Vous êtes sur la page 1sur 8

This section will cover achieving consistency in electrical supply, metal melting, sampling and sampling supplies, electrical

and chemical calibration, and instrumentation so as to achieve the most accurate and repeatable thermal analysis measurements. Thermal analysis (TA) is a method of determining chemistry and microstructure in complex metals by measuring thermal arrests, and energy production during solidification, and sometimes energy production during solid state transformations. The temperatures of multiple arrests and the rate of change in temperature at various points are measured by means of an embedded thermal couple. The thermal couple produces a temperature dependent voltage which is converted to digital format and processed by a Microprocessor or by a personal computer to produce the results. In the production of irons, TA is commonly used to measure arbon !"uivalent, arbon and silicon. #ess common uses include graphite morphology for gray, compacted and ductile irons, the degree of inoculation, the degree of chill, and the degree of pearlite$ferrite in final irons, and the effective magnesium in ductile and compacted graphitic irons. In white irons, where graphite production must be very tightly controlled, TA provides a comparison with production iron vs. %&&' chilled iron. In the production of aluminum alloys, TA is commonly used to measure the degree of inoculation and modification of hypoeutectic(silicon alloys (A)%*, A)+&, A),,, A),etc), and the degree of nucleation of A+&& series and hypereutectic silicon alloys. In the production of opper alloys, TA is commonly used to measure the phosphorus levels for bra.ing alloys, and for high phosphorus copper alloys up to about %,'. In steel production, TA is used to measure low carbon levels in low alloy steels, and to measure correct carbide levels in wear resistant alloys. Applications Thermocouples are most suitable for measuring over a large temperature range, up to %/&& 0 . They are less suitable for applications where smaller temperature differences need to be measured with high accuracy, for example the range &((%&& 0 with &.% 0 accuracy. 1or such applications, thermostats and 2T3s are more suitable. Temperature measurement techni"ues In the ferrous foundries ladle temperature varies above %4&& . At this temperature continuous measurement is difficult. Therefore dip type immersion pyrometer techni"ue is widely used. Immersion pyrometer consist of metal housing, !lectronic Measurement ircuitry (!M ), metal lance, Thermocouple connected to the circuitry by means of compensating cables which passes through the lance. 5hen thermocouple dip into the molten metal, it gives milli(volts output. These milli volts are carried to the !M , where the signal (mv) is amplified and processed to give

direct digital output. The processing of the signal is done by various techni"ues. 6ea7 latching, 6lateau 3etection, ontinuous trac7ing are some of these techni"ues. 6ea7 #atching8 As the name indicates the techni"ue involves latching of the pea7 temperature and displaying it. In an induction furnace due to its nature hot spots are generated that moves randomly across the molten metal. 9enerally the temperature at the hot spot is always higher than the molten metal temperature. If the Thermocouple comes into contact with the hot spot then this reading is latched by the !M and displayed. Thus temperature at the hot spot is not correct representation of the entire bath temperature. Therefore pea7 latching is not world wide accepted for use in measuring molten metal temperature in induction furnaces. ontinuous trac7ing8 In this techni"ue temperature is displayed continuously as it senses. :o you do not get stable display of temperature. :o it is difficult to 7now, when to remove lance. Temperature recording is fully dependent on operator s7ill and it can not separate out hot spot temperature and true temperature. :o it is not suitable for ferrous foundries. 6lateau detection techni"ue8 In this techni"ue transient response of sensor and system is ignored. :teady state response is monitored and confirmed by ta7ing several samples. ;nce it fits in its detection criteria the temperature is displayed. The operator can remove the lance to prevent it from melting in the molten metal bath. This techni"ue displays correct temperature and ignores incorrect readings by giving error messages. Thus there is no possibility of hot spot measurement in this techni"ue. This operation is completed in %() sec in case of single dip paper tube thermocouple tips and )(/ sec in case of Multi dip thermocouple tips. It ensures life of receptacle (connector for tip and lance). Transient response phenomenon is very common term in the field of process instrumentation i.e. how system (including sensor) responds for any sudden change is characteri.ed by this terminology. In case of Immersion 6yrometer, as we are suddenly inserting thermocouple inside the molten metal bath< its response could be li7e shown in 1ig % or 1ig + as it depends on various factors.

=ormal trac7 mode, time based > pea7 latch temperature indicator fails to identify > measure the true temperature. 1or molten metal temperature measurement only, 6lateau detection techni"ue is recommended > accepted worldwide, In case, if there is a ?;T :6;T with molten metal in the furnace, the temperature at ?;T :6;T is more than the rest of the molten metal. If you are using ordinary 6!A@ #AT ? I=3I AT;2 type pyrometer, it displays this ?;T :6;T temperature, which may be more by +&(+, degree centigrade. In such condition you may face the problem of ;#3 6;A2I=9, as the actual temperature is less by +&(+, deg. But in case of 6#AT!AA 3!T! TI;= T! ?=ICA!, as soon as thermocouple is dipped into molten metal, first pea7 temperature is ignored and it ensures 4 constant readings for a period of one second > declares true temperature readings. It ignores false readings due to furnace hot spot, noise, and transient response > once it detects steady state temperature i.e. flat temperature it will be displayed. ?ow 3oes a Thermocouple Tip 5or7s

It is important to note that thermocouples measure the temperature difference between two points, not absolute temperature. In most applications, one of the Dunctions E the cold DunctionE is maintained at a 7nown (reference) temperature, while the other end is attached to a probe. 1or example, in the image below, the cold Dunction will be at copper trac7s on the circuit board. Another temperature sensor will measure the temperature at this point, so that the temperature at the probe tip can be calculated. Thermocouples can be connected in series with each other to form a thermopile, where all the hot Dunctions are exposed to the higher temperature and all the cold Dunctions to a lower temperature. Thus, the voltages of the individual thermocouple add up, which allows for a larger voltage.

?aving available a 7nown temperature cold Dunction, while useful for laboratory calibrations, is simply not convenient for most directly connected indicating and control instruments. They incorporate into their circuits an artificial cold Dunction using some other thermally sensitive device (such as a thermister or diode) to measure the temperature of the input connections at the instrument, with special care being ta7en to minimi.e any temperature gradient between terminals. ?ence, the voltage from a 7nown cold Dunction can be simulated, and the appropriate correction applied. This is 7nown as cold Dunction compensation. Asually the thermocouple is attached to the indicating device by a special wire 7nown as the compensating or extension cable. The terms are specific. !xtension cable uses wires of nominally the same conductors as used at the thermocouple itself. These cables are less costly than thermocouple wire, although not cheap, and are usually produced in a convenient form for carrying over long distances ( typically as flexible insulated wiring or multi core cables. They are usually specified for accuracy over a more restricted temperature range than the thermocouple wires. They are recommended for best accuracy. ompensating cables on the other hand, are less precise, but cheaper. They use "uite different, relatively low cost alloy conductor materials whose net thermoelectric coefficients are similar to those of the thermocouple in "uestion (over a limited range of temperatures), but which do not match them "uite as faithfully as extension cables. The combination develops similar outputs to those of the thermocouple, but the operating temperature range of the compensating cable is restricted to 7eep the mismatch errors acceptably small. The extension cable or compensating cable must be selected to match the thermocouple. It generates a voltage proportional to the difference between the hot Dunction and cold Dunction, and is connected in the correct polarity so that the additional voltage is added to the thermocouple voltage, compensating for the temperature difference between the hot and cold Dunctions. 3ifferent types of Thermocouple

A variety of thermocouples are available, suitable for different measuring applications (industrial, scientific, food temperature, medical research, etc.). Type @ ( hromel (=i( r alloy) $ Alumel (=i(Al alloy))8 The Fgeneral purposeF thermocouple. It is low cost and, owing to its popularity, it is available in a wide variety of probes. They are available in the G+&& 0 to H%+&& 0 range. The type @ was specified at a time when metallurgy was nowhere near as advanced as today and conse"uently characteristics vary considerably between examples. There is another problem in that one of the constituent metals is magnetic (=ic7el). The characteristic of the thermocouple undergoes a step change when a magnetic material reaches its urie point. This occurs for this thermocouple at ),I0 . :ensitivity is approximately I% JK$0 .

Type ! ( hromel $ onstantan ( u(=i alloy))8 Type ! has a high output (-/ JK$0 ) which ma7es it well suited to low temperature (cryogenic) use. Another property is that it is non(magnetic. Type L (Iron $ onstantan)8 #imited range (GI& to H4,& 0 ) ma7es type L less popular than type @. The main application is with old e"uipment that cannot accept modern thermocouples. L types cannot be used above 4-& 0 as an abrupt magnetic transformation causes permanent de(calibration. Type LMs have a sensitivity of N,+ JK$0 . Type = (=icrosil (=i( r(:i alloy) $ =isil (=i(:i alloy))8 ?igh stability and resistance to high temperature oxidation ma7es type = suitable for high temperature measurements without the cost of platinum (B, 2, :) types. They can withstand temperatures above %+&& 0 . :ensitivity is about )* JK$0 at *&&0 , slightly lower than a Type @. 3esigned to be an improved type @, it is becoming more popular. Thermocouple types B, 2, and : are all noble metal thermocouples and exhibit similar characteristics. They are the most stable of all thermocouples, but due to their low sensitivity (approximately %& JK$0 ) they are usually only used for high temperature measurement (O)&& 0 ). Type B (6latinum(2hodium$6t(2h)8 :uited for high temperature measurements up to %/&& 0 . Anusually type B thermocouples (due to the shape of their temperature(voltage curve) give the same output at & 0 and I+ 0 . This ma7es them useless below ,& 0 . Type 2 (6latinum $6latinum with 4' 2hodium)8 :uited for high temperature measurements up to %-&& 0 . #ow sensitivity (%& JK$0 ) and high cost ma7es them unsuitable for general purpose use. Type : (6latinum $6latinum with %&' 2hodium)8 :uited for high temperature measurements up to %-&& 0 . #ow sensitivity (%& JK$0 ) and high cost ma7es them unsuitable for general purpose use. 3ue to its high stability type : is used as the standard of calibration for the melting point of gold (%&-I.I) 0 ). Type T ( opper $ onstantan)8 :uited for measurements in the G+&& to ),& 0 range. The positive conductor is made of copper, and the negative conductor is made of constantan. ;ften used as a differential measurement since only copper wire touches the probes. Type T thermocouples have a sensitivity of N I) JK$0 . Thermocouples are usually selected to ensure that the measuring e"uipment does not limit the range of temperatures that can be measured. =ote that thermocouples with low sensitivity (B, 2, and :) have a correspondingly lower resolution. Type B, :, 2 and @ thermocouples are used extensively in the steel and iron industry to monitor temperatures and chemistry throughout the steel ma7ing process. 3isposable, immersible, Type : thermocouples are regularly used in the electric arc furnace process to accurately measure the steel temperature before tapping. The cooling curve of a small

steel sample can be analy.ed and used to estimate the carbon content of molten steel. !lectrical 9rounding

The TA instrument is measuring &.& to &.&%/ volts when using : or 2 type thermocouple and &.& to &.&,, volts when using @ type thermocouple. The raw precision is generally listed at &.% degree which can be improved through software filtering to about &.&, degrees . To achieve this degree, it is important to have a good ground for reference. =eutral lines typically have as much as P volt of induced energy in them. That is, the electrical noise in the lines can be e"ual to or greater than the signal we are trying to measure. #ong runs of thermocouple wire mean more susceptibility to pic7ing up electromagnetic energy. The power pic7up shoes of an overhead crane were arcing between rail segments in one foundry causing a pic7up of energy e"uivalent to %& degrees of temperature. 2unning a shielded thermocouple wire solved the problem. Metal onsistency

Thermal Analysis loo7s at the transitions between li"uid and solid metal. To measure the effect of any element in the metal, that element must be dissolved in the li"uid. Andissolved crystals, while they may promote nucleation, can throw off the hemistry Analysis. As a general rule, iron should be heated above +,,& 1 or %I&& before it is sampled. This allows the silicon to go into solution. ;nce the iron has been heated above this temperature, it can be cooled down to lower temperatures and still successfully measured as all the silicon is now in solution. :imilar problems exist with arbon and :ilicon arbide additions. :ilicon arbide dissolves over time and cannot be measured by TA until it is dissolved. ;ne foundry was adding I' silicon carbide to a cupola that was feeding into a %& ton holding furnace. The cupola bed was low, and the holding furnace was almost empty when the TA and the chill wedges showed low silicon. The spectrometer reported the silicon as being within range. It was suspected that the silicon carbide had not finished dissolving, and when the castings turned out hard at sha7eout, they were sent bac7 to the cupola for re(melting rather than try to machine them with undissolved silicon carbide in the iron. #i7ewise arbon floating on the surface of the furnace can create a arbon rich layer if the furnace is not powered on. 9enerally try to not sample through a carbon or slag layer. lear a clean spot on the surface, and wash the sample spoon clean of any remains left geometric center of cup due to rapid cooling in the base of the cup. :ampler consistency

:ome types of analysis depend on measuring the cooling rates of the sample. These rates and even the arrest points can be influenced by how full the sample cup is. :ample cups that are filled to less than % cm or %$) inch below the top are suspect. The thermocouple needs to be in the center of heat of the metal mass. This point is slightly above the

geometric center of cup due to rapid cooling in the base of the cup. :amplers can also fail. This usually means that the thermal couple can melt, or come in contact with the metal. The @ type of thermal couple is subDect to melting when :ampling molten iron. The T melts at +,I& 1 or %)*& . :ince Iron is generally much hotter than that, we depend on a certain amount of temperature loss in transferring the sample , and in temperature loss to the cup before the thermocouple reaches its max. The twisted thermal couple has good mechanical bonding and will only fail if the ceramic tube crac7s or the core wash over the T is thin or missing, or the melting point is reached. The "uart. tube style cup can fail if the "uart. brea7s, or if the wire softens enough. 3ue to the connection with the cup stand, the wire is in tension, and can pull apart if the Doint softens enough. The wire will pull away from one side of the cup about P centimeters. This will typically terminate the analysis. alibration consistency

Many foundries ma7e life hard for themselves by over calibration. 3eming pointed out that this can be a maDor source of error, and TA proves the point. 1irst the calibrator has an internal temperature sensor that corrects for differences between room temperature and the melting point of ice. This sensor is embedded inside of the instrument and can be fooled. If the calibrator was stored in an air conditioned (summer) or heated (winter) office, the cold Dunction sensor can be off by %& to %, degrees . Both the calibrator and the metal pins$rails of the cup stand need to be at room temperature to avoid errors. 6eople who calibrate too often end up ta7ing shortcuts and introduce errors into the calibration by not letting the calibrator and or stand heads come to room temperature8 an operation that can ta7e up to )& minutes depending on the instrument. It is best if proper care is ta7en in the calibration, and then no further calibration is done until there is an assignable cause (damage or replacement of components) or the time period is expired for the calibration (usually - months). Instrument calculation consistency

Thermal analysis is limited by some mathematical constraints called the degrees of freedom. Typically for example, in iron, we calculate chemistry from two data points and try to solve for arbon and :ilicon. Actually arbon, :ilicon, 6hosphorus, Manganese and hromium all affect the arrest points. To calculate silicon, it has to be assumed that phosphorus, manganese and chromium remain relatively consistent or that the change in those elements is insignificant. Manganese is the usual culprit when the silicon analysis varies. The handboo7 e"uation is .!. Q arbon H :ilicon$) H 6hosphorus$+ H Manganese$, ( hromium$ *. In actuality we generally find the silicon factor to vary between +., and ).) depending on iron type. 9eneral practice is to determine the best silicon e"uation for each maDor metal type. A maDor metal type would be those that had significantly different levels of manganese, phosphorus, or chromium. arbon analysis agrees with combustion analysis to within the typical error of combustion analysis, so it










arbon !"uivalent analysis is defined by the TA measurement. That is .!. is determined by the #i"uidus temperature. :ome people confuse the issue by thin7ing that by using the spectrometer chemistry in a formula that they can calculate the .!. and therefore what the #i"uidus temperature should be. That is not so. The calculations using chemistry are approximant. The actual measurement of the #i"uidus is exact. In the old days the #i"uidus temperature was used to measure the fluidity of the metal (lower temperature implied more fluidity). This was an improvement on the fluidity spiral which could be affected by temperature or wetness of the mold. onclusion Mathematically the standard error of a system is the s"uare root of the sum of the s"uares of all the errors. It might be daunting to some to see all the sources of error possible. But with good engineering, a good instrument, and a good lab, it is possible for most foundries to benefit from the use of TA. In practice the amount of error varies from foundry to foundry. But generally a standard error of &.&)' carbon and &.&)' silicon can be obtained if these guidelines are followed. This compares with lab results of &.&),' for combustion carbon and &.&+' for spectrometer silicon.