铝合金轧制

On-Line Sensor for Aluminum Making Process Investigations1Richard F. Conti, 2Kazuharu Hanazaki1Heraeus Electro-Nite Co., LLC, Langhorne PA USA2Heraeus Electro-Nite Japan, Ltd., Ichikawa, JapanKeywords: inclusions, electric-sensing zone, aluminumAbstractIncremental improvements to already efficient and clean aluminum production processes can be guided by metallurgical on-line sensors that provide in-situ measurements of key process variables. A few such sensors are examined in detail correlating aspects of their measuring accuracy under actual plant conditions measuring non metallic particles in molten aluminum.IntroductionA measuring sensor develops from observations that exposure to a specific stimulus results in a response to that stimulus in a distinctive manner. Bundling this distinctive response for the purpose of measuring a physical quantity by converting the response into a signal which can be read and interpreted by an observer is a measuring system. During a manufacturing process for the, measuring systems that provide information describing the status of the process or are readily available are termed, “on-line” due to the immediacy of their response. One segment of on-line sensors that have become common in the manufacturing process of converting raw materials into finished shapes are sensors designed to be immersed into the liquid metal. In many instances, the primary responsive principle of these sensors and their early development exists in a context removed from the target application. Immersion Particle analyzers, LiMCA, (ABB), and ESZ-PAS, (Heraeus Electro-Nite), provide non metallic particle detection in liquid metals based on the electric sensing zone (ESZ), or Coulter Counter principle first utilized in aqueous solutions [1].Inclusion DetectionMolten aluminum is frequently contaminated to some extent by non-metallic inclusions that give rise to defects in the final products. Prior to the development of the initial prototype LiMCA system for molten aluminum measurement at McGill University [3], it was not possible to measure inclusions, in-situ, enabling metallurgical process decisions at key moments during product manufacturing. Typically, in order to determine the quality of cast aluminum shapes regarding inclusions as close to the production process as possible, operators using optical equipment determined the number and sizes of inclusions in a small surface section. Other techniques utilizing sedimentation, filtration and metallographic combinations of scanning electron microscopes and optical scans of polished surfaces to report observed surface irregularities require considerable amounts of time and labor. Compared with these techniques, the electric sensing zone method has the advantage of providing quantitative information on the volume concentration of inclusions, and also on their size distribution from a considerable larger sample size virtually on-demand.Electric-Sensing Zone MeasurementThe electric sensing zone method of particle size analysis involves an electronically conductive liquid induced to flowfrom one vessel to another through a non-conducting orifice bythe alternating use of vacuum and pressure. The vessels are electrically insulated and a constant current, regulated by a ballast resistor, is applied across this orifice by two electrodes, Aand B, connected to a battery.Figure 1. Passage of a non-conducting particle, d, through aninsulating orifice of diameter D gives rise to avoltage pulseΔV.The presence of a non-conducting particle in the fluid flowing through this orifice causes a change in the electrical resistance detected at the aperture as a voltage pulse as shown schematically in Figure 1. The change in resistance of the fluidis proportional to the volume of fluid displaced by the passing particle given by [4]∆(1)where ρe is the electrical resistivity of the liquid, d is the particle diameter, and D is the orifice diameter at its minimum point.When d is not small compared to D, distortion of the currentflow field leads to an increase in effective resistance ∆R AB ,which should then be modified by a correction factor, f (d/D)[5,6].∆f(d/D)(2)1 2 3 4 51 2 3 4 5⁄ 1 0.8 ⁄ (3)Voltage pulses generated in the presence of the applied constant current can then be measured, and the number and size of the particles counted:∆∗ 1 .08 ⁄(4)The passage of a particle through an ESZ of a molten metal measuring system where 25 10 Ω with an applied current of 20 to 60 amperes is very different from those electrolyte fluids of an aqueous Coulter Counter. These high currents result in strong radial electromotive forces accelerating non-conducting particles toward the sidewall of the ESZ [7]. Particles failing to transit the orifice impinge upon the orifice wall leading to unsteady fluid flow, a partially clogged orifice resulting in incorrectly measured ∆V, high system electronic noise, a low particle pass through fraction and early measurement termination from a clogged orifice. The motion of spherical particles in a current carrying liquid metal flowing through a circular pipe are characterized by trajectories dependent upon hydrodynamic and electric fields, particle size, particle density and the property of the fluid [8]. Large particles reach the sidewall faster than smaller particles and dense particles travel further along the orifice before hitting the sidewall as compared to less dense particles. The percentage of particles in the fluid that successfully pass through the orifice and counted is called the “pass through fraction.”The problem of orifice blockage has been extensively studied and numerous solutions proposed to avoid clogging the orifice by large particles. In order to detect small particles, the orifice should be reduced to the lowest possible and the current increased to the highest possible elevate the transient voltage above the background noise of the measuring instrumentation. Both of these strategies increase the likelihood of large particles clogging the measuring orifice. In much of the prior art, the distribution of numerous smaller inclusions rather than the occurrence of few large inclusions justified the elimination, filtering and separation of large inclusions to allow the measurement to proceed consisting of particle diameters typically as low as 10 percent of the orifice diameter. However, when desiring knowledge of the relatively sparse larger particles present in the liquid metal, this requires much larger sample volumes and a new sampling strategy promoting detection, not elimination.Experimental SystemThe ESZ sensor, similar to that developed by Heraeus Electro-Nite for use in molten steel [9], as shown in figure 2 is a metal sampling chamber comprised of a 12 mm ID, 470 mm long quartz tube. A 1200 micron cylindrical orifice with softened contours at the metal inlet surface is laser drilled into the tube sidewall. The unusually larger orifice passage is selected to satisfy the maximum detection for a particle size approaching 800 microns. The chamber tube is coated on all surfaces with a nearly transparent boron nitride coating. The internal electrode, is a steel coupled to a copper bushing and steel chill block which are crimped to a hollow contact pin. The internal pin assembly is used in conjunction with an argon gas supply to purge the chamber and an applied vacuum to draw metal into the chamber.A low resistance electrical contact is maintained between the inner electrode and an air cooled receptacle via a high temperature copper alloy spring. The outer electrodes, designed for immersions up to an hour, not shown in the follow figure, are a pair of titanium alloy rods attached to a steel tube surrounding the upper portion of the chamber makes electrical contact with another air cooled electrical contacting member in the receiving receptacle. For shorter duration immersions of the same sensor design, steel outer electrodes have been employed. All electrodes at their point of contact at the receptical were kept below 150°C by the air cooling system. The sensor is a single use device with no reuseable portions.Figure 2. Inner electrode arrangement and quartz chamber. The outer electrode is not shown.The contact receptacle is part of a measuring lance holder, containing a signal conditioning amplifier connecting to the remote instrumentation by use of a 10 meter detachable shielded cable that houses power, signal, and gas conduits. Power to the ESZ measuring circuit is a 12 volt high ampere capacity battery that is continuously charged during all sampling cycles except during the actual measuring time when all AC power is isolated from the measuring circuit. A custom LabVIEW program operates the electrical and pneumatic systems while ESZ voltage, current and system pressure data are acquired at 32 kHz and analyzed in separate custom software that displays the acquired data waveforms, measures the actual current, identifies the inclusion peaks, evaluates the quality of the data in regard to the electronic noise and reports the size and size distribution of the combined measurement segments. Although the resistivity of the molten aluminums varies according to the specific alloy measured, and this value will change the calculated particle equivalent sphere, the resistivity of pure aluminum is used for the calculation of size.SamplingA Heraeus Electro-Nite Electric Sensing Zone Particle Analysis System, (ESZ-PAS) was installed at a casting factory of Toyota to measure the non-metallic particles in A356 /A319 metal in the holding furnace.To begin the automatic ESZ sampling process argon is flushed through the purge/vacuum lines prior to immersion. The immersion depth was set by a lance holder so that the orifice was between 150-200mm below the bath surface. Approximately 40g of metal is drawn into the sampling chamber with a vacuum pressure of -40 kPa to achieve a flow rate of 4m/s. Since the metal flow is not instantaneous at its target flow, the measuring current is switched on only after the metal pressure set-point is reached. In this fashion, data used in forparticle analysis is acquired only during the middle 2.5 secondsor, ~25g of aluminum. Although more liquid aluminum has passed the orifice before and after the analysis period, it has been found that the quality of that data is low for low fluid velocity. After data acquisition, the sampling vacuum is relieved and the obtained metal and senor components equilibrate during a 30 second rest period after which this metal is purged from the chamber by argon. The liquid around the orifice entrance is stirred by a 30 second argon flush at 7 kPa from the open orifice. The alternate, high current cycle begins immediately following this purge time.Metal coupons were taken during some ESZ sampling for comparison.(3a)(3b)Figure 3. Instrumentation screens for sampling procedure, (3a) and sampling analysis, (3b).Figure 4. Fractured surface of a metallurgical coupon. SEM analysis of acquired coupons during an earlier measurement determined predominately Al2O3 and Al-Si particles.Transient pulse detection by the ESZ-PAS software is a complex combination of peak characteristics translated into a set of logic statements. In order to be identified as a peak, a set of data points collected at 32000/s, must pass predetermined criteria. The baseline data points are filtered then scanned point by point looking for a set number of consecutive data points each achieving a minimum voltage rise over the back averaged baseline that could potentially signal the start of a voltage peak. Once identified other criteria such as, but not limited to, the rate of rise of the leading edge, time of peak half height, total peak duration and undershoot at the foot of the failing edge must all meet their acceptance criteria. After qualification, the height is measured over the local baseline average and ∆V is obtained for use in equation 4. Particle counts of the low current cycle and the high current cycle are detected by the ESZ-PAS peak detection algorithm. An equivalent sphere diameter is calculated from the transient voltage peak height, grouped according to a particle size index.DiscussionThe dual current ESZ-PAS was operated in Toyota at various times for different conditions duplicating scenarios that would occur under actual plant conditions. The number of low and high current measurement cycles and the time between each cycle are part of the parameter setting of the instrumentation. Typically, the number of cycles is selected depending upon the rate of erosion of electrodes which is a function surface refreshing rate and total contact time with the aluminum. For sampling 1kg of liquid metal, 40 cycles of 1m11sec each will take nearly 50 minutes. Both electrodes, the boron nitride coating of the quartz and the stability of the orifice dimension were adequate for the duration of the measurement.Numerous evaluations at different were reviewed to establish that the distribution pattern of sized groups developed quickly with approximately 10 cycles of measurement, figures 5 and 6.Figure 5. Assembly of accumulated inclusion distribution patterns after 2, 10, 20, and 40 cycles of low and high, (18-47A), current measuring segments from the holding tank. Figure 6. Assembly of accumulated inclusion distribution patterns after 2, 4, 6, 8, 10 and 18 cycles of low and high, (21-58A) current measuring segments support observation that for these experimental conditions, the distribution reaches a steadystate after~250g of material.Since the accumulated particle count is a combination of two applied measuring currents there could be an excess of detected particles where the low detection limit of the high current cycle overlaps the upper limit of the low current cycle. In practice, the ratio of high to low current levels can coincide with the pass through fraction so that the low detection limit of the low current segment approximates a built in division as in figure 7. When this is not possible, selection criteria for mutual cut-off levels can be easily applied.Figure 7. Accumulation of particles detected by the same sensor, D=1200, in the same metal, grouped by applied current.Even with the most noise free environment possible, there is a minimum peak height above the baseline which could reasonably be considered a “peak,” in other words, the minimum reported inclusion. There can arise a circumstance where a “debate” occurs as to the amount of small particles counted and the amount of baseline noise improperly identified as particles. In the ESZ-PAS system a cut-off of counted peaks, is preset by the pre-calculated particle the size of 15% of the orifice at the measurement current. On the contrary, for large particles because the peak voltage is so much greater than the baseline voltage, the ease at which large particles can be identified leads to no debate.The high currents employed by traditional ESZ systems to detect small particles will promote a low success rate for the detection of large particles due to the influence of the electromagnetic “pinch effect” on the particle trajectory passing, and in most cases, failing to pass through the measuring orifice. The conspicuous absence of large particles detected at 68A under identical conditions alternated in time with a low current sensor detecting large particle clearly support the earlier modeling work of Mei and Guthrie [8] that low pass through fractions are current created operating conditions even in the situation of a large orifice. Additionally, the lack of positive baseline jumps, most likely the consequence of particle attachment or fluid detachment from the orifice walls becomes apparent as current is reduced. Although it is suggested that a large orifice will lead to coincidence counting, [10] the chance that two particles enter the orifce at the same time, this is observed in water testing where high concentrations of particles are purposefully added, but in actual industrial melts only to a limited extent.When the measuring current is set at the low current setting the previously identifiable small particles vanish into the baseline but the larger particles stand clearly above the baseline, not deflected away from the ESZ entrance allowing for a high pass through fraction. A high current can now be applied to measure small particles which have a high pass-through fraction.The presence of a low pass-through fraction for large particlesresults in a skewed ESZ distribution and due to their lowoccurrence rate on a metal sample surface, support this under reporting. An accurate assessment of large particles, when this is the intended measurement, would necessarily mandate a measuring conditionConclusionsThe on-line sensors described here are immersion type sensors for use in molten aluminum and molten salt. Success in this specific market depends upon adaptations to the measuring apparatus through understanding of the unique requirement of the measured media. Particle size and size distribution measuring system for in molten aluminum where a high electromagnetic fields in the electric sensing zone results in “pinch” effects lowering the ability to detect large particles has been modified to provide dual measuring currents for the same size orifice providing alternate detection cycles of large and small particles.AcknowledgementThe authors wish to thank; Toyota Motor Corporation for providing assistance and support for ESZ measurements in an industrial environment.References1.R.I.L Guthrie and D. Doutre: Proc. Int. Seminar onRefining and Alloying of Liquid Aluminum and Ferro-Alloys, Trondheim, Norway, 1986, pp. 146-163.2.H. Takada, T. Shibata, K. Nihei, T. Inoue, ImmersionThermometer for Molten Metal, Japanese PatentJP62019727, 19873. D. Doutre, R.I.L. Guthrie, “A Method and Apparatus forthe Detection and Measurement of Particulates in LiquidMetals”, U.S. Patent 4,555,662, November 26, 1985.4.R.W. Deblois, and C.P. Bean: Rev. ScientificInstruments, 1970, vol. 41 (7), pp 909-915.5.R.W. Deblois and C.P. Bean and R.K.A. Wesley: J.Colloid Interface Sci., 1977, vol. 61 (2), pp. 323-3356.W.R. Smythe: Phys. Fluids, 1961, vol. 4 (6), pp 756-7597.Mei Li and R.I.L Guthrie: “On the Detection andSeparation of Inclusions in Liquid Metal CleanlinessAnalyser (LiMCA) Systems”, Metallurgical and MaterialsTransactions B, Vol. 32B, Aug. 2000, pp 767-777.8.Mei Li and R.I.L Guthrie: “In-situ Detection of Inclusionsin Liquid Metals: Part I. Mathematical Modelling of theBehaviour of Particles Traversing the Electric SensingZone”, Metallurgical and Materials Transactions B, Vol.32B, Dec. 2001, pp 1067-1081.9.R.P. Stone, C.C.Liu. P.C. Glaws: “Experience With anInnovative On-Line Inclusion Determination System forLiquid Steel”, AIST Conf. Proc.Vol 1, AISTech 200810.M. Isac, A. Chakraborty, L. Calzado, R.I.L. Guthrie:“Development of an Aqueous Particle Sensor (APSIII),System as a Research Tool for Studing the Behaviour ofInclusions in Water Models of TUndish Operations”,Sensor, Sampling and Simulation for Process Control.TMS Conf. Proc. 2011。