (b) non-uni form w a l l thickness and (c) var iable taper on the four faces can also generate variable heat transfer i n the meniscus region. E . Minimizing Quality Problems in Billet Casting [1] T h e major recommendations to m i n i m i z e laps and bleeds i n b i l le t cast ing are: (a) reduce metal l eve l fluctuations, (b) adopt "co ld m o u l d " practice, (c) ensure un i fo rm o i l f l ow and dis t r ibut ion, (d) mainta in negative strip t ime between - 0 . 1 2 to - 0 . 1 6 and (e) operate w i t h a M n : S ratio greater than 25:1 to m i n i m i z e tearing o f so l id shel l [76] and hence, b leed format ion. [2] The maj or recommendations to m i n i m i z e rhomboid i ty i n b i l le t casting are: (a) reduce metal l eve l fluctuations, (b) adopt "co ld m o u l d " practice, (c) ensure un i fo rm o i l f l ow and dis t r ibut ion, (d) ensure un i fo rm spray coo l ing . [3] The current design o f metal l eve l control system c o m m o n l y emp loyed i n b i l l e t cast ing, w h i c h maintains metal l eve l at a desired posi t ion by altering the cast ing speed, is not desirable since changes i n casting speed influence osc i l l a t ion characteristics (negative strip t ime and m o u l d lead) and also, disturb the metal l eve l . The system that is desired includes a constant metal l eve l and a constant casting speed that is isolated f rom the osc i l l a t ion frequency. W i t h such a system, it w i l l be possible to mainta in a constant negative strip t ime and m o u l d lead dur ing the casting operation. 256 F. On-line Detection of Process Upsets and Defects [1] M e t a l l eve l sensor is adequate for moni tor ing g loba l changes i n metal l eve l . H o w e v e r , these cannot moni tor loca l disturbances i n metal l eve l fluctuations. M o u l d w a l l thermocouples instal led above the meniscus, on the four faces as w e l l across a g iven face can be employed to detect l oca l fluctuations i n metal l eve l . M o u l d w a l l thermocouples instal led on the four faces above the meniscus, also can provide informat ion on the a l ignment o f the tundish stream. [2] Standard devia t ion o f metal l eve l s ignal (derived f rom metal l eve l sensor and thermocouples) provides a good idea o f the state o f metal l eve l fluctuations. Furthermore, it m a y be useful to moni tor the rise and fal l i n metal l eve l w h i c h are important i n the generation o f transverse depressions [21] and laps. In the "intell igent m o u l d " , each rise and fa l l i n metal l eve l w i l l be moni tored and decisions on each bi l le t cast w i l l be made on the basis o f the magnitude o f this fluctuations. [3] L a p s can be detected us ing 3 to 4 thermocouples instal led i n the m o u l d w a l l at the midface locat ion , w i t h the first one located at least 100 to 200 m m be low the meniscus. L a p s are manifested as "val leys" i n the m o u l d temperature s ignal w h i c h travel d o w n the m o u l d at a speed equal to the wi thdrawal rate. A s imi la r strategy w i l l be adopted for transverse depression. [4] W i t h respect to bleeds, m o u l d thermocouples located at the midface may not be adequate for the detection. Thus , it may be preferable s imp ly to moni tor metal l eve l fluctuations and m o u l d w a l l temperatures and mainta in values w i t h i n a desired range where bleed format ion is m i n i m a l . 257 [5] In the case o f rhomboid i ty , the average difference between m a x i m u m and m i n i m u m temperature for the four wa l l s w i l l be computed for one or more locat ions situated 25-50 m m be low the meniscus ( in the v i c in i ty o f peak hot face temperature o f mou ld ) , for a t ime per iod corresponding to the entire b i l le t length. T h e peak hot face temperature w i l l be computed us ing mathematical models embedded i n the "intell igent m o u l d " and decis ions w i l l be made us ing these informat ion and the nature o f metal l eve l fluctuations. G. Transient Thermal Response of Mould [1] W i t h respect to moni tor ing defects on-l ine, an understanding o f the transient response o f the m o u l d is necessary. It was found that the response t ime for the m o u l d to reach steady state is about 10 s. [2] M o u l d osc i l l a t ion was simulated i n the mathematical mode l by m o v i n g the heat f lux profi le up and d o w n the m o u l d (for a desired osc i l la t ion frequency and stroke length). It was found that the thermal fluctuations due to m o u l d osc i l la t ion is m a x i m u m at the meniscus. In addi t ion, the disturbances at the mid-thickness thermocouple locat ions, are quite sma l l compared to those generated by metal l eve l fluctuations. [3] M e t a l l eve l change (rise and fall) was s imulated i n the mathematical m o d e l by s imp ly m o v i n g the average measured heat f lux profi le ax ia l ly up and d o w n the m o u l d for a desired duration. The relat ionship between measured temperature change and corresponding metal l eve l change depends on the m o u l d peak hot face temperature, the distance o f the thermocouple f rom the average meniscus posi t ion and also, the distance o f the 258 thermocouple tip f rom the hot face o f the m o u l d . The t ime durat ion o f the change i n metal l eve l is also important since there is a delay i n the response o f the m o u l d to thermal disturbances generated on the hot face. [4] S imula t ions o f transverse depressions (manifested as "val leys" i n m o u l d heat f lux) indicate that the size o f depressions (depth and w i d t h i n the casting direct ion) , the spacing (distance between consecut ive depressions) between them, and cast ing speed are factors that inf luence the temperature drop registered by the thermocouple. Thus , to translate the measured temperature drop into the size o f the depression on the b i l le t surface, these factors must be accounted for i n the calculat ions. In the "intel l igent m o u l d " , a neural-network-based module is required to q u i c k l y estimate the size o f the depression f rom the temperature drop recorded by the thermocouples; the f ina l dec i s ion on the magnitude o f the temperature drop w i t h regard to the severity o f the p rob lem, w i l l be made us ing a reasoning based on fuzzy log ic . [5] Ca lcu la t ions suggest that the presence o f transverse depressions o f depth - 3 . 0 m m can cause the m o u l d heat f lux to drop by - 8 0 pet. Var ia t ions i n m o u l d temperature were computed for 46 transverse depressions present on the surface o f a b i l le t (#2 i n t r ia l D - 2 ) ; there is reasonable agreement between the measured and calculated temperatures. Thus , each "va l ley" (both partial and complete) i n the m o u l d temperature corresponds to a transverse depression on the bi l le t surface. Furthermore, the shape o f the "va l leys" are affected by the spacing between consecutive depressions. 11.1 Generation of New Knowledge and Major Contributions T h e f o l l o w i n g is a l ist o f "new knowledge" generated dur ing this study and the major contr ibutions o f this w o r k to the understanding and advancement o f the b i l le t cast ing process: 259 [1] T h i s study has uncovered the mechanisms o f defects (laps, bleeds and rhomboid i ty ) format ion i n the meniscus region and the importance o f metal l eve l f luctuations. P r io r to this work , there was no clear explanat ion on the role o f metal l eve l fluctuations i n generating these p rob l ems . It is now clear h o w metal l eve l fluctuations can create adverse heat transfer and lubr ica t ion condit ions i n the meniscus region and cause defects such as laps, bleeds and rhomboid i ty i n bi l lets . The randomness associated w i t h the severity o f several problems that originate i n the v i c in i ty o f the meniscus, is n o w easy to comprehend on the basis o f metal l eve l fluctuations. [2] T h e importance o f peak hot face temperature o f the m o u l d relative to the b o i l i n g range o f lubr ica t ing o i l , has been highl ighted w i t h respect to the formation o f defects such as laps, bleeds and rhomboid i ty . It was also shown that the greatest enhancement i n heat transfer due to o i l b o i l i n g and pyro lys i s , occurs when the peak hot face temperature l ies i n the b o i l i n g range o f o i l . [3] A new mechan i sm based on meniscus fluctuations, peak m o u l d hot face temperature, and o i l dis t r ibut ion has been proposed to expla in the formation o f laps and bleeds i n h igh carbon steels. The effect o f variables such as steel grade, m o u l d taper at the meniscus, c o o l i n g water ve loc i ty and quali ty, osc i l la t ion frequency and maintenance o f o i l dis t r ibut ion system on the severity o f laps and bleeds observed i n the plant trials at the various companies is consistent w i t h the proposed mechanism. [4] W i t h respect to rhomboid i ty , the difference between the two diagonals was measured a long the entire b i l le t length and reported for the first t ime. It was shown that the magnitude and orientation o f rhomboid i ty can change s ignif icant ly i n a bi l le t . Thus , the exis t ing procedure i n v o l v i n g a single measurement per heat, may be mis lead ing . 260 [5] It was found that a combina t ion o f h igh m o u l d heat transfer and short freezing range is important w i t h respect to the severity o f the rhomboid i ty , i n addi t ion to var iable heat transfer i n the meniscus region. T h i s new f inding can expla in the severity o f rhomboid i ty i n m e d i u m carbon steel that was evident i n this study and also reported i n several previous ones [25-27]. Calcula t ions show that variations i n heat transfer i n the meniscus region, have the greatest impact on the m e d i u m carbon steels as compared to l o w carbon and h igh carbon steels. [6] T h i s study shows that by adopting a "co ld m o u l d " practice, it is possible to m i n i m i z e , i f not e l iminate quali ty problems that are caused by non-uni form heat transfer and adverse lubr ica t ion condi t ions i n the meniscus region. [7] W i t h respect to the development o f the "intell igent mou ld" , strategies for the on- l ine , hot inspect ion o f defects such as laps, bleeds and rhomboid i ty , were out l ined for the first t ime based on a comprehensive analysis o f measured m o u l d thermocouple response and b i l le t qual i ty evaluat ion. [8] S i m u l a t i o n o f transient events i n the m o u l d were carr ied out us ing an adapted vers ion o f the m o u l d heat transfer mode l [7-10]. T h i s study elucidates the effect o f events such as metal l eve l fluctuations, m o u l d osc i l la t ion and presence o f defects (transverse depressions) on m o u l d thermal response. In the case o f transverse depressions, the characteristics o f transverse depressions were measured on an entire b i l le t and correlated w i t h corresponding measured m o u l d thermal response. 11.2 Recommendations for Future Work It is clear that metal l eve l f luctuation i n the continuous casting m o u l d is the biggest r andom upset that occurs i n the meniscus region and triggers several b i l le t defects. A l t h o u g h metal l eve l 261 f luctuat ion is an important variable, data on metal l eve l fluctuations are not recorded for each b i l le t i n most m i n i steel plants. Thus , i n the absence o f any data on metal l eve l fluctuations, it becomes extremely diff icul t to trouble-shoot qual i ty problems. It is therefore, r ecommended that data on metal l eve l fluctuations and tundish stream shape be recorded and reported for each bi l le t . A l t h o u g h it was qual i tat ively shown (wi th the help o f the S C A D A system) that a turbulent tundish stream generated metal l eve l fluctuations, an empi r i ca l relat ionship between tundish stream shape and temperature variations recorded above the meniscus c o u l d not be established i n this study. Thus , a series o f tests must be carr ied out to f i l m and photograph the meniscus and the tundish stream and measure m o u l d temperatures above the meniscus to characterize the roughness o f the tundish stream i n terms o f var ia t ion i n m o u l d temperature above the meniscus. T h e S C A D A system where m o u l d temperatures can be moni tored on- l ine , can be used to col lec t the required temperature data. W i t h regard to rhomboid i ty , it is necessary to test the mechanisms proposed i n this study. Thus , it is important to obtain a series o f rhomboid i ty values (one every foot) for the entire b i l le t together w i t h data on m o u l d temperature and metal l eve l fluctuations. In addi t ion to these measurements, mathematical mode l l i ng o f thermal distort ion o f b i l le t (having non-un i fo rm s o l i d shell) dur ing spray c o o l i n g is required to quantify the effect o f non-un i fo rm s o l i d shel l thickness on the f inal rhomboid i ty value. T h i s w i l l also test the earlier mechan i sm [22] on the generation o f rhomboid i ty due to non-uni form shel l thickness. F o r the detection o f defects i n the "intelligent mou ld" , a neural-network modu le is required to translate the measured temperature drops to the actual size o f transverse depressions (wid th and span). T h i s module w i l l be based on variables such as depth and span o f "va l leys" i n temperature, distance between consecutive "val leys" , casting speed and m o u l d hot face 262 temperature. W i t h respect to the setpoints to be employed i n the "intell igent m o u l d " , further w o r k is required to refine the current values implemented i n the in i t i a l vers ion that was recently tested i n C o m p a n y D . Th i s task o f refinement o f the knowledge base, can proceed i n paral le l w i t h the development and testing o f the "intell igent mou ld" . 263 REFERENCES 1 J . K . B r i m a c o m b e : "Empowerment w i t h K n o w l e d g e - T o w a r d the Intell igent M o u l d for the Cont inuous Cas t ing o f Steel B i l l e t s " , M e t a l l u r g i c a l Transactions B . 1993, V o l . 2 4 B , N o . 6 , pp.917-935. 2 S. K u m a r , J . A . M e e c h , L V . Samarasekera and J . K . B r i m a c o m b e : " K n o w l e d g e Eng inee r ing an Exper t Sys tem to Diagnose Qua l i ty Problems i n the Cont inuous Cas t ing o f Steel B i l l e t s " , I ron and Steelmaker. 1993, V o l . 2 0 , N o . 9 , pp.29-36. 3 I . V . Samarasekera, J . K . B r i m a c o m b e and K . W i l d e r : "The Pursui t o f Steel B i l l e t Qua l i t y " , I ron and Steelmaker. 1994, V o l . 2 1 , N o . 3 , pp.53-63. 4 I . V . Samarasekera, S. Chandra , A . Z . C a m p b e l l and J . K . B r i m a c o m b e : " M o u l d Shape, Heat Transfer and M o u l d - S t r a n d Interaction i n the Cont inuous Cas t ing o f Steel B i l l e t s " , Proceedings o f Conference on Cont inuous Cas t ing o f Steel i n D e v e l o p i n g Countr ies . Ch inese Socie ty o f Me ta l s , B e i j i n g , C h i n a , 1993, pp. 183-199. 5 S. Chandra : "Heat Transfer, O i l Lub r i ca t i on and M o u l d Taper i n Steel B i l l e t Cas t ing M a c h i n e s " , P h . D . Thesis . The Unive r s i ty o f B r i t i s h C o l u m b i a , V a n c o u v e r , Canada , 1992. 6 I .V.Samarasekera , J . K . B r i m a c o m b e , P . V . B a r r and J . A . M e e c h : "Smart Cont inuous Cas t ing and Smart Reheat Furnace", Research Proposal Submit ted to N S E R C Canada . T h e Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada , 1992. 7 I . V . Samarasekera: "Thermal Dis to r t ion o f Cont inuous Cas t ing M o u l d s " , P h . D . Thes is . T h e Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouver , Canada , 1980. 8 I . V . Samarasekera and J . K . B r i m a c o m b e : "Thermal and M e c h a n i c a l B e h a v i o u r o f Cont inuous-Cas t ing B i l l e t M o u l d s " , I ronmaking and S tee lmaking . 1982, V o l . 9 , pp.1-15. 9 I . V . Samarasekera and J . K . B r i m a c o m b e : "The T h e r m a l F i e l d i n Con t inuous -Cas t ing M o u l d s " , Canad ian M e t a l l u r g i c a l Quarter ly. 1979, V o l . 1 8 , pp.251-266. 10 I . V . Samarasekera, D . L . Ande r son and J . K . B r i m a c o m b e : "The T h e r m a l Dis to r t ion o f Cont inuous-Cas t ing B i l l e t M o u l d s " , M e t a l l u r g i c a l Transactions B . 1982, V o l . l 3 B , pp.91-104. 264 11 L V . Samarasekera and L K . B r i m a c o m b e : "The Influence o f M o u l d B e h a v i o u r on the Produc t ion o f Con t inuous ly Cast Steel B i l l e t s " , M e t a l l u r g i c a l Transactions B . 1982, V o l . l 3 B , pp.105-116. 12 R . L Dippenaar , L V . Samarasekera and L K . B r i m a c o m b e : " M o u l d Taper i n Cont inuous Cas t ing B i l l e t M a c h i n e s " , 1986, ISS Transact ions. V o l . 7 , pp.31-43. 13 S . C h a n d r a , L K . B r i m a c o m b e and I. V . Samarasekera: " M o u l d Strand Interaction i n the Cont inuous Cas t ing o f Steel B i l l e t s . Part III - M o u l d Heat Transfer and Taper i n Steel B i l l e t Cas t ing Mach ines" , I ronmaking and S tee lmaking . 1993, V o l . 2 0 , N o . 2 , pp. 104-112. 14 L V . Samarasekera and L K . B r i m a c o m b e : "Heat Ex t rac t ion Capab i l i t y o f Cont inuous-Cas t ing B i l l e t M o u l d s " , Proceedings o f W . O . P h i l b r o o k M e m o r i a l S y m p o s i u m Conference. I S S - A I M E , Warrendale , P A , 1988, pp. 157-171. 15 LV.Samaraseke ra , L K . B r i m a c o m b e and R . Bommara ju : " M o u l d B e h a v i o u r and So l id i f i ca t ion i n the Cont inuous Cas t ing o f Steel B i l l e t s , Part II - M o u l d Heat Ex t rac t ion , M o u l d - S h e l l Interaction and Osc i l l a t ion M a r k Format ion" , 1984, ISS Transact ions. V o l . 5 , pp.78-94. 16 L K . B r i m a c o m b e , L V . Samarasekera and R . Bommara ju : " O p t i m u m D e s i g n and Opera t ion o f M o u l d s for the Cont inuous Cas t ing o f Steel B i l l e t s " , Proceedings o f the 5th International Iron and Steel Congress . I S S - A I M E , Warrendale , P A , 1986, pp.409-423. 17 L K . B r i m a c o m b e and L V . Samarasekera: "Fundamental A n a l y s i s o f the Cont inuous Cas t ing Process for Qua l i ty Improvements", Proceedings o f I n d o - U S W o r k s h o p on Pr inc ip les o f So l id i f i ca t ion and Mater ia l s Process ing. Hyderabad , India , January, 1988, O x f o r d and I B H Pub l i sh ing C o m p a n y , N e w D e l h i , India , pp. 179-222. 18 S. K u m a r : " A n Exper t Sys tem to Diagnose Qua l i ty Prob lems i n the Cont inuous Cas t ing o f Steel B i l l e t s " , M . A . S c . Thesis . The Un ive r s i t y o f B r i t i s h C o l u m b i a , V a n c o u v e r , Canada , 1991. 19 L K . B r i m a c o m b e and K . Sor imach i : "Crack Fo rma t ion i n the Cont inuous Cas t ing o f Steel" , M e t a l l u r g i c a l Transactions B . 1977, V0I .8B, pp.489-505. 20 I . V . Samarasekera and L K . B r i m a c o m b e : " A p p l i c a t i o n o f Mathemat i ca l M o d e l s for the Improvement o f B i l l e t Qua l i ty" In M o u l d Operat ion for Qua l i t y and Produc t iv i ty . E d s . A . W . C r a m b , and E . S . Szekeres, I S S - A I M E , Warrendale , P A , 1991, pp.55-67. 265 21 I . V . Samarasekera, J . K . B r i m a c o m b e , H . M . Adje i -Sarpong , P . K . A g a r w a l and B . N . W a l k e r : "Bas i c K n o w l e d g e for the Intelligent Cont inuous Cas t ing M o u l d " , Proceedings o f 2nd Canada-Japan S y m p o s i u m on M o d e r n S tee lmaking and Cas t i ng Technique . E d s . J .J . Jonas, J . D . B o y d and N . Sano, 33rd Conference o f Meta l lu rg is t s , M e t a l l u r g i c a l Socie ty o f C I M , Toronto , Canada, 1994, pp. 169-187. 22 R . Bommara ju , J . K . B r i m a c o m b e and I . V . Samarasekera: " M o u l d B e h a v i o u r and So l id i f i ca t ion i n the Cont inuous Cas t ing o f Steel B i l l e t s , Part III - Structure, So l id i f i ca t ion Bands , C r a c k Format ion and Off-Squareness", 1984, ISS Transact ions. V o l . 5 , pp.95-104. 23 I . V . Samarasekera, R . Bommara ju and J . K . B r i m a c o m b e : "Factors Inf luencing the Fo rma t ion o f Osc i l l a t i on M a r k s i n the Cont inuous Cas t ing o f Steel B i l l e t s " , Proceedings o f International Conference on Cont inuous Cas t ing o f Steel B i l l e t s . 24th Conference o f Meta l lurg is t s , M e t a l l u r g i c a l Society o f C I M , Vancouve r , Canada , 1985, pp.33-58. 24 I . V . Samarasekera and J . K . B r i m a c o m b e : "Investigation o f M o u l d - R e l a t e d Qua l i t y Prob lems i n Cont inuous B i l l e t Cas t ing U s i n g Mathemat i ca l M o d e l s " , Proceedings o f 3rd P T P Conference on A p p l i c a t i o n o f Mathemat ica l M o d e l s i n the I ron and S tee lmaking Industry. I S S - A I M E , Warrendale , P A , 1982, pp.283-295. 25 D . P . Loren to : "Developments i n M o u l d Techno logy" , Paper presented to Globetrotters M e e t i n g , 1988. 26 V . Kru je l sk i s and J . C o o k : "The influence o f M o u l d Tube Taper and Di s to r t i on o f Cast B i l l e t Qua l i ty" , S tee lmaking Conference Proceedings. V o l . 7 1 , I S S - A I M E , Warrendale , P A , 1988, pp.349-352. 27 V . Kru je l sk i s and J . C o o k : "Init ial Exper iences w i t h a N e w B i l l e t / R o u n d s Caster - Factors Inf luencing B i l l e t R h o m b o i d i t y " , Proceedings o f the 4th International Conference on Cont inuous Cas t ing . M a y 1988, Brusse ls , B e l g i u m . 28 L A . B a k s h i , J . L . B rend zy , B . N . W a l k e r , S. Chandra , I . V . Samarasekera and J . K . B r i m a c o m b e : M o u l d Strand Interaction i n the Cont inuous Cas t ing o f Steel B i l l e t s . Part I - Industrial Tr i a l s " , I ronmaking and Stee lmaking . 1993, V o l . 2 0 , N o . l , pp.54-62. 29 J . L . B r e n d z y , L A . B a k s h i , B . N . W a l k e r , I . V . Samarasekera and J . K . B r i m a c o m b e : M o u l d Strand Interaction i n the Cont inuous Cas t ing o f Steel B i l l e t s . Part II - L u b r i c a t i o n and Osc i l l a t i on M a r k Format ion" , I ronmaking and S tee lmaking . 1993, V o l . 2 0 , N o . l , pp.63-74. 266 30 L A . B a k s h i , J . L . B r e n d z y , L V . Samarasekera and J . K . B r i m a c o m b e : "Measurement o f M e c h a n i c a l Interaction Be tween Strand and M o u l d i n the Cont inuous Cas t ing o f Steel B i l l e t s " , Proceedings W e i n b e r g S y m p o s i u m on So l id i f i ca t ion Process ing . 29th Conference o f Meta l lurg is t s , M e t a l l u r g i c a l Socie ty o f C I M , H a m i l t o n , Canada , 1990, pp. 146-156. 31 M . S . J e n k i n s , B . G . T h o m a s , W . C . C h e n and R . B . Mahapat ra : "Investigation o f Strand Surface Defects U s i n g M o u l d Instrumentation and M o d e l l i n g " , S tee lmaking Conference Proceedings. V o l . 7 7 , I S S - A I M E , Ch icago , I L , 1994, pp.337-345. 32 F . Haers and S . G . Thornton: " A p p l i c a t i o n o f M o u l d T h e r m a l M o n i t o r i n g on the T w o - S t r a n d Slab Caster at S idmar" , I ronmaking and S tee lmaking . 1994, V o l . 2 1 , N o . 5 , pp.390-398. 33 W . H . E m l i n g and S. D a w s o n : " M o u l d Instrumentation for Breakou t Detec t ion and C o n t r o l " , In M o u l d Operat ion for Qua l i ty and Produc t iv i ty . E d s . A . W . C r a m b , and E . S . Szekeres, I S S - A I M E , Warrendale , P A , 1991, pp .161-181. 34 M . R . O z g u : " M o u l d and Strand G u i d e Instrumentation for Cont inuous S lab Casters", Proceedings o f 13th P T P Conference. I S S - A I M E , N a s h v i l l e , T N , 1995, pp. 177-200. 35 L D . S o m m e r v i l l e and E . L M c K e o g h : "Reoxida t ion o f Steel by A i r En t ra ined D u r i n g Cas t ing" , Proceedings o f 2nd P T P Conference. I S S - A I M E , Warrendale , P A , 1981, pp.256-268. 36 L J . Heas l ip , A . M c L e a n and I . P . Sommerv i l l e : " C h e m i c a l and P h y s i c a l Interactions D u r i n g Transfer Operations", Cont inuous Cas t ing V o l . 1. I S S - A I M E , Warrendale , P A , 1983. 37 S . M . G o k l u and K . W . Lange : " A i r Entrainment b y O p e n Stream Cas t ing" , Proceedings o f 6th P T P Conference. I S S - A I M E , Warrendale , P A , 1986, pp.1135-1146. 38 P . M a s s a r d and K . W . Lange : "Breakup L e n g t h o f Unstable P o u r i n g Jets", A r c h i v . Eisenhuttenwesen. 1977, V o l . 4 8 , N o . 2 , pp.65-69. 39 P . M a s s a r d and K . W . Lange : "Characteristics Var iab les o f the Fragments A r i s i n g P u r i n g Jet Breakup" , A r c h i v . Eisenhuttenwesen. 1977, V o l . 4 8 , N o . 4 , pp. 199-202. 4 0 K . W . L a n g e and P . Massa rd : " O n the Phase Bounda ry Peca rbur i za t ion o f P o u r i n g Jets a n d P r o p l e t s " , A r c h i v . Eisenhuttenwesen. 1977, V o l . 4 8 , N o . 7 , pp .367-372. 267 41 K . W . L a n g e and P . Massa rd : "Intake o f O x y g e n by Pour ing Jets F a l l i n g i n A i r " , A r c h i v Eisenhuttenwesen. 1977, V o l . 4 8 , N o . 10, 1977, pp.521-526. 42 K . Iwata, T . C h o h and M . Inouye: " A M o d e l Exper imen t on the G a s Entra inment w i t h P o u r i n g L i q u i d Steel Stream", Transactions o f the Iron and Steel Institute o f Japan. 1983, V o l . 2 3 , N o . 3 , pp.218-227. 43 E . J . M c K e o g h and D . A . E r v i n e : " A i r Entrainment Rate and D i f fu s ion Pattern o f P l u n g i n g L i q u i d Jets", C h e m i c a l Engineer ing Science . 1981, V o l . 3 6 , pp .1161-1172. 44 E . V a n D e Sande and J . M . Smi th : "Surface Entrainment o f A i r by H i g h V e l o c i t y Wa te r Jets": C h e m i c a l Engineer ing Science . 1973, V o l . 2 8 , pp. 1161-1168. 45 E . V a n D e Sande and J . M . Smi th : "Jet Break-up and A i r Entra inment by L o w V e l o c i t y Turbulent Wate r Jets", C h e m i c a l Engineer ing Science . 1976, V o l . 3 1 , pp.219-224. 46 J . K . B r i m a c o m b e , I . V . Samarasekera, B . N . W a l k e r , L A . B a k s h i , R . B o m m a r a j u , F . W e i n b e r g and E . B . H a w b o l t : " M o u l d Behav iou r and So l id i f i ca t ion i n the Cont inuous Cas t ing o f Steel B i l l e t s , Part I - Industrial Tr i a l s " , ISS Transact ions. 1984, V o l . 5 , pp.71-77. 47 S.Petry: " A d v a n c e d T h e r m a l M o u l d M o n i t o r i n g " . Proceedings o f 13th P T P Conference . I S S - A I M E , N a s h v i l l e , T N , 1995, pp.209-215. 48 J . S u n i and H . Hene in : " A n a l y s i s o f S h e l l Thickness Irregularity i n Con t inuous ly Cast M i d d l e C a r b o n Steel Slabs U s i n g M o u l d Thermocouple Pa t a " , S tee lmaking Conference Proceedings . V o l . 7 7 , I S S - A I M E , Ch icago , I L , 1994, pp.331-336. 49 B . N . W a l k e r , I .V.Samarasekera and J . K . B r i m a c o m b e : "Measu r ing H o r i z o n t a l M o v e m e n t s o f B i l l e t Cas t ing M o u l d " , Unpub l i shed W o r k , The U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1995. 50 W . R . I rv ing: "Cont inuous Cas t ing o f Steel" , The Institute o f Mate r i a l s , L o n d o n , E n g l a n d , 1993. 51 Plant T r i a l Report to C o m p a n y A , Unpub l i shed W o r k , T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1992. 52 Plant T r i a l Report to C o m p a n y A , Unpub l i shed W o r k , T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1994. 268 53 Plant T r i a l Report to C o m p a n y C , Unpub l i shed W o r k , T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1991. 54 Plant T r i a l Repor t to C o m p a n y D , Unpub l i shed W o r k , The U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1993. 55 Plant T r i a l Report to C o m p a n y E , Unpub l i shed W o r k , The U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1991. 56 R . B e r r y m a n , L V . Samarasekera and L K . B r i m a c o m b e : " C o o l i n g Wate r F l o w i n Cont inuous-Cas t ing B i l l e t M o u l d s " , ISS Transactions. 1988, V o l . 1 , pp .69-77. 57 L A . B a k s h i , E . O s i n s k i , L V . Samarasekera and L K . B r i m a c o m b e : "Measurement o f Osc i l l a t i on M a r k s on Cont inuous ly Cast B i l l e t s w i t h an Au toma ted Prof i lometer" , Proceedings o f International S y m p o s i u m on Di rec t R o l l i n g and H o t C h a r g i n g o f Strand Cas t B i l l e t s . 27th Conference o f Meta l lurg is t s , M e t a l l u r g i c a l Socie ty o f C I M , M o n t r e a l , Canada , 1988, pp.60-69. 58 O . K o n d o , K . Hamada , T . Kur ibayash i , Y . T a n i z a w a , A . Sato, and T . Inoue: " N e w D y n a m i c Spray C o n t r o l Sys tem for Secondary C o o l i n g Z o n e o f Cont inuous Cas t ing M a c h i n e " , S tee lmaking Conference Proceedings. V o l . 7 7 , I S S - A I M E , C h i c a g o , I L , 1994, pp.309-314. 59 M . L . T E K T O R - M o u l d Lubr i ca t i on Detector, P B K Installat ion Brochure , 1984. 60 V . R a k o c e v i c , J . A . M e e c h , S. K u m a r , L V . Samarasekera and L K . B r i m a c o m b e : "Computa t iona l Intell igence i n a R e a l - T i m e S C A D A Sys t em to M o n i t o r and C o n t r o l Cont inuous Cas t ing o f Steel B i l l e t s " , Presented at I P S A 1995, V I International F u z z y Systems Assoc i a t i on W o r l d Congress , Sao Paulo , B r a z i l , J u l y 1995. 61 V . R a k o c e v i c , S. K u m a r , R . Gur ton , J . A . M e e c h , L V . Samarasekera and L K . B r i m a c o m b e : "Real- t ime Exper t Sys tem for the Cont inuous Cas t ing o f Steel B i l l e t s " , Presented at the 16th C I M Dis t r ic t 6 A n n u a l M e e t i n g , C I M , V a n c o u v e r , Canada , 1994, paper 131. 62 V . R a k o c e v i c : "Development o f a S C A D A Sys tem to M o n i t o r and C o n t r o l Cont inuous Cas t ing o f Steel B i l l e t s " , M . A . S c . Thes is . The Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1995. 63 R . B . Mahapat ra : " M o u l d Behav iou r and Product Qua l i ty i n Cont inuous Cas t i ng o f Slabs" , P h . D . Thesis . The Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada , 1989. 269 64 R . B . Mahapat ra , J . K . B r i m a c o m b e , I . V . Samarasekera, B . N . W a l k e r , E . A . Paterson and J . D . Y o u n g : " M o u l d Behav iou r and Its Influence on Qua l i t y i n the Cont inuous Cas t ing o f Steel Slabs: Part I. Industrial T r i a l s , M o u l d Temperature Measurements , and Mathemat i ca l M o d e l l i n g " , M e t a l l u r g i c a l Transactions B . 1991, V o l . 2 2 B , pp.861-874. 65 R . B . Mahapat ra , J . K . B r i m a c o m b e and I . V . Samarasekera: " M o u l d B e h a v i o u r and Its Influence on Qual i ty i n the Cont inuous Cas t ing o f Steel Slabs: Part II. M o u l d Heat Transfer, M o u l d F l u x Behav iour , Format ion o f Osc i l l a t i on M a r k s , L o n g i t u d i n a l Of f -Corne r Depressions, and Subsurface Cracks" . M e t a l l u r g i c a l Transact ions B . 1991, V o l . 2 2 B , pp.875-888. 66 B . N . W a l k e r , I .V.Samarasekera and J . K . B r i m a c o m b e : "Des ign o f Thermocouples for the Intelligent M o u l d " , U n p u b l i s h e d W o r k , The Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1995. 67 T h e Temperature Handbook , O m e g a Engineer ing Inc., U S A , 1989. 68 K . W i l d e r : U n p u b l i s h e d work , B i l l e t Cas t ing G r o u p , The Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1993. 69 R . Gur ton , B . N . W a l k e r , I . V . Samarasekera and J . K . B r i m a c o m b e : U n p u b l i s h e d work , B i l l e t Cas t ing G r o u p , The Unive r s i ty o f B r i t i s h C o l u m b i a , V a n c o u v e r , Canada , 1995. 70 U B C Cont inuous Cas t ing Lubr i ca t i on Sys tem, A p p l i c a t i o n for U S Patent, T h e Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada , 1990. 71 C o m p a n y D : Private C o m m u n i c a t i o n , 1995. 72 Plant T r i a l Report to C o m p a n y B , Unpub l i shed W o r k , T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1989. 73 B . Carnahan, H . A . Luther , and J . O . W i l k e s : A p p l i e d N u m e r i c a l M e t h o d s , W i l l e y , N e w Y o r k , N Y , 1960. 74 Plant T r i a l at C o m p a n y A , Unpub l i shed W o r k , The Un ive r s i t y o f B r i t i s h C o l u m b i a , Vancouve r , Canada, 1995. 270 75 I. V . Samarasekera, J . K . B r i m a c o m b e , S. K u m a r and A . M c L e a n : "The T u n d i s h P o u r i n g Stream and Cast B i l l e t Qua l i ty" , Proceedings o f International S y m p o s i u m on Near-Net -Shape casting i n the M i n i m i l l s . 34rd Conference o f Meta l lu rg is t s , M e t a l l u r g i c a l Socie ty o f C I M , Vancouve r , Canada , 1995, pp.135-160. 76 J . K . B r i m a c o m b e : "The D e s i g n o f Cont inuous Cas t ing M a c h i n e s B a s e d on a H e a t - F l o w A n a l y s i s : State-of-the-Art R e v i e w " , Canad ian M e t a l l u r g i c a l Quarter ly . 1976, V o l . 1 5 , pp .163-175. 77 J . K . B r i m a c o m b e , P . K . A g a r w a l , S. H i b b i n s , B . Prabhaker and L . A . Bapt is ta : "Spray C o o l i n g i n the Cont inuous Cas t ing o f Steel" , S tee lmaking Conference Proceedings . V o l . 6 3 , I S S - A I M E , Warrendale , P A , 1980, pp.235-232. 271 APPENDIX A - Data Acquisition Hardware 1. Met raby te ' s U n i v e r s a l Expans ion Interface. M o d e l E X P - 1 6 M e g a b y t e ' s U n i v e r s a l Expans ion Interface, M o d e l E X P - 1 6 is an expans ion mul t ip lexer /ampl i f ie r system that can be used w i t h any data acquis i t ion system. E a c h E X P - 1 6 concentrates 16 differential analog input channels into one analog output channel and also provides s ignal ampl i f ica t ion , f i l ter ing and condi t ion ing . In addi t ion, the instrumentation ampl i f ie r provides gains o f 0.5, 1.0, 2.0, 10.0, 50.0 and 100 as w e l l as programmable gain capabi l i ty . T h e 16 differential input channels are selected by a s o l i d state 4 bi t T T L / C M O S compat ib le address. P r o v i s i o n is made on the board for f i l ter ing, attenuation and measur ing current instead o f voltage. A l l analog input connections are made o n miniature screw connector strips. Co ld - junc t ion sensing, compensat ion c i rcui t ry and a b ias ing resistor for open thermocouple detection exist i n the system. The E X P - 1 6 can be direct ly connected to Met raby te ' s D A S - 8 or sets o f E X P - 1 6 s can be cascaded by ident ical cables to a total o f 128 channels o f standard voltage and 112 o f thermocouple measurement. W h e n used w i t h D A S - 8 , channel select ion is v i a the O P 1 - 4 digi ta l outputs o f D A S - 8 . 2. Met raby te ' s D A S - 8 Sys tem Met raby te ' s D A S - 8 is an 8 channel , 12 bit h igh speed A / D converter and t imer/counter board designed for the I B M - P C . The D A S - 8 board is 5" l ong and can be fitted into a "half-slot" o f an I B M - P C . A l l connections are made through a standard 37 p i n D male connector that projects through the back o f the computer. The D A S - 8 is a successive approximat ion A / D converter w i t h sample/hold . T h e fu l l scale input o f each channel is +/-5.0 V w i t h a resolut ion o f 2.44 m V and the inputs are single ended w i t h a c o m m o n ground. The A / D convers ion t ime is around 25 microseconds (35 microseconds max) . The 8254 programmable counter t imer w h i c h provides 272 per iodic interrupts for the A / D converter has 3 separate 16 bit d o w n counter one o f w h i c h is connected to the sub-mult iple o f the system c l o c k and a l l the I /O functions o f the remain ing two are accessible to the user. Input frequencies o f up to 2.5 M H z can be handled by the 8254 programmable timer. The 7 bits o f T T L d ig i ta l I /O provided are composed o f one output port o f 4 bits and one input port o f 3 bits. E a c h output handles 5 standard T T L loads (8 m A sink current). A prec is ion o f +10.00 (+/0.1 V ) reference output is der ived f rom the A / D converter reference. The output can source/sink 2 m A . 273 APPENDIX B Chemical Composition of Heats Monitored at the Plant Trials 274 C O M P A N Y A ( T r i a l #1) Heat No. Grade Chemical Analysis (weight pet.) C Mn S. P Si Cu Cr Ni Mo V Nb Sn Pb Zn Al E-25982 9707A 1.05 0.56 .020 .015 0.14 0.28 0.19 0.12 .028 .019 .023 E-25984 9707A 1.00 0.63 .016 .020 0.18 0.29 0.19 0.12 .029 .021 .014 E-25987 9707A 1.00 0.59 .024 .018 0.16 0.25 0.23 0.14 .046 .019 .026 E-25990 9707A 1.02 0.57 .027 .019 0.20 0.28 0.18 0.11 .034 .022 .017 D-25967 8795F 0.92 0.63 .019 .017 0.17 0.27 0.39 0.15 .062 .014 .015 D-25968 8795F 0.89 0.60 .026 .014 0.17 0.24 0.39 0.17 .072 .013 .014 D-25969 8795F 0.90 0.68 .022 .019 0.18 0.22 0.40 0.18 .078 .014 .013 D-25973 8795F 0.92 0.59 .019 .010 0.17 0.26 0.38 0.14 .064 .016 .014 E-25954 8795H 0.93 0.65 .017 .017 0.18 0.23 0.49 0.15 .066 .020 .012 E-25955 8795H 0.90 0.67 .016 .014 0.18 0.21 0.48 0.14 .066 .024 .010 E-25956 8795H 0.89 0.67 .022 .015 0.19 0.20 0.49 0.15 .070 .021 .012 E-25957 8795H 0.91 0.73 .019 .017 0.21 0.25 0.50 0.12 .042 .021 .012 E-25958 8795H 0.91 0.71 .030 .014 0.26 0.27 0.49 0.11 .043 .021 .015 E-25959 8795H 0.93 0.76 .032 .017 0.19 0.25 0.50 0.14 .047 .019 .021 E-25968 8795H 0.90 0.73 .021 .013 0.21 0.21 0.49 0.13 .053 .019 .014 E-25970 8795H 0.88 0.69 .024 .018 0.15 0.25 0.49 0.12 .046 .021 .017 E-25962 8693A 0.89 0.82 .017 .023 0.21 0.27 0.20 0.12 .031 .002 .016 E-25963 8693A 0.87 0.76 .019 .020 0.19 0.30 0.26 0.17 .045 .002 .015 E-25964 8693A 0.89 0.81 .021 .026 0.20 0.30 0.25 0.13 .039 .002 .024 E-25965 8693A 0.92 0.72 .026 .018 0.18 0.28 0.19 0.11 .032 .001 .024 E-25966 8693A 0.89 0.78 .021 .023 0.19 0.29 0.23 0.11 .034 .002 .024 E-25973 7785A 0.79 0.68 .021 .018 0.20 0.29 0.58 0.19 .062 .019 .014 E-25979 7785A 0.82 0.82 .015 .021 0.20 0.24 0.58 0.14 .047 .023 .014 E-25980 4045A 0.43 0.85 .020 .020 0.32 0.22 0.85 0.12 .176 .037 .010 E-25981 4045A 0.41 0.90 .027 .025 0.28 0.26 0.86 0.14 .181 .032 .020 E-25991 4045A 0.41 0.90 .018 .017 0.25 0.22 0.87 0.11 .178 .037 .018 E-25967 2833A 0.28 0.77 .022 .015 0.25 0.23 0.51 0.54 .194 .037 .012 E-25995 1823A 0.19 0.76 .027 .014 0.33 0.31 0.54 0.52 .187 .036 .010 E-25992 1520A 0.15 0.80 .027 .022 0.18 0.27 0.33 0.16 .060 .001 .023 E-25993 1520A 0.17 0.72 .016 .013 0.20 0.27 0.23 0.21 .054 .001 .016 E-25994 1520A 0.14 0.72 .018 .014 0.21 0.37 0.38 0.21 .059 .001 .026 Note: Al l of the above heats were cast with oil lubrication 275 C O M P A N Y A ( T r i a l #2) Heat No. Grade Chemical Analysis (weight pet) C Mn S P Si Cu Cr Ni Mo V Nb Sn Pb Zn Al E-30708 - 0.43 0.67 .025 .014 0.19 0.21 0.20 0.11 .044 .001 .031 .012 E-30709 - 0.31 0.74 .023 .010 0.20 0.19 0.16 0.10 .028 .001 .001 .009 E-30710 - 0.19 0.53 .022 .008 0.22 0.27 0.13 0.09 .026 .001 .023 .011 E-30711 - 0.19 0.46 .015 .008 0.21 0.23 0.13 0.11 .028 .001 .023 .009 E-30712 - 0.19 0.48 .017 .007 0.17 0.17 0.14 0.08 .027 .001 .025 .007 E-30713 - 0.19 0.50 .016 .008 0.18 0.22 0.12 0.08 .021 .000 .026 .008 E-30714 - 0.19 0.50 .041 .015 0.19 0.24 0.12 0.08 .035 .001 .023 .008 E-30715 - 0.16 0.84 .024 .016 0.27 0.20 0.15 0.08 .023 .001 .026 .009 E-30716 - 0.41 0.83 .023 .012 0.24 0.21 0.86 0.12 .196 .004 .034 .010 E-30717 - 0.88 0.75 .031 .019 0.19 0.32 0.24 0.14 .050 .002 .003 .011 E-30718 - 0.93 0.70 .026 .020 0.21 0.29 0.22 0.11 .039 .000 .002 .013 E-30719 - 0.90 0.68 .020 .011 0.20 0.24 0.15 0.09 .028 .000 .002 .017 E-30720 - 0.88 0.75 .020 .020 0.20 0.26 0.16 0.11 .037 .001 .002 .020 E-30727 - 0.71 0.82 .024 .014 0.24 0.23 0.88 0.11 .049 .003 .019 .012 E-30728 - 0.72 0.81 .016 .015 0.16 0.22 0.89 0.16 .065 .003 .021 .009 E-30729 - 0.68 0.83 .015 .014 0.15 0.23 0.23 0.12 .185 .001 .024 .011 E-30730 - 0.67 0.79 .011 .014 0.11 0.24 0.27 0.16 .186 .002 .022 .016 E-30737 - 0.79 0.83 .016 .011 0.16 0.24 0.59 0.13 .056 .002 .025 .014 E-30738 - 0.81 0.79 .017 .015 0.17 0.22 0.61 0.12 .081 .001 .021 .014 E-30743 - 0.93 1.14 .016 .021 0.16 0.27 0.25 0.13 .046 .004 .001 .014 Note: Heat # E30708 to E30718 were cast with mould powder lubrication while Heat # E30719 to E30743 were cast with oil lubrication 276 C O M P A N Y B ( T r i a l #1) Heat No. Grade Chemical Analysis (weight pet) C Mn S P Si Cu Cr Ni Mo V Nb Sn Pb Zn Al B24636 - .046 0.53 .038 .012 0.16 0.11 0.07 0.08 .020 .006 .004 B24637 - .040 0.42 .039 .017 0.12 0.09 0.07 0.08 .020 .006 .004 B24638 - .068 0.46 .026 .009 0.14 0.09 0.04 0.06 .020 .005 .003 B24639 - .045 0.36 .022 .009 0.08 6.10 0.03 0.08 .020 .006 .003 B24640 - .035 0.37 .026 .011 0.09 0.12 0.04 0.07 .020 .006 .004 B24641 - • .041 0.48 .020 .008 0.09 0.10 0.03 0.07 .020 .005 .004 B24642 - .043 0.44 .020 .008 0.12 0.09 0.03 0.07 .020 .005 .003 B24643 - .170 0.83 .018 .018 0.20 0.10 0.08 0.08 .020 .006 .004 B24644 •- .180 0.75 .013 .010 0.24 0.17 0.07 0.10 .020 .000 .005 B24645 - .180 0.85 .018 .018 0.28 0.14 0.07 0.09 .020 .009 .005 B24646 - .400 0.78 .015 .014 0.24 0.13 0.10 0.10 .020 .008 .005 B24647 - .420 0.81 .019 .012 0.21 0.13 0.10 0.09 .020 .008 .004 A23408 - .051 0.35 .021 .014 0.10 0.11 0.04 0.09 .020 .007 .004 B24649 - .050 0.40 .020 .011 0.11 0.09 0.07 0.09 .010 .007 .005 A23409 - .150 0.35 .026 .017 0.10 0.27 0.07 0.11 .020 .010 .004 B24650 - .120 0.47 .018 .013 0.10 0.21 0.05 0.10 .020 .011 .004 A23412 - .090 0.41 .025 .014 0.12 0.35 0.06 0.10 .020 .011 .004 B24653 - .094 0.46 .020 .010 0.11 0.19 0.04 0.08 .020 .009 .004 B24654 - .120 0.40 .022 .008 0.10 0.24 0.05 0.10 .020 .010 .004 A23413 - .120 0.42 .020 .008 0.12 0.23 0.04 0.10 .020 .009 .004 B24655 - .110 0.35 .023 .007 0.08 0.20 0.04 0.11 .020 .012 .004 B24657 - .400 0.73 .018 .011 0.24 0.21 0.07 0.06 .021 .005 .004 Note: All of the above heats were cast with oil lubrication. 277 C O M P A N Y C ( T r i a l #1) Heat No. Grade Chemical Analysis (weight pet) C Mn S P Si Cu Cr Ni Mo V Nb Sn Pb Zn Al D-6122 1045 0.48 0.72 .018 .007 .24 .16 .05 .05 .006 .025 .006 D-6123 1045 0.49 0.69 .018 .008 .24 .15 .05 .05 .006 .023 .007 C-7653 1045 0.47 0.73 .017 .014 .26 .11 .05 .04 .006 .023 .004 C-7654 1045 0.46 0.82 .018 .014 .29 .12 .08 .05 .005 .024 .007 C-7655 1045 0.45 0.72 .014 .014 .28 .18 .09 .05 .047 .022 .007 C-7658 5160 0.57 0.77 .008 .008 .17 .11 .79 .05 .050 .015 .006 C-7659 5160 0.58 0.81 .013 .013 .20 .14 .80 .05 .059 .016 .007 C-7660 5160 0.57 0.81 .014 .014 .19 .14 .85 .06 .068 .018 .009 C-7661 5160 0.57 0.81 .014 .014 .19 .13 .86 .07 .004 .018 .008 C-7663 1141 0.38 1.42 .020 .020 .24 .11 .11 .03 .011 .027 .006 C-7664 L-325 0.21 0.88 .010 .010 .28 .39 .89 .70 .009 .009 .009 A-28184 L-325 0.24 0.88 .019 .019 .34 .46 .87 .62 .013 .010 .006 D-6131 1084 0.86 0.71 .012 .012 .24 .22 .10 .07 .007 .023 .008 A-28187 L-20 0.21 1.10 .021 .021 .21 .15 .07 .05 .004 .056 .007 A-28188 1045 0.46 0.69 .021 .021 .27 .13 .08 .04 .004 .024 .006 D-6135 1045 0.46 0.69 .010 .010 .24 .16 .05 .04 .009 .023 .006 A-28191 L-17C 0.21 0.95 .017 .017 .18 .10 .06 .05 .008 .004 .006 A-28192 L-17C 0.19 0.98 .015 .015 .19 .05 .05 .04 .012 .004 .003 A-28193 L-17C 0.19 1.00 .016 .016 .25 .27 .09 .06 .005 .004 .008 D-6143 L-17C 0.21 0.98 .008 .008 .19 .14 .02 .05 .00 .003 .006 Note: All of the above heats were cast with oil lubrication. 278 C O M P A N Y D ( T r i a l #1) Heat No. Grade Chemical Analysis (weight pet) C Mn S P Si Cu Cr Ni Mo Ti B Sn N Ca Al 531142 - 0.12 0.84 .028 .019 0.21 0.26 0.10 0.10 0.02 .016 531146 - 0.32 0.71 .023 .023 0.20 0.30 0.10 0.10 0.02 .024 531147 - 0.32 0.86 .009 .022 0.24 0.27 0.09 0.10 0.02 .033 .0020 .016 .001 .010 531148 - 0.32 1.31 .006 .022 0.24 0.29 0.24 0.10 0.02 .033 .0032 .022 .002 .005 531149 - 0.84 0.71 .022 .023 0.24 0.35 0.09 0.10 0.02 .026 Note: All of the above heats were cast with oil lubrication. 279 C O M P A N Y D ( T r i a l #2) Heat No. Grade Chemical Analysis (weight pet.) C Mn S P Si Cu Cr Ni Mo Ti B Sn N Ca Al 541260 0.11 0.85 .021 .010 0.17 0.32 0.10 0.09 0.02 .013 541261 0.12 0.85 .022 .009 0.18 0.31 0.09 0.09 0.02 .014 541262 0.11 0.85 .021 .012 0.16 0.39 0.08 0.15 0.01 .019 541263 0.11 0.80 .023 .018 0.17 0.35 0.09 0.11 0.01 .018 541264 0.12 0.88 .020 .013 0.19 0.35 0.10 0.08 0.01 .018 541265 0.12 0.85 .021 .018 0.15 0.38 0.11 0.09 0.01 .016 541266 0.12 0.84 .022 .010 0.15 0.33 0.09 0.11 0.02 .021 541277 0.14 0.84 .020 .010 0.17 0.28 0.09 0.09 0.02 .014 541278 0.13 0.82 .021 .010 0.17 0.30 0.09 0.09 0.02 .015 541279 0.14 0.85 .019 .010 0.18 0.27 0.09 0.08 0.02 .015 541280 0.13 0.89 .020 .009 0.18 0.29 0.09 0.10 0.02 .017 541281 0.13 0.86 .021 .012 0.20 0.26 0.10 0.09 0.02 .015 541282 0.13 0.88 .024 .015 0.19 0.30 0.09 0.08 0.01 .014 541286 0.14 0.82 .017 .006 0.17 0.30 0.07 0.10 0.02 .013 541297 0.12 0.86 .018 .010 0.16 0.31 0.10 0.09 0.02 .012 541298 0.13 0.82 .018 .008 0.17 0.33 0.08 0.09 0.02 .014 541299 0.12 0.84 .019 .008 0.16 0.35 0.08 0.09 0.02 .017 541300 0.13 0.88 .020 .010 0.19 0.33 0.09 0.09 0.02 .016 541311 0.33 1.29 .007 .011 0.18 0.29 0.24 0.10 0.01 .038 .0027 .016 .0094 .002 .007 541312 0.30 1.33 .009 .010 0.20 0.31 0.24 0.09 0.01 .038 .0037 .016 .0086 .002 .007 541313 0.33 1.29 .011 .009 0.18 0.30 0.23 0.09 0.02 .038 .0028 .015 .0086 .002 .006 541314 0.31 1.26 .008 .009 0.17 0.30 0.24 0.12 0.03 .036 .0025 .014 .0095 .002 .006 541315 0.32 1.36 .006 .010 0.19 0.25 0.25 0.09 0.02 .035 .0024 .014 .0081 .002 .007 541316 0.32 1.31 .007 .008 0.20 0.24 0.26 0.08 0.02 .040 .0027 .014 .0079 .002 .007 541333 0.32 1.28 .005 .007 0.18 0.30 0.24 0.09 0.01 .038 .0023 .015 .0087 .003 .007 541334 0.32 1.26 .008 .009 0.19 0.33 0.23 0.09 0.01 .035 .0028 .015 .0081 .003 .006 541351 0.80 0.65 .012 .009 0.18 0.28 0.07 0.08 0.01 .014 541352 0.81 0.65 .017 .007 0.20 0.27 0.08 0.08 0.01 .015 541353 - 0.80 0.63 .024 .013 0.20 0.29 0.10 0.09 0.01 .016 Note: Heats 541260 to 541286 and 541311 to 541316 were cast with mould powder lubrication while Heats 541297 to 541300 and 541333 to 541353 were cast with oil lubrication 280 C O M P A N Y E ( T r i a l #1) Heat No. Grade Chemical Analysis (weight pet) C Mn S P Si Cu Cr Ni Mo V Nb Sn Pb Zn Al 26493 1056 0.57 0.79 .025 .010 0.23 0.09 0.78 0.05 .012 .029 .003 .007 .005 .002 .003 26494 1056 0.57 0.78 .029 .012 0.22 0.07 0.77 0.05 .011 .026 .002 .006 .003 .001 .003 26495 1056 0.57 0.82 .029 .008 0.22 0.08 0.78 0.06 .014 .024 .000 .006 .003 .002 .003 26501 1018 0.21 0.73 .025 .023 0.23 0.10 0.08 0.06 .011 .002 .002 .008 .004 .011 .003 26503 1018 0.19 0.85 .018 .007 0.25 0.11 0.06 0.05 .013 .003 .002 .007 .005 .010 .003 26504 1018 0.17 0.83 .021 .007 0.24 0.11 0.07 0.06 .014 .002 .002 .007 .007 .009 .003 26505 1018 0.17 0.79 .018 .010 0.24 0.11 0.07 0.06 .013 .003 .002 .008 .005 .007 .003 26507 1018 0.18 0.85 .024 .015 0.19 0.11 0.14 0.06 .023 .004 .003 .008 .006 .010 .003 26508 1146 0.45 0.85 .098 .008 0.23 0.12 0.08 0.06 .018 .023 .003 .008 .005 .006 .003 26509 1146 0.44 0.79 .101 .009 0.21 0.11 0.09 0.06 .012 .021 .002 .008 .004 .002 .002 26510 1090 0.87 0.95 .025 .014 0.46 0.11 0.11 0.06 .014 .003 .023 .008 .007 .001 .004 26511 1090 0.86 0.92 .028 .014 0.47 0.12 0.10 0.05 .011 .003 .022 .008 .005 .001 .004 26512 1090 0.87 0.86 .036 .016 0.43 0.12 0.11 0.05 .015 .003 .019 .007 .005 .002 .004 26514 1080 0.83 0.74 .022 .010 0.21 0.08 0.08 0.04 .009 .021 .001 .005 .003 .003 .002 26515 1080 0.85 0.75 .024 •010 0.24 0.07 0.08 0.04 .008 .020 .002 .005 .004 .002 .002 26516 1080 0.85 0.77 .025 .010 0.23 0.08 0.09 0.04 .011 .020 .002 .006 .005 .002 .003 26519 1080 0.86 0.75 .027 .013 0.23 0.09 0.09 0.04 .010 .020 .003 .006 .003 .001 .002 26520 1080 0.89 0.73 .028 .013 0.23 0.09 0.10 0.05 .012 .022 .003 .013 .009 .001 .003 26521 1080 0.87 0.73 .028 .010 0.21 0.08 0.09 0.05 .011 .018 .003 .006 .003 .001 .003 26535 1541 0.40 1.44 .036 .018 0.22 0.11 0.09 0.05 .012 .024 .002 .007 .005 .001 .001 26538 1050 0.54 0.98 .029 .008 0.22 0.11 0.11 0.07 .013 .003 .001 .004 .003 .001 .016 26539 1050 0.53 0.98 .030 .008 0.22 0.09 0.11 0.05 .011 .003 .001 .005 .003 .001 .018 26540 1050 0.52 1.04 .030 .009 0.23 0.10 0.11 0.06 .014 .004 .001 .006 .005 .001 .017 26541 1045 0.51 1.00 .031 .010 0.21 0.10 0.10 0.06 .013 .003 .001 .005 .002 .001 .014 26555 1045 0.46 0.76 .028 .015 0.26 0.14 0.10 0.07 .013 .004 .001 .008 .003 .001 .013 Note: Al l of the above heats were cast with oil lubrication. 281 APPENDIX C - Billet Quality Index T h e current w o r k has addressed several issues related to on- l ine detection o f qual i ty problems us ing the "intell igent mou ld" . F o r example , the severity o f rhomboid i ty was found to be strongly related to temperature variations - 2 5 - 5 0 m m b e l o w the meniscus w h i l e the characteristics o f "val leys" i n m o u l d temperature were related to the severity o f transverse depressions and laps on b i l le t surfaces. In addi t ion, since m o u l d thermocouples located b e l o w the meniscus may not be adequate for detection o f surface bleeds, informat ion o n metal l eve l fluctuations and m o u l d hot face temperature can be employed as an indirect w a y to infer the severity o f bleeds and also, certain smal l laps, especial ly i n h igh carbon steels. T h e "intell igent m o u l d " appears to be a p romis ing technology w i t h respect to the on- l ine hot inspect ion o f bi l lets . Howeve r , to make the sensor data useful to plant operators, there is a need to combine informat ion related to bi l le t quali ty into an overa l l " B i l l e t Qua l i t y Index" ( " B Q I " ) . O n the basis o f such an index, bil lets can be accepted, set aside for manua l inspect ion and modi f ica t ion (scarfing or gr inding) , or rejected. Furthermore, for a b i l l e t that is set aside and subjected to scarfing or gr inding , a new and improved " B Q I " can be assigned, once the offending defect is removed. It is envisaged that this nove l qual i ty index w i l l emerge as a "quali ty standard" for companies deal ing w i t h cont inuously cast steel bi l lets . F o r example , bi l le ts for forging or w i re -d rawing applications w i l l require a higher "bi l let quali ty index" as compared to those processed into rebars. A l t h o u g h the assignment o f a quali ty index to each bi l le t appears to be quite attractive, assembl ing the various components o f b i l le t qual i ty into a single index is a formidable task. F o r 282 example , the qual i ty index must take into account informat ion o n the state o f the process variables and b i l le t qual i ty , that is der ived f rom sensor signals and the effect o f l i q u i d steel issues (compos i t ion and superheat) on bi l le t quali ty as w e l l as the knowledege on the sensi t ivi ty o f steel grades to var ious quali ty problems. The index should also account for those problems that are di f f icul t to detect us ing sensor signals; there is no knowledge on detection o f internal cracks as w e l l as surface cracks us ing m o u l d sensors, eventhough mechanisms o f c rack format ion i n b i l le t cast ing are fa i r ly wel l -understood. T h i s is an important issue i n the development o f the " B Q I " , par t icular ly since bi l lets w i t h surface cracks are rejected. Thus , indirect ways must be found to incorporate those qual i ty problems for w h i c h there is no knowledge for direct, on- l ine detection. W i t h respect to surface cracks and other quali ty problems for w h i c h no knowledge can be obtained f rom m o u l d sensors, informat ion on the state o f the bi l le t cast ing process can be u t i l i zed together w i t h an understanding o f the mechanisms o f generation o f these problems. F o r such defects, the expert system i n the "intelligent m o u l d " must operate i n a "predict ive" mode to p rov ide input for the " B Q I " . The "predict ive" expert system assumes that i f a l l o f the process parameters are strictly w i t h i n control l imi ts (pre-defined by experts on b i l le t casting), for a m o u l d operation that was op t imized w i t h respect to design and operating variables, generation o f qual i ty problems should be m i n i m a l . T h i s statement is v a l i d on ly i f a l l the variables affecting the generation o f qual i ty problems, can be measured, on- l ine . I f for some reason, there is a l ack o f knowledge on detection, or an absence o f any pract ical instrumentation, one or more variables cannot be measured and so the conclusions o f the "predict ive" expert system w i l l be l imi ted . O b v i o u s l y , the predict ive capabi l i ty o f the expert system w i l l have to be thoroughly tested before the "intel l igent m o u l d " is put into operation. The first step i n developing a " B Q I " requires examinat ion o f the basic components o f b i l le t qual i ty . Qua l i ty problems i n b i l le t casting consist o f three major units: surface qual i ty (SQ) , 283 internal qual i ty (IQ) and rhomboid i ty ( R H ) . Surface defects inc lude cracks, depressions, laps, bleeds and deep osc i l l a t ion marks wh i l e internal quali ty problems inc lude cracks, segregation, cast structure (columar/equiaxed) and inclus ions . So , the " B Q I " to be adopted i n the "intell igent m o u l d " must inc lude the three quali ty parameters g iven by the f o l l o w i n g equation: BQI = Z,(5Q) + UIQ) + ^(RH) (a) where, BQI is the overa l l b i l le t quali ty index, SQ is the index (0-100) for surface qual i ty, IQ is the index (0-100) for internal qual i ty , and RH is the qual i ty index (0-100) for b i l le t shape. Z j , £3 are mathematical functions T h e variables SQ, IQ and RH i n the above equation are defined as fo l lows : SQ = F ^ surface cracks, depressions, laps, bleeds, oscillation marks) (b) IQ = F2(internal cracks, segregation, porosity, inclusions) (c) RH = F^difference between two diagonals) (c) where, F , , F2, F 3 are mathematical functions T h e definit ions o f the three sub-indices are based on knowledge o f format ion and detection o f qual i ty problems. F o r example , SQ includes information on m o u l d hot face temperature, 284 "val leys" i n m o u l d thermal response, metal l eve l fluctuations, tundish stream shape and misa l ignment , m o u l d lubr icat ion and steel chemistry. O n the other hand, IQ includes informat ion on l i q u i d steel qual i ty (superheat l eve l and chemistry) as w e l l as some pertinent design and operating issues i nc lud ing taper i n the lower part o f the m o u l d and cast ing speed for off-corner cracks. The RQ is a function o f variations i n m o u l d temperature i n the meniscus reg ion and m o u l d hot face temperature relative to the b o i l i n g range o f lubr ica t ing o i l . A l t h o u g h a single quali ty index appears to be s imple conceptual ly , it is extremely dif f icul t to combine the three, very different, quali ty parameters into a single qual i ty index. In addi t ion, it is also di f f icul t to ascertain the condi t ion o f a bi l le t based on a single index. F o r example , two rejected bi l le ts w i t h the same quali ty index can be very different - one may have poor surface qual i ty w h i l e the other may be rhombo id (off-square). Thus , an alternative approach has been examined and three separate indices (having an alphabetic scale o f A to F ) , instead o f a s ingle index, are proposed for the three quali ty states. In this method, three letters, each hav ing a range o f A to F where A is excel lent and F is poor, w i l l be assigned to each bi l le t . The first letter denotes surface qual i ty , the second, internal qual i ty and the third is indicat ive o f b i l le t rhomboid i ty . F o r example , i n the case o f the two rejected bi l lets discussed earlier, the one w i t h poor surface qual i ty w i l l have an index " F B B " w h i l e the r h o m b o i d b i l le t w i l l have an index " B B F " . It is expected that w i t h this method, ascertaining the quali ty o f cast bi l lets w i l l be more useful. 285 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078578/manifest

Comment

Related Items