Đo nhiệt độ P19 doc

20 Temperature Measurement in Industrial Appliances 20 .1 Introduction The majority of industrial appliances, and especially industrial heating appliances are equipped not only with temperature measurement and recording installations, but also with temperature control systems . The control systems are necessary for optimal, or at least correct, operation of technological processes such as heat treatment of metals, ceramic firing, sintering, melting, drying etc . For these reasons it has been decided to treat also the problems of temperature measurement for control purposes in this chapter . Such an approach is also justified by the common practice of using one temperature sensor, either as a single or double assembly, for temperature measurement, recording and control as previously discussed in Chapter 12 . Although a really detailed analysis of temperature control problems is not possible, it is important to emphasise the significant influence which the type, location and dynamic behaviour of a temperature sensor has upon the results of temperature control (Michalski and Eckersdorf, 1987) . These aspects will be considered in detail in this chapter . Temperature measurement and control should be jointly analysed to give the highest accuracy and to grant the high infallibility and reliability of the heating appliances, especially those for the heat treatment of very precious charges such as occur for example in semiconductor production . 20 .2 Chamber Furnaces 20 .2 .1 General information A leading example of this joint treatment occurs in the temperature measurement and control of electric or gas chamber furnaces . A simplified block diagram of temperature measurement and control system of an electric chamber furnace is shown in Figure 20 .1 . A chamber furnace of 5 to 100 kW heating power and working temperature up to 1300 °C has the heating elements mounted on side walls . The charge is placed on a steel bottom plate . The single or double thermocouple, described in Section 3 .3, is connected to a controller Temperature Measurement Second Edition L. Michalski, K. Eckersdorf, J. Kucharski, J. McGhee Copyright © 2001 John Wiley & Sons Ltd ISBNs: 0-471-86779-9 (Hardback); 0-470-84613-5 (Electronic) 398  TEMPERATURE MEASUREMENT IN INDUSTRIALAPPLIANCES (al  (b) SENSOR LINING  1 I 1 I CHARGE RECORDER I I FURNACE  YP HEATING  V  STEEL PLATE ELEMENTS  I POWER UNIT  I I I  CONTROLLER  I I  I I  '9 r  I ._ . -~ J Figure 20 .1 Block diagram of temperature measurement and control in electric chamber furnace : (a) classic arrangement ; (b) microprocessor based configuration and recorder, which are most likely combined with a microprocessor system . An on-off controller with a contactor or a continuous PID controller with a thyristor power unit are usually used (Michalski and Eckersdorf, 1987) . 20 .2 .1 Internal furnace temperature The internal furnace temperature has a certain equivalent value measured by a temperature sensor, which shouldnot hinder or interfere with the furnace charging in any way . For this reason it is usually inserted through a hole in the furnace dome or in the upper part of the back wall . However, the charge temperature, which is the most important variable in a thermal technological process, is also difficult to measure . Hence, in most cases, it is assumed to be equal to the internal furnace temperature in the thermal steady state, . Assuming that the heat transfer inside the furnace chamber is predominantly by radiation, it can be shown (Michalski, 1966) that the indicated sensor temperature does not depend upon the position of the sensor in the chamber, provided the sensor is no nearer to the heating elements than about 5 to 7 cm . This is true for all sensors ofvery small junction dimensions, such as bare or MI thermocouples . Sheathed sensors, which have to be placed a little further away from heating elements, must also be inserted sufficiently deeply into the chamber to eliminate any errors caused by heat conduction along the sheath as described in Chapter 17 . In the thermal steady state, or when the furnace temperature changes slowly, Michalski(1966) has shown that the temperature of a minute, point-like sensor junction is given by : vt  v1 vt 4  4 TT = 4 Ew 1Tn ( PT,, + Ew (I - -w)1 1Ti ( PT >n  T )n  (20 .1) n=l  n=t k=l kin CHAMBER FURNACES  399 where E W is the total emissivity of the chamber walls, n is the consecutive number of the six chamber walls, k is the consecutive number of the five chamber walls, with the sixth wall excluded, T n and T k are the temperatures of a particular one of the six chamber walls, tpT-,n is the angle factor of the sensor, T, relative to wall n (Jakob, 1953), and (P,,-,k is the angle factor of wall n, relative to wall k (Jakob, 1953) . In equation (20 .1), it is assumed that all of the chamber walls have the same emissivity c~,  which most frequently has a value e, = 0 .9 . For approximate calculations it can be assumed that the emissivity of all of the walls is e W =1 . The chamber can also be regarded as a cube for the relative dimensions of the chamber walls confined from 1 :1 :1 to 1 :2 :4 . Equation (20 .1) is then modified to : T T = a 6 ~Tn  (20 .2) n=t Consider a furnace with a charge in its chamber . The temperature measuring device indicates a certain mean value of temperature between that of the walls and the charge . As the charge temperature approaches that of the walls ( T, = TW ) the indicated value of the temperature is nearly equal to the charge temperature . Thus the calculated temperature, TT, is near the chamber temperature, which is defined as the temperature measured by a sensor of negligibly small thermal inertia placed in the geometrical centre of the chamber . 20 .2 .3 Charge temperature The temperature of the charge inside the furnace chamber can be measured either by contact or pyrometric methods . In the contact method, the sensor, which is placed on the charge surface, or inside the charge, in the manner described in Chapter 16, has its conductors led outside the furnace chamber . Hence, the contact method is rarely used . However, when this method is appropriate, thin, bare thermocouples of wire diameter about 0 .5 to 2 mm with flexible insulation, or thin MI thermocouples, are usually used . They are fed through holes in the walls or in the door, to the outside . The method is used for short-time measurements such as in the determination of charge through-heating time . Once this time is determined it is then repeated in batch production . Fibre optic thermometers with a black body sensor, as described in Chapter 6, can also be used for charge temperature measurement inside a furnace . The contactless, pyrometric method can be applied for continuous as well as for spot measurements . In most cases a protecting sighting tube is used, as described in Chapter 11 . 20 .2 .4 Measurements for temperature control Measurements for temperature control are usually based on internal furnace temperature, as described above . However, the sensors used should have as small a time constant N T as possible (see Chapter 15) . 400  TEMPERATURE MEASUREMENT IN INDUSTRIAL APPLIANCES The static and dynamic properties of a furnace, which is the controlled system, can be described in an approximate way by its transfer function : K g -sLs G s (s) ~  + sNs  (20 .3) 1 where K S is the system gain (°C/W), L S is the system lag, or delay (s), and N S is the system time constant (s) . The dynamic properties of the temperature sensor should also be taken into account when choosing the settings of closed loop temperature control . As given in Chapter 15, these are usually described by the equivalent transfer function, GT(s) : G T (s) =  1  (20 .4) 1 + sNT where N T is the sensor time constant . Michalski and Eckersdorf (1987) made some simple transformations to obtain the combined, equivalent transfer function of the furnace and the sensor, GST(s), as : K e sGsT GST (s) ~ 1 + SNST  (20 .5) where L ST is the equivalent lag of the furnace and sensor together, NsT is the time constant of the furnace with the sensor and the gain K S is the same as in equation (20 .3) . The equivalent time constant N sT ; t N s because for the majority of cases : N S > I OON T (20 .6) and the equivalent time lag of the furnace and the sensor together is simply calculated as : L ST :L L s + N T (20 .7) Sensors with as small a time constant as possible should be used in very popular on-off control . This results from the fact that L ST and thus also N T , are the main factors determining the control quality of on-off control, which is estimated by the peak-to-peak value of the temperature oscillations in quasi-steady state as shown in Figure 20 .2 . The smaller the peak-to-peak the better is the control quality . In continuous P, PI and PID control, a smaller value of L ST improves the system dynamics . In this way the first overshoot is smaller and the heating-up settling time as well as the settling time after a disturbance are both shorter as shown in Figure 20 .3 (Michalski and Eckersdorf 1987) . CONTINUOUS FURNACES  401 6+3 =32°C T60  A-5 =B°C  AS-_3,4°C 750 - 740 - 730- N T =140s  N T =20s N T =4s 60s TIME Figure 20 .2 Peak-to-peak oscillation, AO, in on-off control of a chamber furnace for different sensor time constant, N T N T =0 20 50 100  150  300 9i5- 810-805 u .o °  u 905  ~~ a  ° G  k' 900 795 I  '  I  I I ~ ~  ~  ~ U tOs TIME Figure 20 .3 Transients in continuous PID temperature control of a chamber furnace for different sensor time constant, N T 20 .3  Continuous Furnaces Continuous heating appliances comprise furnaces, dryers, etc . i n which the charge is moved along by a transporter or conveyor system . In most cases continuous appliances take the form of multi-zone furnaces as in Figure 20 .4 . The temperature of each zone, k, is measured by a separate sensor T k , whose output signal is used for recording as well as for control purposes . In long multi-zone furnaces, used for heat treating metallic, ceramic or glass charges, the number of heating zones may be as high as even 50 . The charge, passing through the consecutive zones, each with a different temperature set point, is the subject of programmed temperature processing . The pyrometers, described in Chapter 10 are used for charge temperature measurement at different points along the furnace . In bigger furnaces, microprocessor systems are usually employed for temperature measurement, recording and control in multi-input/multi-output control configurations . During start-up of a new production run, it is sometimes necessary to measure the temperature ofa moving charge in a continuous way using two different methods, which consist of drawn through thermocouples or the thermally insulated storing devices . 402  TEMPERATURE MEASUREMENT IN INDUSTRIAL APPLIANCES (o)  ZONE No :  k-t  k  k .1 P  Tk_ ~  P  Tk  P k-1  k  Y .1 G CHARGE V HEATING i LINING  i TRANSPORTER (b) ELEMENTS 4(0- CHAMBER TEMPERATURE ;-SENSOR OF ZONE k TEMPERATURE 8lU-TEMPERATURE CHARGE  P FIBRE OPTIC PYROMETER FOR c  COME TEMPERATURE IN ZONE k Figure 20 .4 Temperature measurement of each zone and charge in a continuous multi-zone furnace Drawn-through thermocouples are insulated, elastic, thin thermocouples of about 0 .5 to 2 mm diameter, whose measuring junction is fastened to the moving charge . They are connected by elastic conductors to a stationary recorder . After passing along the whole length of the furnace, the thermocouple, most conveniently a K-type, is cut off . Unfortunately thermocouples of this type also break sometimes . A thermally insulated storing device, moving along the furnace together with the charge, stores the temperature values, at pre-set time intervals . After passing the whole length of the furnace, the digitally stored data are read out and recorded . The temperature data are taken from thermocouples fastened to the surface or inside the charges . For example the microprocessor based Furnace Tracker from DATAPAQ Ltd (1999), shown in Figure 20 .5, stores the measuring data in 6 or 10 channels . This system is contained in a thermally insulated housing, equipped also in internal cooling systems . E Figure 20 .5 Thermally insulated storing and recording system for longitudinal temperature distribution of a continuous furnace (Courtesy of DATAPAQ Ltd .) SALT-BATH FURNACES  403 Technical parameters of the system are : "  accuracy : ±0 .4% or ±1 .1 ° C, "  thermocouples : K, N, R, S, or B-type, "  channel number : 6, 8 or 10, " resolution : 0 .1 °C, "  sampling rate : up to 0 .1 s, "  storing capacity : up 11000 points per channel, "  start of recording : set by hand or at preset temperature or time . The permissible times and temperatures, the system can remain in the furnace, are as follows : "  Heat treatment of TV kinescopes : 210 min, up to 450 °C . " Aluminium heat treatment : 600 min, up to 550 °C . "  Glass products : 20 min, up to 600 °C . "  Steel heat treatment : 180 min up to 1300 °C . After completing the recording, the PC software supplied enables the display of the following data and functions : "  Temperature vs time in particular points . "  Maximum temperatures and rate of temperature changes . "  Comparison of any two temperature profiles . "  Three-dimensional display of dynamic temperature field . 20 .4  Salt-Bath Furnaces Salt-bath furnaces, which operate at temperatures up to 1300 °C, are used mostly for heating steel tools in hardening processes . They require continuous temperature measurement . Some special alloy steel tools require very precise temperature control of about t2 °C . When the working temperatures are very high and also when the heat transfer between the molten salt and steel charge is very intense, sufficient precision may be achieved by only measuring the salt temperature . Methods employed depend upon the specific conditions . For example, K-type immersion thermocouples are used up to 1000 °C, while S-type and B-type are used up to 1300 °C, B-type is the most stable one . When there is a highly corrosive salt influence with high operating temperatures, correct choice of the sheath material and sheath design are critical, deciding factors, as well as proper technique and area of application . Heat resisting alloy steels and mullit are the most commonly used sheath materials . Non-porous ceramic protection tubes are necessary to prevent contamination of the thermocouple which could lead to changes in the emf versus temperature characteristics . Metal sheaths should either be cast or drilled from full cylinders . Welded tubes are unsuitable because welds are not corrosion resistant . Before immersion they should be heated up to nearly normal operating temperature and then, for example, a drilled CrNi-steel sheath of diameter 27 mm could stand continuous operation at 1250 oC in a special hardening salt (91 % BaCl, 2 .5 % MgF2, 404  TEMPERATURE MEASUREMENT IN INDUSTRIALAPPLIANCES Borax) for about 7 to 10 days . To minimise the highly corrosive influence of salt vapours, all electrical equipment should be well protected or placed in another room . Salt temperature can also be measured by total radiation pyrometers or photoelectric pyrometers . Although the equivalent emissivity of molten salts equals unity, most errors are caused by the slag layer on its surface which is always cooler than the salt itself . Immersing and withdrawing of charges also temporarily obscure the field of view of the pyrometer . The best results are obtained by applying pyrometers with peak-pickers(Chapter 12) . In indirect measurements by pyrometer the instrument is directed inside a ceramic tube with a closed end, immersed in the molten salt . At 1/d>6 (Section 8 .2 .) the interior of the tube can be regarded as a black body . Application of such a sighting tube, which limits the disposable bath volume to some extent, requires a certain air-pressure to be kept inside the ceramic tubes to prevent salt from diffusing into it . Sighting tubes, equipped with a prism, which permits the pyrometer to be mounted horizontally, are also used . 20 .5  Glass Tank Furnaces Characteristic points for temperature measurements in a glass tank, are described in an application sheet of Land Infrared Ltd (1997) . They are either relevant for the technological process or for the work-time of the tank lining . Pyrometers and thermocouples may bothbe used . Thermocouples, which are used for the temperature measurement ofmolten glass at temperatures up to 1300 °C, may, be either B-typeor more rarely S-type . Protecting tubes which should resist corrosion and have sufficient mechanical strength, should neither colour nor pollute the molten glass in any way . Greenberg (1975) asserts that molybdenum or platinum are most commonly used . The thermocouples are mounted through holes in the bottom . MI thermocouples, when also used, are placed in permanently mounted A1203 protecting tubes, permitting easy exchange of the thermocouple from the outside (Greenberg, 1975) . The application of photoelectric pyrometers, aiming at the inside of A1203 sighting tubes, mounted from below through the bottomand protected outside by the molybdenum tubes, is a competitive solution . Water-cooling of the pyrometer housing is usual . The temperature of the furnace roof can be measured by sheathed thermocouples inserted from above . As their thermoelectric characteristics change rapidly in the given operating conditions, measuring errors can amount to about 70 °C after only one monthof operation . A certain extension of their lifetime can be achieved by maintaining nitrogen at high-pressure inside the sheath . Total radiation or photo-electric pyrometers give better performance . They are directed inside a closed end ceramic sighting tube, inserted through the roof for a length of about three times diameter . Gas diffusion into the tube must be prevented by maintaining an air over-pressure inside it . The best method of measuring the roof temperatureuses a pyrometer which is directed at the roof through a hole in one of the walls . A proper choice of the spectral response of the pyrometer can only be made after the presence of flames in the furnace has been considered . Early detection of places, where the wall lining may fail, requires periodic checking of the temperature of the outside wall surface . Portable contact thermometers, pyrometers, infrared imagers and sometimes thermocouples placed in the wall-lining during bricklaying might be used for this purpose . INDUCTION HEATEDCHARGES  405 For measurement of the temperature of furnace partitions, photoelectric pyrometers of high distance ratio are used . They have to operate at effective wavelength outside the absorption bands of C02 and air vapour as pointed out in Chapter 11 . For temperature measurement of the outflowing glass at up to 200 °C ambient temperature Si-photoelectric fibre optic pyrometers, which do not need any water cooling but only the air purge of the lens (Ircon Inc ., 1997), are used . 20 .6 Induction Heated Charges In induction heating, where metallic or semiconductor charges are placed in an alternating magnetic field with a frequency ranging from 50 Hz to about 1 MHz, the induced a .c . currents cause charge heating . Electromagnetic waves penetrating the charge are damped, with heating confined mainly to a depth 8, called the penetration depth . This depth, which depends on frequency and material properties, is given by : S = 5050  f  cm  (20 .8) where p is the charge resistivity, in Qcm, Pr is the relative magnetic permeability of the charge, and f is the frequency in Hz . In contact methods only thermocouples are used in a similar way as that shown in Figure 20 .6 . The presence of alternating magnetic fields causes some additional problems due to induction heating of the thermocouples themselves, especially at medium frequencies from about 500 to 10 000 Hz or high frequencies in the range 60 kHz to 1 MHz . Michalski and Eckersdorf (1981) assert that the use of thin thermocouple wires can prevent this effect . The specific power density in a thermocouple wire, when considered as a cylinder, is : Ps = H Z P Fr W/cm 2 (20 .9) where H is the magnetic field strength in A/cm, p is the resistivity of thermocouple wires in Qcm, S is the penetration depth in cm given by equation (20 .8), and F r is the shape factor of a cylinder as described in Davies and Simpson (1979) . When the wire diameter d < 0 .5 8 and for F r -~ 0 , parasitic thermocouple heating can be neglected . As platinum has the lowest specific resistivity of all the thermocouple CHARGE INDUCTOR THERMOCOUPLE OO O b C~~ i ~>s OO O O O°° Figure 20 .6 Thermocouple sensor in an induction heated charge . 406  TEMPERATURE MEASUREMENT IN INDUSTRIALAPPLIANCES materials, it also has the smallest penetration depth at a given frequency . Consequently platinum wires should have the smallest of all diameters . For example at 1000'C the penetration depth of platinum is 3 .6 cm at 8 kHz and 0 .5 cm at 300 kHz . Evidently the condition d < 0 .5 S is easily met for all thermocouple materials, especially as u, =1 in most cases . Michalski and Eckersdorf (1981) and Rosspeinter et al ., (1972) indicate that another way of avoiding parasitic thermocouple heating is to place them in the charge at a depth, 1, larger than the penetration depth S as shown in Figure 20 .6 . Parasitic emfs, induced in a thermoelectric circuit, may cause both measuring errors and damage to the indicating instrument . To prevent these the following precautions can be taken (Chakraborty and Berezovich, 1980) : " placing the thermocouple deeper than the penetration depth, as mentioned above, " shielding of the thermocouple, for example by using a MI thermocouple with isolated measuring junction, " applying grounded shields along the compensating and connecting leads, "  applying transposed connecting leads, "  arranging thermocouples along the equipotential lines of the magnetic field, "  applying RC filters at 50 Hz and RL filters at high frequency across the terminals of the measuring instrument to remove the a .c . component . Filters cause a slight increase in the inertia of the indications . The charge, the shield and the measuring instrument essentially require a single point ground only at the shield of the thermocouple, irrespective of whether the charge is to be grounded or not . Contactless, pyrometerc methods, which in most cases present the only possible solution for temperature measurement of induction heated moving charges, are increasingly popular . The application of photoelectric fibre optic pyrometers is advised for the following reasons : 1 . Easy targeting of the charge by the optical system of very small field of view, d=1 to 3, between the inductor coil windings, from a distance of about 10 to 20 cm . 2 . The small optical head in ceramic housing is not heated in the h . f . magnetic field . 3 . The length of the light guide, over about 50 cm enables the electronic system of the pyrometer to be kept outside the strong h .f . magnetic field . 4 . The 95% response time of about 1 .5 to 50 ms permits the measurement of rapid temperature variations, which are characteristic for induction heating . To avoid the influence of emissivity on the readings it is advised that two-wavelength ratio pyrometers are used . A typical example is the Fiber Optic Ratio Thermometer MRIF from Raytek Corp . Good results can also be obtained using two-wavelength photoelectric lens pyrometers of small distance ratio, which enable the target temperature to be measured between the inductor coil windings . A good example is the TempMatic 8000 Series stationary, two- wavelength photoelectric pyrometer by Williamson Corp (1997) (given in Figure 20 .7) which has the distance ratio Ild = 130 . The pyrometer readings are independent of target emissivity variations and the presence of steam or smoke . An advantage of the pyrometer is also that its readings are correct even in the case of 95% obscuring of the view field . It