THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING EFFECT OF BLOWING CONDITIONS ON DEPHOSPHORIZATION IN THE BASIC OXYGEN FURNACE Franklin E. Rote A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the University of Michigan Department of Chemical and Metallurgical Engineering 1967 September, 1967 IP-789

Preface I wish to express my gratitude to all those who have aided me in this investigation and particularly to the following: Professor Richard Ao Flinn, Chairman of the Doctoral Committee, for his continuing interest, encouragement and helpful suggestions throughout the course of this investigation. Professors J. H. Current, J. J, Martin, R. D. Pehlke, and P. K. Trojan, and Mr, L, J. Johnson, members of the Doctoral Committee, for their interest, advice and critical analyses. The Crucible Steel Company of America, Ford Motor Company, Jones and Laughlin Steel Company, McLouth Steel Corporation, and the United States Steel Corporation for kindly contributing chemical analyses. The Jones and Laughlin Steel Company for fellowship support. Professor R. H. Kadlec for his helpful suggestions. Messrs. Peter Severn and Douglas Connell for their help in the design and fabrication of the sampler probes. Mr. Wallace Koebnick for his help in conducting the experiments. My fellow students, Mr. Walter Bargeron, Mr. Robert Blossey, Mr. Philip Guichelaar, Mr. John Hartwig, Mr. John Herron, Mr. David Jepson, and Mr. Rand Winters, for their generous assistance. Miss Sandee Thomson for typing this thesis. The Industry Program of the College of Engineering for reproduction of this thesis. My wife, Ann, without whose patience and encouragement this work could not have been completed. ii

Table of Contents Preface ii List of Tables iv List of Figures v Abstract ix Introduction and Objectives 1 Review of the Literature 3 1. Commercial Operating Practice 3 2, Laboratory In svestigations and Hot-Model Studies 9 Planning the Experiments 1 5 Experimental Procedure 17 1 Development of Techniques 17 2, Conduct of the Experiments 32 Results and Discussion 39 1. General Features 39 2. Phosphorus 42 3. Silicon 51 4. Carbon 60 5. Manganese 68 6. Temperature 74 7, Slag Composition 81 8. Special Heats 87 9, Summary of the Results 92 10o Mathematical Analyses 98 Conclusions 109 Appendix I, Data 111 Appendix II, Calculations 130 Bibliography 143 iii

List of Tables Table Page I Internal Dimensions of Several Basic Oxygen Converters 4 II Lance Operating Variables, Jet Penetration and Jet Penetration - Bath Depth Ratio 5 III Composition and Weight of Steelmaking Slags at the End of the Blow 8 IV Reproducibility of Sample Compositions for Standard Blowing Conditions 24 V Representative Weights, Source and Composition of Charge Components 30 VI Compilation of Blowing Conditions 35 VII Composition of Homogeneous (Average, S-2) Slag Samples for all Blowing Conditions 82 VIII Composition of Probe Samples Taken from a Homogeneous Slag and Comparative Average Compositions 84 IX Composition of Bath Blown Under Standard Experimental Conditions but Without Flux 87 X Distribution of Droplets in the Converter 89 XI Accountable Element J for all Blowing Conditions 98 XII Oxygen Efficiency and Utilization for all Blowing Conditions 99 XIII Comparison of Equilibrium and Actual Phosphorus Contents for all Blowing Conditions 103 XIV Blowing Time - Metal Composition Data for all Blowing Conditions 112 XV Blowing Time - Slag Composition Data for all Blowing Conditions 129 XVI Calculation of Mole Fraction J 0b in the Slag 134 a b iv

List of Figures Figure Page 1. Configuration of a Basic Oxygen Converter 3 2. Sequence of Oxidation of Elements in a Basic Oxygen Converter 6 3. Physical Design of the Experiments 16 4. Tube Opening Time Versus Melt Temperature 19 5. Metal Sampler Probe 22 6. Temperature Probe after Installation 25 7. Schematic Thermocouple - Switch Wiring Diagram 27 8. Hot-Model Basic Oxygen Converter Used in These Experiments 28 9. Locating a Sampling Hole 28 10. Shape of a Sampling Hole 29 11. Depth of the Bath Indicated on a Steel Bar 33 12. Detail Drawing of the Lance with the 0. 316 inch Diameter Nozzle 34 13. Addition of Flux During a Run 36 14. Insertion of a Metal Sampler Probe 37 15. Droplet Collecting Trays 38 16. % Phosphorus Versus Blowing Time for all Sampling Locations (6 ino bath, 7 ino blo hto 20 psig, 0. 316 in, nozzle) 43 17. Constant % Phosphorus Versus Bath Location at 1 Minute Blowing Time (6 ino bath, 7 ino blo ht., 20 psig., 0, 316 ino nozzle) 44 18. % Phosphorus Versus Blowing Time for all Sampling Locations (6 in. bath, 21 in. bl, hto, 20 psig, 0. 316 ino nozzle) 45 19. Constant % Phosphorus Versus Bath Location at 1 Minute Blowing Time (6 ino bath, 21 ino bl, hto, 20 psigo, 0, 316 ino nozzle) 46 v

20. % Phosphorus Versus Blowing Time for all Sampling Locations (10 ino bath, 7 ino bl. hto, 20 psig, 0. 316 in. nozzle - solid lines) and (10 ino bath, 7 in. bl. ht, 35 psigo, Oo 316 in. nozzle - open points) 48 21o % Phosphorus Versus Blowing Time for all Sampling Locations (6 ino bath, 7 ino bl. hto, 25 psig, 0, 162 ino nozzle) 50 22 %T SiliconVersus Blowing Time for all Sampling Locations (6 ino bath, 7 ino bl. ht,, 20 psig, 0, 316 ino nozzle) 52 23, Constant % Silicon Versus Bath Location at 1/2 Minute Blowing Time (6 ino bath, 7 ino blo hto 20 psig, 0o 316 ino nozzle) 53 24, % Silicon Versus Blowing Time for all Sampling Locations (6 ino bath, 21 ino bl. hto 9 20 psig, 0, 316 in, nozzle) 54 25, Constant % Silicon Versus Bath Location at 1/2 Minute Blowing Time (6 ino bath, 21 ino blo ht, 20 psig., 0. 316 ino nozzle) 55 26o %o Silicon Versus Blowing Time for all Sampling Locations (l0 ino bath, 7 ino bl. ht., 20 psig., 0 316 ino nozzle - solid line) and (10 in. bath, 7 ino blo hto 35 psigo, o 316 ino nozzle - open points) 57 27, o Silicon Versus Blowing Time for all Sampling Locations (6 ino bath, 7 ino blo hto, 25 psig., 0. 162 ino nozzle) 59 28, % Carbon Versus Blowing Time for all Sampling Locations (6 ino bath, 7 ino bl. hto 20 psigo Oo 316 in, nozzle) 61 29, Constant %o Carbon Versus Bath Location at 2-1/2 Minutes Blowing Time (6 ino bath, 7 in, bl. hto, 20 psigo 9 0, 316 ino nozzle) 62 30 %o Carbon Versus Blowing Time for all Sampling Locations (6 ino bath, 21 ino blo ht,, 20 psigo, 0o 316 in, nozzle) 63 vi

31. o Carbon Versus Blowing Time for all Sampling Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0. 316 in. nozzle - solid lines) and (10 in. bath, 7 in. bl. ht., 35 psig., 0. 316 in. nozzle - open points) 65 32. o Carbon Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0. 162 in. nozzle) 67 33. o Manganese Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0. 316 in. nozzle) 69 34. %o Manganese Versus Blowing Time for all Sampling Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0. 316 in. nozzle) 70 35. o Manganese Versus Blowing Time for all Sampling Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0. 316 in. nozzle - solid lines) and (10 in. bath, 7 in. bl. ht., 35 psig., 0. 316 in. nozzle - open points) 71 36. o Manganese Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0. 162 in. nozzle) 73 37. Temperature Versus Blowing Time at 4 Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0. 316 in. nozzle) 75 38. Temperature Versus Blowing Time at 4 Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle) 76 39. Temperature Versus Blowing Time at 4 Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle) 77 40. Temperature Versus Blowing Time at 4 Locations (10 in. bath, 7 in. bl. ht., 35 psig., 0.316 in. nozzle) 78 41. Temperature Versus Blowing Time at 4 Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0. 162 in. nozzle) 80 vii

42. % Phosphorus Versus Blowing Time at Sampling Location 61 for Initial % Phosphorus of 0. 76% and 0. 24% 89 43. 0. 6 cm. Diameter Droplet Showing Ferrite, Pearlite and Gas Holes, 2% Nital Etch, 20X 90 44. % Phosphorus Versus Blowing Time for Protected and Unprotected Samples Taken at Sampling Location 61 91 45, Summary Plots of Pg Si and C Contents at 2 Sample Locations Versus Blowing Time for all Blowing Conditions 93 46. Summary Plots of Bath Temperature at 2 Locations Versus Blowing Time for all Blowing Conditions 94 47. Differential Oxygen Efficiency and Utilization Versus Blowing Time with Standard Conditions 100 48. Differential Oxygen Efficiency and Utilization Versus Blowing Time with 21 inch Blowing Height 101 49. Average Bath % Phosphorus Calculated from Mass Transfer Models Compared to the Data - Standard Conditions 104 50. Surface % Phosphorus Calculated from Mass Transfer Models Compared to the Data - Standard Conditions 106 51. Volume Plot Used in Calculating Average Bath % P 131 52. Schematic Drawing of Mass Transfer Conditions of Model 2b 137 viii

Abstract The objectives of this investigation were to determine the effect of different operating conditions on dephosphorization and to determine the mechanism of dephosphorization during the steelmaking process inabasic oxygen converter. For this purpose the chemical and thermal gradients in the bath were determined under the different operating conditions which were reported to affect dephosphorization such as lance height above the bath, bath depth and nozzle diameter. To obtain the necessary data, special probes, developed for this investigation, were inserted through the side of the converter. Substantial chemical gradients were found for P, C, Si and Mn, despite the fact that the bath was only 1 5 inches in diameter and 6 inches deep. The magnitude of the difference in chemical content was affected most by the degree of penetration of the jet into the bath. For example, with shallow penetration, the P content varied from 0. 17% to.0. 30%ocompared to.0. 14% to 0. 20% P with deep penetration. Temperature differences also varied with blowing conditions. With the poor mixing accompanying shallow jet penetration a temperature range from 3000oF. to 2500~F. was encountered while with jet penetration to the center of the bath, the temperature range was 3000~F. to 29000F. The data indicated marked differences in rate of phosphorus, silicon, carbon and manganese removal depending principally upon the depth of penetration of the jet. If the penetration was relatively shallow (due to increased blowing height, greater bath depth or smaller nozzle diameter) steep compositional gradients were obtained and refining was delayed. The chemical analyses and the thermocouple data all indicated a circulation pattern which was upward at the center and radially outward at the surface. ix

With supplementary experiments in which metal droplets were collected during the run, the following mechanism of dephosphorization was developed. 1) Silicon is rapidly removed and some iron oxidized inthe early stages of the run. This produces, along with the CaO added as flusx a fluid basic slag containing CaO, SiO and FeO. 2) Metal droplets are lifted above the slag by the force of the jet, fall back into the slag and are refined while passing through the slag. Mathematical models of refinement by bath diffusion, circulation and droplet transfer were presented. In general, the data reverse the conclusions of many earlier publications recommending avoidance of deep jet penetration in refining. x

Introduction The importance of the basic oxygen steel converter is best illustrated by its growing use in the steel industry, In 1954 the McLouth Steel Corporation installed the first converter in the United States. In January of 1965, 1 5. 4% of the steel made in the United States was produced by the BOF (Basic Oxygen Furnace). The percentage was increased to 21o 2% by December of 1965o During 1965 worldwide production by the BOF was 102, 000, 000 tons of steel. Despite the importance and growing use of this process, important refining reactions such as dephosphorization and decarburization are not well understoodo Further, the proper operating conditions to best accomplish refining are established by empirical methods. The objectives of this research were to determine the effect of different operating conditions on dephosphorization and to determine the mechanism of dephosphorization in a basic oxygen converter. To accomplish these objectives the chemical and thermal gradients in the bath of an operating converter were determined under several different sets of blowing conditions. New sampling techniques were used to obtain the necessary data on metal composition, metal temperature and slag composition while the converter was in operation. A wide range of operating practices was employed to determine the effect on the conversion process of 1) blowing height (7 inches to 21 inches), 2) bath depth (6 inches to 10 inches), and 3) nozzle diameter (0. 162 inches to 0. 316 inches). A standard operating practice was established which reflected commercial practice as closely as possible. The 250 lb. to 400 lb. hot model used in this research was distinctive in that it was not externally heated after it was charged with bot metalo 1

2 Other novel experimental techniques were developed to measure the amount of metal spray above the bath and to determine the effect of this spray on bath dephosphorization.

Review of the Literature The literature is reviewed in two sections: 1) Commercial Operating Practice and 2) Laboratory Investigations and Hot-Model Studies. There is some repetition due to the absence of a clear distinction between a small operating converter and a large hot-model. 1) Commercial Operating Practice This section is divided into five groups: a) Converter Design, b) Charged Materials, c) Operating Practice, d) Refined Materials, and e) Summary. a) Converter Design The designs of basic oxygen converters reported in the literature vary; however, the general configuration is the same. In Figure 1 a converter is shown schematically with the generalized dimensions I. D., height and bath depth. In Table I these dimensions are tabulated for converters which are described in the literature. I BATH ""H Fig. 1. Configuratioia of a Basic Oxygen Converter. 3

4 It is evident from the table that the bath depth-diameter ratio is independent of the converter capacity and varies from 0. 16 to 0. 45 and that only 1 5%o to 20% of the converter volume is occupied by the bath. Blowing practices reported by several authors are tabulated in Table II in terms of lance height, (height of the oxygen nozzle above the bath), blowing pressure, nozzle diameter, number of nozzles, oxygen jet penetration depth and penetration depth-bath depth ratio. While blowing practices do vary, the jet penetration depth-bath depth ratio is 0. 6 to 0. 8 for single nozzle lances. Table I Internal Dimensions of Several Basic Oxygen Converters Bath Depth to Ht. above Capacity I. D. Height Depth Dia. Ratio Bath Reference (Tons) (in.) (in,) (in.) (in.) 55 180 295 31 0. 16 264 2 60 123 266 55 0. 45 211 3 66 180 328 35 0. 19 293 2 77 168 285 39 0. 23 246 4 880 153 246 46 0.30 200 5 100 157 318 56 0. 35 262 3 120 180 332 56 0. 31 276 6 150 204 304 53 0.26 251 7 205 202 334 65 0. 33 283 8 230 232 360 65 0.28 295 6 250 234 372 66 0. 28 306 6 250 322 324 55 0. 17 269 9 300 322 327 62 0.19 265 9 335 280 360 68 0 24 292 10

5 Table II Lance Operating Variables, Jet Penetration and Jet Penetration-Bath Depth Ratio Lance Blowing Nozzle No. of Jet Height Pressure Diameter Nozzles Peno * Ratio Reference (in.) (psig.) (in ) (in.) 88 190 2.05 1 32 0.82 4 40 --- 1,62 1 32 0.68 11 80 165 2.78 1 35 0o63 6 90 165 2,6 1 33 o 62 7 79 95-105 2.18 3 28 0.41 10'Jet penetration was calculated according to the empirical relationship developed by Flinn et. al. (11). b. Charged Materials The composition of charged hot metal varies depending principally on the source of the iron ore and blast furnace practice. Based on compositions reported by several authors (2,4, 8, 12 - 18), a representative hot metal charge composition is 4. 1% C, Oo 9% Si, 1. 0% Mn, 0. 2% P, and 0. 04% So The quantity of steel s crap averaged 22% of the hot metal charge, The range in composition of the charged hot metal is 3. 6 to 4, 3% C, 0, 2 to 1. 2% Si, Oo 2 to 3. 0% Mn, 0. 07 to 0. 21% P and 0, 02 to 00 05% So The proportion of steel scrap added with the hot metal charge varies between 1 5% and 30%o. Oxygen consumption has been reported by several authors (3, 4, 12, 13) to be about 2000 cuo ft. per net ton of steel. The oxygen used is at least 99% pure. Flux compositions were reported by (2, 3, 7, 13, 16, andl8)o The average burnt lime addition is 6% of the hot metal chargeo The average addition of fluorspar, an agent which speeds lime

6 solution, is 0. 4% of the hot metal charge. The ranges of flux additions are: burnt lime: 4. 5 to 6. 5% and fluorspar: 0. 2 to 0. 5% of the hot metal charge. c. Operating Practice. Cuscoleca, Rosner and Kuhnelt (16) reported changes in bath composition with blowing time. Their results are summarized in Figure 2. The sequence of oxidation of bath impurities shown in Figure 2, silicon and manganese followed by phosphorus and carbon, is typical of the general oxidation sequence in a converter bath (19). %P C,Mn,Si 4.'.06 3..04 2. HMn — CP —-^ —-.02 1.. 2 4 6 8 10 12 14 16 18 2022 BLOWING TIME (MIN.) Fig. 2. Sequence of Oxidation of Elements in a Basic Oxygen Converter.

7 Specific operating practices reported in the literature are summarized below. Dephosphorization is an essential part of the steelmaking process. Several authors (20-26) were associated with the installation in Austria of the first commercial basic oxygen converters. A summary of their recommendations on an operating practice intended to dephosphporize the metal is to blow pure oxygen gently onto the bath surface, producing large quantities of FeO which in turn fluxes the CaO forming a fluid, basic slag. Behrens and Koenitzer (14) report starting a blow with a 71 inch lance height to enhance the formation of FeO. Silicon oxidation is not a problem in the converter. It is the first element oxidized (See Figure 2) and the oxidation of silicon can proceed without a basic slag by reaction with dissolved oxygen in any part of the bath. Decarburization is the main oxidation process. Most authors report conditions which give deep jet penetration (Table II) in conjunction with rapid decarburization and high oxygen efficiencies. Suess (27) specified shallow penetration throughout a run in his American patent on this process. Behrens and Koenitzer (14) report starting a blow with a 71 inch lance height; however, the lance is lowered to 32 inches to hasten decarburization. Manganese oxidation, like silicon oxidation, is not a problem in the converter. Bath temperature is controlled by the timing of the addition and amounts of flux and scrap. Petrilli (15) reports adding no flux during the first two minutes of the blow and then adding 80% - 100% of the flux with any remainder being added as a coolant after eighteen minutes of blowing. Cuscoleca et. al. (16) report the addition of flux and scrap after fourteen minutes of blowing to cool the bath.

8 d. Refined Material Steel compositions which are produced in the converter range from 0. 02% P to 0. 05% P, negligible Si, 0. 04% C to 0. 7% C, 0. 1% to 0. 4% Mn and 0. 02 to 0. 03% S (4, 12, 16). Substantial amounts of all the elements except sulfur are oxidized from the bath. Slag compositions vary with flux and metal contents and with operating procedure. Slag compositions reported by several authors are shown in Table III. Table III Composition and Weight of Steelmaking Slags at the End of the Blow SiO CaO MgO P205 FeO AL3 FeZ0 Wt. Ref.'~c~~~~~ -52 3 %o % % N% % % o% lb./T........... —.- 250 2 2.7to3.2 --- --- --- --- --- 1 V=2. 7 to 3.2.. 1 2 V=2.7 to 3.2 --- --- 18-23 - --- --- 15 12-20 27-40 3-7 0.6-1. 5 8-12 --- 5.5 350 16 16 48 --- --- 15 --- 4 --- 17 16 43 0.8 4.7 18 2.8 4 300 18 10-20 27-40 3-10 0.6-1.5 10-15 --- -- --- 19 V = o CaO/%o SiC e. Summary Over the wide range of converter capacities which have been reviewed, the following similarities are noted: the bath depth-bath diameter ratio of 0. 16 to 0. 45, the occupation by the bath of 15%o to 20%o of converter volume and jet penetration of 60%o to 80% of the bath depth. Thus although the Suess patent and some operating instructions specify shallow penetration for dephosphorization, the cornrriercial operations are conducted with penetration to the center of the bath and beyond.

9 2) Laboratory Investigations and Hot-Model Studies This section is a review of a) physical phenomena of jet penetration and bath circulation and b) chemical phenomona of mechanisms of bath oxidation, dephosphorization, silicon oxidation, decarburization, and manganese oxidation. a) Physical Phenomona Jet Penetration The question of jet penetration into the bath was mentioned in the previous section. Rellermeyer et. al. (28) and Holden and Hogg (29) maintain that on impact the jet does not penetrate deeply into the bath, Krainer et. al. (30) and Kootz (31) maintain that under certain conditions the jet does penetrate deeply. Philbrook (32) shows deep penetration with a low blowing height and shallow penetration with a high blowing height. Flinn et. al. (11) show that the depth of penetration varies directly with blowing pressure and nozzle diameter and inversely with the square root of the lance height. Results which lead to the same conclusions as Flinn et, al. (11) were reported by several authors. Krainer et. al. (30) show that the impact momentum is directly proportional to the nozzle area. The reduction of impact pressure by increased blowing height is demonstrated by Kootz, (31), Philbrook (32), and Smith and Dukelow (33). They show that while the momentum of the jet is retained the jet diverges as a cone with increased lance distance/nozzle diameter. This divergence leads to a reduced impact pressure with increased blowing height or decreased nozzle diameter. Philbrook (32) states that changes in the blowing conditions which lead to reduced impact pressure also reduce jet penetration.

10 Bath Circulation There is general agreement that the metal bath in an operating converter circulates; however, there is some question as to the direction. Suess (34) states that the circulation produced by the impinging jet is a down in the center up at the sides pattern. Several authors (11, 29, 31, and35) maintain that the circulation is in the opposite direction rising at the center of the bath (under the jet impingement area) and flowing outward across the surface. Other authors (16, 28, and 36) place emphasis on the spray of metal droplets thrown up by the jet. Krainer et. al. (30) maintain that the circulation pattern is determined by the blowing conditions. Holden and Hogg (29) state that turbulence in the bath is produced by the mechanical energy of the jet. Others maintain that the turbulence is due to a carbon boil. The carbon boil will be covered later. Flinn et. al. (11) show that circulation patterns are alike when nitrogen or oxygen is blown on the bath. The one difference noted is that the droplets thrown up by the jet burn brightly when oxygen is blown. b) Chemical Phenomona Bath Oxidation Holmes and Thring (36) outline the two theories of the mechanism of bath oxidation. One theory suggests that oxygen reacts at the surface of the jet impingement area forming FeO which is circulated through the bath. The second theory suggests that the iron droplets in the jet spray are oxidized and fall back through the slag into the bath carrying FeO with them. Holden and Hogg (29) point out that the droplet mechanism requires that 40 lb. of 0. 12 in. diameter drops be constantly above a 50 ton bath if oxidation rates are to be maintained. This calculation was based on negligible circulation in the drop.

11 Perbix (37) shows that oxygen efficiency is increased by increasing the oxygen flow rate. Kootz (31), Smith and Dukelow (33), and Perbix (37) all show decreases in oxygen efficiency with increased blowing height. Depho spho rization Equilibrium limits on the extent of bath dephosphorization were established by Winkler and Chipman (38)o Their results show that K, the equilibrium constant is given as shown below. (4Ca0O P205) 40 log K log[ ( 407 15 06 (CaO') (FeO) [P] K is decreased by a factor of 10 by a temperature increase of 100~Fo CaO' is the free, unreacted, lime in the slag. SiO2 and Al 0 react with CaO in 1: 2 molar ratios. As shown in the equation above, P205 reacts with CaO in a 1: 4 ratio. Fe203 reacts with CaO in a 1l 1 ratio. All of these oxides detract from the dephosphorizing power of the slag. MnO and MgO are considered to be basic slag constituents equal to CaO in dephosphorizing power. The importance of the FeO content of the slag is shown by its presence to the fifth power in the equilibrium constant. The mechanism of bath dephosphorization is outlined by Bogandy, Dick and Stranski (39) as 2P + 50 ~ P205 at the slag- 25 metal interface where the reaction 4 CaO + P 05 - 4 CaO P205 can proceed. Winkler and Chipman (38) show that O and (FeO) are both valid sources of oxygen. Kootz and Neuhaus (40) relate bath circulation and the necessity of a slag-metal interface for dephosphorization and conclude that the phosphorus content of the bath varies from one location to another.

12 Bath turbulence was established as important by Kootz eto alo (41) and they point out that the rate of the dephosphorization reaction in a converter is limited by the metal boundary layer; therefore, it is important to have a large slag-metal interface. Cuscoleca et alo (16) state that deep penetration causes turbulence in the metal bath and throws up a spray of metal droplets which increases the slag-metal interface. The effect of bath turbulence on dephosphorization was studied by Kato eto alo (42). They found that the rate and extent of dephosphorization are markedly increased by artificially induced bath turbulence, Bogandy et. alo (39) report that equilibrium is never reached in a basic oxygen converter so that after a blow dephosphorization can continueo They report that a nitrogen purge causes sufficient agitation to reduce the phosphorus content from 0. 04% to 0. 03% after a converter is turned down. Similar results are achieved with carbon deoxidation and the subsequent CO boilo Silicon Oxidation Equilibrium data show that the product [ %Si] [ %0] 3. 6 x 10-5 when an iron bath is equilibrated with pure SiO2 (43). In basic slags of the type found in a converter the mole fraction of SiO2 is about 0. 1 reducing the equilibrium product to 3. 6 x 1 06 if the slag is an ideal solution. The assumption of ideality is shown to be false by Winkler and Chipman (38)o They show that all the SiO is combined with CaO in basic, V > 2, slags. Decarburization Equilibrium between carbon and oxygen is determined primarily by the reaction C + 0 -> CO (g) and to a lesser extent by C + 2 O - CO2 (g)o At one atmosphere pressure of pure CO, 0o 1% C is in equilibrium with 0. 02% 0 in an iron bath at 1 6000~C

13 At higher carbon contents the solubility product is increased due to changes in the activities of the carbon and oxygen (43)o Pearson et. al. (44) report that in a converter bath at 2980~F containing 0. 1% C the oxygen content is 0. 04% to 0. 1% Oo The mechanism of decarburization is C + 0 or C + 1/2 0 going to CO gas. No slag-metal interface is required as a reaction site. Several authors (16, 19, 24, and32) maintain that the decarburization reaction proceeds through a carbon boilo Philbrook (32) discusses the carbon boil in detail concluding: 1. CO volumes of 2 to 2, 5 times the bath volume are evolved every second with usual blowing practice, the bath is virtually a bubble-metal slurry, 2. no upper limit of oxygen absorption by the bath is found, 3. concentration gradients are minimized by the carbon boil, 4. greater mixing of slag and metal including ejection of slag streamers and metal droplets into the oxidizing gas phase is induced by the carbon boil, Krainer eto alo (30) report data showing that decarburization is accomplished principally in an area immediately adjacent to the jet impingement area. Coupling this finding with the fact that the bath volume is not sufficiently increased to accomodate the required boil during the rapid decarburization period, they conclude that no more than one third of the CO is evolved from the batho They contend further that the majority of the CO is evolved from the surface of the impact area, and report temperatures of 4500~F in this region. Flinn eto alo (11) mention the brightly burning droplets of iron thrown up by the oxygen jeto Baker eto alo (45) report data on decarburization of iron drops showing that a 1/4 inch diameter drop could be decarburized from 3. 5% C to 0. 1% by pure oxygen in two or three seconds under the conditions found in a converter, They show further that if C0 is substituted for oxygen the decarburization time is increased 4 fold,

14 Smith and Dukelow (33) and Perbix (37) found that the decarburization rate of a bath is linearly increased by increasing the delivery rate of oxygen. Manganese Oxidation Equilibrium data for Fe, Mn, O0 MnO, FeO show that the equilibrium manganese content of the metal bath is approximated by % Mn = 0. 4 x (MnO)/(FeO) (43). The mechanism of manganese oxidation is the same as silicon oxidation in that it can proceed without a slag and that it begins immediately upon the initiation of blowing.

Planning the Experiments Three areas of substantial disagreement exist in the literature concerning the proper operation of a basic oxygen converter: 1) Dephosphorization - Some authors state that large amounts of FeO produced by shallow jet penetration are necessary to achieve good dephosphorization while other authors contend that bath turbulence produced by deep jet penetration is necessary. 2) Decarburization - Some authors contend that a substantial carbon boil is responsible for decarburization while others state that direct jet-bath interaction either at the jet impingement area or in the metal droplets is responsible for decarburization. 3) Bath Circulation Some authors state that the converter bath is generally turbulent, others contend that a discernible down in the center circulation pattern exists while the majority assert that the circulation is up in the center. If a soft blow is necessary to achieve appreciable dephosphorization the rate and extent of dephosphorization should be greater in soft blown heats. On the other hand, if agitation is most important to good dephosphorization, heats blown hardest should be the most completely dephosphorized. If a carbon boil is responsible for decarburization then the metal bath should expand well beyond its normal limits. If the boil is minimal or non-existent, the bath would retain its normal limits. If bath circulation is down in the center, the center of the bath should be the hotter and lower in alloy content; but if bath circulation is up in the center, the surface of the bath should be hotter and lower in alloy content than the rest of the bath. If the bath is generally turbulent then the temperature and composition should be uniform, 15

16 To resolve these points, it was decided to determine the chemical and thermal gradients in a converter under different operating conditions. To determine the mechanisms of element removal it is far more important to know the distribution of elements in a converter bath than to know only the average bath composition. The plan of the experiments is shown schematically in Figure 3. At the center are the "standard" conditions. Ranging around them are the variations in practice made to test the effects on the conversion process of blowing height, bath depth, blowing pressure and nozzle diameter. BATH DEPTH: 6 in. BATH DEPTH: 6 in. BLOWING HEIGHT: 21 in BLOWING HEIGHT: 7 in. BLOWING PRESSURE: 20psig. BLOWING PRESSURE: 25psig. NOZZLE DIAMETER:.316 in. NOZZLE DIAMER.162in. BLOWING HEIGHT NOZZLE DIAMETER BATH DEPTH: 6 in. BLOWING HEIGHT 7in. BLOWING PRESSURE: 20psig. NOZZLE DIAMETER:.316in, BATH DEPTH BLOWING PRESSURE BATH DEPTH: 10 in. BATH DEPTH: 10 in. BLOWING HEIGHT: 7i. BLOWING HEIGHT 7in. BLOWING PRESSURE:20psig. BLOWING PRESSURE: 35psig. NOZZLE DIAMETER:.316in. NOZZLE DIAMETER:.316 in. Fig. 3. Physical Design of the Experiments.

Experimental Procedure This description of experimental procedure is divided into two sections 1) Development of Techniques, and 2) Conduct of the Experiments. 1) Development of Techniques Before making any heats three techniques were developed: A) Sampler Preparation, B) Converter Preparation and C) Charge Preparation. A) Sampler Preparation A brief description of the development of the sampler may help others to avoid some of the problems experienced in this work. Once the concept of determining chemical gradients was established the problem of gathering the data had to be solved. a) Top Sampling The first thoughts on the matter suggested sampling by extending probes down into the bath from aboveo This technique was abandoned after preliminary tests because it was unsuitable on three points. First, any sampling device which is lowered into the bath passes through the slag taking some down into the metal. This unnatural mixing of slag and metal could cause local dephosphorization thus making the metal sample nonrepresentative. Second, probing from the top involved a distance of more than 38 inches making accurate sampler location very difficult. Third, the area directly over the converter mouth was exposed to the molten slag and metal which were ejected from the converter during a heat. 17

18 b) Metal Sampler Probes When it was clear that the converter would have to be sampled through the side wall the initial sampling probe was developed. Three areas of development were studied simultaneously: a sampling device, a refractory coating to protect the sampling device and a support for the system. The sampling device (sampling tubes) were required to take and retain a metal sample from a specified location in the converter bath. The sample had to be taken after the probe was in place but before it was destroyed. The first technique considered was vycor tubing shielded with a metal foil window. When the foil melted the sample was taken as metal flowed into the open tube. This design was found to be poor due primarily to fabrication problems. The second technique considered was to insert evacuated vycor tubes into the bath. This too was abandoned because the vycor tube slowly collapsed during immersion. The successful design utilized evacuated pyrex tubes which were insulated from the bath by refractory except at a tip which protruded through the insulation. The location of the pyrex tube in the bath was determined with guides on the support and in the converter wall. The opening time of the exposed tip of the sampling tube depended upon bath temperature, exposed area and the thickness of the tip. The dependence of opening time on bath temperature is shown in Figure 4. In the temperature range of experimental interest the opening time was 5 seconds at 2500~F to 1 second at 3300~F. Analysis of movies taken of the sampling procedure indicated that a sampling probe was located in the bath in 1/8 to 1/4 second and tests on the system showed that the probe was not damaged after a 15 second exposure to steel at 3200~F.

19 TUBE OPENING TIME (SEC.) 6 _4 _ ______ 2400 2700 3000 3300 MELT TEMP. (~F.) Fig. 4. Tube Opening Time Versus Melt Temperature. The opening time depended only slightly on the exposed area of the tip; however, care was taken to insure that the exposed area was a circle roughly 9 mm. in diameter. The opening time varied almost directly with tip thickness in the range 0. 01 inches to 0. 03 inches. The data in Figure 4 were. tip thickness of 0. 015 inches + 0.002 Inches This was the specification for tip thickness throughout the investigation. The tip thickness was checked periodically to insure production vwithin thile range specified. The refractory insulation was required to pack around the tubes easily but densely, to resist spalling when inserted at roon temperature into a metal bath at 32000F and to liberate no gas when inserted into the bath. The refractory could be preheated to 000oF (a limit set by the creep properties of pyrex) but had to be cooled to near room temperature immediately before use.

20 A wide range of refractories was tested and most of them failed either because of spalling or gas liberation. It was found that most refractories could successfully be packed around the tubes as a thick slurryo The optimum refractory insulation was found to be a slurry of fines taken from commercial MgO refractory. After the slurry was applied and air dried, the probe was slowly heated to 1000~F, held 48 hours and furnace cooled before use. After this treatment the refractory neither spalled nor liberated gas for at least 15 seconds after immersion into steel at 3200~F. The sup ort of the sampler probe was required to be chemically inert in a steel bath before the sample was taken, to be strong enough to pierce a patch in the converter wall and not to disturb the batho Earlier experience with vycor showed that this material was inert in a bath for long periods so the first supports were fabricated from 25 mmo vycor tubingo While chemically inert, the vycor tubes cracked when the assembly was preheated to 10000F. This cracking was due to the greater thermal expansion of MgO than vycoro This problem was solved by lining the vycor tube with asbestos which served as a cushion for the thermal expansion of the MgOo Vycor supports were not strong enough to pierce the thin refractory patches in the convertero Whenever an attempt was made to pierce a patch with a vycor support it cracked and then shattered, This difficulty was never overcome and vycor supports were abandoned. The strength of steel allowed steel supports to pierce a wall patch easily; however, bath contamination by the probe had to be checkedo

21 Comparative samples were taken for chemical analysis with a steel probe and with a vycor tube and suction bulb from baths at 2850~F. The result of this comparison was that the respective phosphorus contents were 0, 45% and 0, 48% in one test and 0. 18% and 0. 17% in a second test, In the second test a chill cast slug was also poured which contained 0. 18% P. In both tests the silicon contents were found to be 0. 72% + 0. 02% for all samples, The range of carbon contents was 4. 2% versus 4o 3% and 4O 3% versus 4. 3% in the two tests. Manganese contents were the same for all samples. On the basis of these results it was decided that steel probes could be used without fear of sample contamination. The design of the probe was a 1 inch diameter cylinder with a pointed end. The amount ofbath disturbance when a sampler probe was inserted was determined with water-model experiments, Streamline flow of water was established in a steel trough at a velocity of 5 inches/second. Dye was fed into the stream to mark the flow direction. A sampler probe was inserted into the stream with the result that within roughly 1/4 inch of the probe the flow lines were deformed to accommodate the probe. Beyond this distance no disturbance of the flow lines was noted, Initially it was hoped that the samplers could be inserted into the converter and then be removed. Considerable time was spend attempting to transform this concept into a working procedure; however, this was not possible because of one or the other of two recurring problems. The sampler was usually frozen into the converter wall by bath metal which filled the 1/16 inch gap between the sampler and the wall of the sampling holeo On those rare occassions when the sampler was retrieved the resulting hole in the converter wall was so large that substantial metal was lost from the bath before the hole was botted,

22 The final sampler probe design is shown in Figure 5. These probes remained in the converter wall throughout the heat. The preparation procedure and the specific dimensions of the probe are given below. EXTENT OF BATH I A \A SHER NDLE A V T 3 —- 42 ---— 9- ~ 6,,' SECTIONAA T^^^- REFRACTORY INSULATION 9mm. ^ ^-NEST OF SAMPLING TUBES Fig. 5. Metal Sampler Probe. 9 m. m. O. D., tipped sampling tubes were evacuated to less than 1 J Hg after a weighed amount of 24 gauge aluminum wire was inserted. Sufficient wire was used to add 0. 4% Al to the sample. The sampling tubes were nested into the support in a strictly maintained arrangement to insure proper identification of the samples after a run. When the tubes were in place the support was filled with a thick slurry of MgO fines. The handle was filled with a fire-clay

23 slurry. The fire-clay was used to guard against a broken tube causing a leak and to supply a heat sink to solidify the samples rapidly. After fabrication the samplers were slowly heated to 10000F., held 48 hrs. and furnace cooled. The steel support cylinder was a slotted 3/4" low C steel pipe. The slot was 5/8 inch to 7/8 inch wide, A solid steel point was welded into one end of the cylinder and a handle was welded to the other. The handle was fabricated from 3/4" low C steel pipe and a 2" 0. D. washer. Immediately prior to a run a fireclay collar was applied in front of the washer and a cone of fire-clay was applied behind the washer. Reproducibility of sample location was checked first by inserting sampler probes through unpatched holes into an empty, cold converter. By measuring the distance from the converter wall to the tips of sampling tubes the error in radial location was determined. The error in vertical location was determined by measuring the angular deviation of the sampler probes from a horizontal axis after an experiment. Errors in radial sample location were found to be + 1/4". Inclination of the sampler probe averaged 1~ as most were horizontal. The range of inclination was + 50, At + 5~, the worst condition, the location error of the 6 inch tip was + 1/2 inch, the 3-1/2 inch tip, + 3/8 inch and the 1 inch tip, + 1/8 inch. The location of the volume of metal taken as a sample was determined in water-model experiments.. J probes were bent from a length of 9 mo m, pyrex tubing, fitted with suction bulbs and immersed into a water-turpentine batho The turpentine being immiscible in water floated on the bath surfaceo The open end of the J probe was first positioned 6 inches from the water-turpentine

24 interface and moved upward to within 1/2 inch of the water-turpentine interface before it was disturbed when the suction bulb was released and a 15 c. Co sample taken. The experiment was repeated with light machine oil and with alumina particles held afloat by surface tension. In these cases the results were the same, Horizontal tests were made by holding the open leg of the J probe just below the surface of the water-alumina model and taking a sample. In this case the crosssection of the sample volume was roughly a semicircle 1/2 inch in radius. Reproducibility of sample analyses was checked by taking samples from two similar runs at the same location and blowing time. The results of this check are shown in Table IV along with the results of comparative analyses of the same sample. The error in comparative analyses is of the same magnitude as the variation between samples. Table IV Reproducibility of Sample Compositions for Standard Blowing Conditions Element P Mn C Si Similar Samples from Different Heats Average Comp. 0. 17 0. 15 3. 5 0. 07 Average Variation 0. 03 0, 02 0. 25 0, 03 + % Variation + 8% + 7% + 4% + 23% No. of Samples 14 20 16 12 Comparative Analyses of the Same Sample Average Comp. 0.16 0 14 2.0 0 06 Average Variation 0O 027 0. 02 0. 25 0. 02 + % Variation + 8% + 7% + 6% +16% No. of Samples 9 11 9 8

25 c) Slag Sampling Probe The slag sampling probes were fabricated from 3/4 inch low C steel pipe in much the same way as metal samplers. The slotted portion was 7 inches long separated from the handle by a 2-1/2 inch length of spacer pipe. The handle was filled with fire-clay and the spacer pipe was rammed solid with MgO refractory. The slotted portion was left open. Before a run the slotted portion was washed with a zirconia slurry to ease separation of the sample. d) Thermocouple Probe The preparation of the thermocouple tube deserves special attention because of the harsh environment of a converter bath. Figure 6 shows schematically a probe after installation. The SHELL MgO FIRECLAY -I t1 REINFORCING -' MINIMUM Fig. 6. Temperature Probe after Installation.

26 protection tube was high purity alumina (99, 7% A1203 and was covered with a layer of castable alumina (99o 3% Al 031 which mechanically strengthened the structure. The reinforcing layer was applied as a slurry of fines, The tip of the protection tube extended about 1/4 inch beyond the reinforcing layer, for sensitivity of response. After fabrication the assembly was allowed to air set for three hours and then was slowly heated to 800~Fo and furnace cooled. The tubes were located by ramming refractory around them while they were held in placeo The ramming sequence is shown in Figure 6, After ramming was complete the wall was washed with MgO fines to fill any small rifts, The thermocouples were fabricated from 26 gauge Pt-Pt + 10% Rh wires which were fused in a carbon arc without fluxo Response to temperature was tested in a bunsen burner flame which also cleaned any trace of grease from the thermocouple leado The thermocouple wires were insulated with high purity alumina two hole insulators, The assembly was put into the previously installed tubes and sealed in place with refractory cement, Contact to a recorder was made with compensating leads through a switching arrangement which allowed up to twelve thermocouples to be read sequentially at one second intervalso The schematic wiring diagram is shown in Figure 7, Included in the drawing are two constant potentialso the "marker" and the "buck'" Both constant potentials were supplied by potentiometerso The marker was wired into the output circuit to mark a point in the reading sequence so that the output of each of the thermocouples could be identifiedo The buck was wired into the input circuit in opposition to the thermocouples lowering their voltage 303 m, Vo This effectively increased the range of the recorder 500~ F to 2500~ F to 3500~Fo The error in temperature readings was estimated to be + 30~F,

27 MARKER T.C. +/ TC. ~ -3+ RECORDER F." 7.\ Sh aBUCK Fig. 7. Schematic Thermocouple - Switch Wiring Diagram. Errors in thermocouple probe location were + 1/4 inch. The response time of the thermocouple assembly was 1 second, 2000~F. to 3000~F., and limited by the recorder. B) Converter Preparation The 15 inch converter used in the experiments is shown in Figure 8. The lining away from the sampling area was cleaned and patched before each run and the bottom was checked for damage. The lining in the sampler hole area was completely removed so that cast sampler hole forms could be set in place and accurately located. The location of the cylindrical portion of the form was determined as shown in Figure 9. The level assures that the cylinder axis is horizontal. The rule was centered by suspending it directly

28 18 ----- 6'16 6 - - - 2 _- 2-: -STEEL SHELL -- 1 1(nO-I|:L.MgO LINING f/ ^ \\ | ~I'-: BACKUP LINING li cIaI IX 1 1 STANDARD SAMPLE 1AI \ lLOCATIONS A:CONVERTER BODY -'e up 1- lB: REMOVABLE HOOD L71t; 20 CD LANCE 3~-'- __ I 6 ~ D:BATH!*'.1. i.",., \ t ~ 8 Fig. 8. Hot-Model Basic Oxygen Converter Used in These Experiments. Fig. 9. Locating a Sampling Hole.

29 under the nozzle of the lance. With the angle iron it was possible to determine both the cylinder height and the axis orientation relative to the center of the converter. The ramming sequence for the sampler hole area is shown in Figure 10. The shape of the hole is significant. The 1-1/6 inch diameter cylinder was a slip fit with a sampler, allowing it to slide through; the 2 inch length allows only axial motion. The patch was rammed after the rest of the lining was in place. Once the wall was patched it was washed with MgO fines to fill any pores or rifts. SHELL MAGNAMIX FIRECLAY TPATCH i 24 4-I L 6 Fig. 10. Shape of a Sampling Hole. Fig. 10. Shape of a Sampling Hole.

30 The thickness and material used for wall patches were determined by careful experimentationo Hollow refractory cones were prepared and patched with several refractories in thickness from 1/8 inch to 1 incho These cones were immersed in cast iron or steel baths and held for 10 minutes at temperatures up to 3500~Fo. If the patch had not leaked and was still easily pierced it was acceptable. A 1/4 inch patch rammed with MgO cement met both criteria and was usedo C} Charge Preparation The charge for a run consisted of hot metal and fluXo Both components were weighed to accuracies of 1%o Representative weights of all components of the charge are shown in Table V along with the source and reported compositiono Table V Representative Weights, Source9 and Composition of Charge Components Hot Metal Component Weight Source Reported Compo (Ibs) Major Elements pig iron 246 Great Lakes Foundry C - 4o 3% Supply Co Si - 0. 62% Mn - 0o 54% P - 0,08% 50% Fe-Si o 25 Union Carbide Corp. Si - 50% 75o Fe-Mn o 20 Union Carbide Corpo Mn - 75% Fe-P 2, 0 Great Lakes Foundry P - 24% Supply Co. Flux Lime 18o 0 Wyandotte Chemical COo CaO - 91% MgO 3o 5%0 SiLO lo 3%0 Fluorspar o 90 Great Lakes Foundry CaF2 P 88%0

31 The pig iron was received as 35 lb. pigs which were broken into thirds and the pieces were stored in separate containers. The 246 lb. charge was taken from one container. This practice and the anticipated mixing of the pigs prior to delivery produced a uniform charge. The alloy additions were taken from shipping containers; the lime was obtained directly from a carloading facility and the fluorspar from two pound bricks. The flux was weighed out into 6 paper bags each containing 3 lbs.. Two distinctly marked bags each contained 0. 45 lbs. of fluorspar. The weights were increased to 4. 8 lbs. and 0. 72 lbs. respectively when 400 lb. heats were run.

32 2) Conduct of the Experiments a) Melting the Charge The metallic charge was melted in two 3000 c. p. s., 50 KVA induction furnaces with 200 lb. and 110 lb. individual capacities. Both units were MgO lined. The alloy additions were placed in the bottom of the furnace and it was filled with pigs. Neither furnace would hold its full cold charge so additions were made as melting progressed. During the melt down period the molten metal temperature did not exceed 2700~F. Prior to tapping the melts were superheated to 2850~F. When 400 lb. heats were run the additional 160 lbs. were melted as two charges in the 110 lb. induction furnace. The metal was transferred to a clay-graphite crucible in a gas-fired holding furnace. The heating capacity of this unit was such that tap temperatures of 26500~F. to 2680~F. were attained. Charging the Converter Prior to a run the converter was preheated to a 2000~F. surface temperature. This was accomplished in about 18 hrs. with a natural gas-air torch. When all the metallic charge was molten and superheated it was tapped into a 100 lb. ladle and transferred to the preheated converter. b) Pre-Run Checks When the converter was fully charged it was rais'ed to an upright position and the bath depth was measured. A 1 inch by 1/4 inch steel bar was lowered along the lance to the bottom of the converter. The bar was held in position for 6 seconds, lifted out and placed on the lab floor where the bath depth was measured.

3 3 The extent of the bath is clearly evident on the bar as shlown in Figure 11. The bath was adjusted to 01 1 in, of the desired depth, Ps.'_1~~~~~~.4. &~: 4;1: 4......... i Fig. 11. Depth of the Bath Indicated on a Steel Bar. The bath temperature) was followed to 2!5500?\. with an optical pyrometer. During this time any loose slag was removed from the bath surface. When the converter bath reached 2550~ F., satmple M l was ta1ken. Th8e sample swas poured from:i th:e (cnvzxerter into a 5 Ib. t adle and then chill cast in a core sand mold. This procedure produced an average starting analysis of 0 2.4% P, 0. 6% Si, 4. 1% C, and 0 55% Mn. The range of a nalys es was 0. 2% tho 0. 4b% P, 4% to. 70/ Si, l 9l % to 4. 3%, and 0. 4% to optical l~yro~fletei~.During this timemyloose.... sla. was.removed from the K th surface. When the converter l~ath rei ched A 55Q0 l. sample MI I was taken, The sampl was poured from.the.converter it a5l.ld an th chlcsin coes dml ~~i~~~i i~~~~iAln!......... ~~~~~~~~~~~~~~~ w r~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~........................ ~i~P1Q.................................................................. i ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~............ ~~~~~~~~~~~i~~~~~~~~~~~~~~~~~~~~~ t: i:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.... ~~~:~~~~ ~~~~~~d~~~~ B~~~~64~ ~ ~~~...... 3 2~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~VY~~~~~~~~~~~~~~~~~~~~~...... i:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...... p i:~ a:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.......::' B n~~~~~~~~~~~~~~......... ~ ~~ ~~~ ~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~........ i:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...... LE"%g, II, Dcptlr of Ihe asat21 anca~~~~~~~~~~~~~~cated on a Sl.............. ~~ilc t-,atl~~~ tempctralur e was followed to ZiiSO"F'................. optical py~~~~~~ometer. I:Dusing Ihis tirt~~~~~~~~~~~e any I~~~~oose slag tvas................. ffon~~ the bal-.i surface, iiif~~~ien ehe c30nvertcr hath rescfted ~~~......... sample M-f was faken, The sample was puufed frotn tl~~~~~~~~~~~~~~~~e............ iizto a 5 li~~~~, I~~adle anct then clziX1. ca~~~~~~t it2 a core sand t-n....... %'his procedure produced ai~1 avcraac starting analysis of~~~~~~~~~~~~~~~~.........

34 0. 7% Mn. The starting analysis of the individual heats is given in Appendix I, Data. The converter was now turned upright, and a safety catch basin which fitted the contour of the converter and an overhead protective shield were put in place. While these changes were being made the lance was lowered to the proper blowing height. Figure 12 is a detailed drawing of the oxygen lance with the 0. 316 inch diameter nozzle. The other lance used in this investigation was the same except the nozzle diameter was 0. 162 inches. Both lances were fabricated of copper and all joints were brazed. 66 /AV~~~~ * Pressure Tap (Pd) located 3-In. from 56.75 the Nozzle Exit. Drill No.64 (.0360) Dia. 4.25 ---.270.16 —.365Dia.,// \ \, 3 >G.7Do \ \:I.7 Dioa.//I i iSI _ _.316 Dia. 1.625 Dia. Heli Arc Weld - All other indicated Connections Silver Soldered. Dimensions: Inches Material: Copper Tubing Fig. 12. Detail Drawing of the Lance with the 0.316 inch Diameter Nozzle.

35 The effects on converter phenomona of the individual variables which make up a set of blowing conditions were determined by varying each of them individually. In Table VI the blowing practices used in this investigation are tabulated along with sampling locations and the particular variables being studied. Table VI Compilation of Blowing Conditions (1) (2) (3) (4) Blowing Bath Blowing Nozzle Sample Heights Variable Height Depth Pressure Diameter Above Bottom * Studied (in. ) (in. ) (psig.) (in. ) (in.) 7 6 20 00316 6, 5, 3, 1 Standard 21 6 20 0.316 7, 5, 3, 1 (1) 7 10 20 0.316 10, 8, 5, 2 (2) 7 10 35 0.316 10, 8, 5, 2 (3) 7 6 25 0. 162 6, 5, 3, 1 (4) * Heights are measured to sampler centerline. Sample location is 1/2 inch above the centerline. In all cases samples were taken on a radius 1, 3-1/2, and 6 inches from the wall of the converter. The bath radius was 7-1/2 inches. The significant point of Table VI is that metal samples were usually successfully taken only 1/2 inch above the level of a quiet bath; however, when a 21 inch blowing height was used metal samples were taken 1-1/2 inches above the level of a quiet bath. Slag samples were taken 2 inches and 5 inches above the level of a quiet bath. c) Making the Run The run time was begun when the oxygen was turned ono The six bags of flux were added at 0, 5, 10, 15, 20, and 25 seconds. The two bags containing fluorspar were added at 0 and 5 secondso Flux was added with a long-handled trough as shown in Figure 13, Also evident in Figure 13 are the safety catch basin and the shield,

36 Fig. 13. Addition of Flux During a Run. Metal samples were taken by piercing the specially prepared sampling locations with the pointed end of the sampler and continuing to push the sampler into the bath until the locating collar stopped motion. After the sampler was in place the fire-clay cone was rammed around the sampler. This procedure is shown in Figure 14. Slag samples were taken through holes higher in the vessel. In this case the sampler was retrieved at once and the hole packed with clay. Sampling times were recorded to + 1 second. When all the samples were taken and temperature data collected the 02 was turned off and the total blowing time recorded. After the blow the safety catch basin and shield were removed and a pig mold positioned. Before tapping the converter a spoon sample of slag, sample S-2, was taken and another chill cast metal sample, sample M-2. poured.

37 Fig. 14. Insertion of a Metal Sampler Probe. After a run the individual samples were collected and identified. The metal samples were analyzed for P, Si, C, and Mn and the slags for P 0' CaO, SiO, MgO, MnO, ALO3, and in most cases FeO and Fe03 and spot checked for ZrO2. 23 2 d) Special Run A special run was made to determine the amount and size distribution of the droplets thrown up by the oxygen lance under various blowing conditions. Figure 15 shows the collecting device. Prior to use, it was filled with a fire-clay slurry and covered. It was inserted through a 2 inch by 7 inch hole in the converter wall which was 2 inches above the surface of a quiet metal bath. The lance was set at a predetermined height and the 02 jet turned up to a predetermined blowing pressure. The collector was uncovered for 5 seconds, recovered and removed and the jet turned off.

38 Fig. 15. Droplet Collecting Trays. The collected droplets were weighed, sized and in one case examined metallographically to allow an estimate of the carbon content.

Results and Discussion The results aret reported under the following headings: 1) General Features, 2) Phosphorus, 3) Silicon, 4) Carbon, 5) Manganese, 6) Temperature, 7) Slag, 8) Special Heats, 9) Summary of Results, and 10) Mathematical Analyses. 1) General Features Before describing the specific effects of the various blowing conditions on the composition and temperature, the general effects of the different blowing conditions upon the melts should be reviewed as background. a) Standard Conditions The standard experimental conditions (6 inch bath depth, 7 inch blowing height, 20 psig. blowing pressure and 0. 316 inch lance) were scaled from commercial practice. The 0. 4 bath depth to bath diameter ratio and the deep jet penetration were taken directly from commercial practice. The oxygen rate was 380 scfm. per ton which is higher than that used commercially; however, total oxygen consumption of 1900 cubic feet per ton is the same as commercial practice. These standard conditions resulted in slight throw out of slag and metal droplets throughout the heat. Metal samples could be taken no higher than 1/2 inch above the level of the quiet bath. At greater heights only slag was obtained (with a few metal droplets). Experimental runs were complete after 5-1/2 minutes of blowing (5: 30) at which time the final slag sample had the viscosity of thick oil. 39

40 b) Effect of Blowing Height The experimental conditions designed to determine the effect of blowing height (6 inch bath depth, 21 inch blowing height, 20 psig. blowing pressure and 0. 316 inch diameter lance) produced a substantially different appearance. The early portion of the heat was marked by little ejection of slag or metal droplets; however, after approximately 5 minutes and 20 seconds blowing time (5: 20) a foamy slag overflowed from the mouth of the converter. Metal samples were successfully taken 1-1/2 inches above the level of the quiet bath. Runs were halted when the slag overflowed and delays of up to 3 minutes were required before the converter could be safely approached. The final slag sample was gas-filled and very thick. c) Effect of Bath Depth When the bath depth was increased to 10 inches, the runs appeared similar to those conducted under standard conditions. Metal samples were taken 1/.2 inch above the level of a quiet bath. The runs required 8: 30 and the final slag sample was thick and dry. The effect of blowing pressure was determined by modifying the condition above with a 35 psig. blowing pressure (10 inch bath depth, 7 inch blowing height, 35 psig. blowing pressure and 0. 316 inch diameter lance). This change increased metal and slag ejection to the extent that the converter operation was quite violent. Because of this a limited number of samples were taken. Metal samples were successfully taken 1/2 inch above the level of a quiet bath. The run was completed in 5: 30. The final slag sample is fluid and very flat.

41 d) Effect of Nozzle Diameter The heats with a smaller nozzle diameter (0. 162 inches) produced some difficulty in maintaining adequate temperature. The ejection of metal and slag droplets was slight. The flux formed a solid crust on the bath surface completely shielding it from the oxygen jet. This crust had to be manually broken up to continue the heat. Metal samples were taken 1/2 inch above the level of a quiet bath. Heats were run for as long as 13: 38 at which time the slag was very thick. To summarize these observations, the amount of metal ejected from the converter was generally proportional to the depth of jet penetration except in the case of an increased blowing height when slag overflowed the converter. The uppermost metal samples were usually taken 1/2 inch above the level of a quiet bath. The one exception was the 21 inch blowing height condition where the samples were taken 1-1/2 inches above the level of a quiet bath. While there were some differences in bath expansion for different blowing conditions it is significant to note that the bath expansion was usually 1/2 inch, 10%, and in the exceptional case reached a maximum of 25%. In all cases, the final slag samples had a medium to heavy consistency. The slag sampling holes were placed 1-1/2 inches above the uppermost metal sampling holes when the converter was rammed. During a run little metal flowed out of these slag sampling holes when they were opened; therefore, at points only 1-1/2 inches from the metal bath the emulsion of slag and substantial amounts of metal (spoken of by most investigators) is not present in the converter. With these general features produced by the various blowing conditions in mind, the chemical and thermal gradients of a converter bath and the composition of converter slags can now be described.

42 2) Pho sphorus a) Standard Conditions The phosphorus content of the bath under standard experimental conditions is shown in Figure 16 as a function of time and bath location. The points on the individual plots are data points. * The origin of each individual plot is located at a sampling location. For example, the origin of the uppermost, left-hand plot in Figure 16 is located 6-1/2 inches above the bottom and 1 inch from the wall. This is sampling location 61. In designation "61" the "6" indicates approximate height and "t t1 the approximate distance from the wall. Similarly, the origins of the other two plots 6-1/2 inches above the bottom are 3-1/2 inches and 6 inches from the wall, sampling locations 63 and 66 respectively. The other rows of plots are 5-1/2 inches, 3-1/2 inches, and 1-1/2 inches above the bottom. With standard experimental conditions the greatest dephosphorization is found at the top of the bath, 0. 02% P at 4: 1 5, and the least at the bottom of the bath 0. 10% P at 4: 30. The maximum dephosphorization rate is 0. 12% P/min. at the top of the bath near the wall at 1: 00 while 0. 06% P/min. is found at the bottom of the bath at 2: 30. The variations in dephosphorization lead to substantial gradients as shown in Figure 17. The range of phosphorus is 0. 14% to 0. 20% from the top to the bottom of the bath at 1: 00. The constant composition curves are all nearly horizontal. The gradient in phosphorus content is steepest at the top of the bath (0. 03%o P/inch) and very slight in the lower half of the bath. "The data are tabulated in Appendix I, Data.

43 IN. ADOVE BOTTOM 8 I I I I %P BLOWING TIME ~6 i-i,. l2 0I I E IlI, I 1 2 3 4 5 6 7 IN. FROM WALL Fig. 16. % Phosphorus Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

44 IN FROM BOTTOM 7........ _.m. 4 -.20-..20-.22:1 -— 1 2 3 4 —5 6 7 0 1 2 3 4 5 6 7 IN FROM WALL Fig. 17. Constant % Phosphorus Versus Bath Location at 1 Minute Blowing Title (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle). b) Effect of Blowing Height Increasing the blowing height from 7 inches to 21 inches produces the phosphorus pattern shown in Figure 18. The top of the bath dephosphorized earlier and to a greater extent than the bottom, 0. 05% P versus 0. 12% P at 3: 00. Maximum dephosphorization rates of 0. 15% P/min. are found throughout the bath. The difference in phosphorus content is due to the later initiation of rapid dephosphorization lower in the bath. The phosphorus gradients at 1: 00 are shown in Figure 19. The range of phosphorus content in the bath is 0. 17% to 0. 30% P, top to bottom. The constant composition lines curve upward near the wall of the converter. The top of the bath is 50% dephosphorized while the bottom of the bath is virtually unrefined.

45 IN. ABOVE BOTTOM ~/oP.,3 8.2. 1 2 3 4 5 12345 BLOWING T1ME 7 71~ 1 1 1 5 I I I 4 3 2__ 1_ _ 2_,_I f.3 I I5 IN. FROM WALL Fig. 18. % Phosphorus Versus Blowing Time for all Sampling Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle).

46 IN. ABOVE TOM_.1~ ~ ~ ~~.17.20 -—.. —--- —......14 ---—. — 0 —.S^-. —_ 0.25 32.302I3RMWL6, 0 1 2 IN3FROM WALL 5 6 Fig. 19. Constant % Phosphorus Versus Bath Location at 1 Minute Blowing Time (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle). The vertical phosphorus gradient is generally smallest near the top of the bath and nearly constant below the 6 inch depth at 0. 025% P/inch. At the top of the bath near the wall the gradient is increased to 0. 03% P/inch. The effect of blowing height on the dephosphorization is clearly shown when Figures 16 and 18 are compared. At all sampling locations throughout the heat the phosphorus content of the bath blown under standard conditions is lower than that of the bath blown with a 21 inch blowing height. The dotted lines on Figure 19 are standard condition data emphasizing the difference in the two blowing practices. The effect of the deep penetration is especially apparent near the bottom. The upward trend of the constant composition lines with the 21 inch blowing height is in direct contrast to the nearly horizontal lines resulting from the 7 inch blowing height. The differences in phosphorus gradients for these two conditions are

47 caused solely by increased blowing height as all other conditions are constant. c) Effect of Bath Depth In Figure 20 the solid points and curve show the effect of increasing the bath depth to 10 inches (400 lb. of hot metal). Initially rapid dephosphorization is found at the bath surface followed by slow dephosphorization. The major portion of the bath was slowly dephosphorized. This system does not produce a substantial gradient because so little phosphorus is removed. The data points drawn as open circles in Figure 20 show the effect of increasing blowing pressure to 35 psig. with a 10 inch bath depth. The bath is dephosphorized rapidly and extensively, e. g., 0. 03% P at 5: 22 at sample location 51. The rate of oxygen input with 20 psig. and a 10 inch bath depth is 0. 12 scfm/lb. while at 35 psig. the rate is 0. 17 scfm. /lb. or a 40% increase. Increased pressure increases the jet penetration from 4-1/2 to 7-1/2 inches. As was noted earlier the slag on the 10 inch bath blown at 20 psig. was dry and very thick at 8: 30 which accounts for the poor dephosphorization with these conditions. Increasing the blowing pressure to 35 psig. produces a fluid slag at 5: 30. The dotted lines in Figure 20 in the 10-1/2 inch row and the 5-1/2 inch row are data from the standard conditions at the 6-1/2 inch and 5-1/2 inch sampling heights respectively. In the former case the sampling heights are the same distance below the slag-metal interface while in the latter case the sampling heights are the same distance above the bottom. Regardless of whether the reaction location (slag-metal interface) or the bottom of the bath is used as a level the data show that at corresponding locations the rate and extent of dephosphorization

48 %P ALL DATA IN THE TOP ROW IS EQUIDISTANT FROM THE SLAG.3 - 2 Q Q 11r.1t W; 1," —- - - 2 4 6 8 IN. AROVE BOTTOM! 10 Locations (10 in. bat, 7 bl. ht., 20 i 0.36. nole - nozzle - open points).

49 is greater with the standard 6 inch bath than with a 10 inch bath and 20 psig, blowing pressureo Further, the standard data correspond very closely with the data from the heats blown at 35 psigo In other words, as bath depth is increased, penetration and oxygen delivery rate should be correspondingly increasedo d) Effect of Nozzle Dimater The effect of decreasing nozzle diameter to 0, 162 inches and using 25 psig. blowing pressure is shown in Figure 21o The time scale has been changed on the plots to accommodate the longer blowing time. The bath is dephosphorized at all sampling locations with the usual pattern of most rapid and extensive dephosphorization taking place at the bath surface. Dephosphorization rate is 0. 03% P/minute at the surface and as low as 0O 01% P/minute at the bottom of the bath at the initiation of blowingo The dotted lines in Figure 21 represent comparative data from the standard conditions. The oxygen input rate with the 0. 162 inch diameter nozzle is 30% of the rate under standard conditionso As would be expected, decreasing the oxygen input rate has decreased the rate of dephosphorizationo As will be seen later this heat was not fully decarburized in 13: 38 so the maximum dephosphorization is not knowno Summary - Dephosphorization Under all conditions a phosphorus gradient is established with the lowest phosphorus in the upper layers of the bath. The difference in phosphorus content from the top to the bottom of the bath is least in the cases with deep penetration, eo go standard conditions and the 10 inch bath with 35 psigo The high lance, deep bath and small nozzle conditions produce good surface dephosphorization but are less effective in the lower portions of the bath.

50 IN, ABOVE BOTTOM_ 8 7 %P 6 __.2k: I ___-_- _. __ _ 2 4 6 8 10 12 BLOWING TIME 5 4 -- 3 __ 2 ~. 2 3 4 5 6 7_ IN. FROM WALL Fig. 21. % Phosphorus Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht. 25 psig., 0. 162 in. nozzle).

51 3) Silicon a) Standard Conditions The silicon content of the bath blown with standard conditions is plotted in Figure 22 as a function of blowing time and location in the bath using the same locations as for phosphorus. The silicon is oxidizedmost rapidly at the initiation of blowing and finally decreases to less than 0. 05% at all sampling locations. At the iniation of blowing the rate of silicon removal at the top and bottom of the bath are respectively (1. 5%/min. ) and (0. 5%/min. ). Gradients of silicon are established in the bath by these differences in the rate of silicon oxidation as shown in Figure 23. The composition lines are lowest in the bath near the center and then move upward and become horizontal as the lines approach the wall. After 0: 30 the silicon content has been reduced by 2/3 at the bottom of the bath and at the top of the bath the metal is virtually silicon free (<0. 05% Si). The gradient of silicon content is steepest at the top of the converter being 0. 10% Si/inch and is reduced to 0. 02% Si/inch between the 5-1/2 and 1-1/2 inch bath depths. b) Effect of Blowing Height A substantial reduction in the rate of silicon oxidation is caused by increasing the blowing height to 21 inches as shown in Figure 24. The silicon content at the bath surface is reduced at a rate of 0. 5% Si/min or 1/3 the rate found when the blowing height is 7 inches. At the bottom of the bath the rate is 0. 15% Si/min. compared with 0. 5%/min for the 7 inch lance height. Maximum silicon oxidation rates are found at the initiation of blowing. The silicon content is reduced to less than 0. 05% Si at the top of the bath and to 0. 06%7 Si 3-1/2 inches above the bottom of the bath at 5: 30.

52 IN. ABOVE BOTTOM 8 %Si 6" f f 1 2 3 45 BLOWING TIME 6 —' 5 4 3 ""~*.. —. —-. _ 31 2 3 4 5 6 7 IN. FROM WALL Fig. 22. % Silicon Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

53 IN. FROM BOTTOM 7.05 IN. FROM |W-ALL-. (6 in. bath, 7 in. bl ht., 20 psig,. 316 in. nozzle). 0 1 2 3 4 5 6 (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

54 N. AOVE BOTTOM_ 9 %Si 123456 4~ ~.4 2 1 2 3 4 5 6 BLOWING TIME, i.., g.,0.316 i. n.- IN. FROM WALL Fig. 24. % Silicon Versus Blowing Time for all Sampling Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle).

55 Constant composition lines at 0: 30 are shown in Figure 25. The lines higher in the bath have a definite upward curve from the center to the edge of the bath. Lower in the bath the 0. 5% Si line is horizontal. The bottom 3 inches of the bath are at nearly constant composition averaging 0. 54% Si or 90% of starting silicon. At the IN. FROM BOTTOM 2 -.01 --------------- -------- ----- ------ 0 1 2 3 4 5 6 7 IN. FROM WALL Fig. 25. Constant % Silicon Versus Bath Location at 1/2 Minute Blowing Time (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle). centerline, high in the bath the silicon content is reduced to 0. 30% or 50% of starting silicon. The silicon gradient is 0. 05% Si/inch in the uppermost 4 inches of the bath. Below this there is no appreciable silicon gradient. The dotted lines in Figure 25 are

56 standard condition data. The difference in rates of silicon oxidation is shown clearly by comparison. Silicon has been removed three times as rapidly when the bath was blown from 7 inches. c) Effect of Bath Depth In Figure 26 the solid curves and points are the % Si=time relationships for the sampling locations in a 1 0 inch bath blown at 20 psig. The silicon at the bath surface is oxidized very rapidly to 0. 03% Si after 0O 1 5 blowing time yielding a silicon oxidation rate of 2. 3% Si/min, at the start of the blow. The silicon content at the bottom of the bathis reduced to 0. 03% after 4: 00. The initial rate of silicon oxidation is 0. 1 5% Si/min, at the bottom. The open circles in Figure 26 are % Si versus time data from heats run with a 10 inch bath blown at 35 psig. Silicon content is reduced to 0. 01% at the top of the bath at 5: 15, and to less than 0. 05% Si at location 51 at 5: 22. Early in the run silicon at the bath surface is more rapidly oxidized when 20 psig. blowing pressure is used than when 35 psig. blowing pressure is used. Near the bottom of the bath (location 21) the opposite is true. The dotted lines in Figure 26 at the bath surface are data from the standard runs. Silicon at the bath surface of the 10 inch bath blown at 20 psig. is oxidized more rapidly than at the surface of baths blown under standard conditions. Surface oxidation in 10 inch baths blown at 35 psig. is somewhat slower than under standard conditions. The dotted lines in the 5-1/2 inch row of Figure 26 show, as expected, that at the same distance from the bottom of the converter sili0cn in the 6 inch bath is oxidized more rapidly than silicon in a 10 inch bath when both are blown at 20 psig.

57 %oSi r-l 4 lo _.,,,___ 2 p 2 4 6 8 BLOWING TIME IN ABOVE O TTOM 10 8 I "I 6 2 4 6 IN FROM WALL Fig. 26. %o Silicon Versus Blowing Time for all Sampling Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle - solid line) and(10 in. bath, 7 in. bl. ht., 35 psig., 0.316 in. nozzle - open points ).

58 d) Effect of Nozzle Diameter The % Si-time relationships for the heats blown at 25 psig. with a 0. 162 inch nozzle are shown in Figure 27. The usual pattern of most rapid silicon oxidation at the bath surface is apparent. The surface silicon content is reduced to 0. 01% at 11: 30 with an initial rate of 0. 4% Si/min.. The silicon content at the bottom of the bath is reduced to 0. 04% Si at 10: 00 with an initial rate of 0. 1 % Si/min.. The dotted lines on the plots are data from standard runs. The slower rate of silicon oxidation with the smaller lance is clearly apparent. For example, the blowing time required to reach 0. 05% Si at sample location 53 is 2: 00 for standard conditions as opposed to 6: 00 when the small nozzle is used. Summary - Silicon Oxidation In contrast to phosphorus removal, silicon removal depends only on contact with oxygen, either as gaseous or dissolved oxygen or as iron oxide. Accordingly, silicon oxidation is completed under all blowing conditions to below 0. 05% Si in the bath. As expected, the rate is roughly proportional to the oxygen input; however, with shallow jet penetration the rate is slower at the bottom of the vessel. Apparently when the silicon is oxidized at the surface in this case, the oxygen reacts with other elements in this region. By contrast, with deep penetration and stirring by the jet as in standard conditions, the silicon is oxidized throughout the bath giving a smaller gradient. In all cases the most rapid rate of silicon removal is at the start of blowing because the silicon is the most reactive element in the bath. As the silicon content decreases its activity falls and phosphorus and carbon can react with more of the oxygen.

59 IN. ABOVE BOTTOM 8 7 t,6~ 4 8 12 BLOWING TIME 5 4 3 2 1 _ 2 3 4 5 6 IN. FROM WALL Fig. 27.. %o Silicon Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0. 162 in. nozzle).

60 4) Carbon a) Standard Conditions The carbon content-time plots for standard experimental conditions are shown in Figure 28. Decarburization is most rapid at the bath surface. Carbon content is reduced to 0. 04% C at 4: 45. The rate of surface decarubrization is 1. 1% C/min. after 0: 30. Decarburization is slowest at the bottom of the bath. A minimum carbon content of 0. 1% C was found at the bottom corner at 4: 30. The maximum decarburization rate at the bottom of the bath is 0. 8% C/min. at 3: 00. The differences in the time of initiation and the rate of maximum decarburization produce substantial gradients of carbon in the bath. Constant % C lines are shown for standard blowing conditions in Figure 29. The range of carbon content is 1. 0 to 2. 6% C at 2: 30. This gradient is particularly interesting because at 2: 30 the entire bath is undergoing rapid decarburization which is alleged by some authors to minimize chemical gradients by the boiling action. In the upper portion of the bath the constant composition lines are horizontal while lower in the bath the lines curve downward from the edge of the bath to the center. The line representing 2. 6% C is nearly vertical. The maximum carbon gradient is 1% C/inch at the surface of the bath. b) Effect of Blowing Height The % C-time relationship for a bath blown from 21 inches is shown in Figure 30. Of greatest significance is the reduced decarburization. The minimum carbon content of the bath is 1. 98% C at the centerline at the top of the bath after 5: 25 compared to 0. 04% C at the same location and time for standard conditions. The maximum rate of decarburization at the top of the bath is 0o 3% C/min. at

61 IN, ABOVE BOTTOM 8';' \, i X,, i \.., \.,_ S 2 1_ 23456 BLOWING TIME 2I 1 2 3 4 5 6 7 IN FROM WALL Fig. 28. %o Carbon Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

62 IN. FROM BOTTOM 7 0 rs;4 1.0 0 1. 234 5 6 2.0 4 IN.0 1 2 3 4 5 6ALL IN. FROM WALL Fig. 29. Constant % Carbon Versus Bath Location at 2-1/2 Minutes Blowing Time (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

63 IN ABOVE BO TOM _____ 9 %C 4-. 3 2 2 4 6 BLOWING i1ME'7 — --- --- --- ---- ------------------ ---- 6 IV* - I m I 5 4 2' F / i I,_____2t__I_1 _I__I________5__6 _ I I I 1 2 3 4 5 6 7 IN. FROM WALL Fig. 30. % Carbon Versus Blowing Time for all Sampling Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle).

64 5: 00 as compared to 1. 1% C/min. at 0: 30 for standard conditions. Great differences also exist at the bottom of the bath. The extent and rate of decarburization are extremely slow under the 21 inch blowing height condition. c) Effect of Bath Depth Decarburization patterns for heats run with a 10 inch bath depth and 20 psig. are shown by the solid points and curves in Figure 31. As is usual, decarburization is most extensive and rapid at the surface of the bath (0. 2% C at 5: 00 with a maximum rate of 0. 7% C/min. at 3: 30) The bottom of the bath is the least effectively decarburized (0. 3% C at 7: 00 with a maximum rate of 0. 6% C/min. at 4: 30). The open circles on the plots are data from 10 inch baths blown at 35 psig. Decarburization at the bath surface is the same as when the bath was blown at 20 psig. Lower in the bath decarburization proceeds faster when the 35 psig. blowing pressure is used. The dotted lines in the 10-1/2 row of plots and the 5-1/2 inch row of plots are data taken from standard runs. Decarburization is faster with the standard experimental conditions at all bath locations than it is when the bath depth is increased to 10 inches (20 psig. ). The carbon content of the runs made with a 10 inch bath blown at 35 psig. correspond closely to those blown under standard conditions. The specific rate (scfm. /lb. of metal) of oxygen delivery is the highest for the standard conditions (0. 19 scfm. /lb. \ lower for 35 psig. - 10 inch bath conditions (0. 17 scfm. /lb. ) and lowest for the 20 psig. - 10 bath conditions (0. 12 scfm. /lb. ). The rate of decarburization varies directly with the rate of oxygen delivery.

65 %C 4 3 2 2 4 6 8 BLOWING TIME IN ABOVE BOTTOM 10 I II I I 8 6 0< < 4 2 2 4 6 IN FROM WALL Fig. 31. % Carbon Versus Blowing Time for all Sampling Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle - solid lines) and(10 in. bath, 7 in. bl. ht., 35 psig., 0.316 in. nozzle - open points).

66 d) Effect of Nozzle Diameter The effects of blowing a 6 inch bath with a 0. 162 inch diameter lance are shown in Figure 32. Decarburization is slow throughout the bath; however, the pattern of slowest decarburization at the bottom of the bath is maintained. At the bath surface the carbon content is reduced to 2. 3% C after 11: 30. The rate of decarburization is 0. 5% C/min. at 2: 00. The carbon content at the bottom of the bath is reduced to 3% C at 10: 00. The maximum decarburization rate is 0. 2% C/min. at 4: 00. The dotted lines in Figure 32 show the decarburization in a standard run. Oxygen is supplied 3. 4 times as fast under standard conditions as it is when the bath is blown with a 0. 162 inch diameter lance. Ths is roughly the ratio of the decarburization rates of the two conditions. Summary - Decarburization In contrast to Si elimination, C oxidation proceeds more slowly and in some cases the heat is stopped before low levels are reached. In the case of the 21 inch blowing height for example, a violent foaming condition is encountered by the time the carbon reaches 1.o 98% and the heat was discontinued. Similarly, with the small lance the heat was stopped after the silicon was oxidized but before the carbon was eliminated. In this case the highest carbon was found all across the bottom layers of the vessel even directly beneath the nozzle.

67 IN. ABOVE BOTTOM 8 7 I; I I I I I I1 7- \ - - I 1 - - \ BLOWING "1ME 5 1 1 2 3 4 5 6 7 IN. FROM WALL Fig. 32. % Carbon Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0. 162 in. nozzle).

68 5) Manganese a) Standard Conditions The manganese distribution in a 6 inch bath blown with standard conditions is shown in Figure 33. As expected, the curves resemble those for silicon. Manganese content is reduced throughout the bath to 0. 08% Mn by the end of the blow. The rate of manganese oxidation at the start of a run is 0. 7% Mn/min. at the surface of the bath and 0. 3% Mn/min, at the bottom of the bath. Manganese oxidation is always most rapid at the initiation of blowing. b) Effect of Blowing Height In Figure 34 the manganese pattern in a 6 inch bath blown from 21 inches is shown. Similar to silicon, the pattern of most rapid oxidation at the top of the bath is found (3% Mn/min. versus 0. 07% Mn/min. at the bottom of the bath. ). Of greater significance is the extensive reduction in manganese content at the top-center of the bath while the rest of the bath is maintained at much higher manganese levels. As an example, after 3: 00 the manganese content is 0. 07% at the top-center and 0. 38% at the bottom corner. The difference in concentrations reflects the poor circulation from the jet impingement area to the bottom of the converter with shallow jet penetration. The initial rate of manganese oxidation is 0. 6% Mn/min. at the top-center of the bath. c) Effect of Bath Depth The % Mn-time relationships for a 10 inch bath blown at 20 psig. are shown by the solid lines and points in Figure 35. The bath manganese content is reduced to 0. 14% at all locations. The maximum rate of manganese oxidation at the bath surface is 0. 5% Mn/min. while the rate is 0. 3% Mn/min. at the bottom. In both cases the maximum rate is found at the start of the run.

69 IN. ABOVE BOTTOM _ll 8 %/Mn.6 -.4 2~ 2 4 6 BLOWING TIME 6 —, ___- K. k ____ 5 2-*'"- J - I ---- 2 1 2 3 4 5 6 7 ___ IN. FROM WALL Fig. 33. o Manganese Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0. 316 in. nozzle).

70 IN. ABOVE BOTTOM %Mn 6 24__. K-... 2 4 6 BLOWING TIME 7 5 4i 3 ______2.3.4 5 6.7 IN. FROM WALL Fig. 34. % Manganese Versus Blowing Time for all Sampling Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle).

71 %oMn.6 11 —-2 ___ —-, -- s... 44 6 8 4 2 4 6 8 IN. ABOVE BOTTOM 10 8 I 6 -0 I 4 I2 4 6__________ I i6I ___F I IN. FROM WALL Fig. 35. % Manganese Versus Blowing Time for all Sampling Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle - solid lines) and (10 in. bath, 7 in. bl. ht., 35 psig., 0.316 in. nozzle - open points).

72 The open circles in Figure 35 are data from runs with a 10 inch bath blown at 35 psig. In this case manganese oxidation is more extensive and the rate is slightly faster than in a 10 inch bath blown at 20 psig. d) Effect of Nozzle Diameter In Figure 36 the manganese pattern in a 6 inch bath blown with a 0. 162 inch diameter lance is shown. The bath manganese content is reduced to 0. 14% at 13: 38. The maximum rate of manganese oxidation is 0. 3% Mn/minrat the top of the bath and 0. 05% Mn/min. at the bottom of the bath. In both cases the rate of oxidation is greatest at the initiation of blowing. Summary - Manganese Oxidation Manganese is not quite as active as silicon and the gradients found are intermediate between silicon and carbon. Under conditions of deep penetration, e. g., standard conditions, the elimination of Mn is rapid throughout the bath, reaches a maximum rate early and falls to 0. 08% by the end of the blow. By contrast, where circulation is poor, as with the high lance position, there is still appreciable Mn (0. 14%) at the end of the heat.

73 IN. ABOVE BOTTOM %Mn.6 17 —— 1' 2'' __ —BLONI NG TIME 6 i____ _ ___ _ I iI II ________ I - I - 3 1 2 3 4 5 6 7 IN. FROM WALL Fig. 36. % Manganese Versus Blowing Time for all Sampling Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0. 162 in. nozzle).

74 6) Temperature a) Standard Conditions The temperature distribution in a bath under standard conditions is shown in Figure 37. Surface and bottom-center temperatures increase smoothly at 200~F/min. to 3000~F. at 2: 30. The bottom-corner temperature is linearly increased to 3000~F at 3: 30 giving a rate of 150~F. /min. b) Effect of Blowing Height The temperature pattern is markedly changed by increasing the blowing height to 21 inches as shown in Figure 38. The bottom temperatures increase only to 2600~F. in 5 minutes. At the top of the baththe temperature rises slowly at first then rapidly at 150~F/min. to 2950~F. at 3: 00 and then slowly to 3000~F. at 5: 30. c) Effect of Bath Depth When the bath depth is increased to 1 0 inches the temperature patterns shown in Figure 39 are produced. The top-center temperature increases at 300~F. /min. to 2800~F. at 1: 00 and then much more slowly to 2950~F. at 3: 30. The top edge temperature is increased smoothly to 3100~F. at 7: 00 at a rate of 100~F. /min.. The bottom temperatures show a two minute delay before rising. Temperatures at all locations are increased when the 10 inch bath is blown at 35 psig. as shown in Figure 40. The surface rises initially at 250~F. /min. and then more slowly to 31000F. at 3: 00. The bottom of the bath is heated slowly for the first minute and then the temperature is increased at 200~F/min. to 2900~F. at 3: 00.

75 IN. ABOVE BOTTOM 7 ~F 3250 - 6,."30 0 [ -' _- _ 2750- // -./ 2500 2 4 6 BLOWING TME 5 4 3.2 L 4, _ 11 2 4 5 6 7 IN FROM WALL Fig. 37. Temperature Versus Blowing Time at 4 Locations (6 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

76 IN. ABOVE BOTTOM _ 7 oF 32 50 6 3-00 2750 2500 2 4 6. 2 4 6 BLOWING TIM 5 4 3 2 1 2 3 4 5 6 7 IN. FROM WALL Fig. 38. Temperature Versus Blowing Time at 4 Locations (6 in. bath, 21 in. bl. ht., 20 psig., 0.316 in. nozzle).

77 IN, ABOVE BOTTOM 10 ~F 3250 - 3000.2750-. 2500 / 2' 6''I BLOWING I ME 6 8 8 6 4 2 2 4 6 7L IN. FROM WALL Fig. 39. Temperature Versus Blowing Time at 4 Locations (10 in. bath, 7 in. bl. ht., 20 psig., 0.316 in. nozzle).

78 IN ABOVE BOTTOM 10 oF 3250 3000.-' 2750 2500 2500 /22 4 6 8 BLOWING TIME 8 Fig. 40. Temperature Versus Blowin. Time at 4 Locations 6 j ~~~~~~~4 1N FROM WALL Fig. 40. Temperature Versus Blowing Time at 4 Locations (10 in. bath, 7 in. bl. ht., 35 psig., 0.316 in. nozzle).

79 d) Effect of Nozzle Diameter Temperature-time curves are characterized by long holds when the bath is blown with a 0. 162 inch lance as shown in Figure 41. Temperatures are constant or are decreasing slightly for 4 minutes at the bath surface and 6 minutes at the bottom and then the temperatures increase at approximately 125~F. /mino throughout the bath. Summary - Temperature The maximum rate of temperature increase and minimum gradient are encountered with deep penetration conditions. By contrast with shallow penetration, as obtained with a small diameter nozzle, the temperature at the bottom of the bath actually falls for several minutes. These temperature distributions show clearly that the metal which is oxidized and heated near the jet impingement area moves radially outward at the surface not downward at the cente rline.

80 IN ABOVE BOTTOM 7 0F 3250 - ---— 3000 -- 6 2750 - 25 00______.____ 12 BLOWNG TIME 5 4 3 2 2 il I I 1 1 2 3 4 5 6 7 IN FROM WALL Fig. 41. Temperature Versus Blowing Time at 4 Locations (6 in. bath, 7 in. bl. ht., 25 psig., 0.162 in. nozzle).

81 7) Slag Composition The slag samples provide interesting data when considered with the chang e s in metal chemistry just discussed. The analyses of all samples taken are given in Appendix I, Data. It is best to begin the discussion with a review of the average slag composition given by the spoon (S-2) samples and then progress to the probe samples to illustrate composition gradients. First the change in P 0 will be discussed and then the oxygen level as FeO. P 0 Content of Average Slag a) Standard Conditions In Table VII are presented the compositions of four slags taken with a spoon sampler with the vessel turned down after blowing. Under standard conditions these samples represent blowing times in excess of four minutes and were well mixed with the slag sampling spoon. The P20 contents of these slags range between 2. 3% 25 and 3. 3% and represent good bath dephosphorization under a hard blowing condition. b) Effect of Blowing Height By contrast, the analyses of four other slags, with a 21 inch blowing height, are given in Table VII. The range of P205 contents is 2. 6% to 3. 5%. This blowing condition produced nearly the same P20 content as did the standard condition; however, 25 the bath was at a much lower temperature and other conditions were impractical for good steelmaking.

82 Table VII Compositions of Homogeneous (Average, S-2) Slag Samples For All Blowing Conditions Blowing Slag Composition (weight percent) Time SiO Fe203 FeO A2O CaO MgO MnO P20 (min.) S-2 Samples Standard Conditions 4:10 15.8 4.6 7.5 2.8 48.8 11.2 3.3 3. 3 5:15 13.8 7.3 17.0 1.1 43.5 10.5 2.8 2.4 5:30 9.9 8.9 22.2 0.9 40.0 9.8 3.1 2.3 5:32 9.2 6.3 10.5 4.9 50.8 9.8 2.4 2.8 21 inch Blowing Height 3:46 16.5 --- --- 4.8 41.9 11.8 3.6 3.4 4:15 10.4 9.7 28.4 4.6 33.3 6.2 2.7 3.5 5:30 9.2 10.9 31.4 3.9 28.8 9.2 2.8 2.9 5:32 7.2 8.9 39.9 3.2 28.6 6.4 2.0 2.6 10 inch bath, 20 psig. 8:32 12.9 1.8 4.9 0.9 53.6 19.0 2.1 1.8 10 inch bath, 35 psig. 5:30 13.5 7.4 17.7 7.6 24.6 25.2 2.2.8 6 inch bath, 25 psig. 13:38 8.9 7.9 7.7 2.4 57.1 8.4 1.3 2. 6

83 c) Effect of Bath Depth Only one slag sample is listed in Table VII, at 8: 32, for a heat employing a 10 inch bath blown at 20 psig. The slags are lumpy up until this time and at 8: 32 (the end of the blow) the slag is very stiff. The low P205 contnet, 1.8%, reflects the poor dephosphorization of the batho Increasing the blowing pressure to 35 psigo produced a fluid slag in 5: 30. The P05 content listed in Table VII is 1.8%; however, the slag volume is large as indicated by the low CaO content, 24. 6%. The additional volume was produced by attack of the lining as evidenced by 25, 6% MgO. The bath is adequately dephosphorized indicating the importance of deep jet penetration, 7-1/2 inches, as opposed to 4-1/2 inches with 20 psigo d) Effect of Nozzle Diameter The composition of one slag sample, taken at 13: 38, is listed in Table VII for heats blown with the 0. 1 62 inch diameter nozzle. Slags were very lumpy until this time and this is reflected in the poor dephosphorization of the batho P2 0 Content of Probe Samples The probe samples provide an interesting comparison with spoon (average) samples taken from other heats at about the same blowing time. a) Standard Conditions In Table VIII the composition of probe samples, taken at 8 and 11 inches from the bottom of the converter, are presented. The 8 inch sample contains 3. 7% P 0 while the 11 inch sample contains 2. 80/0 P205 indicating higher phosphorus contents closer to the bath. Since the average P205 at this time is about 2o 5% this indicates lower P205 levels higher in the slag.

84 Table VIII Composition of Probe Samples Taken From a Homogeneous Slag and Comparative Average Compositions Blowing Slag Composition (weight percent) Time SiO Fe03 FeO AO CaO MgO MnO P 0 (min.) Standard Conditions - 8 inch height 4: 40 11. 5 5. 5 17. 5 0.9 36.2 14.2 3.9 3.7 - 11 inch height 5: 21 9.8 8.4 26.0 0.8 38.1 9.5 3.5 2. 8 - S-2 for comparison 5:15 13:8 7.3 17.0 1.1 43. 5 10. 5 2.8 2. 4 21 inch Blowing Height - 9 inch height 3:15 5.4 9.7 32.6 2.3 21.3 22.4 1.9 0.7 - 11 inch height 5: 00 5. 6 20.2 27.0 2.2 22.4 16.7 1.4 0.6 - S-2 for comparison 4 15 10.4 9.7 28.4 4.6 33,3 6.2 2.7 3. 5 0. 162 inch diameter nozzle - 8 inch height 12: 15 10. 5 7.0 28 4 2,3 27.9 13. 2 4.7 0.9 - S-2 for comparison 13:38 8.9 7.9 7,7 2.4 57.1 8.4 1.3 2.6

85 b) Effect of Blowing Height Compositions of the 9 inch and 12 inch level samples are 0. 6 - 0. 7 P20 for the 21 inch blowing height conditions, Table VIII. Since the S-2 samples during the same time period contain 2. 6% to 3. 5% P -0 there is a substantial phosphorus gradient very close to the bath surface. This is reasonable because the slags produced under this condition are stiff allowing little slag circulation and phosphorus transport. c) Effect of Nozzle Diameter Similar results are obtained by blowing with the 0. 162 inch diameter nozzle. The 8 inch probe sample contained 0. 9% P 0 at 12: 15 in a very stiff slag. By contrast the average (spoon) sample taken at 13:38 contained 2..6% P205. Here too, a phosphorus gradient is evident between the bath surface and the 8 inch sample height. FeO Composition of Average Slag Pronounced-differences in FeO content are obtained when standard and soft blowing conditions are compared, Table VII, (p. 86). This is illustrated by average (spoon) samples. a) Standard Conditions Under standard conditions the FeO reaches 7. 5% at 4: 00 and 17% at the end of the heat; therefore, with this type of blowing the FeO content of the slag. remains low (8%o) throughout the decarburization period of the bath and finally increases to about 17% which is in the composition range reported for commercial operations in Table III.

86 b) Effect of Blowing Height Increasing the blowing height to 21 inches increases the FeO content of the slag as shown in Table VII. At 4: 15 the % FeO in the slag produced with these conditions is 28. 4% as opposed to 7. 5 with standard conditions.and the FeO content reaches about 35% at 5: 30 as compared to 17% with standard conditions. Therefore, the chief difference in this condition compared with the standard is that the FeO contents of the s.lag are increased before the bath is decarburized. This high reserve of oxygen, coupled with the oxygen being delivered by the lance can lead to very high interface CO evolution and contribute to the slopping condition. Table VII - Partially Repeated Compositions of Homogeneous (Average, S-2) Slag Samples For All Blowing Conditions Blowing Slag Composition (weight percent) Time SiO Fe20 FeO AL0 CaO MgO MnO P20 (min. ) S-2 Samples Standard Conditions 4:10 15.8 4.6 7.5 2.8 48.8 11.2 3.3 3.3 5:15 13.8 7.3 17.0 1.1 43.5 10.5 2.8 2.4 5:30 9,9 8.9 22.2 0.9 40.0 9.8 3.1 2.3 5:32 9.2 6.3 10.5 4.9 50.8 9.8 2.4 2.8 21 inch Blowing Height 3:46 16.5 --- ---- 4.8 41.9 11.8 3.6 3.4 4:15 10.4 9.7 28.4 4.6 33.3 6.2 2.7 3.5 5:30 9.2 10.9 31.4 3.9 28.8 9.2 2.8 2.9 5:32 7.2 8.9 39.9 3.2 28.6 6.4 2.0 2.6

87 8) Special Heats After a review of the preceding data special heats were made to investigate the following points: a) The role of the basic slag in dephosphorization, b) The effect of high initial phosphorus on bath dephospho rization, c) The nature and amount of metal droplets thrown up by the jet, d) The role of droplets in dephosphorizationo a) The Role of the Basic Slag in Dephosphorization Heat R-40 was made without a flux addition and started with a clean metal surface to determine what portion of dephosphorization is caused by MgO from the lining or by vaporization of P or P205. The heat was run under standard conditions although it was stopped intermittently to take metal samples. The results of this heat are summarized in Table IX. Table IX Composition of a Bath Blown under Standard Experimental Conditions but Without Flux Blowing Time % P % Si % C % Mn (min.) 0 0.22 0.28 3.90 0.26 1-1/2 0.22 0.01 3 36 0. 06 3 0. 20 <0. 01 1. 30 0. 06 4-1/2 0.22 <0 01 0. 20 0. 08 Final Slag: 66. 4% SiO, 16.9% FeZ0, 5 5% FeO, 3. 1% CaO, 0. 3% P2O

88 Silicon, carbon and manganese are all reduced in the usual manner while the phosphorus content is virtually unchanged. The final slag is acid with 0. 3% P05 and 16. 9% Fe 0. Three points are evident in these results. Dephosphorization by vaporization of P or P05 is not obtained. MgO from the furnace lining is not a major factor in bath dephosphorization under normal conditions. The Fe20 -FeO ratio in the slag is very large in this case compared to a normal run. This indicates the role of the basic slag in a normal run in dephosphorization. b) The Effect of High Initial Phosphorus on Bath Depho sphorization Heat R-41 was run with 0. 76% initial phosphorus. Four sampling locations were placed with centerlines 6 inches above the bottom of the converter. In all other aspects the heat was run under standard conditions using flux. The results of this heat are compared to the usual experimental data in Figure 42. The dephosphorization rate is higher with higher initial phosphorus, and, as expected, the phosphorus content of the slag is higher. c) The Nature and Amount of Metal Droplets Thrown Up By the Jet Heat R-39 was run to determine the amount and size distribution of the droplets thrown up by the oxygen jet under various blowing conditions. The techniques used for this heat were described in the Experimental Procedure section.

89 o/oP lo P,8 *8 l| SAMPLE LOCATION 61 FINAL SLAG ~/oP205.6 - STANDARD 3% HIGH Po 4/o M-2..2 ESTIMATED 0 1 2 3 4 5 BLOWING TIME(MIN.) Fig. 42. % Phosphorus Versus Blowing Time at Sampling Location 61 for Initial % Phosphorus of 0. 76% and 0. 24%. The weight of droplets thrown up is tabulated in Table X. Table X Distribution of Droplets in the Converter (Values are lb. of droplets thrown up/min.) Blowing Oxygen Bath Radii Ht. (in.) (psig. ) 2-3 3-5-1/4 5-1/4-8 Total 7 10 3 2 1 6 7 20 25 12 3 40 7 30 14 40 6 60 The weight of droplets increases with blowing pressure. During a 5-1/2 minute run with standard experimental conditions the weight of droplets thrown up is equal to the weight of the bath. Under standard experimental conditions the flux of droplets across the bath surface decreases with distance from the center.

90 The droplets collected are found to range in size from 0. 1 cm. to 0. 6 cm. in diameter. Several were subjected to metallographic examination and a typical structure is shown in Figure 43. Fig. 43. 0. 6 cm. Diameter Droplet Showing Ferrite, Pearlite and Gas Holes. 2% Nital Etch, 20X. The large dark interior areas are gas holes. These gas holes are found to account for at least 50% of the sample volume, apparent density = 3. 5 gm/c. c.. Of greater importance is that the droplet is composed of ferrite and pearlite in a ratio indicating about 0. 4% C although the bath contained 3. 6% C at the time.

91 d) The Role of Droplets in Dephosphorization Heat R-38 was run under standard experimental conditions except that all but one of the sampling locations at the bath surface one inch from the wall were protected from falling metal droplets by a refractory shield. Four sampling locations were prepared with centerlines 6 inches above the bottom. The % P at location 61 of this special heat is compared with standard data in Figure 44. Compared to standard conditions, dephosphorization is retarded at the protected locations and not retarded in unprotected sampling location near the bath surface. 0/oP.3 SAMPLE LOCATION 61 LEGEND.2 * *: PROTECTED X: UNPROTECTED X\: EXP. DATA 1 22 3 4 BLOWING TIME (MIN) Fig. 44. % Phosphorus Versus Blowing Time for Protected and Unprotected Samples Taken at Sampling Location 61.

92 9) Summary of the Results The changes in bath chemistry and temperature for all of the blowing conditions are summarized in Figures 45 and 46 by plotting P, Si, and C contents or temperature versus blowing time at two critical sample locations. The sample locations are 56 and 11 for those conditions employing a 6 inch bath depth or 86 and 21 for that condition employing a 10 inch bath depth. By using these summary graphs the effect of a given operating condition upon chemistry and temperature can be seen. a) Standard Conditions Dephosphorization is faster and more extensive at the top of the bath than at the bottom as shown by comparing the curves marked 1 in Figure 45. Similarly, silicon oxidation, curves 2, and decarburization, curves 3, proceed faster and more extensively at the top of the bath than at the bottom, and temperatures are higher at the top of the bath as shown in Figure 46. The more extensive oxidation at the top of the bath is due to the proximity of the slag-metal interface and the falling droplets in the case of dephosphorization or the proximity of the oxygen source (the heat source) in the cases of silicon oxidation, decarburization and bath temperature. The completion of silicon oxidation before dephosphorization or decarburization rates increase is best shown in Figure 45 at location 56. Silicon oxidation is complete at 1: 00 while the maximum rates of dephosphorization and decarburization are found at 2: 00.

%oC %/oP,Si STANDARD 21 in. BLOWING HEIGHT 10in. BATH DEPTH 0.162 in. DIAMETER NOZZLE 3.6 C. _ 242 - — 2 Si ____ 0 56 56 86 56 LOCATION. 11 t1 _____ ___ 2'11,, 0 16 3 23 10 11 P Si.'Si O 1 2 3 4 5 6 1 2 3 4 5 6 12 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 910 11 BLOWING TIMES (min) Fig. 45. Summary Plots of P, Si. and C Contents at 2 Sample Locations Versus Blowing Time for all Blowing Conditions.

94 OF 3200 ------— _^ LEGEND 0 SYMBOL CONDITION S STANDARD 3000 // 10 21 21inch BLOWING HEIGHT 0 110inch BATH-20 psig. 2 0/O 10inch BATH-35 psig. 2800 / - 162 0.162inch NOZZLE 2600 21 1 I 2 2400 - 3200 3000 2800 10 2sool// )0 ^, 16 2600 2 162 -- 2400 ____ 2 4 6 8 10 12 14 BLOWING TIME (min.) Fig. 46. Summary Plots of Bath Temperature at 2 Locations Versus Blowing Time for all Blowing Conditions.

95 b) 21 inch Blowing Height In this case, dephosphorization is faster and more extensive near the bath surface than it is near the bottom as shown in Figure 45. Similarly, silicon oxidation and decarburization proceed faster and more extensively at the surface and the bath temperature at the surface is higher as shown in Figure 46. The reasons for these variations are the same as those for standard conditions. Of greater importance however are the differences produced by increasing the blowing height. Dephosphorization under these and standard conditions is proceeding rapidly near the bath surface, but near the bottom of the bath dephosphorization is much slower than under standard conditions as seen by comparing the curves marked 1 on the plots. This is due to greater jet penetration under standard conditions causing greater circulation which transports low phosphorus material to all parts of the batho The same reason explains the much larger differences in silicon content from the top to the bottom of the bath blown from 21 inches as seen by comparing the curves marked 2. The differences in carbon content from the top to the bottom of the bath blown from 21 inches are large; however, the significant comparison regarding decarburization between the two systems is the extent of decarburization at corresponding sampling locations. Under standard conditions the carbon content at location 56 at 5: 00 is 0. 1% while at the same location and time the carbon content of a bath blown from 21 inches is 3%, curves 3. This difference is not explained by differences in circulation but the data suggest that the character of the et and of the edecarburization mechanism are changed by increasing the blowing height.

96 c) 10 inch Bath Depth Dephosphorization, silicon oxidation and decarburization are faster and more extensive near the top of the bath than near the bottom, Figure 45. The effect of a decreased jet penetration - bath depth ratio on circulation is best shown in Figure 46 by differences in temperature at the two locations. Near the top of the bath the temperature begins to increase at the initiation of blowing and continues to increase throughout the heat. At the bottom of the bath the temperature remains constant for 2 minutes before beginning to increase. This two minute delay is caused by the absence of circulation at the bottom of the bath. Silicon oxidation and decarburization are slower than with standard conditions but equally extensive. The lower rate is due to decreased oxygen flow rate per pound of bath. Dephosphorization is slower and not as extensive as with standard conditions, curve 1. The high phosphorus content in the 10 inch bath is not due simply to the decreased oxygen rate because, as shown, silicon is oxidized to low bath content, curves 2. The high final phosphorus content is due to the dry, low FeO slag produced under these conditions emphasizing the importance of a fluid active slag to dephosphorization. d) 0. 162 inch Diameter Nozzle Dephosphorization, silicon oxidation and decarburization are faster near the top of the bath than near the bottom. The maximum extent of dephosphorization and decarburization is not known because the heats were not run to completion. The effect of decreasing the nozzle diameter is to decrease the oxygen flow rate to 30% of that under standard conditions andto decrease jet penetration to 40% of that under standard conditions thus decreasing

97 the heating rate and circulation of the bath, Figure 46. The temperature near the bath surface remains constant during the first four minutes of the run. During this time the flux is being heated. The temperature near the bottom of the bath remains constant for six minutes. The additional two minutes are the time required to circulate hot metal from the top of the bath to the bottom.

98 10) Mathematical Analyses In this section the data are subjected to four types of mathematical analyses: a) Mass Balance, b) Oxygen Efficiency and Utilization, c) Comparison Between Equilibrium and Actual Phosphorus Contents and d) Comparison Between Dephosphorization Mechanisms and Data. a) Mass Balance The mass balance is performed by calculating the weight of element J which is charged with the hot metal and comparing this with the weight of the element in the metal and slag at the end of a run.* These calculations are made for all blowing conditions. The results are given in Table XI in terms of "accountable element J" which is defined as the ratio of the weight of J at the end of the blow to the weight of J at the beginning of the blow. The weight of phosphorus and manganese in the system calculated from final metal and slag compositions is close to the weight of phosphorus and manganese charged in the hot metal. The calculated final silicon weight is consistently greater than the initial silicon weight for all experimental conditions. Table XI Accountable Element J for All Blowing Conditions bath depth 6 6 10 10 6 blowing height 7 21 7 7 7 blowing pressure 20 20 20 35 25 nozzle diameter 0.316 0.316 0.316 0.316 0.162 Accountable P 1.1 1.1 1.0 1.3 0.9 Accountable Si 1.5 1.3 1.8 1.5 1.5 Accountable Mn 1. 0. 11 0.8 1. 1 0.9 * The mass balance is derived in Appendix II, Calculations.

99 b) Oxygen Efficiences and Utilizations Oxygen efficiences* and utilization are calculated by assuming formation of SiOZ, MnO, P205, and CO. When the oxygen efficiency is less than 70%, CO is replaced by CO2 in the calculation. Oxygen efficiencies and utilizations are tabulated in Table XII for all the blowing conditions, The oxygen efficiencies are generally high except in the case of the 21 inch blowing height experiment where oxygen efficiency is 51% assuming all CO or 69% assuming all CO. Table XII Oxygen Efficiency and Utilization for All Blowing Conditions O0 In OtoP 2 toSi 02toMn 0 toC 02toFe 02Eff. (lb.) (lb.) (lb.) (lb.) (lb.) (lb.) (%) Standard Conditions at 5: 30 23.3 0.53 1.6 0.32 13.3 3.8 84 21 inch Blowing Height at 5: 30 23.3 0.72 1.7 0.27 3.9 5.7 51 23.3 0.72 1.7 0.27 7.8* 5.7 69* 10 inch Bath Depth, 20 psig at 8: 32 36.1 0.58 2. 2 0.32 21 4 0.91 73 10 inch Bath Depth, 35 psig at 5: 30 33.12 0.80 2.7 0.46 20.2 7.4 95 0. 162 inch Diameter Nozzle at 13: 38 12.7 0.48 1.1 0 19 7.2 1.4 82 * Assumed 100% CO. * An oxygen efficiency calculation is given in Appendix II, Calculation,

100 Oxygen efficiences and utilizations are calculated at 30 second intervals for standard and 21 inch blowing height conditions. These calculations are referred to as differential oxygen efficiencies and utilizations. Differential oxygen efficiency and utilization for standard conditions are plotted in Figure 47. Oxygen efficiency is relatively OXYGEN TO CO (x) and C02 (e) P(O),Si(0).Mn( ) OXYGEN EFFICIENCY (lb.) ~/o X, X X 1.0 CO- 100 I~x- ^ -J~XX 90 3..6 60 2. Ai 40 30 Fe 1..2 20 ___Mn- 10 0 1 2 3 4 5 6 BLOWING TIME (MIN.) Fig. 47. Differential Oxygen Efficiency and Utilization Versus Blowing Time with Standard Conditions. high throughout the experiment. The order of occurrence of maximum oxidation rates is silicon and manganese at 0: 30, carbon at 2: 30, phosphorus at 3: 00, and iron at 5: 30. One lb. of oxygen is used to

101 oxidize Si in the first 30 seconds. 2.7 lb. of oxygen are used to oxidize carbon between 2: 00 and 2: 30. 1.9 lb. of oxygen are used to oxidize iron between 5: 00 and 5: 30. The time intervals when oxygen efficiencies are plotted for both CO and CO2 indicate that sufficient oxygen is available for either gaseous product. The differential oxygen efficiency and utilization plot for the 21 inch blowing height condition is shown in Figure 48. Silicon OXYGEN TO P (O), Si(0), C (A), Mn (I ), Fe (o) (Ib.) 1.2 e CO (X) and C02 (e) OXYGEN EFFICIENCY 0/0 1.0 100.2 —-— ^Co- 20 —- 90.8 - ^ L \ ---- ---- 80 1=~ r". 0..2 EFF. 3 70.6 \ 60 -x~ vFe\ -- -- -0- ~~ —" — -- 50.4' X E' 40 30.2 -,,,, pL —---- 20'-'A L- 10 Mn 1 2 3 4 5 BLOWING TIME (MIN.) Fig. 48. Differential Oxygen Efficiency and Utilization Versus Blowing Time with 21 inch Blowing Height. and iron are oxidized first with manganese and carbon not having clear maximum oxidation periods and phosphorus is oxidized most

102 rapidly at 3: 30. 1. 1 lb. and 0. 35 lb. of oxygen are used to oxidize iron and silicon respectively during the first 30 seconds. 0. 76 lb. / minute and 0. 08 lb. /minute of oxygen are used to oxidize carbon and manganese respectively for long periods. 0. 21 lb. of oxygen are used to oxidize phosphorus between 3: 00 and 3: 30. Oxygen efficiencies assuming 100% CO are quite low. Higher efficiencies are attained if all the carbon is assumed to be oxidized to COZ. Even in this case, efficiencies decreased toward the end of the blow. c) Comparison Between Equilibrium and Actual Phosphorus Contents Equilibrium phosphorus content in the bath is calculated on the basis of thermodynamic data reported in the Review of Literature.. The slag compositions used in these calculations are final slag compositions and have been reported earlier. The temperature of the slag is assumed to be less than the temperature of the metal. The specific temperature used is reported in Table XIII along with the calculated equilibrium phosphorus content of the bath surface and the average bath % P. Equilibrium between the bath and slag phosphorus is not attained at the end of a run on the basis of the equilibrium calculations. The slag temperatures assumed are about 300~F below the metal temperature and an increase in slag temperature to metal temperature would increase equilibrium phosphorus content more than 100 times. For example, if the slag on a bath blown under standard conditions reached a final temperature of 3200~F the equilibrium phosphorus content would be 0. 4% or almost twice the initial phosphorus content of the bath.

103 Table XIII Comparison of Equilibrium and Actual Phosphorus Contents for All Blowing Conditions bath depth 6 6 10 10 6 blowing height 7 21 7 7 7 blowing pressure 20 20 20 35 25 nozzle diameter 0,316 0.316 0.316 0.316 0. 162 equilibrium % P 0. 004 0. 002 0. 02 0. 004 0. 01 bath surface % P 0.019 ----.11* 0 022 0.1* average bath %o P 0. 021 0. 017 - 0. 047 0. 07* slag temperature 2900 2800 2800 2900 2700 (~F) * Blowing time is less than that required to fully decarburize. d) Comparison Between Dephosphorization Mechanisms and Data Four mathematical models were developed to test dephosphorization mechanisms* under standard conditions. The mechanisms tested were: 1) bulk diffusion, 2) boundary layer diffusion, 3) dilution by dephosphorized metal droplets, and 4) a combination of boundary layer diffusion and dilution. 1) Bulk diffusion is suggested as a dephosphorization mechanism by the higher phosphorus gradient near the slag-metal interface. The model is calculated with Fick's Second Law by assuming a phosphorus content of 0. 22% at blowing time 1: 00 and calculating the change in the surface concentration with time. * The models are derived in Appendix II, Calculations.

104 This model ( % phosphorus-time curve 1) is compared to average bath % phosphorus data (curve D) in Figure 49. The model %P 0.3 0.2 - X = = - -- LEGEND DATA: EXPERIMENTAL DATA 1: BULK DIFFUSION 2a: BOUNDARY LAYER 2b: CIRCULATING LAYER 3: DROPLET 0.1 \ 3': MODIFIED DROPLET,.a 4: COMBINED 0~_ 1 2 3 4 5 6_DATA 1 2 3 4 5 6 BLOWING TIME Fig. 49. Average Bath % Phosphorus Calculated from Mass Transfer Models Compared to the Data - Standard Conditions. does not fit the data indicating that bulk diffusion is not the dephosphorization mechanism and that bulk diffusion makes little or no contribution to phosphorus transfer. 2a) Boundary layer diffusion from a turbulent bath is suggested by the bath turbulence in the converter. The model is calculated by assuming on a phosphorus content of 0. 22%o at 1: 00 and then calculating the decrease in average bath.o P due to phosphorus transfer across the metal boundary layer.

105 This model is plotted as curve 2a in Figure 49. The model and the data agree in magnitude from 1: 00 to 2: 30; however, the model curve has the wrong slope. 2b) Boundary layer diffusion from a circulating bath is suggested by the bath circulation. The- model is calculated by assuming 0. 22% P at 1: 00 and a surface layer of thickness S through which metal is circulated from the lower bath at the center of the converter outward to the wall. This model checks not only the mechanism but the direction of bath circulation, This model is plotted as curve 2b in Figure 49o Like 2a, this model and the data agree in magnitude from 1: 00 to 2: 30; however, the model curve has the wrong slope. The % P in the surface layer calculated from the model is compared to data in Figure 50 at the three sampling locations in the 6-1/2 inch row. The model, curve 2b, does not fit the data well; however, a decrease in phosphorus content from the center to the wall which is predicted by the model is shown clearly in the data indicating that the outward circulation assumed in the model is correct, 3) Dilution by dephosphorized metal droplets is suggested by the effect of metal droplets on dephosphorization which was shown to be an important part of bath dephosphorization in Run R-38. The model was calculated by assuming 0. 22% P at 1: 00. The droplets are assumed to be thrown up by the jet fall through the slag, and return to the bath containing 0. 04% P from 1: 00 to 4: 30 and 0. 02% P from 4: 30 to 5: 30. The rate at which droplets are thrown up is 100 cc/sec. for the model.

LEGEND D: EXPERIMENTAL DATA 2b: CIRCULATION 3: MODIFIED DROPLET 4: COMBINED %P 61 %P 63 %P 66 3.3 ------.3 —---.2.2.2.1.1 D_ 2b _2.1 0 1 2 4 5 60 1 2 3 4 5 60 1 2 3 4 0 1 234560 1 234560 1 23456 BLOWING TIME BLOWING TIME BLOWING TIME Fig. 50. Surface % Phosphorus Calculated from Mass Transfer Models Compared to the Data - Standard Conditions.

107 This model is plotted as curve 3 in Figure 49. The model and the data agree in magnitude from 1: 00 to 2: 30; however, the slope of the model curve is in the wrong direction. 3') This model differs from model 3 in that the rate at which metal droplets are thrown up is assumed to be proportional to the rate of decarburization. The rate at which droplets are thrown up increases from 100 cc/sec. at 1: 00 to 300 cc/sec. between 2: 30 and 4: 00. This model is plotted as curve 3' in Figure 49. The model does not agree with the data until 4: 30; however, the model is significant in that it produces a curve with the same slopes as the data curve. If the model curve is translated to the right a distance equivalent to 30 seconds it fits the data almost exactly. While the model does not predict changes in surface concentration with distance from the center, it is plotted in Figure 50 as curve 3'. The model is in good agreement with the data at sampling location 66 and lagged behind the experiment data at locations 63 and 61. 4) Combination of boundary layer diffusion (2b) and dilution (3) is suggested by these realities of converter operation. The two models are described above and were combined by determining the flux of droplets across the bath surface. The flux is determined from Run R-39 to be 40/r3, cc. /sq. cm. /sec. or cm/sec. The flux decreases rapidly with increased distance, r, from the center. This model is plotted as curve 4 in Figure 49. The model agrees with the data at 5: 00. Before this, phosphorus content calculated by the model is lower than actual bath phosphorus contents,

108 The surface concentrations at the sampling location calculated from the model are compared to the data in Figure 50. The model is plotted as curve 4. The phosphorus content calculated by the model is lower than the actual phosphorus content at all three sampling locations. As in the case of model 2b, decreasing % P with increasing r is predicted by the model and reflected in the data. To summarize, the relationship between phosphorus content and blowing time is not calculated from any of the models with sufficient accuracy to deem that model a success; however, the models do serve to emphasize two important features of the results. First the decrease of % P in the surface layer from the center to the wall of the converter is predicted by assumed outward circulation in models 2b and 4. Second, the interrelation of the elements in the metal bath is emphasized in model 3' with which the shape of the dephosphorization curve is predicted by assuming that the droplet volume is proportional to the rate of decarburization.

Conclusions 1. Pronounced chemical and thermal gradients are demonstrated in the bath of a pilot size oxygen converter. The chemical gradients are shown for the first time. Both the chemical and the thermal data indicate a circulation pattern which is up at the center and radially outward at the surface. This is contrary to a large group of publications. These gradients are all strongly affected by operating conditions as indicated in the following conclusions. 2. Dephosphorization occurred most rapidly under conditions of deep penetration, e. g., standard conditions. With soft blowing, good dephosphorization at the slag-metal interface was attained but a severe gradient developed due to poor circulation. Experiments with a 10 inch bath indicated that penetration should be increased to accompany the greater metal depth. While dephosphorization was encountered at the bathslag interface with soft blowing, a second mechanism was encountered with hard blowing. The reduced rate of dephosphorization at locations which were protected from metal droplets falling through the slag, showed that these droplets contributed significantly to bath dephosphorization. 3o Silicon oxidation went to completion under all conditions because this reaction did not require direct transfer of the element from the metal to the slag. In other words, silicon could react with oxygen anywhere in the bath to form SiO and then find its way to the slag. Despite this ease of oxidation, silicon gradients were present similar to phosphorus gradients and developed earlier. Manganese behaved similarly but was oxidized more slowly and to a lesser extent than silicon. 109

110 4. Carbon gradients developed later than those for manganese and silicon. It is significant that even at the height of decarburization, steep gradients were present which indicated circulation upward at the centerline. 5. Compositional gradients were also determined for the slag in a few cases and showed a higher phosphorus content near the metal interface. 6. The thermal gradients indicated that the metal moved radially outward from the jet impingement zone. The maximum temperature measured during a heat was 32500F. at the top of the bath, 1-1/2 inches from the centerline. The metal at the bottom of the bath was colder than the metal across the top surface. 7. The data indicate that best operating practice for dephosphorization is to penetrate deeply, i. e., beyond the center of the bath, while adding flux at a rate to insure a fluid slag.

Appendix I, Data 111

112 The data are presented in Tables XIV and XV below. Location AB = A + 1/2 in. above the bottom of the converter and B in. from the wall. Sample M-l is the chill-cast bath sample taken before the run. Sample M-2 is the chill-cast bath sample taken after the run. In Table XV sample S-2 is the slag sample taken after a run. In bath tables the data are tabulated by blowing conditions: standard, 21 inch blowing height, 10 inch bath and 0. 162 inch diameter nozzle conditions in that order. For each condition the data are tabulated from the lowest to the highest samples and from the wall toward the center of the converter. For example, data from sample location 11 are the first tabulated and data from sample location 66 the last. TABLE XIV Blowing Time - Metal Composition Data for all Blowing Conditions Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 11 26 0: 20.24.41 4. 00 36 27 0: 20.20.25 ----.30 23 1 00.21.07 3.88. 14 25 1: 00.22. 08 3.80.13 37 1 15.21 --- -- 20 1 45.25. 02 3. 25. 11 25 3 00 --- -- 1. 19. 13 17 3 15.17 <. 05 1.48 o29 37 3: 30. 16 --- 1. 58. 18 20 3 45.21. 02 1.61. 20 27 4 30. 05 <, 02 0. 10. 11

113 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 13 23 1:00.18 -- 3. 42.17 24 1: 00.23.10 3.85.14 25 1 00.20 --- 3.34.15 37 1: 15.21 <. 05 —- 26 2: 00.07 <. 05 0. 15.13 37 3: 30.18 ------.20 20 3: 45.18 04 1. 33.22 Location 16 27 0: 20 23.35 3. 57.38 25 1: 00.19.08 3. 26. 19 37 1: 15. 18 <. 05 3. 15 26 2: 00.05.10 ----. 11 23 3: 00.09. 06 1.69. 18 25 3: 00.12 04 1.88. 14 37 3: 30.16 -— 1. 38. 19 20 3 45.16 —----. 21 Location 31 26 0: 35.22.17 3.46. 19 37 0:40. 24 ---- 20 1:30.24.03 3.44.10 23 2 16.14 <. 05 2. 80.17 37 2:46.18 —----- 17 3: 00.27 <. 05 1.76.25 27 3:45.03.02 0. 10.10 26 5: 00.02. 01 0. 04.11 Location 33 20 1:30.24. 03 2.98.13 37 2: 46.17 ---- 17 3: 00.18 <. 05 1. 55.33 27 3:45.04 <. 02 0. 10.10 25 4: 00. 09 <. 05 1 16.14 26 5: 00.02.02 <0. 10. I0

114 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 36 37 0:40.20 <.05 3.04.12 37 2:46.14 --- 1.26.17 17 3: 00.17.08 1.75.30 27 3: 45.04 <.01 0. 10.12 25 4:00.10 <.01 1.06.15 26 5:00.02.01 0. 06 <. 10 Location 51 24 0:45. 14. 17 3. 28.15 25 0.45.22.08 3.90.15 25 2: 00.14 <.05 2 92.13 26 3: 00.09.07 2.85.18 37 3: 00.17 --- 1.91.17 26 4:15.02 <.05 <.10 <.10 27 4:15.09.11 ----.13 27 5: 00. 01. 01 0. 04 --- Location 53 24 0 45.20 --- 3.35.12 25 0:45.20.12 3.34.20 37 1: 30. 18. ~ 37 3:00.13 --- --- 26 4: 15.02.06 0. 05 <. 10 27 5: 00.02.02 0. 06 <. 10 Location 56 24 0:45.23.09 3.54.12 23 0: 46.09 <. 05 ----.20 25 2: 00.09 <.05 ----.17 14 2: 45 ---.02 1.81.25 27 5:00.02. 01.06 <. 10 Location 61 23 1:45. 07 <. 01 1.79. 13 37 2: 00.22 --- 2. 31.28 37 3: 16. 15 --- 1. 35.20 27 3: 30.02 <.05 0. 04 <. 10 27 4:45.02 --- 0. 06 <. 10

115 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 63 23 0:34.12 --- 224.15 20 1: 00.20.02 1.72.19 17 2: 30.17 <.05 1.46.25 37 3: 16.15 --- 1.49.21 27 4:45.02.06 0.06 <.01 Location 66 26 0: 05.16.08 3. 06 - 23 0: 34.21. 08 3.70.14 20 1:00. 19.02 ----.20 37 2:00.14 --- 1.25.18 17 2:30.16.05 1.40.24 37 3: 16.18 ------- 27 4:45.02 <.05 0. 04. 10 Sample M1 17.43.55 3.99.60 20. 33.52 3.97.48 23. 22.50 4.16.41 24.24.53 4. 14.48 25.24.53 4. 14.48 26.23.41 4. 00. 36 27. 25. 54 3. 99. 57 37.22.58 4.00.48 38.201.48 4.03. 50 39.03.10 3.96.04 40.218.28 3.90.26 41.760.50 3. 70.48

116 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Sample M2 17 3: 46.192 <.05 1.19.26 27 5: 31.020.003 0. 064.07 25 4: 04. 068.004 0. 91.15 26 5: 04. 021. 005 0. 058.069 37 3: 53.151. 08 1.28.12 38 3:45.14. 045 1.29.18 39 1: 00. 031.01 3.60.05 40 1: 30.218. 01 3.36.06 40 3: 00.198 nil 1.30.06 40 4:30.218 nil 0. 20 nil 41 4: 03.290.02 0. 84. 15 6 in. bath 20 psig. pressure 21 in. blowing height 0. 316 in. dia. nozzle Location 11 12 1: 22.32.45 4.22.47 5 2: 05.28.37 3.73.36 11 2:43.15.23 3.88.38 8 3: 25.20.20 3. 60.36 Location 13 12 1: 22.31.40 4. 16.52 5 2: 05.27.40 3. 93.39 11 2: 43.16.21 3.77.28 8 3: 25.11.25 3.70.34 Location 16 12 1:22.30.45 4.20. 50 5 2: 05.24.45 3.93.43 11 2:43.13.20 3.90.40 Location 31 12 1: 08.35. 50 4.22.50 11 2:28.18.19 3.94.44 13 4:05.20.27 3.65.37

117 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 33 12 1: 08.26.40 4. 06.46 5 1: 55.25.42 3.83.42 9 2 28.18.18 3.80.32 11 2:28.14 --- 3.90.40 13 4:05.11.14 3.46.28 36 5:25. 07 --- ---- Location 36 36 0:45.18 --- ---- 12 1: 08.32.50 4. 10.47 13 1: 08.21.38 3.67.37 5 1: 55 --- -- 3.87 9 2: 28.19.13 3.72.32 11 2: 28.11.10 3.89.40 13 4:05.06.11 3.45.27 36 5:25.07 --- ---- Location 51 36 1: 00.19 --- --- 5 1: 30 27.33 3.84.48 9 2: 13.26 --- 3.70.38 11 2:14.08.15 3.74.32 13 3: 50.12.07 3.48.33 Location 53 12 0: 55.28.40 4. 14.40 36 1: 00.18 --- 9 2: 13.08.08 3. 50.36 13 3: 50. 11. 11 3. 62. 30 Location 56 12 0: 55.28. 35 4. 16.47 36 1:00.23 --- o 5 1:30 --- --- 2.78 9 2: 13.09.05 3.36.24

118 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 63 36 1: 15. 06 --- -— 17 Location 66 36 1: 15.07 --- ----.17 36 5:25.05 --- 1. 98 Location 71 12 0: 42.24.25 3.88.42 5 1:15.09.30 3.80.46 11 2: 00.08.15 3. 52.32 Location 73 12 0:42.24.35 3.64.36 11 2:00.09.10 3.75.30 8 2: 40.10.06 3. 54.24 13 3:35.12.08 3.44.31 Location 76 12 0:42.26.25 3.72.40 11 2: 00.08.10 3.89.34 8 2:40.09.05 3. 52.26 Sample M1 5.279.85 4.24.60 8.300.65 ----.60 9.380.55 4.28.62 11.304 --- 4. 36.62 12.353 --- 4.28.64 13.405.62 3.96.61 15.424. 51 3.91. 54 36.235.55 3.99.48

119 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Sample M2 12 1: 36.297 --- 3.98.43 5 2 15.272.45 3.82.42 9 2: 40. 110. 10 3.82. 32 11 3 00 040. 10 3 85.32 8 3: 46.13.08 3. 56.30 13 4:15.028 <. 05 3. 62.22 36 5: 30.017.055 1.89.10 15 6: 15. 107 <. 05 2.94.21 10 in. bath 20 psig. pressure 7 in. blowing height 0. 316 in. dia. nozzle Location 21 31 0:32.24.52 2.95.40 31 2: 42. 18. 10 2 71.28 29 4: 00. 19.02 2.60.17 29 7:02.13. 01 1. 03 Location 23 31 0:32.16.11 3.10.26 31 2:42. 19.05 2.78.22 29 4:00.20.02 2 70.24 Location 26 31 0: 32.24.43 3.40.37 31 2:42. 19.05 2.97.21 29 4:00.18.02 2.62.25 29 7: 02. 14. 01 0. 24.22 Location 51 31 1: 10.25.20 2.81.22 31 2: 50. 19. 07 2.75.24 29 3: 45. 16.01 1.41.16

120 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 53 31 2:50.17 <. 05 2.82.20 29 3:45.12 <.05 ----.14 29 6: 15.12 <.05 0.29.23 Location 56 31 1:10.24.10 3.02.17 31 2:50.19.09 2.96.23 29 3:45.13.09 1.63 --- 29 6: 15.15 <.05 0.40 Location 81 31 0: 50.23.33 3.23.38 31 2:30.14.10 2.80.21 29 3:30.19.02 2.70.24 29 6: 00.12. 01 0.21 Location 83 31 0: 50.22 --- 3. 55 31 2:30.16 <.05 2.79.19 29 6: 00.13 <.05 0.20.20 Location 86 31 0: 50.23. 16 3.44.22 31 2:30.18 <.05 2.80.23 29 6:00.11.02 0.25 Location 101 31 0:15.16.02 2.74.23 31 1:30.21.02 2.98.20 29 3: 00.19.02 2.15.21 29 5:00.11.07 0.19.31 Location 103 31 0:15.17 <.05 2.76.23 31 1:30.19 <. 05 2.66.18 29 5: 00.11.05 0. 22.29

121 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 106 31 0: 15. 18 <.05 2. 59.20 31 1: 30.16 <.05 2.90.21 29 5: 00. 12 <.05 0.22.16 Sample M1 29.25.48 4.06.47 31.24.38 4. 03.47 Sample M2 31.15 <.01 2.78.14 10 in. bath 35 psig. pressure 7 in. blowing height 0. 316 in. dia. nozzle Location 21 32 0:45.27.12 3 55.25 Location 51 34 5:22.03.02 0. 04.12 Location 101 32 0:30.24.15 3.74.30 Location 103 32 0: 30.23.14 3.47.20 34 5:15.03 --- 0.03.12 Location 106 32 0: 30.27. 12 3.48.23 34 5: 15.02.01 0 08.06 Sample M1 32.26.46 3.86. 55 33.28.44 4, 10.55 34.25.60 3,81. 55

122 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Sample M2 33 2: 00.15.013 3.33.56 34 5: 30.047 <. 01 0. 02.043 6 in. bath 25 psig. pressure 7 in. blowing height 0. 162 in. dia. nozzle Location 11 21 1:45.21 --- 3.97.44 22 1:46.24.44 3.66.37 35 10: 00.15.23 2.62.18 Location 13 22 1:46.20. 37 3.99.36 35 10: 00.17.04 3. 18.39 Location 16 22 1: 46.23.42 3.36.40 35 10:00.16.06 3.18.22 35 13: 00.13 <. 05 2 95.18 Location 31 22 1: 30.23.28 4. 01.37 21 1:32.23 --- 4. 06.51 22 3: 00.19.32. —-- 39 35 8: 20. 15 <.05 2.71.20 35 12: 00.09 <. 05 2. 17.15 Location 33 22 1: 30.20.42 3.91.26 22 3: 00.20.32 3.21.37 35 8: 20.12 <.05 2.22.20 35 12: 00.10 <. 05 2. 13.18 Location 36 22 1:30. 19.28 3. 57.35 22 3: 00.19 --- 3.70 35 8:20.12 <.05 2.95.15 35 12: 00.11 <.05 2. 17.13

123 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 51 22 1:15.17.28 3.65.27 21 1: 17.21.08 3.56.23 21 2: 47. 19. 09 3. 37.13 Location 53 22 1: 15.20.46 3.44.28 21 1:17.20.15 3.39.19 21 2:47.18.09 3.33.13 35 9: 30.14 <.05 2.42.18 35 11:45.09 <.05 2.30.17 Location 56 22 1: 15.18.33 3.48.33 21 1: 17.20.08 3. 52.24 21 2: 47.22 <.05 3.27.23 35 9:30.12.10 2. 15.17 35 11:45.11 <. 05 2. 58.18 Location 61 21 1: 01.21.08 3.67.18 22 1: 01.18.26 3.76.26 21 2: 30.19.07 3.75.11 22 2: 30.20. 19 3. 66.24 35 9 00. 10 <. 05 2. 17.46 35 11: 30. 10 <. 05 2. 08.46 Location 63 22 1: 01.19.33 3.64.26 21 2: 30.19.10 3. 50. 19 35 9:00.12 <.05 2. 56 35 11:30.10. 01 1.98.17

124 Heat Blowing Metal Composition (weight percent) No Time P Si C Mn Location 66 21 1: 01.20.12 3. 27.14 22 1: 01.18. 22 3. 54.22 21 2: 30.17.08 3.75.13 22 2:30.18 --- 2.62.16 35 9: 00.10 <. 05 2.34.17 35 11:30.09.005 2. 28.15 Sample M1 22.253. 68 3.96. 54 21.230.68 4.21.46 35.272.48 4. 29 46 Sample M2 22 4:01. 193.17 3.64.24 21 5:42.182.05 3. 58.14 35 13:45.065.025 2.18.12 Special Heats R-38 Location 31 S 2: 15 21 --- ---- 17 0 2:30.20.004 2,97.17 S 2: 45, 18 --- 1. 53.24 S 3:15.16 --- 1.32.21 Location 33 2:45.14 --- 1.28.34 3: 15.18 --- 1.42.23 Location 36 2: 15. 15 --- ----.25 2:45.17 --- 1.53.24 Location 61 S 0:40.21 --- 3.46.10 0 0: 53. 13 --- 3.25.20 S 1:16.20 --- 1. 19. 12

125 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Location 63 0:53.16 --- 1.36.25 Location 66 0: 53. 16 --- 1. 31.22 2: 00.17 - 1.44.27 Sample M1.210.48 4. 03.50 Sample M2 3: 45.139.045 1.29.18 Special Heat R-39 Sample M1 028.10 3.96.04 Sample M2.031.01 3. 60.05 Special Heat R-40 Sample M1 0: 00.218.28 3.90.26 Sample M2 1:30.218.01 3.36.06 Sample M3 3: 00.198 nil 1. 30.06 Sample M4 4:30.218 nil 0.20.08

126 Heat Blowing Metal Composition (weight percent) No. Time P Si C Mn Special Heat R-41 Location 61 0:45.650 - 1:30.47 --- - 2:38.46 - Location 63 1:30.45 - 2: 38.43 --- Location 66 1:30.46 - 2:38.47 Sample Ml.760.50 3.70.48 Sample M2 4:03.290.02 0.84.15

TABLE XV Blowing Time - Slag Composition Data for all Blowing Conditions Heat Blowing Slag Composition (weight percent) No. Time SiO Tot. Fe FeO0 FeO AZ 0 CaO MgO MnO P20 Tt2 2e F3 32 23 8 in. sample ht. 6 inch bath, 7 inch blowing height, 20 psig. pressure, 0.316 inch. dia. nozzle 26 1:30 17.1 12.9 3.6 13.4 0.75 37.9 15.9 6.2 1.5 23 2:43 22.4 17.0 1.1 20.9 7.7 21.0 15.7 6.0 3.0 27 2:45 23.8 13.1 1.3 15.7 0.8 29.6 12.4 8.4 4.4 17 3: 30 20.1 9.4 3.1 9.3 3.8 33.2 15.9 5.5 5.5 26 3:45 21.1 19.3 4.0 21.2 1.1 27.3 12.8 6.9 3.9 27 4:40 11.5 17.4 5.5 17.5 0.9 36.2 14.2 3.9 3.7 11 in. sample ht. 26 2:17 8.8 15.6 6.0 14.7 6.0 42.5 14.7 2.5 1.9 26 3:15 20.7 23.8 4.2 26.8 1.1 23.5 11.1 7.1 3.4 27 4:00 17.5 22.1 7.4 21.5 0.9 27.3 11.1 6.3 -- 27 5:21 9.8 26.0 8.4 26.0 0.8 38.1 9.5 3.5 2.8 S-2 Samples 24 1:27* 14.2 7.9 4.5 6.1 4.6 49.2 12.2 3.0 4.2 17 3:46* 15.9 5.5 3.6 3.8 4.0 50.5 11.8 3.4 4.9 23 3:48* 13.7 7.4 4.2 5.8 6.0 48.0 14.5 2.8 2.7 20 3: 57* 13.8 5.3 3.9 3.3 3.8 52.4 14.2 2.7 3 4 25 4:10 15.84 9.1 4.6 7.5 2.8 48.8 11.2 3.3 3.3 26 5:15 13.8 18.3 7.3 17.0 1.1 43.5 10.5 2.8 2.4 27 5: 30 9.9 23. 5 8.9 22. 2 0.9 40.0 9.8 3. 1 2.3 18 5:32 9.2 12.6 6.3 10.5 4.9 50.8 9.8 2.4 2.8

Heat Blowing Slag Composition (weight percent) No. Time SiO2 Tot. Fe FeO03 FeO A1203 CaO MgO MnO P20 9 in. sample ht. 6 inch bath, 21 inch blowing height, 20 psig. pressure, 0. 316 inch dia. nozzle 36 1:30 10.0 22.5 7.91 21.9 1.4 33.5 16.5 2.9 1.0 36 3: 15 5.4 32.1 9.7 32.6 2.3 21.3 22.4 1.9 0.7 12 in. sample ht. 36 2:45> 6.1 36.8 7.9 40.2 2.7 19.3 17.3 2.2 0.5 36 5:00 5.6 35.1 20.2 27.0 2.2 22.4 16.7 1.4 0.6 S-2 samples oo 12 1: 36 8.6 17. 5 ---- ---- 1.4 58.9 3.8 2.8 1.6 12 1: 36* 7.3 24.1 10.3 21.7 2.2 49.2 3.8 2.1 2.1 5 2: 15 9.1 26.1 10.1 24.7 2.5 44.8 5.1 2.2 2.0 9 2: 40 10. 6 24.0 ---- ---- 3.8 37.4 8.4 3.3 3.4 11 3: 00 10. 7 26.2 ---- ---- 3. 5 36.3 5.9 2.8 3.0 8 3:46 16.5 13.4 ---- ---- 4.8 41.9 11.8 3.6 3.4 13 4:15 10.4 28.9 9.7 28.4 4.6 33.3 6.2 2.7 3.5 36 5: 30 9.2 32.1 10.9 31.4 3.9 28.8 9.2 2.8 2.9 19 5:32 7.2 37.2 8.9 39.9 3.2 28.6 6.4 2.0 2.6 13 in. sample ht. 10 inch bath, 7 inch blowing height, 20 psig. pressure, 0. 316 inch dia. nozzle 31 1:45 19.5 10.8 2.9 11.4 0.6 38.0 12.0 7.5 1.8 31 3:05 16.7 6.9 2.3 6.7 0.6 48.0 10.1 5.8 1.9 29 4:30 20.1 8.5 4.1 7.3 0.7 34.4 19.9 5.2 3.7 S-2 samples 31 3:26 14.9 5.5 2.5 4.8 0.8 49.8 16.9 3.8 1.5 29 8:32 12.9 4.0 1.8 4.9 0.9 53.6 19.0 2.1 1.8

Heat Blowing Slag Composition (weight percent) No. Time SiO Tot. Fe Fe 0 FeO Al 20 CaO MgO MnO P 0 13 in. sample ht, 10 inch bath, 7 inch blowing height, 35 psig. pressure, 0. 316 inch dia. nozzle 34 4:45 10.9 15.3 5.2 15.0 5.2 40.8 12.4 3.2 3,4 S-2 sample 34 5:30 13.5 18.5 7.4 17,.1 7.6 24.6 25.2 2.2 1.8 8 in. sample ht. 6 inch bath, 7 inch blowing height, 25 psig. pressure, 0.162 inch dia. nozzle 35 8:30 18.4 23,.5 2.4 28.1 5.31 18.0 12.9 9.5 1.3 35 12:15 10.5 27.0 7.0 28.4 2.3 27.9 13.2 4.7 0. 9 11 in. sample ht. 35 10:30 ---- 24.1 --- ---- --- --- ---- --- - S-2 sample 22 4:01 P 17.2 2.0 0.3 2.3 --- ---- ---- 0.8 0.2 21 5:42* 16.4 9.9 0.7 12.1 2.9 39.7 15.7 3.8 0.9 35 13:38 8.9 11.5 7.9 7.7 2.4 57. 1 8,.4 1.3 2.6 Special Heats S-2 38 3:45* 10.4 4,.5 2.2 3.9 5.6 60.2 10.7 2,.2 1.4 40 4:30 69.6 16.1 16.9 5.5 2.0 1.5 3.9 1.4 0.3 41 4:03 9.8 3.5 2.3 2.5 3.3 63,.3 10.8 1,8 4,1 * Sample contained lumps of undissolved lime and was considered inhomogeneous; therefore it was not included in the review of slag data,

Appendix II, Calculations a. Calculation of Average Batch %J b. Mass Balance and Oxygen Utilization c. Equilibrium Calculations d. Mass Transfer Models 130

131 a. Determination of Average Bath %J The average bath content of J is determined from the analysis of final metal samples whenever such data are available. During the early portion of an experiment no data are available so average bath %J is determined as shown by the example below. J = P, t 1: 00 blowing time Conditions: 6 in. bath, 7 in. blowing height, 20 psig. pressure, 0. 316 in. lance Constant composition curves from Figure 17 are redrawn as on Figure 51. The axes are height and distance from the center of the converter squared. The product of the axes is proportional to bath volume. The sum of graph areas between two constant composition lines multiplied by the intermediate %P is proportional to the weight of phosphorous in the bath. The sum of the products is divided by the total graph area yielding average %P. IN. FOM BOTTOM,14 5 ----.,203 2 1 2 3 4 5 6 IN. FROM WALL SQ. IN. FROM i SPACING Fig. 51. Volume Plot Used in Calculating Average Bath % P.

132 b. Mass Balance and Oxygen Utilization Mass balance and oxygen utilization calculations can be made when sufficient data are available on the slag and metal compositions variation with time. The mass balance and oxygen utilization equations are derived below. Mass Balance The slag weight is determined assuming no loss of CaO. In other words, while slag and metal are thrown out of the converter during a run CaO was presumed to be thrown out in amounts proportional to its fraction of the total slag. Slag wt. = lb. CaO in/fraction CaO in the slag Metal wt. at time t - metal wt. charged - wt. by time t of elements lost to the slag including Fe in the slag. wt. of elements in the metal - % element J x metal wt. /100. wt. of elements in the slag - %Ja x slag wt. /(100 x mol. wt. J a( /a x mol. wt. J). Total wt. of all J = J in metal + J in slag. Accountable J = (Total wt. J/initial total wt. ) x 100%. Whenever data were available on the compositions of slag and metal during an experiment mass balance calculations were made at incremental times; otherwise, the calculations were made only at the end of an experiment on the basis of beginning and final metal composition and final slag composition.

133 Oxygen Utilization The oxygen used to oxidize each metal impurity was calculated assuming the formation of specific oxides. Here SiO2, MnO, P20, CO, FeO and Fe 03 were the oxides initially considered. wt. of J in the metal = wt. of metal x %J/100. wt. of J oxidized in time t = initial wt. of J - wt. of J at time t. Oxygen to oxidize J = wt. of J oxidized x b x mol. wt. O/ a x mol. wt. Total oxygen used = Z oxygen to J. J Oxygen efficiency = (total oxygen used/total oxygen in) x 100%. In cases where data were sufficient, differential efficiencies were calculated by subtracting the weight of J in the metal at time t - At from the weight of J in the metal at time t. Oxygen efficiencies were calculated for the time increment At. As a matter of interest whenever the oxygen efficiency was less than 70% an additional calculation was made assuming 100% CO2 2

134 c. Equilibrium Calculations Equilibrium phosphorus concentrations have been calculated from metal and slag composition data and thermodynamic constants. Slag temperatures were assumed to lag behind metal temperature. After 5: 30 blowing time at standard conditions, 2910~F. was assumed to be the slag temperature. The following calculations of equilibrium %P were made using Winkler's (38) thermodynamic data. TABLE XVI Calculation of Mole Fraction J b in the Slag t = 5: 30, T 2910~F. = 1600~C., Sample: 27S-2 Compound SiO2 A203 FezO P 0 MgO MnO CaO FeO molwt. 60 102 160 122 40.3 71 56 72 wt. % 9.9 0.9 8.9 2.3 9.8 3.1 40.0 22.2 mol/lOOlb..165.009.056.019.243.044.715.308 mol frac..10.01.03.01.16.03.46. 20 B = CaO + MgO + MnO = 0.69 A SiO2 + 2P205 + 1/2 Fe 3 + 1/2 A1203 = 0.14. B/A = 4. 9. FeO = 20 mol.%. (P 05) / (%P2) 600. Figure 11 Winkler and Chipman (38) (P205) = 0. 01. 1/2 %P = (0. 01/600) o0. 004%.

135 d. Mass Transfer Models Four models have been developed for dephosphorization under standard blowing conditions. The four models are 1) Bulk Diffusion, 2a) Boundary Layer Transfer from a Turbulant Bath, 2b) Boundary Layer Transfer from a Circilating Bath, 3) Bath Dilution from Metal Droplets Falling through a Dephosphorizing Slag, and 4) Combination of Circulation and Droplet Models. Each of the models is derived separately below. 1) Bulk Diffusion This model is based on a stagnant bath being dephosphorized by diffusion of phosphorus to an infinite surface sink at 0%P. This situation is best described by Fick's Second Law. P = %P, t = time in seconds h r distance from the bath surface in cm D = 1.5 x 10-4 cm2/sec. RHO = density of liquid steel = 7 gm/cc dP d2p (RHO/100) d (RHO/100) D d (1) dh2 B. C. P = 0. 22 at t= 0 and 0 < h < oo a semi infinite bath is assumed SoP = 0 at h = 0 and 0 < t < oo. With these boundary conditions the integrated form of (1) is: (0. 22 - %P) / (0. 22 -0) = 1 - erf( (h)/(2 Vt) ) (2) %P fbr various h and t is calculated with (2). %P is calculated to be 0. 19% after 10, 000 seconds, roughly 3 hrs., at a distance of

136 2. 5 cm., 1 inch, from the bath surface. The dephosphorization rate is 0. 01%/hr.. Clearly, this model is not appropriate. 2a) Boundary Layer Transfer From a Turbulent Bath Wagner (47) has described this general model in detail. In this case phosphorus loss is given by m = (D/6 )A(P- P* ) (3) 6 = diffusion distance =.003 cm. P* = equilibrium phosphorus content (gm/cc. ) P = phosphorus content at time t (gm/cc. ) Equation (3) has been modified for use with %P, X. m = (D/ 6 ) A (RHO/100) (X - X* ) (4) X at t - [ (m x 100%)/(bath wt.) ] At = X at t + At (5) Equation (5) has been used in calculations of X as a function of blowing time, 2b) Boundary Layer Transfer From a Circulating Bath A schematic drawing of this condition is shown in Figure 52. M (r) = flow velocity of the circulating layer at r (cm/sec). I = thickness of the circulating layer (cm.) C1 = M 2ir r = constant (c. c. /sec. ) (6) ( D/ 6 ) (RHO/100) (XB - X*) = phosphorous flux gm. /cm2/ sec. XB (r) = %P in circulating layer at r.

137 I / SLAG M M pTO M IRCULATING M xI SLAG x LAYER! /r Ir+Ar SURFACE ELEMENT TURBULENT BATH Fig. 52. Schematic Drawing of Mass Transfer Conditions of Model 2b.

138 -(RHO/100)(C1XB - CX ) = (D/6)(2TrrAr)(RHO/100)(XB - X*) r r+Ar (7) (dXB)/(dr) = -(D/6) (21/C1) (XB - X*) r (8) (dXB)/(XB -X*) = - (D/6) (Tr/C1) (XB - X) rdr (9) In(XB - X) = -(D/6) (T/C) r2 + K (10) XB X* = K exp[ - (D/6) (Tr/C) ] (11) at r = 2. 54 XB = X (Bulk %P) (12) 2 XB = (X - X*) exp [(D/6) C (i/C) (r -6.452)] + X* (13) XB was calculated at r's corresponding to the 3 sampling locations at the bath surface with equation 13. The change in the bulk phosphorous content is determined by assuming that there is no interaction between the bulk of the bath and the surface layer except when new metal at bulk concentration flows in at r = 2. 54 and when refined metal flows out at the wall. V = 16.51 - Q VX - VX = C1 (X - XB) (14) t t+At XB = surface %P at r = wall - [ (dX)/(X-XB)] = (C1/V) dt (15) (X - XB) = A exp [(- C/V) At] (16) at t = OX = X0 = %P (17) X = (X - XB) exp (-t C/ V) + XB (18)

139 The bulk phosphorus concentration can be calculated for any blowing time with (18). 3) Bath Dilution From Metal Droplets Falling Through a Slag This model is analogous to liquid from a tank being run into a refining unit and then back into the tank. Let SPRAY = volume of droplets thrown up per second. VX - VX = SPRAY (X - XE) (19) t t+At XE = %P in droplets leaving the slag XE > X* (20) X = (X0 - XE) exp ( - SPRAY t/V) + XE (21) 3') This model was modified such that SPRAY is proportional to the rate of decrease of the bath %C, e. g., SPRAY a d%C/dt. 4) Combination of Circulation and Droplet Models The two models, 2b and 3, have been described separately. The combination is derived below. Mass balance on a surface element C - C + 2TrrArL - 2rrrL = 0 (22) r r+Ar surface layer volume is constant C1 is constant L(r) = distribution of droplets over the surface L(r) = C3/r3, C3 = constant (23) Phosphorous balance around surface layer element

140 RHO/00) )(ClXB) - CXB I + 2wrArLX - 2wrArLXB) = r r+Ar (D/ 6)(2TrrAir)(RHO/100) (X - X*) (24) -C1 [(dXB)/(dr)] = 2rr [(D/6)(XB-X*) + L(XB-X )] (25) If X, = XE (26) -C1 (dXB)/(dr) = 2rr4 [(D/6 ) + L] (XB - XE) (27) (dXB)/(XB-XE) = [(-2i)/(Cl)] [(D/6)/L] rdr (28) L C' Cr3 (23) (dXB)/(XBXE) =((-2rr)/Ci) [(Dr/6 ) + (C/r2)] dr (29) in (XB - XE) = (-2T)/(C) [(Dr2/26) - (C3/r)] + K (30) XB - XE = Kexp [(-2r/C1){(Dr2/26) - (C3/r)} ] (31 at r = 2.54 XB = X (bath %P) (32 2 XB = (X-XE) exp (-2rr/C1) (D/26 ) (r - 6. 462).... -C3 [(1/r) - 0.3937] + XE (33, XB (r) was calculated for r corresponding to the samplirg locations at the bath surface. The time dependence %P in the bath is derived below; Phosphorus balance RHO/100 has been cancelled out. VX - Vx: IX I + C1XB t+At t r=2. 54 r=wall -X L2'nrdr + LXB2rrrdr (34)

141 -C1X = -C1X (35) r=2. 54 CXg|l = C[(X-X) exp( —-) +XE] (36) r=wall r=wall C= X exp ( —-) - CXE exp ( —). + C1XE(37 X L2rrdr = 2rXC3 [l/(r2)] dr (38) 2. 54 rrXC3 (0.3937 - 1/r2) (39) LXB 2rrdr = 2-TC3S (1/r2) [(X-XE) exp ( —-) + X] dr (40) = 2ZrC3XS (1/r2) [exp ( —-)] dr - 2 C3XE (1/r2). [(exp ( —-)] dr + 2rC3X. (1/r2) dr (41) Combining equations 34, 35, 37, 39, and 41: vx| - VX | = -CqX + CX exp ( —-)- C1 x ( —-)+... t+At t CXE - 2rCX [0. 3937 (1/r2)] + 2TrC3X (1/r2) exp ( —-) dr - 2rTC3XE(1/r2) exp ( —-) dr. + 2C3XE [(0.3937) - (1/r )] (42) Groupihg terms in 42: B = -Q + C1 exp ( —-) - 2rC3[ 0. 3937 - (1/r )] +.. 2rC3S (1/r2) exp ( —-) dr (43)

142 VdX/dt = BX - BXE (44) (dX) / (X - XE) = (V/B)dt (45) In(X- X) = (V/B) t + K (46) E X- XE = K exp [(V/B) t] (47) at t =, X = X (48) X = (XO - XE) exp[(V/B) t] + X (49) Evaluation of (1/r) exp ( —-) dr was done graphically for r = 2. 54 to r = 19. 05.

Bibliography 1, Ao L. Hodgeo Iron and Steel Engineer. 1966, V. 44, pp. 97-100. 2, P. Wo A. Lanzing and Ro Smito Journal of Metals, 1962, Vo 14, ppo 510-514. 3o To Ikida, op. cito ppo 518-520. 4o Go Grosvenoro Journal of Metals. 1961, Vo 13, pp. 621624, 5o J. N. Albaugho opo cito po 631. 60 Panel Discussiono Open Hearth Proc. 1965, V. 48, ppo 120-135. 70 To J. Murray Fo Eo Williams and E. Co Vice. Journal of Metalso 1964, V. 16, pp. 644-650, 80 Jo Ao Glasgow. Journal of Metals, 1961, V. 13, pp. 625627o 9. R. Bo Morganii op, cito ppo 628-630o 10o G. Massobrio and Fo Santinio Open Hearth Proc. 1965, Vo 48, pp. 1154119. 11, Ro Ao Flinn, Ro Do Pehlke, Ro Do Glass and Po O0 Hayes, Ao Io Mo Eo Trans. to be publishedo 12, Jo Ao Glasgowo Journal of Metals, 1962, Vo 14, pp. 514= 516o 13. Co Go Campbello op. cito pp. 516-517. 14, Ko Fo Behrens and JO Koenitzero Open Hearth Proc. 1964, Vo 47, ppo 54-72. 15, Fo Petrillio opo cito ppo 73-77. 143

144 16. 0. Cuscoleca, K. Rosner and W. Kuhnelt. O. A. M. Bull. Jan. 1952. 17. W. 0. Philbrook. Journal of Metals. 1958, V. 10, pp. 477482. 18. T. P. Calclough. J.I.S.I. 1959, 192, pp. 201-215. 19. H. Trenkler. Revue Universelle des Mines. V. 96, 1953, pp. 644-657. Brutcher Trans. no. 3238. 20. 0. Cuscoleca. Stahl und Eisen. Vol. 72, no. 17, Aug. 14, 1952, pp. 989-992. 21. H. Trenkler. op. cit. pp. 992-997. 22. K. Rosner. op. cit., pp. 997-1004. 23. W. Kuehnelt. op. cit., pp. 1004-1010. 24. G. Tromel. Stahl und Eisen. Vol. 73, no. 1, 1953, pp. 19-22. Brutcher Trans no. 3227. 25. H. Trenkler. Inbetriebnahme Breitbandstrasse Blasstahlwerk Jan. 5, 1953, pp. 17-18, 20-22. 26. H. Hauttman. op. cit., pp. 23-30. Private Translation 27. T. E. Suess "Method of Carrying Out Melting Processes" U.S. Pat. 2,800, 631, July 23, 1957. 28. H. Rellermeyer, H. Knueppel and J. Sittard. Stahl und Eisen. 1957, V. 77, pp. 1296-1303. 29. C. Holden and A. Hogg. J. I. S. I. 1960, V. 196, pp. 318332. 30. H. Krainer, K. Borowski and J. Maatsch. Krupp Tech. Rev. 1965, V. 23, pp. 53-66. 31. T. Kootz. J.I.S.I. 1960, V. 196, pp. 253-258. 32. W. O. Philbrook. Journal of Metals. 1961, V. 13, pp. 613620.

i45 33. G. C. Smith and D. A. Dukelow. Journal of Metals. 1964, V. 16, pp. 357-361, 34. T. E. Suess. Iron and Steel Engineer. 3-1952, V. 29, pp. 91-93. 35. J. Pearson. Iron and Coal Trades Review. 1960, V. 181, pp. 1407-1413. 36. B. S. Holmes and M. W. Thring. J. I.S.I. 1960, V. 196, pp. 259-261. 37. G. Perbixo Journal of Metals. 1966, V. 18, pp. 824-831. 38. T. B. Winkler and J. Chipman. Trans Met. Soc. AIME. 1946, V. 167, pp. 111-133. 39. L. von Bogandy, W. Dick and I. N. Stranski. Archri fur das Eisenhuettenwesen. 1958, V. 29, pp. 329-337. 40. T. Kootz and H. Neuhaus. Stahl und Eisen. 1961, V. 81, pp. 1810-1815. 41. T. Kootz, Ko Behrens, H. Maas and P. Boumgarten. Stahl und Eisen. 1965, V. 85, pp. 902-907. 42. To Kato, J. Imai, and Ko Fujiwara. Tetsu-to-Hagane. 1962, V. 48, pp. 465-467. Brutcher Trans. no. 5675. 43. Physical Chem. of Steelmaking Comm. A. I. M. E. Basic Open Hearth Steelmaking. Chilton Company, New York, N.Y., 1964. 44. T. F. Pearson, A. S. Venkatadri and J. O'Hanlon. J. I. S. I. 1966, V. 205, pp. 997-1006. 45. L. A. Baker, N. A. Warner and A. E. Jenkins. Trans. Met. Soc. AIME. 1967, V. pp. 857-863. 46. C. Wagner. The Physical Chemistry of Steelmaking. The M.I. To Press, Cambridge, Mass., 1958.

*'to 00 "5~~C 1^CI *|ssj~ Isi^~~;~;BS~ ^R<~e~7 SS"