TN295 1 1 1 > No. 9195 i L-f ' ■ M ^ 'T* .< !i.iitB!i«'iiil.T' :• aMi 5 [jrCjji'i i"KUM / ' k 't * J, :^.T* .^'^ % ° J^ -^ J.'J^" ..-". ^^ 'v-^ V »1* • il^' 0^ 'o, ♦.»7;T» A /\ l^*' ^^'\ '^fW.^ /\ -j^^. *b *.t;t*' A 'o. *.T.T»' A 5v: '^^ i^ .^^•'. '-.4 X 0' 0^ **!•' .^,0-..-^. /\.:::^%\, /.^i^..% .•^^.c:;^^^ -^^ »j»^ ^^% iR*. o_ * ^ ov^^i!^'- ^^^^^■' ;^^^'. '^^0^ /''^Sr. -^af ■'m^r.\ ^^^^ <9- * o « ' ^ 0^ ^ *?^T* A 5°,<. c» * vv * o^-'^ c* *: *^^ ^^<^ e bo Bureau of Mines Information Circular/1988 New Steelmaking Technology From the Bureau of Mines Proceedings of an Open Industry Briefing Held in Association With the Electric Furnace Conference, December 8, 1987, Chicago, IL Compiled by Staff, Bureau of IVIines i!(0^^^ UNITED STATES DEPARTMENT OF THE INTERIOR maJM^JjA^, 8//A^f ~~ — ; Information Circular 9195 w New Steelmaking Technology From the Bureau of Mines Proceedings of an Open Industry Briefing Held in Association With the Electric Furnace Conference, December 8, 1987, Chicago, IL Compiled by Staff, Bureau of Mines UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Model, Secretary BUREAU OF MINES T S Ary, Director -r ui)' c\\ ^5 Library of Congress Cataloging-in-Publication Data New steelmaking technology from the Bureau of Mines. (Bureau of Mines information circular ; 9195) Bibliographies. Supt. of Docs, no.: I 28.27: 9195. 1. Steel— Metallurgy— Congresses. I. Electric Furnace Conference (1987 : Chicago, IL). II. United States, Bureau of Mines. III. Series: Information circular (United States. Bureau of Mines) ; 9195. TN295.U4 [TN730] 622 s [669 '.1424] 88-600072 PREFACE On December 8, 1987, the Bureau of Mines held an open industry briefing in association with the Electric Furnace Conference sponsored by AIME's Iron and Steel Society. The papers presented at that briefing are contained in this Information Circular, which serves as a proceedings of the meeting. The papers highlight the Bureau's most recent research aimed at improving steelmaking technology. Areas addressed by this research include arc stability in electric steelmaking furnaces, stainless steel pickling processes, substitutes in steelmaking, steelmaking refractories, and recy- cling of steelmaking dusts and wastes. The open industry briefing used as a forum for the transfer of this research is one of the many mechanisms used by the Bureau of Mines in its efforts to move research developments, technol- ogy, and information resulting from its programs into industrial practice and use. To learn more about the Bureau's technology transfer program and how it can be useful to you, please write or telephone: Bureau of Mines Office of Technology Transfer 2401 E Street, NW. Washington, DC 20241 Telephone: 202-634-1224 CONTENTS Preface i Abstract 1 Introduction 1 Improved Arc Stability in Electric Arc Furnace Steelmaking by Thomas L. Ochs and Alan D. Hartman 2 Preheating of Ferrous Scrap by R. H. Nafziger and G. W. Elger 12 Fluorspar Substitutes in Steelmaking by R. H. Nafziger and G. W. Elger 23 Research on Basic Steelmaking Refractories by T. A. Clancy and J. P. Bennett 28 Basic Research on Corrosion of Iron-Based Materials by David R. Flinn 33 Fundamentals of Stainless Steel Acid Pickling Processes by Bernard S.Covino, Jr 39 Decreased Acid Consumption in Stainless Steel Pickling Through Acid Recovery by G. L. Horter and J. B. Stephenson. ... 45 Recycling of Stainless Steelmaking Dusts and Other Wastes by L. A. Neumeier and M. J. Adam 50 Using Wastes as a Source of Zinc for Electrogalvanizing by V. R. Miller 58 Economic Evaluation of a Technique To Pelletize Flue Dust and Other Waste From the Manufacture of Stainless Steel by Joan H . Schwier 67 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT A ampere lb pound A/dm2 ampere per square decimeter lb/ft3 pound per cubic foot A/min angstrom per minute Ib/min pound per minute atm atmosphere, standard Ib/st pound per short ton at- pet atomic percent Mgal thousand gallons Btu/(lbmol) British thermal unit per pound per mg/(mincm^) milligrams per minute per square mole centimeter Btu/st British thermal unit per short ton min minute Btu/yr Brithsh thermal unit per year mL milliliter cm centimeter mm millimeter cm^ square centimeter MMBm million British thermal units °C degree Celsius MQ-cm megohm centimeter °F degree Fahrenheit mol/L mole per liter dB decibel ms millisecond dm2 square decimeter m/s meter per second ft foot mt metric ton ft/s foot per second mV millivolt g gram nAlcrn^ microampere per square centimeter gal gallon /iP/cm^ microfarad per square centimeter g/cm^ gram per cubic centimeter nm micrometer g/L gram per liter lis microsecond g/m^ gram per square meter nm nanometer g-mol/dm^-h gram mole per square decimeter per Qcm^ ohm centimeter squared hour P poise gr/dscf grain per dry standard cubic foot pet percent h hour ppm part per million Hz hertz ppt part per thousand in inch psi pound per square inch K kelvin s second kcal/mol kilocalorie per mole scfm standard cubic foot per minute keV thousand electron volts st short ton kHz kilohertz st/d short ton per day kW kilowatt st/yr short ton per year kWh kilowatt hour V volt kW-h/lb kilowatt hour per pound V/h volt per hour kW-h/st kilowatt hour per short ton vol pet volume percent kW-h/yr kilowatt hour per year wt pet weight percent kVA kilovolt ampere yr year L liter NEW STEELMAKING TECHNOLOGY FROM THE BUREAU OF MINES Proceedings of an Open Industry Briefing Held in Association With the Electric Furnace Conference, December 8,1987, Chicago, IL Compiled by Staff, Bureau of Mines ABSTRACT This report is a proceedings of a briefing recently sponsored by the Bureau of Mines at which Bureau personnel presented findings from their research efforts to improve steelmaking techology currently used in the United States. The papers contained in this report address many areas of con- cern to the iron and steelmaking industry. Among these are improving arc stability in electric arc furnaces, preheating ferrous scrap to reduce energy consumption, fluorspar substitutes in steel- making, basic steelmaking refractories, corrosion of iron-based materials, improvements in stain- less steel acid pickling processes, recycling of stainless steel dusts and other wastes, and use of wastes as a source of zinc for electro/galvanizing. INTRODUCTION For over 50 years the Bureau of Mines has worked to improve technology used by a major constituent of the U.S. minerals industry — the iron and steelmakers. The most recent promising results of this research are presented in this report. Some of the research described has been completed; other research studies high- lighted are in progress. However, this report focuses on many sig- nificant findings with a probable high positive impact on the indus- try. For instance, the Bureau is studying the fundamental behavior of arcs in electric furnaces in order to improve the efficiency of this steel manufacturing technology. Past research has shown that electrical disturbances caused by the unpredictable arc are respon- sible for power surges and fluctuations and for noise levels in excess of 120 dB. Understanding fundamental arc behavior may enable researchers to control the arc and therefore optimize electric fur- nace operation and efficiency. Already these studies have yielded positive results in the identification of potential areas for change in furnace designs and operating procedures to greatly increase effi- ciency. In addition to this research, the Bureau has also evaluated the feasibility of preheating ferrous scrap charges in electric furnace operations to decrease energy consumption by using furnace off- gases. During a laboratory test, furnace offgases from a 1-st-capacity electric arc furnace were used to preheat continuously charged automotive scrap and metal stampings up to 1,110° F. Results showed that about 7 pet less electrical energy was used in charging preheated scrap than was used when cold scrap was charged. Other research conducted by the Bureau may help reduce costs associated with basic oxygen and electric furnace operations by providing a less expensive substitute for fluorspar fluidizers. Sub- stitutes evaluated as alternatives in basic oxygen flimace operations include colemanite, fused boric acid, synthetic fluorspar, and used aluminum smelter potlining. Alternatives for electric arc furnace operations were synthetic fluorspar, boric acid, hydroboracite, used aluminum poflining and anode tailing wastes, and Soreflux B (ilmenite). The substitute fluidizers did not adversely affect the steel produced in test operations. Along with optimizing the efficiency of fiimace operations, the Bureau is searching for ways to reduce the loss of strategic and critical metals during various phases of steelmaking and to reduce waste generation. This can be accomplished by recycling pickling solutions, wastes, and dusts. The Bureau has experimented with ion-selective membrane technology in developing a means to recycle acid solutions used during the pickling of stainless steels. Disposal of these acid solutions is costly to the steel manufacturing industry and results in the loss of valuable chromium and nickel. Prelimi- nary research has revealed that an electrodialysis cell using ion- selective membranes does have the potential for separating dissolved metals from spent pickling acid solutions while regenerating the acids for return to the pickling process. Through other research studies, the Bureau has developed a process permitting in-plant recovery of about 90 pet of the chromium, molybdenum, nickel, and iron from stainless steelmaking dusts and wastes. By using yet another Bureau-developed process, zinc extracted from electric arc furnace dust can be used for electrogalvanizing. Detailed accounts of the laboratory tests and results for each of these research studies are presented in this report. The report also provides a description of other studies conducted including an economic analysis of the technique to pelletize flue dusts and other waste resulting from steel manufacture in order to recover contained metals. IMPROVED ARC STABILITY IN ELECTRIC ARC FURNACE STEELMAKING By Thomas L. Ochs^ and Alan D. Hartman^ ABSTRACT In order to improve the performance of electric arc furnaces used to manufacture steel, the Bureau of Mines is studying the fundamental behavior of the electric arc in the electric arc furnace. Improvements in control and processes will allow more efficient and quieter operation of the elec- tric furnaces. Presently, electrical disturbances caused by the unpredictable arc are responsible for flicker and surges on the power grid and sound levels in excess of 120 dB in the vicinity of the furnace. It is these disturbances that the Bureau is investigating. In the Bureau experiments, electrical signals are sampled at 50,000 Hz and photographs of the arc are taken at up to 40,000 images per second using high-speed cinematography. These images are then correlated with the electrical signals to study the physical events in the arc plasma. Arcs studied to date indicate that the electrical waveforms have unique signatures preceding some volt- age excursions. Initial Bureau investigations indicate that mathematical techniques of analysis in the field of nonlinear dynamics have characteristics that enable these methods to speed up process- ing of the electrical signals from the arc for use as control parameters. The behavior of the arc in the experimental environment has led to the conclusion that there are new fiirnace design changes and operating procedures possible. The potential areas for change include the furnace shell geometry, continuous feeding methods, electromagnetic pumping of mol- ten metal, electrical control, furnace atmosphere, waste heat recovery, and electrode design. These changes could greatly increase efficiency, which is typically in the 60-pct range, improve furnace operation, and reduce noise. Interactions of these changes are complicated and must be considered together. INTRODUCTION The share of steel produced by electric arc furnaces has increased over the past 20 yr because of the flexibility of the minimill concept and the consequent reduction in costs to produce steel in the environment of rapidly fluctuating demand. There have been many changes in the arc furnace over the past 30 yr of use, includ- ing ladle metallurgy, ultrahigh power operation, water jacket cool- ing, oxyfuel burners, and scrap preheating. Control of the furnaces, however, has remained a initiative process based on the prior experience of the furnace manufacturer and the operator. This is because the high-current arcs used in electric arc furnace steelmaking are violent high-temperature conducting plasmas and have proven very difficult to understand over the past 100 yr of study. Because of this limited knowledge, the arcs are difficult to control (/, pp. 15-16, 2). 3 Currents of 100,000 A are common in commercial arc furnaces. These currents can produce temperatures in the core of the arc of 12,000 to 15,000 K, which is more than twice as hot as the surface of the Sun. ' Mechanical engineer. ' Chemical engineer. Albany Research Center, Bureau of Mines, Albany, OR, ' Italic numbers in parentheses refer to items in the list of references at the end of this paper. Although these arcs have been used in smelting and melting metals since the turn of the century, the fundamental processes taking place inside the arc and at the points where the arcs attach to the electrodes and the charge are poorly understood. Disrup- tions of these high-current arcs during operation can produce fluc- tuations on the power grid (3), ablation of the furnace refractory, and poor heat transfer to the melt. At the present time, control of the furnace is based on the experience of the operators and the control manufacturers, not on any fundamental understanding of the processes taking place in the arc. Presently, control systems react to reverse a past event (2) as opposed to acting to prevent a future event. Prevention of future events is possible only if the future situa- tion is predictable, based upon the past events. However, arc behavior presently is unpredictable. New methods of examining the signals available from the voltage and current waveforms in the transformer secondary circuitry, and therefore in the arc itself, may be able to supply unique signatures useful over the span of a wavelength for indicating disruptive events. If the arc and its inter- actions with the furnace interior were better understood, then there could be improved control methods or modified equipment designs that would result in gains in efficiency and a reduction of disrup- tive electrical and acoustical noise. The electric arcs studied in the Bureau investigations have shown deterministic behavior that is sensitive to the conditions of operation. Extreme sensitivity to operating conditions leads to a lack of predictability of the arc behavior since the furnace operat- ing conditions cannot be measured exactly. This unpredictable arc behavior has been characterized as stochastic, or random, in prior studies, but instead is indicative of chaotic behavior resulting from nonlinear interactions. Viewing the arc as a chaotic, deterministic system of discrete events, it is possible to look at electrical wave- forms and expect short-term precursors (half-cycle) to the seem- ingly random events. Short-term precursors indicate the possibility of anticipation and control of arc disruption. It is disruption of the arc while it is carrying a high current that causes both electrical and acoustical noise. From these investigations, it is becoming clear that traditional transform methods of analysis and statistical analysis of the elec- trical waveforms are of limited use. These methods are normally used on smooth waveforms or waveforms with a small number of periodic discrete events. The waveforms that were obtained in this research are composed of many nonperiodic discrete events on dis- torted square and sine waves. These discrete events must be treated using discrete digital methods. STOCHASTIC VERSUS CHAOTIC BEHAVIOR The electric arc has been described as a random, or stochas- tic, system, with events taking place at unpredictable intervals. The present investigation shows that the arc is not a stochastic system, but rather a nonlinear system that is very sensitive to initial condi- tions. This sensitivity gives the appearance of random behavior since the internal conditions cannot be measured closely enough to pre- dict the next operational state of the arc (4-5). Behavior of these nonlinear chaotic systems is not possible to predict for any length of time from a mathematical solution based on measured system conditions. Instead, the system can be treated using the methods of nonlinear dynamics. Using these techniques, there is the possi- bility of short-term (one ac cycle) prediction of future events based upon the inferred conditions as deduced from the signature analy- sis of the real-time waveforms. The basic premise of this method of predictive control is that the system is mathematically well behaved (continuous and single valued), and over a short term (one cycle) the arc behavior can be predicted. Over the long term (longer than a cycle), the minor var- iations in operating conditions that cannot be measured will pro- duce unpredictable behavior even though the system is deterministic. This means that the control system must take real-time data of the arc waveforms, compare it against a library of waveform signa- tures, and make decisions in a short time frame, typically about a quarter-cycle (4 ms). Recent increases in computational capabil- ities and decreases in cost have made data analysis and system con- trol of the type described feasible. EQUIPMENT An experimental 200-lb-capacity single-phase electric arc fur- nace was used for conducting the experiments. The furnace used two 3-in-diam graphite electrodes. The power was supplied by two single-phase ac welders connected in parallel. Each welder was rated at 1,500-A current and had a rated full-load voltage of 40 V. The primary rating was 440 V and 170 A single-phase. The furnace was modified in order to simplify the data analy- sis. Three types of arc targets were used in the furnace. The arc target block materials were graphite, steel, and copper. These blocks were used to study different system configurations and obtain calori- metric data. The calorimetric data are for assessment of heat transfer rates and efficiency. Initially, electrical signals in the single-phase furnace were taken across both arcs, one from each of the elec- trodes. In this configuration, the two arcs consisted of one with the electrode acting as the cathode and one with the electrode act- ing as the anode. This averaged the events attributable to each arc and made data analysis difficult. Therefore, a simplifying modifi- cation was made to the experimental system. This consisted of threading one of the electrodes into the conductive target block (fig. 1). Threading the electrode into the block eliminated one of the arcs and its associated signals, while it maintained the current path that normally would be taken by the current flowing through the arc. This made the magnetic field in the ftimace similar to that pres- ent in the two-arc system. The second modification to the experimental furnace involved replacing the furnace shell with an airtight enclosure. By using this enclosure, the atmosphere within the ftimace could be controlled (fig. 2). This allowed experimentation to be conducted with gases other than air, as well as gas injection through the electrode. A third modification was the addition of two viewports at 90 ° to each other (fig. 3). Thus two perpendicular images of the arc could be captured simultaneously by a high-speed motion picture camera. The two images of the arc were directed into the lens of a high-speed camera by the use of mirrors. The camera used 450-ft rolls of 16-mm film and operated at up to 1 1,000 frames per sec- ond. The actual image capture can take place at up to 44,000 image pairs per second, which was accomplished by using the camera's internal prism to divide a frame into quarters (fig. 4). After 200 ft of film had been exposed, allowing the camera to reach maxi- mum speed, a waveform analyzer was triggered that recorded the simultaneous voltage and current electrical signals corresponding with the film images. Synchronization between the waveforms and the film is achieved by the use of timing pulses on the film. The waveforms were digitized at 50 kHz per channel. r'^ »', Figure 1.— Electrode threaded into target block to simplify target path. Figure 2.— Experimental furnace shell showing gas Inlets. Figure 3.— Orthogonal views by use of mirror system. Rotating prism assembly Aperture mask -Objective lens Film frame at film gate aperture - Viewf inder eyepiece Focusing prism First field lens and prism assembly Figure 4.— View of single frame showing eight images and their relationship to original image. EXPERIMENTAL PROCEDURE AND RESULTS Areas investigated by the Bureau include (1) electrode tip design, (2) inert gas injection, and (3) arc analysis. Electrode tip design was investigated since the structure of the electrode provid- ing the arc has been related to the unsteadiness of arcs (6). ELECTRODE TIP DESIGN Electrodes providing a concave tip were found to reduce arc flare and maintain the arc under the electrode tip (7). In reference 7, the author described attempts to reproduce this effect by testing hollow versus solid electrodes. Although the hollow electrodes increased the heating efficiency of the test furnace by approximately 10 pet over solid electrodes, the hollow electrode consumption was 2 to 10 pet above that of the solid electrodes. In order to improve energy efficiency while reducing electrode consumption, the Bureau designed alternative electrode tips. A basic electrode tip design used in the Bureau's research was a 1-in-diam tip machined onto the end of the 3-in-diam electrode (fig. 5). The theory behind this tip design is that the arcing between the workpiece in the furnace and the elec- trode will remain concentrated onto the 1-in-diam tip. The tip will become heated while the remaining 3-in-diam section of the elec- trode will remain relatively cool. This phenomenon was demon- strated in the high-speed films of the arc since the tip glowed white hot while the remaining, larger diameter electrode section surface remained black. By making the smaller diameter electrode tip section an expend- able section, the larger outer section of the electrode could become a semipermanent structure, thus reducing the amount of graphite needed for arcing (fig. 6). INERT GAS INJECTION Inert gas injection was investigated to decrease electrode con- sumption by replacing the oxidizing atmosphere with an inert atmosphere. The inert gas was introduced into the furnace at three locations, through each viewport and also through the arcing elec- trode. Holes 1/32 in. in diameter on a 1-in-diam circle encompass- ing the electrode tip were used to introduce the gas (fig. 7). The gas from the 15 exit holes shrouded the arc and maintained it in a vertical direction, thus allowing more of the heat to be directed to the melt. The gas also helped to confine the arc to the tip section while it cooled the outer portion of the electrode, again decreasing electrode consumption. ARC ANALYSIS Electrical waveforms of voltage and current were monitored across the arc with a waveform analyzer. The waveforms differed markedly depending upon the target composition. The differences between arc waveforms when using graphite, steel, and copper tar- gets, are easily visible (fig. 8). These differences are expected since the thermal and electronic properties, such as the melting point and the amount of energy needed to free an electron from the surface of a material, vary dramatically from material to material. Figure 8 shows the voltage as a solid line and the current as a dashed line. The spikes on the voltage waveforms corresponded to the move- ment of the arc as seen in the high-speed films. In figure 8 (top), the positive and negative half-cycles have roughly the same abso- lute amplitude because the arc occurred between a graphite elec- trode and a graphite target. However, in the center and bottom R 1-1/2 R 1/2 TOP VIEW SIDE VIEW Figure 5.— Button electrode tip. Feedable portion of electrode Stationary electrode shell Figure 6.— Center feed electrode tip. R 1/16' TOP VIEW SIDE VIEW Figure 7.— Button electrode tip with gas injection. panels of figure 8, the positive and negative half-cycle amplitudes show asymmetry owing to arcing between a graphite electrode and steel or copper, respectively. This asymmetry is due to the ability of graphite to emit electrons thermionically, whereas metals melt before thermionic emission occurs. In each of the waveforms, the electrode is acting as the anode in the positive half-cycles. Waveforms and the corresponding high-speed films were ana- lyzed together to identify the arc characteristics that were associated with the millisecond events on the waveforms. An example is shown in figure 9. In this case, the furnace atmosphere was 100 pet Ar and the arc was between a graphite electrode and a copper block. In the figure, voltage is the solid waveform and current is the dashed waveform. Positive half-cycles 1 and 3 are at a relatively low volt- 60 < 1 1 N O ^ 40 — — 1— " Voltoqe-i Z Current^ a: a: 20 3 n \ ^ ^ r-\ r^ - O T3 C O ^--^^ _ K V "/ ,/.:-> \ — "/ ^--> K"---/ r::> \ > LJ O-20 V V 1— o > 1 1 80 TIME.s Figure 8.— Waveform of voltage and current for arc from graph- ite electrode to graphite (top), steel (center), and copper (bot- tom) target in 90-pct-He, 10-pct-Ar atmosphere. age as compared to positive half-cycles 2 and 4. Half-cycles 1 and 3 could be matched to short arcs between the graphite tip and cop- per block as the arc target, whereas half-cycles 2 and 4 correspond to long arcs on the film that were between the 3-in-diam section of the electrode and the copper block arc target. Many shorter term discrete events of much less than a half- cycle duration have been identified in both voltage and current wave- forms. In figure 10, four major events on a single-cycle voltage waveform have been correlated with the arc movement in the high- speed films. These events took place in a 90-pct-He, 10-pct-Ar atmosphere. Event A, which is a positive voltage spike, cor- responded to the arc changing positions from arcing a short dis- tance from the electrode tip, to arcing a longer distance, between the 3-in-diam section of the graphite electrode and the copper block as the arc target. The maximum voltage for event A was 53 V and occurred as the arc developed into the long arc column as depicted in sequence A of figure 10. Event B, on the positive side, showed that the rapid fluctuation of the voltage related to a long arc decreas- ing to a short arc on the tip section of the electrode and then back again to a long arc for each fluctuation. Event C, termed a shark's tooth, on the negative half-cycle could be related to the arc's move- ment across the tip section in a right-to-left motion, and finally event D was the opposite motion of the arc moving from the left to the right side on the electrode tip section. A similar experiment with an atmosphere of 90 pet He- 10 pet Ar produced the arc motions as shown in figure 1 1 . Five areas are related to the arc's movement in the high-speed films. Area 1 related to a short arc rotating directly beneath the electrode's tip. Area 2 showed the gradual movement of a long arc, between the 3-in-diam graphite electrode and copper block, from the tip to the outer edge of the electrode. Area 3 is a short, very stationary arc on the nega- tive half-cycle of the voltage waveform. Area 4 was correlated with the quick movement of the short arc from the left side to the right side on the tip. Area 5 was a rotating short arc directly beneath the electrode tip. B Long arc ± T I Copper block Figure 9.— Waveform. A, Long arc; B, short arc. 6,680 10.020 TIME, ps Area 1 1 Area 1 ' 1 r A_^ 2 - - 1 Area Area Area 1 3 1 1 1 4 5 1 3,320 6,640 9,960 TIME, ;js 16,600 Rotary motion Plasma W Copper block' %r Arc motion ^ 1 r 11 Direction |__J X^ mo veme nt ^{) Figure 10.— One cycle with discrete events (A-D) delineated and diagramed. ovement Figure 1 1 .—One cycle with five discrete areas delineated and diagramed. 1, rotating, short arc; 2, arc moving right to left, arc- ing between 3-in-diam section and copper block; 3, very station- ary arc; 4, arc movement; 5, rotating arc under tip. 10 Image analysis is being used to help define the arc's core within the plasma shown on the high-speed film ft-ames. The arc structure is difficult to resolve because of the high luminosity that tends to saturate photographic emulsions, producing a seemingly white arc. Image analysis is helping to resolve the inner structure by digitally filtering the images to reconstruct the actual radiant intensity gra- dients. From these intensity gradients, the electron paths can be used to relate arc parameters to the resistance of the arc. For instance, image analysis is used to map areas of highest light inten- sity so that an arc path of conduction can be defined. By using this technique it has been possible to measure an arc length (fig. 12). SYSTEM INTERACTION The occurrence of a break in the arc at high current or the short- ing of scrap to the electrode is responsible for electrode breakage, increased melt times, and voltage spikes that can damage electri- cal equipment, cause flicker on the electric power grid, and cause excessive noise. The identified events that cause these disruptions of the furnace arc are related to motion of the arc, scrap, electrodes, and/or magnetic fields, as seen in the Bureau experiments. When methods of stabilizing the arc in the furnace environment are con- sidered, it is necessary to consider all of the interactions between the furnace variables, such as slag composition, atmosphere com- position, type of scrap, temperature of the melt, composition of the electrodes, electrode geometry, and control methods. Early in this investigation, it became clear that there is a synergistic rela- tionship between the variables in the furnace interior that is a direct result of the nonlinear relationships of the coupled magneto- hydrodynamic equations governing the electric arc behavior. Because of this interactive behavior, it is not possible to change one variable without affecting the other furnace variables. Presentiy, most of the important parameters in the electric furnace are allowed to float at whatever value they may take. For example, there is no mechanism to control furnace atmosphere composition, instantane- ous voltage, or geometry of the electrodes. These variables have been shown in experimentation to be very important for arc operation. ATMOSPHERE CONTROL One of the most influential variables in the arc experiments is the furnace atmosphere. Diatomic molecules such as oxygen and nitrogen must be disassociated before they can be ionized, and there- fore, they are more difficult to ionize than inert gases. If inert gases are used, then the electrode consumption is dramatically decreased since there is no oxygen or nitrogen to react with the graphite. These reactive diatomic gases also will react with the steel if they are pres- ent, and if they are absent, then the steel composition can be more closely controlled. The gases found to be most promising in this study are mixtures of helium and argon. These mixtures have good heat transfer properties and are easy to ionize. It has also been indicated in the studies of the arc fluid dynamics that shrouding the arc in a flowing gas will help to stabilize the arc as in a plasma torch. This comes about by a combination of a thermal pinch (contraction of the arc due to cooling of the outer arc surface and a subsequent reduction in electrical conductivity, forcing the electron flow into the center of the arc), causing wall stabilization and actual fluid dynamic forces preventing the arc from migrating through the gas shroud (fig. 13). This indicates that properly engineered gas injection through the electrode would help to stabilize the arc. This plasma jet effect also will increase con- vective heat transfer to the melt, thereby increasing thermal effi- ciency. The nonreactive inert gases also will permit the use of a wide range of slags that cannot be used in the traditional air atmosphere, and the gas injection through the electrodes possibly could be used to introduce reductants as needed. Figure 12.— Processed image (actual size) showing arc path between 3-ln-diam section of graphite electrode (upper right- hand attachment point) and graphite block as the arc target (near bottom of 1-in button on electrode). Figure 13.— Gas shroud around transferred arc showing forces tending to stabilize arc (actual size). 11 CONTROL SYSTEM A computerized control system could be operated by using a data base of historical waveform information and pattern match- ing information to determine the real-time furnace operating con- ditions. Using the data such as voltage and current waveforms, atmosphere composition, feed material, and slag type, the control system could adjust the furnace operating conditions for optimal performance. The adjustments would include variables such as atmosphere composition, feed rate, feed composition, voltage, gas injection rate, magnetic stirring, and electrode position. These adjustments could be made at rates that depend on the parameter being adjusted. For instance, if the parameter is voltage, then the adjustments must be made in the course of a half-cycle (8 ms). On the other hand, if the parameter is feed material, then the adjust- ment will be allowed to take place over a period of minutes. For indicators of a catastrophic disruption such as breaking of the arc during high-current conditions, the most simple control strategy would be to turn the current off at a zero current crossing, then position the electrodes to restart seconds later. More complex con- trol would involve electronic tap changing and electrode control to maintain the arc under adverse conditions. This would prevent disruption at high current. The type of system best suited to this process of decisionmak- ing based on data and experience is an expert system. The system could maintain a knowledge base of waveforms and past experience. Over a period of time, the system can be customized to the individual furnace it operates on by logging any uncataloged events and the corresponding results so that future decisions could be made based on this experience. These expert systems are beginning to be seen throughout industry . Adaptive expert systems are now being devel- oped and within the next few years will be corrmiercially available. SUMMARY A new perspective is available for investigation of electric arc behavior through the use of high-speed motion pictures and syn- chronized electrical waveforms. Analysis of the seemingly random occurrences in the arc on an event-by-event basis shows that the arc is deterministic and hence theoretically controllable. High-speed computers now make it possible to economically control factors such as transformer tap settings, furnace atmosphere, and electrode positions. These control factors coupled with new electrode geom- etry can improve efficiency and yield while stabilizing the elec- tric arc. New ways have been developed for investigating electric arc behavior through the use of high-speed motion pictures synchronized with electrical waveform data. REFERENCES 1. Ochs, T. L., A. D. Hartman, and S. L. Witkowski. Waveform Anal- ysis of Electric Furnace Arcs as a Diagnostic Tool. BuMines RI 9029, 1986, 19 pp. 2. Paschkis, M. E., and J. Persson. Open-Arc Furnaces. Ch. in Indus- trial Electric Furnaces and Appliances. Interscience, 2d ed., 1960, pp. 179-228. 3. Schwabe, W. E. Arc Furnace Power Delivery Scoping Study. Elec- tric Power Res. Inst., Palo Alto, CA, EPRI RP-1201-24, 1982, 146 pp. 4. Abraham, R. H., and C. D. Shaw. Chaotic Behavior. Part 2 of Dynamics, The Geometry of Behavior. Aerial Press Inc., Santa Cruz, CA, 1983, pp. 99-105. 5. Crutchfield, J. P., J. D. Farmer, N. H. Packard, and R. S. Shaw. Chaos. Sci. Am., v. 255, No. 6, 1986, pp. 46-57. 6. Schwabe, W. E. Lighting Flicker Caused by Electric Arc Furnaces. Iron and Steel Eng., Aug. 1958, pp. 93-100. 7. . Experimental Results With Hollow Electrodes in Electric Steel Furnaces. Iron and Steel Eng., June 1957, pp. 84-92. 12 PREHEATING OF FERROUS SCRAP By R. H. Nafzigeri and G. W. Elger^ ABSTRACT Energy conservation is an important consideration in all steelmaking operations. Energy con- sumption impacts on the productivity and costs of producing steel. Accordingly, the Bureau of Mines has evaluated the feasibility of preheating ferrous scrap charges in basic oxygen furnace (BOF) and electric furnace steelmaking operations to decrease energy consumption. Offgases gener- ated during oxygen blowing of a molten charge in a !4-st-capacity BOF were passed through a static bed of shredded auto scrap. Final bed temperatures ranged from 1,150° to 1,650° F. The thermal energy recovered can contribute up to 44 pet of the energy necessary to melt the scrap. Furnace offgases from a 1-st-capacity electric arc furnace were used to preheat continuously charged automotive scrap and metal stampings up to 1,1 10° F. Approximately 7 pet less electrical energy was used compared with that consumed in continuously charging cold scrap. Conventional back- charging techniques also were used for comparison purposes. INTRODUCTION In 1985, electric arc furnaces produced 34 pet or nearly 30 million st of raw steel in the United States (1).^ Most of the remainder (59 pet) was produced in BOF's. Electric furnace steel- making operations use cold ferrous scrap nearly exclusively as charge materials, and a considerable amount of cold scrap is used by the BOF. Energy consumption is one of the primary concerns and a major cost in domestic steelmaking operations. For exam- ple, approximately 535 kW-h/st of steel produced is required in electric arc furnace steelmaking (2). This represents approximately 1.8 X 10* Btu/st of steel. The fuel and electrical energy consump- tion in a BOF is approximately 0.9 x 10* Btu/st (3). This represents 1.14 X 10'" Btu/yr consumed in domestic steelmaking. A 1-pct decrease would save 1.14 x 10'^ Btu/yr or 334 x 10* kWh/yr. In both types of steelmaking operations, the Bureau is striving to increase the recycling of scrap. The BOF experiments were con- ducted to assess the feasibility of increasing the proportion of scrap used in the charge mixture. In addition, the use of hot offgases for preheating offers the potential of decreasing energy consumption and costs. The additional amount of scrap used could offset the limited availability of hot metal owing to blast furnace shutdowns, for example. The evaluation of preheated scrap in electric arc fur- nace steelmaking was aimed at promoting scrap use by making elec- tric arc furnace steelmaking more competitive through decreased electrical energy consumption and costs. Early Bureau research involved the use of waste heat in electric arc furnace offgases to preheat prereduced iron ore pellets during continuous charging of the furnace (4-5). ' Research supervisor. 2 Research chemist. Albany Research Center, Bureau of Mines, Albany, OR. ^ Italic numbers in parentheses refer to items in the list of references at the end of this paper. Others have discussed the preheating of scrap in BOF opera- tions. Several methods have been described, including (1) in- vessel preheating, (2) separate vessel preheating with the thermal energy provided by oxygen-natural gas or oxygen-fuel oil burners (6-7), and (3) waste gas preheating with the heat derived from offgases generated during oxygen blowing of the BOF charge (8). In one application, natural gas-oxygen burners preheated scrap charges prior to the molten iron addition. Increased productivity and lower ingot costs were realized, but excessive scrap oxidation was noted (9). In another study, increased scrap could be charged to a BOF when it was preheated to 1,700° F. This decreased lime and flux consumption, decreased slag volume, and decreased metal blow- ing time. However, refractory consumption was increased (]0). Developments in preheating scrap charges for electric furnace steelmaking occurred as early as 30 yr ago. In nearly all cases, some means of external heating was used. Three techniques for preheat- ing have been used. The first utilizes a special vessel for preheat- ing. Typically, external fuel burners supplement the heat from the offgases used. After preheating, the scrap is transferred to a charging bucket prior to introduction into the furnace (11). Considerable scrap handling, large space requirements, and high costs for the vessel pit are cited as disadvantages. A second method involves placing the charge bucket into a pre- heating vessel. In this case, the scrap cannot be preheated to a high temperature, and there is more dust (12). Variations in this tech- nique involve the direction of hot furnace offgases to preheating stands or to a preheating chamber to heat the scrap in a charging bucket (13-14). In the third technique, a bucket lined with castable refracto- ries serves as a preheating vessel and a charging bucket. Potential bucket distortion, dust losses, and a weight that requires a high crane capacity can cause problems with this method. 13 Oil burners either mounted in the charging bucket or in the furnace, natural gas lances, the use of charging buckets with lou- vers placed over hot billets or ingots, or the use of preheat cham- bers are additional preheating techniques that have been used (15-22). All of these techniques involve backcharging methods for feed- ing the furnace. Relatively high capital costs have precluded the adoption of fuel-fired preheaters. Other disadvantages include (1) distortion or warpage of the charging bucket doors, (2) uneven heat distribution within the charge, and (3) scrap oxidation. Objectives of the Bureau research reported herein include (1) an evaluation of preheating BOF scrap charges using recovered hot offgases that are passed through a scrap bed in a separate cham- ber to eliminate oxygen-fuel preheating, and (2) a determination of the feasibility of continuously charging fragmented scrap into an electric arc furnace whereby the scrap is heated by countercur- rent hot furnace offgases to realize decreased electrical energy requirements compared with those necessary in conventional back- charging tests. BOF EXPERIMENTS EXPERIMENTAL EQUIPMENT, MATERIALS, AND PROCEDURES All of the tests were performed in a 'i-st capacity BOF, shown in figure 1 . Offgases generated during oxygen blowing of the fur- nace charge were drawn through an adjacent preheat vessel by a blower located downstream from the preheater in the dust collect- ing system. The preheat vessel contained the scrap charge. A schematic diagram of the system is depicted in figure 2. Further details have been published previously (23-24). The furnace charge consisted of shredded automobile scrap hav- ing pieces no larger than 3 by 3 in, with a bulk density of 92 lb/ft\ This material was fed into the preheat vessel prior to blowing the BOF metallic charge. After the blow, the heated scrap was used in the next test as a replacement for the normal cold scrap charge. Offgas flow was controlled by a manual slide damper located down- stream from the preheater. The BOF charges contained 100 to 180 lb of preheated automobile scrap and 270 to 350 lb of molten pig Damper Duct Lance Scrap preheater Insulated duct Basic oxygen furnace Figure 1 . — Basic oxygen furnace and preheater system. Figure 2.— Schematic diagram of the BOF-preheater system. 14 iron. Typical charge compositions are summarizeci in table 1 (23-24). After the preheated scrap-hot metal mixtures were charged, the oxygen lance was positioned, and blowing began after 25 lb of lime and 1 lb of fluorspar were added to the furnace. Oxygen blowing, at a fixed rate of 27 scfm, was terminated when visual observations of flame height and color indicated a carbon level of 0. 10 pet. Blow times varied from 12 to 17 min and were depend- ent upon the quantity of scrap added. RESULTS An initial series of tests showed that cold scrap could consti- tute up to 28 pet of the charge. Cold scrap additions above this level resulted in a significantly lower oxygen efficiency and deteriorat- ing furnace operations (23-24). A second series of experiments were conducted in which pre- heated scrap charged to the BOF ranged from 22 to 40 pet of the total charge. Data from these tests are shown in table 2. Average preheated scrap temperatures ranged from 1,050° to 1,650° F, with the lower amount of preheated scrap resulting in the highest scrap temperature. Scrap temperatures were dependent upon the quan- tity of scrap in the preheater and upon the length of the oxygen blowing period. With 40 pet preheated scrap, the average scrap temperature increased as a result of a significantly lengthened blow time with lower oxygen efficiency. At 1,650° F, sufficient heat is stored in the scrap to yield approximately 44 pet of the energy required for melting. Results indicated that a 40-pct preheated scrap charge was the maximum that could be tolerated by the Bureau's Table 1 .—Typical chemical analysis of BOF metallic charge and product, weight percent Description C Cu Mn P Si Scrap .... Hot metal. SteeM .... <0.1 4.0 0.12- .15 0.12 .25 .21 0.062 .80 :5.1 0.011 .07 ^.01 0.32 1.0 ^.1 Steel product from 33-pct scrap addition. Table 2.— Average temperature and heat recovery values of preheater scrap charges Preheated Av scrap Heat Total heat Heat scrap, temp, content, ' recovered, recovered. pet of °F Btu/(lb'mol) Btu pet required BOF charge for melting^ 22 1,650 14,240 25,500 44 27 1,265 9,770 21,000 30 29 1,175 8,630 20,100 27 33 1 ,045 7,330 19,700 23 36 1 ,050 7,380 21,100 23 40 1,150 8.380 27,000 26 ' From BuMines B 476, 1949, p. 85. Heat content, Btu/db-mol) 2 Calculated = oo ,o., o. .»,k.' u X 100. BOF to maintain satisfactory oxygen efficiency and sufficient metal temperature for tapping. Therefore, preheating the scrap increases scrap utilization from 28 pet of the charge to 40 pet, an increase of 43 pet. On the basis of these tests, 36 pet of preheated scrap appeared optimum with respect to final steel temperatures ( = 3,000° F) and oxygen efficiency. Typical temperatures of preheater entrance and exit gases and scrap under these conditions are shown in figure 3 (23-24). Scrap preheating was enhanced by the oxidation of the CO, formed in the BOF, to CO2 before reaching the preheater. This was caused by secondary air infiltration around the BOF hood. Signifi- cant scrap oxidation occurred only when scrap bed temperatures exceeded 1,800° F. A tighter fitting hood over the BOF would decrease the air infiltration and perhaps allow higher offgas inlet temperatures in the preheater without excessive scrap oxidation (23-24). However, the heat content of the inlet gas would be lower and the scrap preheat temperature therefore would be lower. 1- < cr UJ a. UJ 2,550 2,250 L950 - .650 1,350 1,050 750 450 I I I I ' I r KEY A Offgas enfrance temp • Average scrap temp ■ Offgas exit t( 4 8 12 16 OXYGEN BLOWING TIME, min 20 32,130 Btu/(lb-mol) (Denominator is the heat content of steel at 2,786° F.) Figure 3.— Effect of oxygen blowing time on preheater offgas and scrap temperatures. Charge condition: BOF = 160 lb scrap + 290 lb hot metal, preheater = 160 lb scrap. 15 ELECTRIC ARC FURNACE EXPERIMENTS EXPERIMENTAL EQUIPMENT, MATERIALS, AND PROCEDURES All tests were conducted in a conventional electric arc steel- making furnace of 1-st capacity, lined with a basic brick and cov- ered with a rammed-alumina roof, as shown in figure 4. The electrical energy was provided by a 1,200-kVA transformer through three 4-in-diam graphite electrodes. A stainless steel chute was attached to the furnace. At the opposite end of the chute was a rotating feeder-preheater that consisted of six 12-in-diam sections of stainless steel tubing connected at 90 ° to each other in a zigzag fashion. Figure 5 shows a schematic diagram of the furnace, scrap feeder-preheater, and furnace offgas ductwork, including gas sam- pling location and dust removal units. In use, the charge was fed through the chute and preheater, with the hot exhaust gases pass- ing through the preheater (3)* countercurrent to the direction of the charge, thereby heating the charge and cooling the gases. The offgases exited directly into a long vertical section of duct that assisted in the removal of those smaller sized dust particles escaping a stainless steel cyclone. The gases then were directed horizontally to a second cyclone (1) and an adjacent baghouse for final cleaning before being exhausted to the atmosphere. Gas and particulate samples were taken in this horizontal section (2). This furnace has been described previously in more detail (5, 25). The components in the sampling train for particulate stack sampling. " Bold numbers in parentheses refer to components identified in illustrations. shown in figure 6, included the probe with attached nozzle (1), a particulate filter (4), a cooling and/or gas collector with four impingers (5), flow-measurement devices, and a vacuum pump (13). Other components shown in figure 6 include a cyclone (2), flask (3), thermometers (6), check valve (7) connecting cord (8), vacuum gauge (9), coarse adjust valve (10), fine adjust valve (11), oiler (12), filter (14), dry-gas meter (15), orifice tube (16), incline manometer (17), solenoid valves (18), pitot (19) thermocouple (20) and temperature recorder (21). This equipment and the procedures used for gas sampling also have been described in more detail previ- ously (25). The shredded scrap used for these tests was purchased from a local scrap processor and consisted of three separate batches, each purchased at a different time and with a different composition. The metal stampings used in the charges were purchased from the same source. Only pieces of scrap with largest dimensions of less than 4 in were used. The chemical analyses of these materials are shown in table 3 . The analyses were obtained by a direct-reading spectro- graph for cast samples melted in separate 800-lb wash heats with- out quartz and lime additions to provide a slag cover. Scrap meltdown without a slag cover was conducted to keep the alumi- num in the metal phase rather than transfer that constituent to the slag phase. KEY / Exhaust gas cleaner 2 Gas sampling J Preheater 4 Electric-arc furnace Figure 4.— One-short ton electric arc furnace and feeder- preheater used for steelmaking tests. Figure 5.— Schematic diagram of electric arc furnace, feeder- preheater, and offgas ductwork (not to scale). 16 Table 3. — Chemical composition of shredded automotive seraph and metal stampings used in three types of scrap charging tests in an electric arc furnace, weight percent Batch 1 Batch 2 Batch 3 Stamp- Large Small ings Al ... 0.46 0.83 NA 2.15 0.046 C... .41 .56 0.051 1.66 .67 Cr... .16 .40 .050 .47 .49 Cu... .18 .81 .20 1.18 .087 Fe... . 98.5 95.7 99.5 91.5 97.2 Mn .. .12 .43 .036 .69 .41 Ni . . . .17 .30 .11 .31 .70 P.... .008 .034 .011 .070 .013 S... .030 .045 .037 .048 .008 Si... <.01 .85 .011 1.86 .26 Sn... NA .019 .009 .037 .005 Ti .. . NA <.01 <.01 <.01 .003 NA Not analyzed. 1 Pb and Zn contents of the scrap samples were not analyzed since these constituents volatilized from the furnace during charge meltdown and were recovered in the dust product. For all tests, the furnace was preheated, and the initial charge of 450 lb of shredded auto scrap, metallurgical coke reductant, and slag formers (pebble lime and quartz) was topcharged to the fur- nace by means of a charging bucket. After the initial charge was melted, continuous feeding commenced. For the tests in which cold scrap was continuously charged, the feed material was fed into an opening in the side of the feed chute between the furnace and the preheater, as shown in figure 7. The cold scrap to be preheated was fed directly into the preheater shown in figure 8. Preheated scrap temperatures were calculated from data obtained by taking samples of scrap as they entered the furnace from the preheater and immediately immersing the samples into a measured amount of water. During the backcharging tests, only a portion of the initial 450-lb charge was melted. At that point, the furnace roof was swung aside, and the first backcharge, consisting of 675 lb of shredded scrap, was loaded as shown in figure 9. When this was melted suffi- ciently to allow another 675 lb of shredded scrap to be added, the same procedure was followed. The remainder of the test was iden- tical to the continuously charged tests. / Probe i'Cyclone J Flask ^ Particulate filter 5 Impingers 6' Thermometer 7 Check valve 8 Connecting cord 9 Vacuum gage 10 Coarse adjust valve // Fine adjust valve 12 Oiler 13 Vacuum pump 14 Filter 15 Dry-gas meter 16 Orifice tube 17 Incline manometer 18 Solenoid valves 19 Pilot - 50 CO O 40 o tn > 30 20 ~i 1 \ r More fluid slags 1 I I r Less fluid slogs Low Apporent solidification High 1,300 1,340 1,380 1,420 TEMPERATURE, 1,500 1,540 Figure 2.— Viscosity-temperature profiles for natural and syn- thetic fluorspar additions to BOF slags (8), A,, Natural fluorspar (ceramic grade), 2.7 pet F; A2, natural fluorspar (ceramic grade), 2.2 pet F; B,, synthetic fluorspar (improved product), 3.2 pet F; B2, synthetic fluorspar (improved product), 2.9 pet F; C, synthetic fluorspar (commercial product), 2.9 pet F; Di, synthetic fluor- spar, 2.7 pet F; D2, synthetic fluorspar, 1.9 pet F. 100 90 80 70 Q- ^ 60 p- >- 50 30 20 10 - "I i 1 1 r More fluid slogs "I 1 1 r Less fluid slogs Low Apparent solidification temperature -High 1,280 1,320 1,360 1,400 TEMPERATURE, 1,480 1,520 Figure 3. — Viscosity-temperature profiles for pilot-scale BOF steelmaking slags to which natural and synthetic fluorspars were added (8). Ai, Overblown BOF slag with natural fluorspar, 0.7 pet F, 35.9 pet total Fe, 3.6 basicity; A2, normal practice BOF slag with natural fluorspar, 1.4 pet F, 19.7 pet total Fe, 3.8 basic- ity; B, BOF slag with synthetic fluorspar (improved product) 1 .3 pet F, 17.6 pet total Fe, 3.3 basicity; C, BOF slag with commer- cial synthetic fluorspar, 1 .4 pet F, 13.3 pet total Fe, 4.4 basicity. 26 Table 3.— Chemical analyses of BOF slags and metals (8), weight percent Table 4.— Steel analyses from evaluation of used potlining in a BOF (10), weight percent Natural Bureau snythetic Commercial syn- fluorspar fluorspar thetic fluorspar Metal Slag Metal Slag Metal Slag AI203 .... NAp 0.35 NAp 0.35 NAp 0.36 c 0.28 NAp 0.05 NAp 0.07 NAp CAO NAp 42.1 NAp 50.5 NAp 53.2 F NAp 1.5 NAp 1.3 NAp 1.2 Fe2+ .... NAp 16.8 NAp 12.6 NAp 13.3 MgO .... NAp 4.9 NAp 2.6 NAp 2.7 Mn .23 6.1 .26 4.6 .29 5.5 P <.01 .28 <.01 .36 <.01 .51 S .011 .096 .014 .076 .016 .100 Si .12 NAp .24 NAp .12 NAp SiOa NAp 12.8 NAp 15.1 NAp 14.8 Total Fe . . NAp 20.7 NAp 16.5 NAp 16.0 NAp Not applicable. The results of Kilau (8) also indicated that a fluidizer substi- tute need only be effective in the early stages of a blow to solubi- lize the dicalcium silicate. Toward the end of the blow, increasing iron contents of the slag could serve to provide the required fluidity. Hence, a relatively unstable synthetic fluorspar could be effective in BOF steelmaking. In some cases, a BOF could be operated with- out fluorspar, especially for low-carbon steels in which the longer blowing time could permit adequate refining and produce sufficient iron oxide in the slag to effect satisfactory fluidization. Chemical analyses of the steel produced in the BOF using alu- minum smelter podining material and fluorspar are shown in table 4 (10). Spironello found that higher sulfur levels resulted from the use of potlining material because the sulfur content was higher in the hot metal charge. All slags were fluid and permitted satisfac- tory sampling and tapping. Slag analyses are presented in table 5, also from Spironello (10). Based on the MgO contents of the slags, there were no significant differences in refractory lining attack when potlining was used. The higher AI2O3 and Na20 contents resulted from the aluminum and sodium in the potlining material (10). Heat Mn Fluorspar: 1 2 Potlining lump:' 1 2 Potlining pellets:2 1 2 0.46 .47 .35 .57 .42 .33 <0.10 <.10 <.10 <.10 .10 .10 0.42 .45 .36 .34 .46 .58 0.006 .013 .010 .010 .010 .010 0.008 .009 .019 .015 .023 .017 0.10 .15 .10 .10 .10 .18 1 Minus % in plus 10 mesh. 2 Minus % plus V2 in. Table 5. — Slag analyses from evaluation of used potlining in a BOF (10), weight percent Heat AI2O3 CaO F Fe^* MgO Mn NazO P SiO: Total Fe Fluorspar: 1 0.44 53.6 1.4 12.4 5.4 4.3 <0. 02 0.55 0.063 13.5 17.3 2 73 56.4 1.8 11.0 5.0 5.0 <.02 .69 .072 17.5 11.1 Potlining lump:' 1 2.8 50.5 .91 11.8 4.7 5.5 .95 .47 .087 14.8 16.2 2 3.2 53.1 .85 14.5 4.5 5.4 .95 .52 .084 13.2 17.5 Potlining pellets:^ 1 1.7 56.6 .67 8.6 4.7 4.6 .47 .56 .100 15.1 11.6 2 1.7 53.9 .65 11.6 4.8 5.1 .50 .57 .110 14.2 14.6 1 Minus % in plus 10 mesh. 2 Minus % plus V2 in. The product steels were hot- rolled and mechanically tested. Spironello found that the results showed yield and tensile strengths comparable to conventionally produced steels (10). ELECTRIC ARC FURNACE EXPERIMENTS EQUIPMENT AND PROCEDURES All tests were conducted in a 1-st-capacity, three-phase ac elec- tric arc furnace at the Bureau's Albany (OR) Research Center. Shredded automotive scrap, lime, and quartz were used as charge materials. After the feed had entered the furnace and the bath was molten, the slag conditioner was added. After about 5 min of heat- ing, the slag was removed from the bath. Then, 10 lb of silicoman- ganese was added, the bath was adjusted to the proper temperature, and the furnace was tapped. Both metal and slag samples were taken after slag removal and at the tap (12). RESULTS Of the three types of slag conditioners evaluated on the basis of visual observations (e.g., stickiness of the slag and ease of slag- metal separation), the Gerstley borate and B2O3 group appeared the most effective in increasing the fluidity of the individual slags. The resultant slags contained up to 0.5 wt pet B, while the tapped metal products contained 0.001 to 0.002 wt pet B. Boron at these levels is not considered a harmful constituent in most low- or medium-carbon steels. The constituent markedly increases the hardenability of steels at higher levels approaching 0.007 wt pet. The fluorine-containing materials were also effective in increas- ing slag fluidity. In the tests made with natural and synthetic fluor- spar slag conditioners, the tapped metal products contained between 0.03 and 0.05 wt pet P. The synthetic fluorspar made from fluosi- licic acid obtained from a phosphate plant was not a source of phos- phorus contamination in the tapped metal products. The sodium constituent in aluminum potlining waste presented a slight fuming problem during scrap meltdown. In the third group, Sorelflux B was the least effective under the conditions used. Only visual com- parisons can be given since slag viscosity data were not obtained. Moreover, the slag compositions varied from test to test owing to erosion of furnace lining material. The furnace refractories suffered considerable damage owing to open-bath conditions (12). 27 SUMMARY AND CONCLUSIONS Experiments have shown that boron-containing compounds such as colemanite and fused boric acid fluidize BOF slags better than fluorspar on the basis of laboratory-scale viscosity determinations. The comparisons were made on equivalent amounts of boron and fluorine concentrations in the slags. Kilau demonstrated that the boron-containing slags showed no boron losses, whereas slags con- taining fluorspar were more unstable and lost fluorine by volatili- zation, resulting in increased slag viscosities and higher apparent solidification tempjeratures {7). Increased slag basicity increased the fluorine-containing BOF slag viscosity {7). The stabilities and fluidizing abilities of synthetic fluorspars in BOF slags varied con- siderably, depending upon their method of preparation, as shown by Kilau (8). Improved preparation methods {8) yielded synthetic fluorspars that compared favorably with natural fluorspar with respect to stability and fluidizing ability. On the other hand, a com- mercially prepared synthetic fluorspar was slightly inferior to nat- ural fluorspar (8). Tests in a pilot-scale BOF demonstrated that synthetic fluor- spar can be used as a substitute for natural fluorspar as a flux fluidizer without decreasing the quality of the metal produced. No excessive refractory wear was evident. Fluorspar fluidizers may need to be added only in the early stages of a blow since increasing iron oxide contents of the slag as the blow progresses in a BOF may provide the required fluidity. Low-carbon steels may require no fluidizer. Use of aluminum smelter potlining material as a sub- stitute fluidizer resulted in fluid slags and acceptable metal qual- ity. Refractory wear was nominal. Three groups of slag conditioners that contained boron (hydroboracite and fused boric acid), fluorine (natural and synthetic fluorspar and aluminum potlining and butt tailings), or titanium (Sorelflux B) were used as substitutes for natural fluorspar in elec- tric arc furnace steelmaking. The boron- and fluorine-containing additives were effective in increasing the fluidity of the slags, with the boron-containing materials more effective. Use of the boron- containing conditioners did not increase the boron levels in the melt- down metal. The tapped metal products contained 0.001 to 0.002 wt pet B. Open-bath conditions caused refractory lining erosion. REFERENCES 1. Pelham, L. Fluorspar. BuMines Minerals Yearbook 1985, v. I, 1987, pp. 419-428. 2. . Fluorspar. Sec in BuMines Mineral Commodity Summaries 1987, pp. 52-53. 3. Buxton, F. M., and P. A. Sandaluk. Fluorspar Substitutes in Steel- making. Ind. Heat., v. 40, 1973, pp. 288, 290, 292. 4. Balding, P. C, D. R. Augood, and R. J. Schlager. Making Steel Using Spent Potlining Flux. Trans. Am. Foundrymen's Soc, v. 91, 1983, pp. 493-498. 5. Nash, B. D., and H. E. Blake, Jr. Fluorine Recovery From Phos- phate Rock Concentrates. BuMines RI 8205, 1977, 16 pp. 6. Balgord, W. D. Recycle of Potlining in the Primary Aluminum Indus- try. Opportunities for Technical Improvements. Paper in Proceedings of the Sixth Mineral Waste Utilization Symposium. IIT Res. Inst., Chicago, IL, 1978, pp. 324-333. 7. Kilau, H. W., V. R. Spironello, and W. M. Mahan. Viscosity of BOF Slags Fluidized With Fluorspar, Colemanite, and Fused Boric Acid. BuMines RI 8292, 1978, 25 pp. 8. Kilau, H. W., V. R. Spironello, I. D. Shah, and W. M. Mahan. Evaluation of Synthetic Fluorspar in BOF Slags. BuMines RI 8558, 1981, 28 pp. 9. Drost, J. J., C. B. Daellenbach, W. M. Mahan, and W. C. Hill. Thermal Energy Recovery by Basic Oxygen Furnace Offgas Preheating of Scrap. BuMines RI 7929, 1974, 8 pp. 10. Spironello, V. R., and I. D. Shah. An Evaluation of Used Alumi- num Smelter Potlining as a Substitute for Flurospar in Basic Oxygen Steel- making. BuMines RI 8699, 1982, 11 pp. 11. Spironello, V. R. An Evaluation of Aluminum Smelter Potlining as a Substitute for Fluorspar in Cupola Ironmelting and in Basic Oxygen Steel- making. BuMines RI 8775, 1983, 18 pp. 12. Elger, G. W., R. H. Nafziger, J. E. Tress, and A. D. Hartman. Utilization of Scrap Preheating and Substitute Slag Conditioners for Elec- tric Arc Furnace Steelmaking. BuMines RI 9130, 1987, 26 pp. 28 RESEARCH ON BASIC STEELMAKING REFRACTORIES By T. A. Clancy^ and J. P. Bennett^ ABSTRACT The Bureau of Mines conducted research studies over a 10-yr period on basic refractories com- monly used in steelmaking processes. These studies dealt with properties of refractory raw materials (magnesia, dolomite, and natural flake graphite) and formulations containing these materials. High- temperature properties are reported for periclase (MgO) produced from seawater, brines, and magne- site (MgCOs). Improvements in high-temperature strength and slag resistance prop)erties are described for refractories produced from periclase grain altered by various additions. The properties of 14 calcined domestic dolomites are described. The role of natural flake graphite in dolomite-carbon refractories and the feasibility of substitution by synthetic carbons are discussed. INTRODUCTION The Bureau of Mines has conducted extensive research on basic refractories commonly used in steelmaking processes. These labora- tory studies have focused on raw materials (periclase, dolomite, and natural flake graphite) and refractory formulations (magnesia and dolomite-carbon). Most of these studies have been directed at conserving natural resources via substitution or by improved materials performance. In the case of the refractory raw materials, laboratory studies generally consisted of chemcal, physical, and mineralogical characterization of materials from domestic sources. Refractory mix formulations, however, involved studies of high- temperature physical properties of new or modified mix formula- tions. High-temperature tests such as flexural strength, deforma- tion under load, and slag resistance evaluations are the most useful for comparing the high-temperature properties of various refrac- tories. Basic refractories are the preferred material in most steelmak- ing operations such as the basic oxygen furnace (BOF), the elec- tric arc furnace (EAF), the argon-oxygen decarburizing process (AOD), and ladles. As mentioned in a 1983 paper by Van Dreser and Neely (1),^ the BOF produces the greatest tonnage of steel in the United States, as in most of the world. Much attention has been focused on refractories used in this furnace, extending service life of refractory linings from 200 to 300 heats of earlier days to an average of 1,500 heats. The types of materials used with current BOF practice are shown in table 1 . Magnesite and periclase in the table distinguish only grades. Over 80 pet of the world's produc- tion of steel is produced in BOF's and EAF's (2). The EAF is con- tinually undergoing changes in steelmaking practices, bringing about changes in refractory requirements. These include changes such ' Ceramic engineer, Tuscaloosa Research Center, Bureau of Mines, Tuscaloosa, AL. This paper is based upon work done under an agreement between the Univer- sity of Alabama and the Bureau of Mines. ^ Italic numbers in parentheses refer to items in the list of references at the end of this paper. as scrap preheating, water-cooled panels for sidewalls and roofs, and EAF use as a feed source for AOD vessels. Material usage on EAF sidewalls and slaglines are shown in table 1 , roof refrac- tories, being mainly alumina-based materials, are not included. With the advent of water-cooled panels, the EAF is rapidly becoming a diminishing market for refractories. There is, however, a new market in gun mixes for water-cooled sidewall panels as main- tenance mixes, at slaglines and bottoms. At the slagline, the pre- dominant usage is magnesia-carbon (MgO-C) brick. Shop option and performance generally determine the type of MgO-C brick (amorphous graphite or flake graphite) and the carbon level (8-20 pet residual carbon) to be used. As in the BOF, the principal mode of wear is decarbonization of the matrix. Consequendy, the major Table 1 .—Basic brick usage In steelmal behavior \ following , Cr addition "PP \ Corrosion potentiol 1 JL 10"' 10° 10' 10' 10' TIME, min 10* 10' Figure 2.— Schematic showing time of CI addition and its effect on potential-time behavior of passivated Fe-18Cr and 430SS. Epp is the primary passivation potential. Data from Covino (12). chloride ion in the minimum potential region and in the increasing potential region, respectively. It was found, as expected, that increasing amounts of chloride ion decreased the induction times. However, very significant increases in induction time were observed in cases where the chloride ion was added after approximately 100 min. In the previously mentioned study (11) it was found that pas- sive film became thinner and richer in chromium at times in solu- tion of 100 min or longer. Investigations of the type reported here (11-12) provide significant insight into the roles that environmen- tal species play in the formation and breakdown of protective pas- sive films. EFFECTS OF ALLOY COMPOSITION AND STRUCTURE The Bureau has conducted detailed studies of the effects of alloying elements on the corrosion behavior of steels. Ion implan- tation was one method used extensively for preparing iron-based alloys containing preselected amounts of the desired alloying ele- ment (13-18). Ion implantation is a process whereby ions of the desired alloy- ing element are accelerated in a vacuum to energies in the range of 20 to 100 keV and allowed to impact a target of the base metal (13). The ions penetrate this target to a mean depth on the order of a few hundred angstroms, with an approximately Gaussian depth distribution. This technique offers numerous advantages over other alloy preparation methods, including speed of preparation and rela- tively low cost for small amounts of experimental alloys, the very small amount of alloying element required, the wide range of alloy concentrations that can be formed, the preparation of alloys with optimum surface and bulk properties for a given application, and the preparation of certain metastable alloys that cannot be formed by conventional techniques. These attributes make ion implanta- tion a very useful tool for preparation of model alloys for corro- sion research. In the 1970's the Bureau conducted pioneering studies on the use of ion implantation as a technique for the preparation of alloys for corrosion studies. It was proven (14-15) that nickel or chro- mium implanted into iron produced Fe-Ni or Fe-Cr alloys having corrosion properties identical to those of bulk alloys of similar com- position. Because of the thinness of the ion implanted layer, these "surface" alloys obviously would lose their corrosion resistance if exposed to aggressive environments for extended times. Even so, for the period of time that this alloy layer is intact, it can be expected to behave in a manner similar to that of a bulk alloy of the same composition. In another study (16), it was shown that chromium-implanted iron exposed to air at 320 ° C oxidized at a rate identical to an equiva- lent bulk alloy for times as great as 1,000 h, and that the oxide thickness was identical on the two alloys. In the same study (16) it was shown that aluminum implantation into titanium produced an alloy that, when coupled to aluminum in a NaCl solution, greatiy reduced the galvanic corrosion loss of aluminum compared to that produced when aluminum was coupled to unimplanted titanium. The implanted aluminum was found to reduce the rate of oxygen reduction on the titanium by as much as a factor of 40. The effects of ion implantation on stress corrosion cracking (17) and on corrosion fatigue (18) of iron-based alloys have also been examined. Very significant effects were noted on the stress corrosion cracking (SCC) behavior of AISI Type 316 stainless steel (316SS) in boiling MgClz by implanted Si, N, or Ar. Argon implan- tation reducted the time to failure by 30 to 40 pet compared to unim- planted samples for a total implant dose of 3x10'* ions/cm (approximately equivalent to 20 at. pet in the implanted volume). This detrimental effect of argon was caused by ion damage to the 36 surface region during implantation (approximately 40 nm in depth). Implantation of nitrogen into the 316SS resulted in a similar decrease in time to failure by SCC. Examination of the nitrogen-implanted surface by scaiming elec- tron microscopy (SEM) following these tests showed that portions of the surface had been ruptured along slip planes by an explosive escape of gas from below the surface. An elemental depth profile, obtained by AES, of the surface of a nitrogen-implanted sample after SCC failure is shown in figure 3. A significant amount of nitro- gen remained in the sample following the test, indicating that much of the nitrogen must have been present within the alloy rather than just in grain boundaries. Implanted silicon had a much different effect on the SCC behavior of 316SS. As shown in figure 4, the resistance to SCC increased approximately linearly with the fluence (atoms/cm^) of implanted silicon. Examination of the silicon-implanted samples fol- lowing exposure to the MgClj showed that the surface was cov- ered by a film that was rich in silicon and magnesium. This film passivated the surface, reducing the general corrosion rate. Sili- con also reduced the SCC crack propagation rate. This effect was not anticipated because of the very thin implanted region. In another study (18), the effects of implanting Ti or a combi- nation of Mo and Ta on the corrosion fatigue behavior of 1018 car- bon steel were examined. The Mo plus Ta implanted steels behaved no differently from unimplanted steel in the 0. lA^ H2SO4 solution, with apparently identical corrosion potentials and surviving about UJ o UJ o Z) < 50 100 SPUTTERING TIME, min 150 the same number of cycles in a rotating beam test prior to failure. The steels were adversely affected by the titanium implantation, however. The corrosion rate of the steel was approximately dou- bled, and the fatigue life decreased linearly with the titanium implant dose. An approximate 15 at. pet concentration of titanium decreased the cycles to failure by about 50 pet compared to the unimplanted steel . Examination by SEM of the surface of the titanium-implanted steel following failure revealed the presence of a film containing titanium and iron, with small iron crystallites penetrating this film. The reduced fatigue resistance is thought to be caused by an acceler- ated corrosion at these small crystallites. Another technique that has been used in Bureau research to prepare experimental alloys with controlled composition and struc- ture is vapor deposition (19). This technique involves sputtering a target (or targets) composed of the elements to be incorporated in the desired alloy. Energetic inert gas ions are used to remove atoms from the target by impact. The sputtered atoms are quenched onto a substrate to form the alloy of interest. This quenching proc- ess is equivalent to cooling a molten alloy so rapidly that virtually no movement of the deposited atoms occurs following deposition when the substrate is cold. Depending on such factors as the sub- strate temperature and the rate of deposition, the alloy produced may be either crystalline or amorphous. One alloy system produced by vapor deposition that showed extremely good corrosion resistance for a wide range of composi- tions was Fe-Zr (19). Vapor deposited Fe-Zr alloys were deter- mined to be amorphous (no long-range crystalline ordering) over the range of composition from Fe9oZr,o through Fq^^Zt^t. Elec- trochemical polarization studies showed the Fe9oZr,o to be approx- imately as corrosion resistant as Fe-18Cr in lA^ H2SO4. Figure 3. — Elemental depth profile as obtained by Auger elec- tron spectroscopy for nitrogen-implanted 316SS indicating retained nitrogen after exposure to boiling MgCl2. Data from Walters (17). 1 2 3 SILICON IMPLANT FLUENCE, 10" atoms/cm^ Figure 4.— Relationship of time to failure of 316SS in boiling MgClz as a function of amount of silicon implanted. Data from Walters (17). 37 The electrochemical polarization response for the ¥t-i-iZr(,^ alloy is compared in figure 5 to that of Fe, Fe-18Cr, and Zr in 1^112804 (19). Over the range of potentials from open circuit corrosion (ca. -0.2 V) to the high positive potentials where oxygen evolution occurs (ca. 1.2 V), the FejaZrev alloy has approximately the same corro- sion resistance as zirconium. AES analysis of the Fe-Zr alloys sub- sequent to the polarization studies showed that the surface corrosion film is composed predominately of zirconium and oxygen, with the iron to zirconium ratio in the film being significantiy reduced com- pared to its value in the interior of the alloy. An additional technique that has been applied to the prepara- tion of experimental alloys is laser processing, in which a laser is used to melt a thin layer of a surface. When a thin coating is pres- ent on the material being processed, the laser beam can provide a means of melting and mixing the coating into the surface of the substrate. The electrochemical behavior of a series of Fe-Cr alloys prepared using a Nd-YAG pulsed laser has recently been reported (20). The electrochemical behavior of these alloys exhibited small, but important differences from those observed for similar bulk alloys. Some, but not all, of these differences were found to be caused by the presence of a very protective oxide film on the sur- face resulting from the incomplete exclusion of oxygen during the laser processing. A three-dimensional chromium concentration plot, obtained by electron microprobe measurements of a cross section of the laser processed region, is shown in figure 6. The original bulk material was a Fe-5Cr alloy. Laser mixing of a thin coating with the sub- strate produced an approximately Fe-15Cr alloy to a depth of from 25 to 30 fim. On top of this alloyed region is a chromium-rich oxide film, approximately 5 fim thick that was produced during process- ing. This oxide layer is clearly visible by optical microscopy or SEM examination. Based on the microprobe resolution on the order of 1/im, only slight variations in chromium concentration are observ- able within the laser-alloyed region. However, chemical etching of the cross-sectioned sample resulted in the upper portion of the laser alloy, equivalent to the region in figure 6 from approximately 5 to 15 /im on the depth axis, becoming extremely roughened. From approximately 15-^m depth within the alloy and continuing into the unalloyed substrate, the etched surface was very smooth, with only minor grain boundary dissolution. Because of the large amount of irregular dissolution that occurred within grains in the upper por- tion of the laser-alloyed region, with features having dimensions . 1.4 - / ) - ro - /^ 1 .6 i'^ / Fe/ .2 ■^ 1 / \ ■ / \ FelSCr \ / / / Fe33Zr67/ J ,^^ -.2 - --^- -.6 I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 LOG CURRENT DENSITY, M.A/cm" Figure 5.— Polarization curves in deaerated 1N H2SO4 for FessZre? amorphous alloy compared to Fe, Fe-18Cr, and Zr. Data obtained at polarization rate of 60 V/h. Data from McCormick f79J. of the order of 0. 1 /^m, it was determined that an elemental analy- sis method having better resolution than the electron microprobe (fig. 6) was required to establish whether the laser-alloyed regions were compositionally homogeneous. Scanning transmission electron microscopy combined with energy dispersive X-ray spectroscopy (STEM-EDS) was utilized to investigate at high resolution the microstructure and microchemis- try within the laser-alloyed region. For this smdy a second alloy, also prepared by the laser processing technique, was used. This alloy contained approximately 18 wt pet Cr in the laser-processed region and the substrate contained 9 wt pet Cr. The alloy and sub- strate regions were analyzed as a function of depth into the sample for a thinned cross section. For these STEM-EDS analyses, a 2(K)-nm analysis interval was used with a 50-mn electron beam (probe) size. The results of these analyses are shown in figure 7. ■^rr, Figure 6. — Three-dimensional chromium concentration plot of a typical laser-processed alloy showing surface oxide, laser- alloyed region, and bulk substrate. Chromium concentrations in bulk and laser-processed region are approximately 5 and 15 wt pet, respectively. Data from Molock (2Q). UJ o o 1.250 2,500 3,750 5,000 6,250 1 1 1 1 1 500 .000 1,500 2,000 2,500 3,0 — .!_ 1 1 ,4 4 c *" 1 1 * 500 1,000 - DISTANCE, nm — 1,500 Figure 7.— STEM-EDS chromium concentration profiles. A, Pro- file typical of upper portion of laser-processed region, indicat- ing extensive microsegregation; B, profile typical of lower portion of laser-processed region; C, profile of Fe-9Cr substrate. Data from Molock (20). 38 By using this very small probe size it was possible to detect inhomogeneities in elemental concentration of a few percent over a distance as short as 100 nm. The chromium concentration in the upper portion of the laser processed region is shown in figure 7 A. Referring to figure 6, this analysis trace was taken in a region cor- responding on the depth scale from about 7 to 13 /zm. It is evident that the chromium concentration is highly erratic in this region, and is the cause of the unusual electrochemical polarization and chemical etching behavior mentioned earlier. It should be pointed out that if the data shown in figure 7 A were grouped in sets of 10 points and the average concentration was plotted, the result would be an apparently uniform chromium level of about 1 8 wt pet for the six calculated averages, similar to what was measured by the microprobe method. For the lower portion of the laser processed region (fig. IB) and the substrate (fig. 7C), the chromium concen- tration appears to be uniform based on the STEM-EDS analysis using the 50-nm probe diameter and 200-nm sampling interval. The cause of the microsegregation of chromium shown in fig- ure lA was attributed (20) to impurities migrating in the liquid ahead of the solidification front subsequent to the laser processing, and to supercooling and formation of nonequilibrium phases in the last of the liquid that solidifies. Whatever the cause of this microsegre- gation, its effects on the corrosion properties were much more sig- nificant than would have been expected considering the scale of variation. CONCLUSIONS During the past few years, several new techniques have become available for preparing new alloys with interesting and potentially useful properties. There also is an increasing availability of ana- lytical instruments having much greater sensitivity and spatial reso- lution capabilities. This combination has already resulted in an improved understanding of the roles of composition and structure in controlling alloy properties and will be of increasing importance in the future. These techniques are just as applicable to the improve- ment in properties of existing classes of alloys as they are for development of new materials. Examples of how the improved understanding of corrosion processes and materials performance have been applied to real prob- lems are given in reference 4. REFERENCES 1. Bennett, L. H., J. Kruger, R. L. Parker, E. Passaglia, C. Reimann, A. W. Ruff, H. Yakowitz, and E. B. Berman. Economic Effects of Metal- lic Corrosion in the United States, Part I. NBS Spec. Publ. 511-1, 1978, 65 pp. 2. National Materials Advisory Board — Commission on Engineering and Technical Systems— National Research Council. New Horizons in Elec- trochemical Science and Technology (U.S. Dep. of Energy contract B- M44SS-A-Z). NAS, Publ. NMAB 438-1, 1986, 147 pp. 3. Battelle Columbus Laboratories. Research Needs for Corrosion Con- trol and Prevention in Energy Conservation Systems (U. S. Dep. Energy contract DE-ACO6-76RL01830/MPO-B-F6814-A-K). Feb. 1985, 74 pp. 4. Flinn, D. R. Optimization of Materials Selection Through Corro- sion Science. Paper in Chromium-Chromite; Bureau of Mines Assessment and Research. Proceedings of Bureau of Mines Briefing Held at Oregon State University, Corvallis, OR, June 4-5, 1985, comp. by C. B. Daellen- bach. BuMines IC 9087, 1986, pp. 115-124. 5. Tomashov, N. D. Corrosion-Resistant Alloys and Prospects for Their Development. Protection of Metals (Engl. Transl. of Z. Metallov), v. 17, No. 1, 1981, pp. 11-25. 6. Uhlig, H. H. Passivity in Metals and Alloys. Corros. Sci., v. 19, No. 11, 1979, pp. 777-791. 7. Sato, N., and G. Okanoto. Electrochemical Passivation of Metals. Ch. in Comprehensive Treatise of Electrochemistry, ed. by J. CM. Bockris, B. E. Conway, E. Yeager, and R. E. White. Plenum (New York), v. 4, 1981, pp. 193-245. 8. Hashimoto, K. Passivation of Amorphous Metals. Paper in Proceed- ings of Fifth International Symposium on Passivity: Passivity of Metals and Semiconductors, ed. by H. Froment (Proc. Conf. Bombannes, France, May 30-June 3, 1983). Elsevier, 1983, pp. 235-246. 9. DriscoU, T. J., B. S. Covino, Jr., and M. Rosen. Electrochemical Corrosion and Film Analysis Smdies of Fe and Fe-18Cr in IN H2SO4. BuMines RI 8378, 1979, 33 pp. 10. Rosen, M. Electrochemical Corrosion of Iron-Chromium Alloys Under Ultra-High-Purity Conditions. BuMines RI 8425, 1980, 66 pp. 11. Covino, B. S., M. Rosen, T. J. Driscoll, T. C. Murphy, and C. R. Molock. TheEffectof Oxygen on the Open-Circuit Passivity of Fe-18Cr. Corros. Sci., v. 26, No. 2, 1986, pp. 95-107. 12. Covino, B. S., and M. Rosen, Induction Time Studies of Fe-18Cr and 430SS Under Open Circuit Conditions in Chloride-Containing Sulfuric Acid. Corros., v. 40, No. 4, 1984, pp. 141-146. 13. Sartwell, B. D., A. B. Campbell IH, B. S. Covino, Jr., and P. B. Needham, Jr. Characterization of Alloys Formed by Ion Implantation. BuMines RI 8434, 1980, 29 pp. 14. Covino, B. S., B. D. Sartwell, and P. B. Needham. Anodic Polari- zation Behavior of Fe-Cr Surface Alloys Formed by Ion Implantation. J. Electrochem. Soc, v. 125, No. 3, 1978, pp. 366-369. 15. Covino, B. S., P. B. Needham, and G. R. Conner. Anodic Polari- zation Behavior of Fe-Ni Alloys Fabricated by Ion Implantation. J. Elec- trochem. Soc, V. 125, No. 3, 1978, pp. 370-372. 16. Sartwell, B. D., A. B. Campbell, B. S. Covino, and T. J. Driscoll. Applications of Ion Implantation to Metallic Corrosion. IEEE Trans. Nucl. Sci., V. NS-26, No. 1, 1979, pp. 1670-1676. 17. Walters, R. P., N. S. Wheeler, and B. D. Sartwell. The Effects of Surface Modification on the Stress Corrosion Cracking Behavior of 316 Stainless Steel. Corros., v. 38, No. 8, 1982, pp. 437-445. 18. Sartwell, B. D., R. P. Walters, N. S. Wheeler, and C. R. Brown. The Effect of Ion-Implanted Alloy Additions on the Linear Polarization and Corrosion Fatigue Behavior of Steel. Paper in Corrosion of Metals Processed by Directed Energy Beams, ed. by C. R. Clayton and C. M. Preece. Trans. Metall. Soc.-AIME, Warrendale, PA, 1982, pp. 53-73. 19. McCormick, L. D., N. S. Wheeler, C. R. Molock, and C. L. Chien. Corrosion Properties of Amorphous Iron-Zirconium Films in IN Sulfuric Acid. J. Electrochem. Soc, v. 131, No. 3, 1984, pp. 530-534. 20. Molock, C. R., R. P. Walters, and P. M. Fabis. Effect of Laser Processing on the Electrochemical Behavior of Fe-Cr Alloys. J. Electrochem. Soc, V. 134, No. 2, 1987, pp. 289-294. 39 FUNDAMENTALS OF STAINLESS STEEL ACID PICKLING PROCESSES By Bernard S. Covino, Jr.^ ABSTRACT Research on the pickling of stainless steels has been conducted by the Bureau of Mines in cooper- ation with the American Iron and Steel Institute (AISI). The objectives of the research were to reduce the loss of the strategic and critical metals chromium and nickel, to reduce the use of HNO3 and HF acids, and to reduce the quantity of waste pickling solutions generated. The model used to design the research consisted of removal of scale (pickling) from the stainless steel by under- cutting (dissolving) the metal beneath the scale. The dissolution behavior of AISI Type 304 stain- less steel (304SS) and of three experimental alloys (Fe-4Cr-13Ni, Fe-12Cr-12Ni, Fe-12Cr-17Ni) representative of the chromium-depleted metal beneath the scale was studied as a function of tem- perature and the concentrations of HNO3, HF, and dissolved Fe, Cr, and Ni. Results indicated that the dissolution process was activation controlled, linearly dependent on HF, Fe, and Cr con- centrations, and nonlinearly dependent on HNO3 concentration. A comparison of the behavior of 304SS to that measured for the experimental alloys indicated that it was possible to optimize the pickling of 304SS in terms of metal lost, acids used, and waste generated. A pickling liquor com- position of 0.8M-1.3M HNO3 plus 0.5M HF at 50° C can optimize the pickling of 304SS. INTRODUCTION Some of the basic steps used in the formation of stainless steel sheet are given schematically in figure 1 . The hot- and cold-forming operations used in this process leave the steel in an unusable work- hardened state. A short-term high-temperature annealing operation is used to soften the metal to allow ftirther rolling of the metal. The effects of annealing are that the bulk microstructure is altered, the surface is oxidized to form a thick scale, and the region just below the oxidized surface is compositionally altered thus forming the chromium-depleted region. While the first effect is desirable and controllable, the final two effects are not and result in the loss of metal. The oxide scale and chromium-depleted region formed during annealing must be totally removed during the pickling oper- ation and the metal removed is usually not recovered. In fact, the metals lost end up concentrating in the pickle liquor, making it dif- ficult to use for ftirther pickling operations, and necessitating the disposal of the spent pickle liquor. An in-depth understanding of the pickling process and of all of the factors affecting it could help to minimize the loss of metals from the stainless steel, the use of pickling acids, and the generation of spent pickle liquor. To accom- plish this, a cooperative program between the Bureau of Mines and the American Iron and Steel Institute (AISF) was initiated. The AISI member companies aided research by providing materials and back- ground information. Pickling is defined here as the act of soaking in a solution for the purpose of cleaning or conditioning. For stainless steels, the ' Supervisory research chemist, Albany, OR Albany Research Center, Bureau of Mines, surface that is eventually pickled has already been influenced by the annealing and scale-conditioning processes. The first process, annealing, determines the initial state of both the surface and of the chromium-depleted region. Temperature, time at temperature, and oxygen partial pressure during the annealing operation affects the thickness, structure, composition, and adhesion of the oxide scale to the stainless steel. All of these factors affect the ease of pickling and if left uncontrolled may cause a variability in the pick- ling process. The depth and composition of the chromium-depleted region are interrelated with the scale formation. That is, the chromium that's depleted from the bulk metal goes to form the oxide scale. Temperature, time at temperature, and oxygen partial pressure, the same factors that affect the annealing scale, affect the chromium- depleted region. This depleted region has to be removed during pickling to assure full corrosion resistance and it may, in fact, play a key role in the pickling process. Conditioning of the scale prior to pickling, which is not always done commercially, is the second process that affects the pickling of stainless steel. This consists usually of mechanical (shot blast- ing), chemical (acid), electrolytic (sulfate), or molten salt treatments that are aimed at facilitating the pickling process. Each of the con- ditioning techniques has in common the alteration of the chemical or physical structure of the oxide scale. While it is easy to recog- nize the usefulness of such conditioning techniques it may be that a proper understanding of the annealing and pickling processes would make them unnecessary. The actual process of pickling stainless steels is complex because of the environment that is usually used. The environment 40 consists of solutions of nitric and hydrofluoric acids at tempera- tures of 50° to 80° C containing large quantities of dissolved Fe, Cr, and Ni. To truly understand the process of pickling, and thus to control it, it is necessary to understand the chemistry of this com- plex environment. This is an environment where HF, a weak acid and a strong complexing agent, is combined with a strong oxidiz- ing acid that can be relatively easily converted into diverse other oxidized or reduced forms. The dissolved metals tend to tie up the HF, making it necessary to continually add more HF, while the high-temperature and chemical dissolution reaction tend to expel the nitric acid from solution as a nitrogen oxide. The knowledge that is needed consists of activity coefficients, solubilities, and sta- bility constants for a large number of stable and metastable com- pounds and complexes. Ingot Hot work . . „., Conditioning: mechanical or chemical ' Pickling Cold work Anneal Conditioning: chemical or electrolytic or salt bath Pickling As necessary Finished material Some smdies in most of the areas relevant to the pickling of stainless steels have been done. These include studies by the AISI members on the effects of annealing and conditioning on the pick- ling of stainless steels. Some basic properties of HNO3-HF solu- tions and methods to remove dissolved metals from these solutions are being studied at the Mackay School of Mines. The work to be reported in this paper consists of the effect of temperature, acid, and dissolved metal concentrations on the pickling of 304SS. The model of the pickling process used to design the experimen- tal approach consisted of an undercutting of the scale through dis- solution of the chromium-depleted region beneath the scale. This is shown schematically in figure 2. The chromium-depleted region on 304SS, which formed because of a short high-temperature anneal, has been characterized (1).'^ On the basis of this characterization, alloys were developed and used in the test program that is being reported here. Pickling was thus assumed to occur by dissolution of the chromium-depleted region as represented by three experimen- tal alloys described in the following section. Optimization of the pickling process would occur when the chromium depleted region dissolved at its fastest rate and the bulk 304SS dissolved at its slowest rate. Under these conditions optimization would result in the mini- mum amount of Fe, Cr, and Ni dissolved, the minimum amount of acids used, and the minimum amount of spent pickle liquor generated. 2 Italic numbers in parentheses refer to items in the list of references at the end of this paper. HNO3-I-HF Figure 1 .—Steps used in processing stainless steel strip. Figure 2.— Schematic of assumed pickling process. 41 EXPERIMENTAL The commercial alloy used in all experiments was 304SS sup- plied by the Allegheny Ludlum Steel Corp., Leechburg, PA, and having the nominal composition Fe-18Cr-9Ni. The three experimen- tal alloys were nominally Fe-4Cr-13Ni, Fe-12Cr-12Ni, and Fe-12Cr-17Ni. All of the alloys were exposed to the HNO3-HF solutions which were made from reagent grade HNO3 and HF and high-purity (18 Mfi-cm) water. Only the 304SS was exposed to solutions con- taining dissolved Fe, Cr, and Ni. These solutions were made using reagent grade Fe(N03)3.9H20, Cr(N03)3*9H20, and Ni(N03)2«6H20. All solutions were deaerated using ultra-high- purity (UHP) nitrogen for 4 to 16 h prior to running a test. All weight-loss tests were conducted by exposing the sample for a period of time, removing from solution, rinsing with high- purity water, and drying with UHP nitrogen. Samples were subse- quently weighed to determine the weight loss, and then reexposed to the solution for additional periods up to a total of 90 min exposure time. All corrosion rates were determined by fitting the linear weight loss versus exposure time data to a linear equation. Auger electron spectroscopy (AES) measurements were taken using a Physical Electronics^ model 10-155 cylindrical mirror analyzer. The Auger electrons were induced by a 5-keV electron beam incident at 60° with respect to the sample normal, and a 2-V peak-to-peak modulation signal was applied to the analyzer. To obtain element depth profiles, a 2.5-keV Ar+ beam incident at 10° with respect to the sample normal was used to sputter away small increments of oxide thickness. RESULTS Data for 304SS are plotted in figures 3 through 6 to show the general effects of temperature, nitric, and hydrofluoric acid con- centrations, and dissolved iron, chromium, and nickel concentra- tions on the dissolution of 304SS. Buildup of dissolution products during the 90-min exposure had no effect on any of the dissolution rates reported here. This was based on the fact that the metals lost weight linearly with time. If dissolution products affected the dis- solution process then the data would tend to deviate from linearity. The activation energy necessary for dissolution of 304SS as calcu- lated from curves similar to figure 3 apj)eared generally to range from 5 to 10 kcal/mol in solutions with 1.3M nitric acid and from 9 to 14 kcal/mol in solutions with lower concentrations of nitric acid. These categories applied also when dissolved metals were present. The data in figure 4 show that the dissolution rate of 304SS at constant temperature and nitric acid concentration varies approx- imately linearly with hydrofluoric acid concentration. This was the same relationship observed for the three experimental alloys. Nei- ther nitric acid nor dissolved iron affected this linear relationship. The data in figure 5 show that the dissolution rate of 304SS at con- stant temperature and hydrofluoric acid concentration passes though a maximum as a function of nitric acid concentration. This maxi- mum occurs at approximately 0AM to 1.5M nitric acid. Hydrofluoric acid concentration affected only the magnitude of the dissolution rate and not the shape of this curve. For the three experimental alloys, the shape of this curve was the same but the location of the maximum was shifted to higher concentrations com- 3.5M HNO, 3.5M HNO-,- 2- -2.7M HF 0.0030 RECIPROCAL TEMPERATURE, K pared to that for 304SS and the overall height of the curve increased with decreasing chromium concentration. A typical pickling solution usually has large quantities of dis- solved iron, chromium, and nickel, but because of the composi- tion of the stainless steels the dissolved iron is always more concentrated than either dissolved chromium or nickel. A compar- ison of the effect of dissolved iron, chromium, and nickel on the dissolution rate of 304SS in HNO3-HF solutions is shown in figure 6. Both iron and chromium cause a decrease in dissolution rate. Iron was the most effective in reducing the dissolution rate of the 304SS, dissolved chromium was less effective than iron, while dis- solved nickel had no apparent effect. Iron was also the only dis- solved metal species to shift the intercept of the curve. Films formed as a result of exposure to HNO3-HF were ana- lyzed using AES. Profiles were obtained by alternately doing AES analyses and argon ion sputtering at a rate of 1 .5 A min (calibrated using Ta205). Profiles of 304SS samples exposed to different HNO3-HF solutions show an enrichment in Cr in the film com- pared to the metal substrate, and a small quantity of fluorine in the region of the film-environment interface. The fluorine level was ' Reference to specific products does not imply endorsement by the Bureau of Mines. E -^ < a: A - /^8M HNO3 - ♦/ ^^^ 3.5M HNO3 ■^ 1 O.OM HNO3 1 Figure 3.— Arrhenius plots for the dissolution of 304SS in HNO3-HF solutions. 12 3 HF CONCENTRATION, mol/L Figure 4.— Effect of HF concentration on dissolution rate of 304SS in HNOa-HF solutions at 50« C. 42 ^ 1.00 u c ~^ .75 E .50 - o ^ - / \ \ ^ - 1 o " 1 D - 1 y^ ' □ ^^ c - /S ^~~~~~ ---~-__7o; c o 1/ n sn" r — — — — ~s_ (Lh — PT A 30- c s ^ — - P, , 1 1 , 1^, ,M 1 1 1 1 1 1 1 1 12 3 4 5 HNO3 CONCENTRATION, mol/L Figure 5.— Effect of temperature and HNOa concentration on dissolution rate of 304SS in HNO3-HF solutions with approxi- mately 0.5M HP concentration. 3.5M HNOj, 70'C No metal added .2M Fe - 12 3 4 HF CONCENTRATION, mol/L Figure 6.— Effect of equimolar concentrations of dissolved Fe, Cr, and Ni on dissolution rate of 304SS in HNO3-HF solutions at TO' C containing approximately 3.5Af HNOa. independent of HF concentrations between 0.5M and 1.5M HF. There was some evidence that the level of fluorine was dependent on the solution at concentrations below 0.5M HF. The main differ- ence was a variation in film thickness by about a factor of 2 between the thickest and thinnest films. This variation in thickness was not correlated with any of the test parameters such as temperature or concentration. The fluorine profile in figure 7 shows that a fraction of a monolayer of fluorine exists on the oxide surface below the carbon surface contamination. The surface carbon consists primarily of hydrocarbons picked up from exposure to the atmosphere follow- ing the HNO3-HF treatment. The conclusion that the fluorine exists on the oxide surface is based on many profiles identical to that shown in figure 7. There was a general correlation between the time required to sputter away the carbon surface contamination and the fluorine. It appeared that the fluorine was rapidly removed only after most of the surface carbon was removed. The relatively flat portions of the fluorine profiles represent regions where only sur- face contamination was being removed. If all of the fluorine was concentrated at the film surface (i.e., at the film-carbon contami- nation interface), then the fluorine could represent as much as one- quarter to one-half a monolayer of fluorine at that surface. SPUTTERING TIME Figure 7.— AES sputter depth profile of film formed on 304SS exposed to 3.5Af HNOs-O.SAf HF-0.2Af Fe solution for 90 min at 70" C. The sputter rate was approximately 1 .5 A min. DISCUSSION The first part of this discussion will be applied to understand- ing the mechanism of dissolution of metals in HNO3-HF solutions. The second part will concentrate on applying the data to optimiz- ing the pickling process. The range of activation energies (5-14 kcal/mol) measured here are similar to those reported (2-8) for the dissolution of rapidly corroding metals in solutions other than HNO3-HF. This is in con- trast to those activation energies (15-22 kcal/mol) reported (7-8) for the dissolution of metals that passivate. It therefore appears that 304SS in the HNO3-HF solution is not inhibited from dissolving by a passive film. The evidence obtained from the AES measurements suggests that there is a film on the surface of the metal in solution and that this film is not a passive or protective type of film. Research (9) done elsewhere on 304SS in similar solutions supports the conclu- sion that this is not a passive film. A truly passive film in HNO3 solutions would exhibit a more intense chromium peak than that shown in figure 7. The presence of the fluoride on the outer sur- face of this film suggests that fluoride is intimately involved in the dissolution mechanism. This is supported also by the observed lin- ear increase in dissolution rate with increasing HF concentration. The dissolution rate vs HF curves for HNO3-HF solutions were extrapolated to zero dissolution rate at an HF concentration of zero. This is in agreement with the reported (10) zero dissolution rate of 304SS in low-temperature HNO3. The HN03-HF-Fe solutions do not, however, pass through zero dissolution rate at a zero con- centration of HF. This is probably related to the formation of iron- fluoride complexes. These complexes are reported (11) to be very 43 stable and would be able to tie up the available fluoride. The inter- cept of the curves in figure 6 corresponds to approximately a 3:1 molar ratio of F to Fe, which suggests the formation of FeFj The iron, chromium, and nickel additions to the HNO3-HF solu- tions cause various degrees of reduction of the dissolution rate of the 304SS. Dissolved iron has the greatest inhibiting effect, fol- lowed by dissolved chromium, which is less effective, and dissolved nickel, which has little effect on the dissolution rate. The effec- tiveness of these metal ions in inhibiting the dissolution of 304SS appears to be due to the specific cation's ability to tie up the avail- able fluoride anions. This can be seen by considering the stability constants for the metal-fluoride ligand involving one metal and one fluoride ion. The stability constants (12) are 1.5x10^ for FeF^*, 2.3 X 10" for CrF2^ and 6.3 for NiF^ Thus, the order of effective- ness of the metal ions is the same as the order of the stability cons- tants. The negligible effect of the dissolved nickel is reasonable because the stability constant (II) of 6.3 for NiF^ is very similar to the value of 3.9 for the most prevalent (12) ionized fluorine species, HF2. The role played by dissolved metals in reducing the dissolu- tion rate of 304SS appears to be one of reducing the amount of fluo- ride available for participation in the dissolution reaction. These dissolved metals do not participate directly in the dissolution reac- tion, but rather they alter the solution chemistry. This emphasizes the importance of the fluoride component of the pickling bath. The finding, by AES measurements, of the fluoride on the surface of the metal's film becomes a key to the mechanism of dissolution of the metal. Others (13-14) have suggested that HF has a cata- lytic effect in the formation of metal-fluoride complexes at the film- solution interface. It is postulated that HF allows a more rapid trans- fer of these complexes into solution as compared to species formed in non-fluoride-containing solutions. The data in figure 5 provide evidence of the catalytic nature of the dissolution reaction. The shape of the dissolution rate versus nitric acid concentration curve is characteristic of a reaction that proceeds through heterogeneous catalysis. In such a case the cata- lyst typically adsorbs to the surface where the reaction occurs and the catalyst remains unchanged. For the dissolution of 304SS in HNO3-HF solutions it appears that the fluoride ion adsorbs to the nearly passive film formed in solution, enhances the dissolution of this film, and then complexes with the dissolved metal species. For this mechanism to continue it is necessary for the film to continu- ally reform and this can more than adequately be done by the reac- tion of the nitric acid with the metal. The data shown in figure 5 make it possible to consider optimiz- ing the pickling process. The maximum and minimum represent two areas where dissolution can be very fast or very slow. It was already noted that a similar type of behavior was observed for the three alloys representative of the chromium-depleted region. Thus to optimize the pickling process in terms of HNO3 concentration it is only necessary to superimpose the graphs for 304SS and one or more of the experimental alloys. This has been done schemati- cally in figure 8 for 304SS and the Fe-12Cr-17Ni alloy. The other alloys had basically the same shape but a greater height. The region on figure 8 labeled as optimum represents the range of nitric acid concentrations over which the pickling of 304SS should be opfimized. At 50° C that range is 0.8M to 1.3M HNO3 in solu- tions with 0.5Af HF. It is in this range that the dissolution of the chromium-depleted region will be maximized and the dissolution of bulk 304SS will be minimized. It can be seen that being to the left of this region would sacrifice base metal for dissolution (pick- ling) speed while being to the right will result in a very slow pickling. The other factors to consider are temperature and HF concen- tration. The HF concentration should linearly increase pickling rate and loss of base metal so that it is best to use as low a concentra- tion as practical. Choice of temperature is another variable that is determined somewhat by practicality. For the activation energies reported here, there is an order of magnitude increase in dissolu- tion (pickling) rate of 304SS for approximately each 20 ° to 40 ° C rise in temperature. The dissolution rate data can also be used to develop an equa- tion describing the overall response of dissolution rate to time, tem- perature, and acid concentrations. The equation would take the form of a heterogeneous catalysis rate equation and be similar to the fol- lowing equation: Dissolution rate = k [HFp [HN03]« 1 -I- [HF]" [HN03]= exp (-AH/RT), where k, a, B, and R are constants, AH is the activation energy, T is the absolute temperature, and [HF] and [HNO3] are acid molar- ities. Such as equation will help to give a better insight into the dissolution mechanism (depending on the fitted values of a, B, and AH) and can also be used to predict dissolution (pickling) rates for acid concentrations and temperatures between those actually measured. Equations such as that described for 304SS and for the three experimental alloys can be used further to model the amount of metal lost during the pickling reaction. This would be accomplished in terms of the overall basic assumption made in this paper. That is, that pickling occurs by a dissolution of the chromium-depleted region, undercutting the scale and thus removing it. The modeling would proceed by assuming a thickness of chromium depletion, an amount of scale formed during annealing, a preset pickling time, and the number of pickling-working-annealing steps in the life of a coil of stainless steel. The result would be the minimum amount of time needed to remove the chromium-depleted region (and thus the scale) and the minimum amount of metal lost (and metal lost unnecessarily). There are certain quantities that would make this predictive model more accurate. The quantities that are relatively unknown at present are represented by the chromium-depleted region and the annealing scale. The needed information would be in the form of equations expressing the effect of annealing time, temperature, and oxygen content of the environment on both the chromium- depleted region and the oxide scale composition and thickness. These would be rewarding areas in which to conduct research. HNO3 CONCENTRATION ► Figure 8.— Schematic of optimum range of HNO3 concentra- tions for piclding 304SS in HNO3-HF solutions. 44 FUTURE WORK Additional work in this area at the Bureau's Albany (OR) Research Center will consist solely of data analysis. Data analysis will be completed and equations developed to model the dissolu- tion rate as a function of acid concentrations and temperature. These will then be used to model the loss of metals during pickling with the hope of being able to best predict optimum conditions for pick- ling stainless steels. This will also include completing the analysis of work on 430SS and experimental alloys used to represent its chro- mium depleted region. CONCLUSIONS 1. Increasing hydrofluoric acid concentration causes a linear increase in the dissolution rate of 304SS. 2. Increasing nitric acid concentration up to 0AM to 1.5M HNO3 causes an increase in the dissolution rate of 304SS, and a decrease in the dissolution rate for higher HNO3 concentrations. 3. Dissolved iron and chromium reduce the dissolution rate of 304SS in HNO3-HF solutions. Dissolved nickel has little or no effect. 4. The surface films present on 304SS in HNO3-HF solutions did not change as a function of dissolution rate and contained a frac- tion of a monolayer of fluoride on the outer surfaces of the films. 5. Optimum pickling conditions exist at nitric acid concentra- tions of O.SMto I.3MHNO3 at 50° C in solutions with 0.5MHF. REFERENCES 1. Fabis, P. M., and B. S. Covino, Jr. Near Surface Elemental Con- tration Gradients in Annealed 304 Stainless Steel as Determined by Ana- lytical Electron Microscopy. Oxid. Met., v. 25, Nos. 5/6, 1986, pp. 397-407. 2. Muralidharan. V. S.. and K. S. Rajagopalan. Kinetics and Mecha- nism of Corrosion of Iron in Phosphoric Acid. Corros. Sci., v. 19, 1979, pp. 199-207. 3. Alexander, B. J., and R. T. Foley. Anion Dependence of the Acti- vation Energy for Iron Corrosion. Corrosion, v. 31, No. 4, 1975, pp. 148-149. 4. Altura, D., and K. Nobe. Activation Energy for the Corrosion of Iron in Sulfuric Acid. Corrosion, v. 20, No. 11, 1973, pp. 433-434. 5. Makrides, A. C, and N. Hackerman. Solution of Metals in Aque- ous Acid Solutions. II Depolarized Solution of Mild Steel. J. Electrochem. Soc, V. 105, 1958, pp. 156-162. 6. Riggs, O. L. Activation Energy from Carbon Steel Corrosion in Phos- phoric Acid. Corrosion, v. 24, No. 5, 1968, pp. 125-126. 7. Covino, B. S., Jr., J. P. Carter, and S. D. Cramer. The Corrosion Behavior of Niobium in Hydrochloric Acid Solutions. Corrosion, v. 36, No. 10, 1980, pp. 554-558. 8. Ishikawa, T., and G. Okamoto. Potentiostatic Response of Passive Metals to the Rate of Temperature Change. Electrochimica Acta, v. 9, 1964, pp. 1259-1268. 9. Asami, K., and K. Hashimoto. An X-ray Photoelectron Spectroscopic Study of Surface Treatments of Stainless Steels. Corros. Sci., v. 19, 1979, p. 1007. 10. Wilding, M. W., and B. E. Paige. Survey on Corrosion of Metals and Alloys in Solutions Containing Nitric Acid. Allied Chemical Corp., ICP-1107, 1976, 56 pp. 11. Smith, R. H., and A. E. Martell. Critical Stability Constants, Vol- ume 4— Inorganic Complexes. Plenum (New York), 1976, pp. 96-103. 12. Pourbaix, M. Atlas of Electrochemical Equilibria in Aqueous Solu- tions. Pergamon (New York), 1966, p. 587. 13. Lochel, B., and H. H. Strehblow. Breakdown of Passivity of Iron by Fluoride. Electrochim. Acta, v. 28, No. 4, 1983, pp. 565-571. 14. Lochel, B. P., and H. H. Strehblow. Breakdown of Passivity of Nickel by Fluoride. II Surface Analytical Studies. J. Electrochem. Soc, v. 131, 1984, p. 713. 45 DECREASED ACID CONSUMPTION IN STAINLESS STEEL PICKLING THROUGH ACID RECOVERY By G. L. HorteM and J. B. Stephenson^ ABSTRACT Acid pickling of stainless steel annually generates approximately 30 million gal of spent nitric acid-hydrofluoric acid solutions as a byproduct. Disposal of these acid solutions significantly increases the cost of manufacturing stainless steel and results in the loss of the acid, Cr, and Ni values. Recent advances in ion-selective membrane technology have opened new avenues to regenerate these acid solutions as an alternative to disposal. Experimental work by the Bureau of Mines has indicated that an electrodialysis cell utilizing ion-selective membranes has potential for separating dissolved metals from spent pickling acid solutions, while regenerating the acids for return to the pickling process. INTRODUCTION The Bureau of Mines is conducting research to conserve mineral values and reduce waste generation from spent nitric acid- hydrofluoric acid pickling solutions. The stainless steel industry annually generates approximately 30 million gal of these solutions, which currentiy have to be neutralized and sent to waste disposal (4).' Hot working and annealing of stainless steels cause a strati- fied scale to form on the surface of the steel . The scale consists of metal oxides in the top layer, an intermediate layer of spinel oxide [FeO(Fe(2-x)CrJ03, where 070 pet passed a 35-mesh screen— usually two passes through the rolls. Only the minus 35-mesh fraction was blended into the pellet mix used in smelting tests; the oversize fraction was saved and added with the pellets as furnace charge. A flow diagram of this procedure is presented in figure 1. The four-waste mixture was pelletized in a 36-in-diam drum pelletizer. The blended mix required the addition of about 12 pet water in order to form %- to %-in-diam pellets. The pellets produced were first air dried for 24 h and then oven dried at 250° F for 6 h. Drying at higher temperatures tended to result in spalling. The pellets had 5- to 30-lb crushing strength, which was sufficient for the limited amount of handling required. When the wastes listed in table 1 were mixed in the indicated proportion and pelletized, the resultant pellets plus oversize mill scale analyzed roughly, in percent, 10 Cr, 4 Ni, 1 Mo, and 2 Mn. Experience showed, however, that it is not always necessary to oven dry the pelletized waste mix. Most of the laboratory heats made in the latter part of the testing program were made using pellets that were air dried only. The experiments included adding pellets to the furnace that were only 2 h old and very wet. These were rolled at a uniform rate down a conveyor and dropped through the furnace top onto the melt surface. As much as 13 pet of the melt charge (the most tried) was added in this manner without problem. Other efforts were made to simplify the pelletizing procedure. For some trials, the cement binder was completely eliminated from the pellet mix (could decrease slag volume). No difficulty was 52 Col 90 pet. Metal Value of Pelletized Wastes Technically and mechanically, this recycling scheme has been shown to work well. The question naturally arises as to whether it is also economical. The Bureau's Process Evaluation Group con- ducted internal studies of capital and operating costs for a plant addi- tion producing pellets from wastes such as flue dusts, mill scale, and/or oily swarf. The estimated fixed capital cost for a 15-st/d pelletizing capacity was approximately $974,000 for oven drying of pellets and approximately $560,000 for air drying. Estimated annual operating costs per ton of pellets based on one-shift-per- day, 5-day -per-week operation (20-yr life) was approximately $117/st for oven drying and $40/st for air drying. This can be contrasted to the contained value of the Cr, Ni, Mo, and Mn in the pellets. In addition to the particular propor- tions of the metals in the wastes, the value depends on the current price of ferroalloys or of appropriate scrap for which the pellets Table 8.— Recovery of Cr, Ni, and Mo from commercial pellet-plus-scrap heats 3 through 7, percent Heat Type of heat Cr Ni Mo 3 . . 4 . . 5 .. 6 .. 7.. Scrap plus 14 pet low-scale pellets. . . . Scrap plus 19 pet low-scale pellets. . . . Scrap plus 15 pet high-scale pellets. . . Scrap plus 14 pet high-scale pellets. . . Scrap plus 19 pet high-scale pellets. . . 93.0 93.7 97.3 90.8 91.4 99.7 89.7 99.1 91.9 92.3 99.9 95.0 93.2 82.8 84.5 may substitute, particularly stainless steel (18 Cr-8 Ni) scrap. In some periods, a charge is also added for Fe units; this has not been the case when the scrap market is relatively depressed. On the basis of the Cr, Ni, Mo, and Mn for a waste mixture similar to those described for low-scale pellets (in percent, 9.5 Cr, 4.0 Ni, 0.8 Mo, and 2.0 Mn), with approximate contained metal values of $0.42/lb for Cr, $2.10/lb for Ni, $3.20/lb for Mo, and $0.33/lb for Mn (mid- 1987 ferroalloy contained-metal values), the pellets would have a value of over $3I0/st. Deducting the approximate net operating cost of $1 17/st for oven drying or $40/st for air drying indicates a net gain of some $0.10 or $0.14/lb, respectively— a significant economical potential. CONCLUSIONS It has been shown that stainless steelmaking wastes such as flue dusts, mill scale, and grinding swarf can be pelletized and reduced in the arc furnace. This provides a means of recovering the con- tained scarce and valuable metals, while coincidentally solving prob- lems of storage and waste disposal. The recovery procedure utilizes the reduction of metal oxides with carbon during the arc furnace melting, followed by a scavenging slag reduction of additional chro- mium oxide with ferrosilicon or Al. One variation of the processing involves preparation of all-pellet smelting heats to produce master alloy ingot for recycle. This may be appropriate if furnace capacity is available in slack periods and large waste backlogs exist. Alternatively, and recommended as being more economical, the waste-bearing pellets can be added directly to the arc furnace as some 10 to 20 pet of the total charge for production heats in lieu of part of the usual scrap or alloy charge required. The addi- tion rate will depend on factors such as the rate of waste genera- tion, waste backlog accumulation, and alloy product mix at a particular plant. The dusts, scale, and swarf can be mixed and pelletized with little difficulty, providing both a means for adding carbon to the mix and a vehicle for charging to the furnace. Numerous varied combinations of pellet mix have been shown to be possible. Only conventional equipment is needed for agglomeration. Usual recoveries of substantially greater than 90 pet of the Ni and Fe have been attained, and some 90 pet of the Cr and Mo appear consistently recoverable with proper control of variables. Other metals such as Mn are coincidentally recovered. Conventional arc furnaces were used throughout the testing. The fact that this technology is readily transferable to an indus- trial scale was shown by the successful making of a number of demonstration heats ranging in size from 12.5 st for an all-pellet heat to about 19 st for a series of pellet-plus-scrap heats. The mas- ter alloy ingot from the all-pellet demonstration heat was used to make commercial stainless steel products without difficulty. No problems were encountered in these commercial stainless produc- tion heats to which up to 19 pet pellets were added in lieu of the normal scrap charge. When the particular waste combination outlined in this paper is pelletized for recycle as a scrap substitute charge material, the pellets have a net value of more than $O.IO/lb for oven drying or $0.14/lb for air drying, which implies an economically viable process. Test results also indicate that a wide compositional variation of specialty steelmaking wastes can be incorporated into pellets for furnace charging. 57 REFERENCES 1. Federal Register. U.S. Environmental Protection Agency. Hazard- ous Waste Regulations. V. 45, No. 98, 1980, p. 33127. 2. Price, L. E. Tensions Mount in the EAF Dust Bowl. 33 Met. Produc- ing, Feb. 1986, pp. 38-41. 3. Barnard, P. G., A. G. Starliper, W. M. Dressel, and M. M. Fine. Recycling of Steelmaking Dusts. BuMines TPR 52, 1972, 10 pp. 4. Dressel, W. M., P. G. Barnard, and M. M. Fine. Removal of Lead and Zinc and the Production of Prereduced Pellets From Iron and Steel- making Wastes. BuMines RI 7927, 1974, 15 pp. 5. Higley, L. W., Jr., and M. M. Fine. Electric Furnace Steelmaking Dusts— A Zinc Raw Material. BuMines RI 8209, 1977, 15 pp. 6. Higley, L. W., Jr., and H. H. Fukubayashi. Method for Recovery of Zinc and Lead From Electric Furnace Steelmaking Dusts. Paper in Proceedings of the Fourth Mineral Waste Utilization Symposium. IIT Res. Inst., Chicago, IL, 1974, pp. 295-302. 7. Powell, H. E., W. M. Dressel, and R. L. Crosby. Converting Stain- less Steel Furnace Flue Dusts and Wastes to a Recyclable Alloy. BuMines RI 8039, 1975, 24 pp. 8. Barnard, P. G., W. M. Dressel, and M. M. Fine. Arc Furnace Recy- cling of Chromium-Nickel From Stainless Steel Wastes. BuMines RI 8218, 1977, 10 pp. 9. Higley, L. W., Jr., L. A. Neumeier, M. M. Fine, and J. C. Hart- man. Stainless Steel Waste Recovery System Perfected by Bureau of Mines Research. 33 Met. Producing, Nov. 1979, pp. 57-59. 10. Higley, L. W., Jr., R. L. Crosby, and L. A. Neumeier. In-Plant Recycling of Stainless and Other Specialty Steelmaking Wastes. BuMines RI 8724, 1982, 16 pp. 11. Neumeier, L. A., and M. J. Adam. In-Plant Recycling of Chromium- Bearing Specialty Steelmaking Wastes. Paper in Chromium-Chromite: Bureau of Mines Assessment and Research. Proceedings of Bureau of Mines Briefing Held at Oregon State University, Corvallis, OR, June 4-5, 1985, BuMines IC 9087, 1986, pp. 85-91. 12. Pargeter, J. K. Operating Experience With the Inmetco Process for the Recovery of Stainless Steelmaking Wastes. Paper in Proceedings of the Seventh Mineral Waste Utilization Symposium. IIT Res. Inst., Chicago, IL, 1980, pp. 118-126. 13. Fosnacht, D. R. Recycling of Ferrous Steel Plant Fines: State-of- the-Art. Iron and Steelmaker, v. 8, No. 4, Apr. 1981, pp. 22-26. 14. Peckner, D., and I. M. Bernstein. Handbook of Stainless Steels. McGraw-Hill, 1977, pp. 3-1-3-35. 15. Mueller, C. P. Recovery of Metallics From Specialty Steel Slags and Wastes. Pres. at AISI Symposium on Recovery of Alloys From Spe- cialty Steel Wastes, Pittsburgh, PA, Oct. 21-22, 1981; information availa- ble from International Mill Service, Inc., Philadelphia, PA. 16. Lehigh University. Characterization, Recovery and Recycling of Elec- tric Arc Furnace Dusts. Final rep. prepared for U.S. Dep. Commerce under project 99-26-09885-10, Feb. 1982, 313 pp.; NTIS PB 82-182585. 58 USING WASTES AS A SOURCE OF ZINC FOR ELECTROGALVANIZING By V. R. Milleri ABSTRACT The Bureau of Mines investigated the use of Zn extracted from electric arc furnace (EAF) dust as a source of Zn for electrogalvanizing. The prepared sulfate electrolytes were used to coat steel sheet in flow cells at current densities up to 150 A/dm^ to provide 90-g/m2 Zn deposits. Elec- trochemical and physical properties of waste-derived coatings were compared with those of coat- ings produced from electrolytes prepared from pure ZnO and from an industrial process. These studies showed the properties to be similar in most cases. INTRODUCTION A dramatic increase in the demand for electrogalvanized steel has resulted from the construction, appliance, and especially the automotive industry needs (1).^ The amount of slab zinc used in 1986 for galvanized steel was estimated to be 47 pet of the total slab zinc consumption of 975,000 mt. An estimated 73 pet or 710,000 mt of this slab zinc was imported. Over 20 yr ago, it was estimated that 235,000 st of zinc could be recovered from domes- tic stack dusts (2). In 1980, 75,000 st of Zn was contained in EAF dust alone (3). This quantity of Zn was contained in about 400,000 st of EAF dust which has been declared a hazardous waste by the Environmental Protection Agency (4). Several studies and surveys have been carried out in recent years to examine the options for recovery of resources from EAF dust. Several processes, both pyrometallurgical and hydrometallurgical, have been proposed for treating EAF dust and these have been well documented in the literature. The bulk of these processes were aimed at recovering pure metals such as Zn. In the case of Zn, this requires stringent purification to produce electrolytes with impurities in the parts per billion range. However, for electrogalvanizing, some of the impurities have been shown to be beneficial (5). Cobalt and chromium at low concentrations in electrogalvanizing baths have produced more corrosion-resistant coatings and Cd has enhanced wire drawability. The Bureau of Mines demonstrated in previous research the feasibility of using Zn recovered from brass smelter flue dust to electrogalvanize steel wire in an industrial pUot plant (6). This paper describes research on electrogalvanizing steel sheet with electro- lytes produced from EAF dust and an EAF dust oxide fume produced from flash smelting the dust. The oxide fume differs from the EAF dust in that it contains the more volatile metals. Solutions from leaching the two products were purified as needed for elec- trogalvanizing and used in flow cells designed to simulate the hydrodynamics of an industrial line. Properties of deposits made with waste dust electrolytes were compared with the properties of deposits made using pure ZnS04 as well as industrial deposits. EXPERIMENTAL WASTE MATERIAL AND ELECTROLYTE Table 1 contains a partial chemical analysis of the dusts used in the experimental work. The electrolyte from the EAF dust was prepared by mixing concentrated H2SO4 with the dust to sulfate it, then water leaching. The leach step involved mixing the sulfated dust with de-ionized water at 90° C for 1 h. The mixture was filtered; the liquor was reserved for electrolyte purification and the filter cake was combined with pure water for the wash step. The Table 1.— Partial chemical analysis of dusts used, weight percent ' Supervisory research physicist, Rolla Research Center, Bureau of Mines, Rolla, MO. ^ Italic numbers in parentheses refer to items in the list of references at the end of this paper. Element Oxide fume from flash smelter Zn . Fe. Pb . Cd . Cu . Mn Mg, Ca . CI. F.. Cr. Electric arc furnace (EAF) dust 35.4 31.3 7.5 19.6 6.5 5.2 .8 .1 .6 .2 .9 2.8 .4 1.3 .9 3.6 6.8 3.5 2.5 .1 .2 .3 wash step consisted of 30-min agitation at ambient temperature fol- lowed by filtration. The filter cake was then set aside for sampling and analysis. The wash water was returned to the leach reactors for starting the leach cycle on the next batch of sulfated dust. Typical combined extractions of Zn and Fe in the two steps were 92 and 11 pet, respectively. The oxide fume electrolyte was prepared by direct H2SO4 leach- ing of the fume. The procedure was similar to that for the sulfated dust except that 450-g/L H2SO4 solution was used rather than de- ionized water. The wash step utilized pure water that was used, following filtration, for making the next acid leach solution. Typi- cal leach liquor compositions are listed in table 2. The necessity of a degree of purification for the solutions is also evident when table 2 is examined in light of the need for low levels of Fe, Cu, Cd, CI, and F. Table 2.— Typical chemical analyses of leach solutions from oxide fume from flash smelter and electric arc furnace dust, grams per liter Element Oxide fume EAF dust Zn . Fe. Cu Cd CI. F.. Pb . 58 155 35.3 10.1 3.0 <.01 .5 .4 11.7 18.6 5.2 .6 <.01 <.01 The Fe concentration was reduced to less than 0.1 g/L using a phosphate purification technique. The technique consisted of reducing the pH to 1.0, oxidizing the Fe with hydrogen peroxide, adding H3PO4 as a phosphate ion source, and raising the pH to 1.9 with lime. The Fe was subsequently precipitated as readily filtera- ble FeP04. The CI levels were reduced using CUSO4 and metallic Cu powder at 90° C, pH 2, for 2 h to form CuCl. Copper, cad- mium, and lead were removed by cementation with Zn dust. Table 3 lists the chemical compositions of the electrolytes that were used to electrogalvanize the steel sheet. The CI content of the oxide fume was high because of insufficient purification, which was not detected because of analytic error. It was detected when the electrolyte was rechecked after electrogalvanizing. Table 3 includes the pure zinc sulfate electrolyte that was prepared by dissolving French process ZnO in H2SO4 and H2O. Zinc dust was added to the hot solution to insure the removal of Cd and Pb. Table 3.— Electrolyte compositions used for electrogalvanizing steel sheet, grams per liter Element Pure ZnSOj EAF dust Oxide fume Zn . Fe. Cu . Cd . Pb . CI. Co. Ni. Mn. 99 91 92 <.001 .025 .050 <.001 .004 .004 <.001 <.001 .002 <.002 .002 .002 <.001 .035 1.510 <.001 <.001 .230 <.001 <.001 .140 <.001 2.94 1.41 59 vent, rinsed with ethanol, and dried in an airstream. The sheets were weighed to the nearest 0.0001 g. The sheets were then elec- trocleaned anodically in commercial alkaline cleaner for steel. Experimentation showed no significant change in weight, < 0.001 g, before and after electrocleaning the sheets. The procedure follow- ing electrocleaning was to rinse the sheets in water, dip them in a 10-pct H2SO4 solution for 15 s to activate the surface, rinse in H2O, place in the flow cell, and start the flow of electrolyte. ELECTROGALVANIZING Electrogalvanizing was done in two different size flow cells that used pumps to move the electrolyte past the stationary elec- trodes to simulate a moving line. The small cell used 4- by 5-cm sheet to yield a coated area of 20 cm^. The anodes were Pb-1 pet Ag and were also 4 by 5 cm in size. The electrode spacing was 9 mm and with available flow rates, velocities at the cathode sur- face of up to 6 m/s were possible. Face velocities were used for flow rates to allow comparison with industrial electrogalvanizing line speeds. Five liters of electrolyte was required for operation of the pump to produce the desired flow rates. The large cell coated 10- by 20-cm sheets and required 100 L of electrolyte. It also used Pb-1 pet Ag anodes with the electrode spacing being 9 mm. Time and current were adjusted for the cur- rent densities of 50 and 150 A/dm^ to produce coatings of about 90 g/m^. After the sheets were electrogalvanized, they were rinsed in H2O, then ethanol, dried in air, and weighed to determine the amount of coating. EXPERIMENTAL DESIGN In conducting the research on the pure zinc sulfate electrolyte, the variables that were considered were current density, tempera- ture, acid concentration (pH), Zn concentration, and face velocity of the electrolyte. Table 4 lists the variables and the upper and lower limits used in the experiments. A factorial design was used to gain the maximum information with a minimum of experimental work. The responses analyzed were current efficiency and preferred crys- tallographic orientation. Selected specimens were also subjected to electrochemical corrosion evaluation in addition to formability testing. Table 4. — Experimental variables and test ranges Variable Range Current density A/dm^. . . 50-150 Temperature °C . . . 30-50 Acid concentration pH . . . 4.0-1 .5 Zn concentration g/L. . . 90-150 Flow rate m/s . . . 2-5 The deposits prepared with electrolytes made from the waste dusts were evaluated in a similar manner using a factorial design. The two parameters, acid and Zn concentration, however, were held at 1.5 pH and 90 g/L, respectively. STEEL PREPARATION The steel sheet was supplied by Inland Steel and was 0.79-mm thick AKDQ alloy. As received, it had been sheared to size for the flow cells and oiled. The specimens were degreased with sol- ELECTROCHEMICAL CORROSION EVALUATION Coupons were stamped from the electrogalvanized sheet to fit a flat specimen holder and yield a 1-cm^ surface area. The coupons were degreased in boiling trichloroethylene, rinsed in ethanol, and dried in a filtered airstream. 60 The cell used was a commercially available 1-L glass vessel with various necks for electrodes, gas inlet and outlets, and ther- mometer (7). The counterelectrode was Pt mesh, and the reference was a saturated calomel electrode. The medium was analytical rea- gent grade (NH4)2S04 of 1 mol/L concentration and pH of 6+0. 1 . Prior to sample immersion, the medium was deaerated with oxygen- free N, and the gas purge continued throughout the experiments. All experiments were carried out at 25° + l ° C. Impedance measurements were made at the open circuit poten- tial (OCP) using a lock-in amplifier over a frequency range of 10 to 20,000 Hz having a peak amplitude of 5.15 mV (3.64 mV RMS). Measurements from 0.005 to 1 1 Hz were obtained in the time domain by an EG&G' fast fourier transform technique. The soft- ware performs the experiment and calculates the data points. The samples were immersed in the test medium for 1 h before measur- ing the electrochemical impedance. FORMABILITY TESTS Both compression and tension bend tests were used to evalu- ate the deposits. The compression samples were bent 180°, with the coating on the inside of the bend, and then straightened. Scotch brand tape was applied to the distorted area and removed. The rela- tive amount of coating powdering was recorded. The tension sam- ples were bent 180°, with the coating on the outside of the bend, and examined under low magnification to determine if peeling or flaking occurred in the deformed area. Then, the specimen was bent repeatedly back and forth over a mandrel until the steel fractured. The fracture area was examined under low magnification for sepa- ration or peeling of the coating. Prying with a sharp knife was used to indicate unsatisfactory adhesion by lift off of the coating (8). In addition to bend tests, drawing and ball punch deformation tests were conducted on coated sheet. Round 92-mm-diam blanks were punched from coated sheet and drawn into flanged cups approximately 43 mm in diameter and 42 mm deep. The ball punch deformation test was conducted in accordance with ASTM E643-78 (9) procedures. As an added indication of adhesion, a reverse ball punch test was also conducted on electrogalvanized sheet. In this test, the sheet was placed on the ram of the ductility tester, Zn side down, to form a 7.8-mm cone. The specimen was then removed and turned over with the top of the cone centered on the ram. The cone was then pushed back through the plate for a total of 14 mm. The condition of the Zn coating in the deformed area was then exam- ined for flaking or peeling. RESULTS AND DISCUSSION CURRENT EFFICIENCY AND ORIENTATION Current efficiency was calculated from the amount of electric charge used for a given sample and the weight of the deposit. Three specimens were coated for each set of conditions to obtain an aver- age value. In the experimental work using the pure zinc sulfate elec- trolyte and small flow cell, the current efficiencies ranged from 95.2 to 99.0 pet. Thirteen out of the sixteen experimental combi- nations produced current efficiencies within 1 pet of each other in this range. This indicates little difference in the effects of the vari- ables over the test ranges in table 4. Similar results were obtained with the electrolytes prepared from the oxide fume and the EAF dust. The current efficiencies were in the same range of upper 90's values indicating no serious decrease due to the impurities present, especially the CI. High levels of CI did result in increased attack of the Pb-Ag anode producing nonconducting surface coatings, which resulted in higher voltages to hold the desired current settings. The percentage of a particular crystallographic orientation that was used in the factorial analysis was arrived at by dividing the total count in the X-ray diffraction peak for that crystallographic plane by the total count for all planes and multiplying by 100. Table 5 lists some typical results obtained for selected conditions. For the pure Zn electrolyte, the predominant orientations were 002 and 103. It was noted that current density and Zn concentration played only a very minor role in either the orientation or current efficiency analysis. Deposits produced from oxide fume electrolytes were differ- ent than those from the pure Zn electrolyte in that little (002) orien- tation was produced under any of the conditions. The majority of the deposits were 1 12 and 101 . The same was true for the electro- lyte from the EAF dust where there were higher percentages of all orientations other than 1 12 and 101 . Examples of typical deposits are shown in the SEM photomicrographs in figure 1 . The grain size of the deposits produced from waste electrolytes were also smaller than those from pure zinc oxide and from the industrial sample. Preferred orientations of the industrial sample used for com- parison were primarily 104, 103, 004, and 002, in order of descend- ing percentage. Similar orientations were obtained for deposits prepared from pure electrolyte in the large cell. ELECTROCHEMICAL CORROSION Table 6 contains the ac data (electrochemical impedance spec- troscopy) for selected samples from the tests on electrogalvanized samples. The values of polarization resistance (Rp) and double- layer capacitance (C^l) have been shown in previous research (10) to be useful in monitoring the performance of electrogalvanized Table 5.— Preferred crystallographic orientation of coated deposits prepared from different electrolytes at 500 A/dm^, 90 g/L Zn, 1.5 pH, 60° C, 2 m/s, percent of total Deposit source 002 004 101 102 103 104 110 112 Industrial 10 20 28 36 Pure ZnO 73 6 2 14 2 EAF dust 14 26 11 17 13 9 Oxide fume . . . 39 4 1 5 45 Table 6.— Alternating current electrochemical impedance data for electrogalvanized steel in 1M (NH4)2S04 at 25° C after 1 h ^ Reference to specific products does not imply endorsement by the Bureau of Mines. Polarization Double-layer Deposit Plating current resistance, capacitance, source density, A/dm^ (Rp), ncm2 (C^l), /iF/cm2 Industrial 60 1,600 15 Pure ZnO 50 1,625 17 150 800 28 EAF dust 50 1,200 13 150 600 20 Oxide fume .... 50 800 28 150 500 42 61 ""W^ mv^- t. 19 V.:-^- '..^^ ' :-.^*> ..^^'^r^ 2 .1 ■is ^ >" i*» Figure 1 .— SEM photomicrographs of electrogalvanized samples. (A) Industrial sample and samples electrogalvanlzed with (B) pure ZnO electrolyte, {C) EAF dust electrolyte, and (D) oxide fume electrolyte. wire. It demonstrated that the corrosion rate of electrogalvanized steel was related to the ac impedance value of the polarization resis- tance in deaerated molar (NH4)2S04 under near-neutral conditions. The value of the double-layer capacitance was also reported to be related to the corrosion rate and to the surface condition of the metal coating. The corrosion rate (icorr^) varies inversely as the polari- zation resistance. Thus, the larger the value of Rp the lower the corrosion rate. Low values of Cjl are usually obtamed when there are surface films such as oxides or hydroxides of Zn and when the deposit is slowly corroding. Thus, low values indicate low corro- sion rates when accompanied by large values of Rp. It is seen in examining the electrochemical data in table 6, with the preceding in mind, that the corrosion rates are similar for the industrial, pure ZnO and EAP dust derived deposits that were plated at the lower current density. The deposit from the oxide fiime exhibits a higher corrosion rate, which may be the result of the high CI level in the electrolyte and the increased grain boundary area due to smaller grain size. The data for the high-current-density deposits show what would amount to an increase in corrosion rate over the low-density plated samples. The EAF dust derived and the pure ZnO deposits are simi- lar, with a possible advantage to the EAF dust deposit. Again, the 62 oxide fume exhibits values indicating higher corrosion rates over the pure ZnO and EAF dust deposits. Results of salt spray corrosion tests by an independent labora- tory on pure ZnO deposits indicated a similar trend. Only current density was significant in influencing salt spray results. Deposits plated at 50 A/dm^ produced ASTM ratings of 8 to 9.5 (11), while those at 150 A/dm^ ranged from 4 to 6. The apparent reason is that current density influences nucleation and growth during elec- trodeposition and has a pronounced effect on grain size, with more active grain boundary material becoming predominant. FORMABILITY TESTS The results of selected bend tests are shown in table 7. All speci- mens bent in compression and straightened exhibited a few small cracks. However, only one specimen had some coating material removed with the tape. That sample was prepared with pure ZnS04 at a current density of 150 A/dm^. The tension bend tests resulted in small cracks on the indus- trial and oxide fume deposits. All others exhibited smooth bends. The examination of the coatings, after they were repeatedly bent through 180° until failure of the steel, indicated no separation or peeling of the coatings. Attempts to pry the coatings loose with a knife edge were unsuccessful, indicating good adhesion. Microscopic examination of the drawn cup samples showed no cracking, flaking, or peeling of the Zn coatings, which indi- cates good ductility and adhesion. Examples of the drawn samples are shown in figure 2, which illustrates this. The ball punch deformation test, used to evaluate the forma- bility of sheet materials, was also used on the coated sheet. Sam- ples plated from pure electrolyte at 60° C had cup heights and maximum loads similar to bare steel and commercial electrogal- vanized sheet, while those plated at 35° C, with coarser-grained deposits, had lower cup heights and loads. Ball punch tests on sam- ples prepared from EAF dust electrolyte indicated good formabil- ity as evidenced by cup heights and maximum loads similar to industrial samples. No peeling or flaking of the coating occurred in the area-of-rupture as shown in figure 3. The reverse ball punch specimen shown in figure 4A is from a commercially coated sheet. The inner "ring," and just above it, is the area of severest deformation. There was no cracking, peel- ing, or flaking of the coating in this area. The specimens prepared with pure electrolyte were very similar to the commercial speci- men. Figure 4fi shows the EAF specimen, which exhibits some cracking at the ring. However, it was not possible to peel the coat- ing at the cracks using a sharp knife point. This amount of crack- ing would not prevent the coating from being acceptable. Table 7. — Bend test results on electrogalvanlzed steel sheet Deposit Plating current Compression Tsnsion Bend rupture source density, A/dm^ Crack Powdering cracks Bends Peeling Industrial .... 60 Yes . No Small . 12 No. Pure ZnO ... 50 Yes No No ... . 10 No. 150 Yes. Moderate. No . . . . 9 No. EAF dust 50 Yes . No No ... . 9 No. 150 Yes . No No 10 No. Oxide fume.. 50 Yes. No Small. 10 No. 150 Yes . No No 11 No. CONCLUSIONS It has been shown that Zn from steel wastes such as EAF dusts can be used to successfully electrogalvanize steel sheet. Purifica- tion of the leach liquor is necessary, however, to reduce coextracted impurities Fe, Cu, Cd, and CI. Electrogalvanizing tests in flow cells using pure and waste-derived electrolytes yielded good deposits at comparable current efficiencies which were in the 94- to 99-pct range. Deposits from waste electrolytes exhibited smaller grain size and less basal plane orientation than pure ZnS04 electrolyte. Electrochemical corrosion evaluation of coated deposits showed that deposits from EAF dust, pure ZnO, and an industrial line have similar corrosion rates, while deposits from oxide fume with high CI content have higher corrosion rates. Electrochemical data corre- late with salt spray corrosion tests in showing that current density influences corrosion rate. Bend, draw, and ball punch tests on coated steel sheet prepared from pure ZnO, EAF dust, and bend tests on oxide fume derived deposits indicated that the adhesion and ductility of the coatings were as good as those of an industrial sample. 63 Figure 2.— Cups drawn from electrogalvanized samples. (A), Industrial sample and cups drawn from samples electrogalvanized with (B) pure ZnO electrolyte and (C) EAF dust electrolyte. 64 B i Figure 3.-Specimens from ball punch tests of electrogalvanized samples. (A) Industrial sample and (B) sample from EAF dust electrolyte. A 65 -'i^l B ,^ y y dus^elect^ytr''"'"' ''°'" '''"'' ''" P""^*^ ^^^*^ °" electrcgalvanized samples. W Industrial sample and iB) sampi e from EAF 66 REFERENCES 1. Jolly, J. H. Zinc. Sec. in BuMines Mineral Commodity Summaries 1987, pp. 180-181. 2. Carrillo, F. V., M. H. Hibpshman, and R. D. Rosenkranz. Recov- ery of Secondary Copper and Zinc in the United States. BuMines IC 8622, 1974, 58 pp. 3. Krishnan, E. R., and W. F. Kenner. Recovery of Metallic Values From Electric Arc Furnace Steelmaking Dusts. Paper in Proceedings of Symposium on Iron and Steel Pollution Abatement Technology for 1982. EPA Center for Environ. Res. Inf., Research Triangle Park, NC, 1983, 687 pp. 4. U.S. Code of Federal Regulations. Title 40— Protection of Environ- ment; Chapter I— Environmental Protection Agency; Subchapter I— Solid Wastes. Part 261— Identification and Listing of Hazardous Waste; Subpart D— List of Hazardous Wastes, July 1, 1983. 5. Adaviya, T., M. Omura, K. Matsudo, and H. Naemura. Develop- ment of Corrosion-Resistant Electrogalvanized Steel. Plat. Surf. Finish., v. 68, No. 6, 1979, pp. 96-99. 6. Dattilo, M., E. R. Cole, Jr., and T. J. O'Keefe. Recycling of Zinc Waste for Electrogalvanizing. Conserv. & Recycling, v. 8, No. 3-4, 1985, pp. 399-409. 7. American Society for Testing and Materials. Standard Recommended Practice for Standard Reference Method for Making Potentiostatic and Poten- tiodynamic Anodic Polarization Measurements. G5-78 in 1982 Annual Book of ASTM Standards. Part 10 — Metals — Mechanical, Fracture, and Corro- sion Testing; Fatigue; Erosion and Wear; Effect of Temperature. Philadel- phia, PA, 1982, pp. 906-916. 8. . Standard Test Methods for Adhesion of Metallic Coat- ings. B571-79 in 1982 Annual Book of ASTM Standards. Part 9, Metallic and Inorganic Coatings; Metal Powders, Sintered P/M Structural Parts. Philadelphia, PA, 1982, pp. 419-422. 9. .Standard Methods for Conducting a Ball Punch Defor- mation Test for Metallic Sheet Material. E643-78 in 1982 Annual Book of ASTM Standards. Part 10, Metals — Mechanical, Fracture, and Corro- sion Testing; Fatigue, Erosion and Wear; Effect of Temperature. Philadel- phia, PA, 1982, pp. 758-761. 10. Dattilo, M. The Use of AC Impedance to Determine the Corrosion Rate of Electrogalvanized Steel. Mater. Perf., v. 35, No. 11, Nov. 1986, pp. 18-22. 1 1 . American Society for Testing and Materials. Standard Recommended Practice for Rating of Electroplated Panels Subjected to Atmospheric Exposure. B537-70 (Reapproved 1981) in 1982 Annual Book of ASTM Standards. Part 9, Metallic and Inorganic Coatings; Metal Powders, Sin- tered P/M Structural Parts. Philadelphia, PA, 1982, pp. 364-374. 67 ECONOMIC EVALUATION OF A TECHNIQUE TO PELLETIZE FLUE DUST AND OTHER WASTE FROM THE MANUFACTURE OF STAINLESS STEEL By Joan H. SchwieM ABSTRACT Contained in this paper is an economic evaluation of a method to pelletize flue dust and other wastes from the manufacture of stainless steel. Bureau of Mines personnel have demonstrated that the alloying elements contained in this waste can be recovered if these pellets are used to replace a portion of the scrap fed to an electric arc furnace producing stainless steel. This evaluation con- siders two pellet drying options— heat dried and air dried. The fixed capital cost for a plant addition required to produce 15 st of pellets per day is esti- mated to be about $974,000 for the heat-dried option and $560,000 for the air-dried option, based on second quarter 1987 equipment costs. The estimated annual operating cost per short ton of pellets to pelletize these wastes is approximately $117/st with the heat-dried option and $40/ st with the air-dried option. The value of the pellets, based on the value of the contained alloying elements as ferroalloys, is estimated to be about $313/st. This indicates that the proposed pelletizing tech- nique has economic potential. INTRODUCTION The manufacture of stainless steel results in the production of several wastes such as grinding swarf, mill scale, and flue dust from electric arc and argon-oxygen decarbonization furnaces. In an effort to recycle these wastes, the Bureau of Mines has investigated a tech- nique to pelletize the wastes recovered from an electric-arc fur- nace. The original research^ centered on smelting these pellets separately then recycling the recovered metal alloy as a replace- ment for the scrap portion of a stainless steel production heat. Additional research has shown that it is also feasible to replace a portion of the scrap charge for stainless steel production heats with the pelletized waste. The pellets can compose from 10 to 20 pet of the total charge to the furnace. This enables recycling of the waste without the intermediate smelting stage. In this manner, it is expected that the alloying elements present in the waste can be recycled at minimum cost to the manufacturer. This evaluation is the latest in a series of evaluations that has been used to help guide the research. The research work is described in another paper in this Information Circular, titled "Recycling of Stainless Steelmaking Dusts and Other Wastes," by L. A. Neu- meier and M. J. Adam. PROCESS DESCRIPTION A plant addition has been designed to produce 15 st/d of pellets from stainless steel wastes, operating one shift per day, 5 days per week. Wastes that are assumed to be available for the production of the pellets are grinding swarf, mill scale, and flue dusts from elec- tric arc and argon-oxygen decarbonization furnaces. The flue dusts, mill scale, and grinding swarf are stored in open piles and moved ' Cost evaluation program assistant. Process Evaluation, Bureau of Mines, Washington, DC. 2 Powell, H. E., W. M. Dressel, and R. L. Crosby. Converting Stainless Steel Fur- nace Flue Dust and Wastes to a Recyclable Alloy. BuMines RI 8039, 1975, 24 pp. daily to storage bins with a front-end loader. Mill scale is screened to separate it into three fractions — plus % in, minus % in plus 35 mesh, and minus 35 mesh. The plus %-in fraction is large enough to be recycled directly and is not included in the pellets. It is con- veyed to a storage bin untU needed in the furnace. Minus %-in plus 35-mesh mill scale is conveyed to a ball mill and crushed. The ball mill product is returned to the screen. Minus 35-mesh mill scale from the screen is conveyed to a storage bin. In a zig-zag mixer, the mill scale is combined with grinding swarf and flue dusts from electric arc and argon-oxygen decarboni- zation furnaces. Coke breeze and portland cement are also added to the mixer. The coke breeze is required to reduce the oxidized 68 portion of the waste when the pellets are smelted. Portland cement is added as a binding agent. The combined wastes are then conveyed to a balling drum for pelletization. The resulting green pellets, containing about 12 pet moisture, are dried by one of two methods, the heat-dried option using truck dryers or the air-dried option using the same type dry- ing trucks, but without a dryer. In the heat-dried option, the pellets are air dried for 24 h before being completely dried in a heated dryer. Using the air-dried option, the pellets are dried in the open for several days. Either drying method produces pellets suitable for feeding to an electric arc furnace. The only difference in the plant designs for the two options is the addition of two truck dryers for the heat-dried option. The evaluation is based on pellets produced from the materials presented in table 1 . The composition of the dry pellets is shown in table 2. There is no reason to believe that the raw material ratios used in this study could not be changed to permit the use of vary- ing quantities of each waste. Table 1 .—Materials used to make up the pellets, percent Argon-oxygen decarbonization furnace flue dust 13.0 Electric arc furnace flue dust 17.4 Grinding swarf 39.1 Mill scale 17.4 Portland cement 4.4 Coke breeze 8.7 Total 100.0 Table 2.— Pellet composition (dry basis), percent Cfiromium 9.5 Nickel 3.96 Molybdenum .84 Manganese 2.0 Iron 41 .8 Carbon 11.8 Silicon 3.5 Other 26.6 Total 100.00 ECONOMICS The intent of an economic evaluation is to present capital and operating cost estimates of a commercial-size plant. In the prepara- tion of any economic evaluation, it is necessary to make many assumptions. In general, the assumptions that are made are expected either to apply to the majority of the potential plants or to have only a small effect on the process capital and operating costs. An example of such an assumption is that the plant operates one shift per day, 5 days per week. If an assumption would be necessary that may not apply to a majority of plants or may have a major effect on capital or operat- ing costs, then it is generally not included in the evaluation. An example of such an exclusion is that land cost and pond construc- tion costs have not been included in the capital or operating cost estimates. When an assumption has been made or deliberately excluded, this fact is documented. A detailed description of the estimating techniques used in this evaluation has been published.^ CAPITAL COSTS The capital cost estimate is of the general type called a study estimate by Weaver and Bauman." This type of estimate, prepared from a flowsheet and a minimum of equipment data, can be expected to be within 30 pet of the actual cost. Equipment costs are from informal cost quotations from equipment manufacturers and from capacity-cost data. The costs of the major items of equipment and their accessories are tabulated in the appendix to this paper. The estimated fixed capital costs for plant additions capable of producing 15 st/d of pellets, on a second quarter 1987 basis (Mar- shall and Swift (M and S) index of 808.0), are approximately $974,000 for the heat-dried option and $560,000 for the air-dried option, as shown in table 3. Because this is a plant addition, the cost to hook up to existing plant facilities and utilities is estimated as 2 pet of the total section costs. Factors for piping, etc., except for the electrical factor, are assigned to each section, using as a basis the effect fluids, solids, or a combination of fluids and solids may have on the process equip- ment. The electrical factor is based on the motor horsepower ' Peters, F. A. Economic Evaluation Methodology. BuMines IC 9147, 1987, 21 pp. ' Weaver, J. B., and H. C. Bauman. Cost anci Profitability Estimation. Sec. 25 in Perry's Chemical Engineers's Handbook, ed. by R. H. Perry and C. H. Chilton. McGraw-Hill, 5th ed., 1973, p. 47. requirements for each section. A factor of 10 pet, referred to as miscellaneous, is added to each section to cover minor equipment and construction costs that are not shown with the equipment listed. For each section, the field indirect cost, which covers field supervision, inspection, temporary construction, equipment rental, and payroll overhead, is estimated at 10 pet of the direct cost. Engineering cost is estimated at 10 pet, and administration and over- head cost is estimated at 5 pet of the construction cost. A contin- gency allowance of 10 pet and a contractor's fee of 5 pet are included in the section costs. The costs of plant facilities and plant utilities are estimated as 2 pet each of the total process section costs and include the same field indirect costs, engineering, administration and overhead, con- tingency allowance, and contractor's fee as are included in the sec- tion costs. Included under plant facilities are the cost of material and labor for auxiliary buildings such as offices, shops, laborato- ries, and cafeterias, and the cost of nonprocess equipment such as office furniture, and safety, shop, and laboratory equipment. Also Table 3.— Estimated capital cost,^ heat-dried and air-dried options Heat dried Air dried $142,000 357,600 Fixed capital: Mill-scale preparation section $142,000 Mixing and pelletization section 726,900 Subtotal Plant facilities, 2 pet of above subtotal. Plant utilities, 2 pet of above subtotal . . Total plant cost Land cost Subtotal Interest during construction period Fixed capital cost Working capital: Raw material and supplies inventory. . . Product and in-process inventory Accounts receivable Available casfi Working capital cost Capitalized startup cost Subtotal Total capital cost 1 ,096,300 868,900 17,400 17,400 499,600 10,000 10,000 903,700 519.600 903,700 70,700 519,600 40,400 974,400 560,000 2,400 39,600 39,600 30,600 2,200 14,800 14,800 9,500 112,200 9,700 41,300 5,600 121,900 46,900 606,900 ^ Basis: M and S equipment cost index of 808.0. 69 included are labor and material costs for site preparation such as clearing, grading, drainage, roads, and fences. The costs of water, power, and steam distribution systems are included under plant utilities. Working capital is defined as the funds in addition to fixed cap- ital, land investment, and startup costs that must be provided to operate the plant. Working capital, also shown in table 3, is esti- mated from the following items: (1) Raw material and supplies inventory (cost of raw material and operating supplies for 30 days), (2) product and in-process inventory (total operating cost for 30 days), (3) accounts receivable (total operating cost for 30 days), and (4) available cash (direct expenses for 30 days). Capitalized startup costs, are estimated as 1 pet of the fixed cap- ital costs, and are shown in table 3. The cost of land is not included in this estimate. OPERATING COSTS The estimated annual operating cost is based on the average of 260 days of operation per year, one shift per day, over the life of the plant. The operating costs are divided into direct, indirect, and fixed costs. Direct costs include raw materials, utilities, direct labor, plant maintenance, payroll overhead, and operating supplies. Raw materials and utility requirements per short ton of pellets are shown in the appendix. The shipping charge must be added to the cost of the raw material because the plant location has not been selected. Payroll overhead, estimated as 35 pet of direct labor and main- tenance labor, includes vacation, sick leave, social security, and fringe benefits. Plant maintenance is separately estimated for each piece of equipment and for the buildings, electrical system, piping, plant utility distribution systems, and plant facilities. The indirect costs include the expenses of control laboratories, accounting, plant protection and safety, plant administration, mar- keting, and company overhead. These costs are estimated as 40 pet of the direct labor and maintenance costs. Research and overall com- pany administrative costs outside the plant are not included. Fixed costs include the cost of taxes (excluding income taxes), insurance, and depreciation. Depreciation is based on a straight- line, 20-yr period. The net operating cost per short ton of pellets is $1 17 for the heat-dried option; and $40 for the air-dried option. These costs are presented in table 4. Included in this cost is a credit of 1 cent per pound ($20/st) for the reduced landfill requirements. If the dusts Table 4.— Estimated annual operating cost, heat-dried and air-dried options Heat-dried costs Annual Per St pellets Air-dried costs Annual Per St pellets Direct cost: Raw materials: Portland cement at $60/st Coke breeze at $32/st Argon-oxygen decarbonization dust at $0.00/st . Electric arc furnace flue dust at $0.00/st Grinding swarf at $0.00/st Mill scale at $0.00/st Replacement balls for grinding at $0.27/lb .... Total Utilities: Electric power at $0.05/kW h Process water at $0.25/Mgal Natural gas at $5.25/MMBtu Total Direct labor: Labor at $10.50/h Supervision, 20 pet of labor Total Plant maintenance: Labor Supervision, 20 pet of maintenance labor Materials Total Payroll overhead, 35 pet of above payroll Operating supplies, 20 pet of plant maintenaee . . Total direct cost Indirect cost, 40 pet of direct labor and maintenaee Fixed cost: Taxes, 1 pet of total plant cost Insurance, 1 pet of total plant cost Depreciation, 20-yr life Total operating cost Credit: Reduced landfill requirements at $0.01 /lb Net operating cost $10,300 13,400 100 23,800 2,600 100 202,000 204,700 65,500 13,100 78,600 12,400 2.500 12,400 27,300 32,700 5,500 372,600 42,400 9,000 9,000 48,700 481,700 23,700 458,000 $2.64 3.44 .00 .00 .00 .00 .03 $10,300 13,400 100 6.11 23,800 ,67 .03 51.79 900 100 NAP 52.49 1,000 16.79 3.36 43,700 8,700 20.15 52,400 3.18 .64 3.18 6,400 1,300 6,400 7.00 14,100 8.38 1.41 21 ,000 2,800 95.54 115,100 10.87 2.31 2.31 12.49 123.52 6.08 117.44 26,600 5,200 5,200 28,000 180,100 23,700 156,400 $2.64 3.44 .00 .00 .00 .00 .03 6.11 .23 .03 NAP .26 11.21 2.23 13.44 1.64 .33 1.64 3.61 5.38 .72 29.52 6.82 1.33 1.33 7.18 46.18 6.08 40.10 NAP Not applicable. 70 are declared as hazardous waste, the disposal cost would be at least $2(X)/st. The grinding swarf and mill scale may have a market value; but a cost for them is not included in the operating cost. PRODUCT VALUE To estimate the value of the pellets, it is assumed that the con- tained alloying elements will have a value equal to their price as a ferroalloy or metal. These values, per pound are, nickel, $2.10; chromium as ferrochrome, $0.42; molybdenum, $3.20; and man- ganese as ferromanganese, $0.33. Based on the composition listed in table 2, the value of the pellets is about $313/st. Based on the chromium and nickel values alone, the pellet value is about $246/st. In either case the value of the metal contained in the pelletized waste is much higher than the cost to pelletize it. It should be noted that the pellet value listed in the preceding paragraph is only an estimate and at best will only be representa- tive of pellets with the same composition. For a particular loca- tion, with its own unique wastes, the value of pellets will vary significantly from the value presented. It is expected, however, that the values given are sufficiently representative to allow anyone interested in the Bureau's recycling technique to make a decision as to whether additional consideration is warranted. TECHNICAL EVALUATION The technique to pelletize the stainless steel waste, as presented in this paper, utilizes standard agglomeration techniques and should present no problems in scale-up to a commercial size. Pellets produced at the Bureau's Rolla (MO) Research Center were used to replace part of the scrap charge to an electric arc furnace at a commercial stainless steel manufacturer and were successfully smelted. The results of this testing can be obtained from the research personnel at the Rolla Research Center. It appears, therefore, that the proposed recycling technique has been sufficiently developed to allow serious consideration of it for adaptation on a commercial scale. The use of the proposed process has two potential advantages. First, the alloying elements previously lost in the wastes will be recycled, which will lower the overall operating cost for the stain- less steel manufacturer. Also, because chromium and nickel are almost totally imported, their recycle will reduce the U.S. depen- dence on these imports. The second advantage will be a reduction in the landfill requirements of the stainless steel manufacturer. Processing the wastes can only increase the environmental accept- ability of a plant. Wastes produced by a stainless steel plant will vary from plant to plant as well as from day to day. This is due to the variety of alloys produced and to variations in equipment and procedures. The Bureau's recycling technique, however, should be applicable to any fine stainless steel manufacturing waste. APPENDIX.— HEAT-DRIED AND AIR-DRIED OPTIONS Table A-1 . — Raw material and utility requirements per short ton of pellets 71 Heat dried Air dried Raw materials: Portland cement st . , Coke breeze st . , Argon-oxygen decarbonization dust st . Electric arc furnace flue dust st . Grinding swarf st . Mill scale st . Replacement balls for grinding lb . Utilities: Electric power kWh . Process water Mgal . Natural gas MMBtu . 0.044 0.044 ,107 .107 .130 .130 .174 .174 .391 .391 .174 .174 .067 .067 3.467 4.533 .133 .133 9.867 Nap Nap Not applicable. Table A-2.— Equipment cost summary, mill scale preparation section, heat-dried and air-dried options Item Equipment Labor Total Mill-scale hopper $800 $300 $1 ,100 Belt conveyor 7,700 1 ,300 9,000 Vibrating screen 10,700 1,400 12,100 Bucket conveyor 1 ,400 400 1 ,800 Storage bin 1 ,800 700 2,500 Bucket elevator 3,000 900 3,900 Belt conveyor 7,000 1 ,200 8,200 Ball mill 3,500 200 3,700 Mill-scale storage bin 300 100 400 Total 36,200 6,500 42,700 Total equipment cost x factor indicated: Foundations, x 0.695 25,200 Structures, x 0.080 2,900 Instrumentation, x 0.050 1,800 Electrical, x 0,362 13,100 Piping, X 0.200 7,200 Painting, x 0.020 700 Miscellaneous, x 0,100 3,600 Total 54,500 Total direct cost 97,200 Field indirect, 10 pet of total direct cost 9,700 Total construction cost 1 06,900 Engineering, 10 pet of total construction cost 10,700 Administration and overhead, 5 pet of total construction cost 5,300 Subtotal 122,900 Contingency, 10 pet of above subtotal 12,300 Subtotal 135,200 Contractor's fee, 5 pet of above subtotal 6,800 Section cost 1 42,000 ^ Basis: M and S equipment cost index of 808,0. 72 Table A-3.— Equipment cost summary, mixing and pelletization section, heat-dried option Item EquipmenV Labor Total Argon-oxygen decarbonization dust storage bin $2,000 $700 $2,700 Argon-oxygen decarbonization dust feeder 1,000 100 1,100 Electric arc furnace flue dust storage bin 2,000 700 2,700 Electric arc furnace flue dust feeder 1,000 100 1,100 Grinding swarf storage bin 2,000 700 2,700 Grinding swarf feeder 1,000 100 1,100 Portland cement storage bin 4,200 1 ,200 5,400 Portland cement feeder 1 ,000 200 1 ,200 Coke breeze storage bin 3,000 1 ,200 4,200 Coke breeze feeder 1,000 100 1,100 Belt conveyor from storage 10,000 2,100 12,100 Mixer (zig-zag) 1,300 100 1,400 Belt conveyor 7,700 1,400 9,100 Pelletizer 48,000 600 48,600 Belt conveyor 5,800 900 6,700 6-truck dryer2 68,500 2,400 70,900 9-truck dryer2 97,000 2,600 99,600 Pellet storage bin 2,600 1 ,000 3,600 Bucket conveyor 6,200 1,900 8,100 Total 265,300 18,100 283,400 Total equipment cost x factor indicated: Foundations, x 0.228 60,400 Structures, x 0.080 21 ,200 Insulation, x 0.020 5,300 Instrumentation, x 0.050 13,300 Electrical, x 0.109 29,000 Piping, X 0.200 53,100 Painting, x 0.020 5,300 Miscellaneous, x 0.100 26,500 Total 214,100 Total direct cost 497,500 Field indirect, 10 pet of total direct cost 49,800 Total construction cost 547,300 Engineering, 10 pet of total construction cost 54,700 Administration and overhead, 5 pet of total construction cost 27,400 Subtotal 629,400 Contingency, 10 pet of above subtotal 62,900 Subtotal 692,300 Contractor's fee, 5 pet of above subtotal 34,600 Section cost 726,900 1 Basis: M and S equipment cost index of 808.0. 2 Includes cost of truck. 73 Table A-4.— Equipment cost summary, mixing and pelletization section, air-dried option Item EquipmenV Labor Total Argon-oxygen decarbonization dust storage bin $2,000 $700 $2,700 Argon-oxygen decarbonization dust feeder 1,000 100 1,100 Electric arc furnace flue dust storage bin 2,000 700 2,700 Electric arc furnace flue dust feeder 1,000 100 1,100 Grinding swarf storage bin 2,000 700 2,700 Grinding swarf feeder 1,000 100 1,100 Portland cement storage bin 4,200 1 ,200 5,400 Portland cement feeder 1 ,000 200 1 ,200 Coke breeze storage bin 3,000 1 ,200 4,200 Coke breeze feeder 1 ,000 100 1 ,100 Belt conveyor from storage 10,000 2,100 12,100 Mixer (zig-zag) 1 ,300 100 1 ,400 Belt conveyor 7,700 1,400 9,100 Pelletizer 48,000 600 48,600 Belt conveyor 5,800 900 6,700 Pellet storage bin 2,600 1 ,000 3,600 Bucket conveyor 6,200 1,900 8,100 Total 99,800 13,100 112,000 Drying trucks 16,500 Total equipment cost x factor indicated: Foundations, x 0.463 46,200 Structures, x 0.080 8,000 Instrumentation, x 0.050 5,000 Electrical, x 0.241 24,100 Piping, x 0.200 20,000 Painting, x 0.020 2,000 Miscellaneous, x 0.100 10,000 Total 1 15,300 Basis: M and S equipment cost index of 808.0. Total direct cost 244,700 Field indirect, 1 pet of total direct cost 24,500 Total construction cost 269,200 Engineering, 10 pet of total construction cost 26,900 Administration and overfiead, 5 pet of total construction cost 13,500 Subtotal 309,600 Contingency, 1 pet of above subtotal 31 ,000 Subtotal 340,600 Contractor's fee, 5 pet of above subtotal 17,000 Section cost 357,600 U.S. GOVERNMENT PRINTING OFFICE: 1988 - 547000/80,059 INT.-BU.OF MINES,PGH.,PA. 28759 380 C T U.S. Dapartmant of th« Intarior BwMM of MifiM-Prod. and Dtotr. Cochrans Mill Road P.O. Box 18070 Pittsburgh, Pa. 15236 OFFICIAL BUSINESS PENALTY FOR PniVATE USC. S300 \ I Do not wi sh to recei ve thi s material, please remove from your mailing list* I I Address change* Please correct as indicated* AN EQUAL OPPORTUNITY EMPLOYER V?^'\/ V*^*/ v^*/ v^*/..v^v v*^ ^h^^^^^ "V'^^Py '^'V^^^V "^^'^^^V^/ '^^V^'^^V \r^^ >•*' • »^^ si^- % /^'M\ /^''M^\ /*:aX /•M^- V/^^-X ,.^ >*4r V^^Vv'^ %-«*\o^ V^^V V'*''*/^ v^^^v l\.^"V-| ,#^ v^v v*^*/ V^^V V^V .V^/ "-^*^ .♦"^^ ?^*i> V •'£!**. 6^ v,*^?r^v^6^ «^.^^*\/' "^^^^wV V^'** ♦^ ««. A'i-:^-^ cp^.s^^^'^^o A*i^*X <^® -^i: A y^*'>^%%. >°\'^ '^^ *•••• A • ««' 8^ '»»■»■ . ,^''t.. , ^»( .^^^^^ v . ^;^ ^ .. /°- V-^^ ^y ..^'•. '- '^oV^ c» * °,. *•'■>•* aP •*i ^^^U.^^' A" '^ -OHO- ^ *^ ** k' o *b>? ^r?>. <.^ f** 4' 4>- '' ^<>>. ."^^ y^» . •.ss:55^vfc' o l» «J» ■M\ ^'^.Z ••^'- X/ *»•- ^*, x<5> ,••'•- '<^, "• .*^°-- V 'bv' ^ ''o.o' V ""-u.^"^' '> V'---- > V^^>*^ %--T/.^T<\,**' V^^-* %/--?^'\o** BOOKBINCMNC CrantviHe. Pa Seat 'Oct. 1988 « , . ^ 'W^ i$^i 7^ ''k'l < ,» /^r