Obtaining accurate temperature readings in a high-temperature application is difficult. Thermocouples and pyrometers have advantages and disadvantages. This article will address the use of pyrometers in applications where the need for accurate and precise thermometry is crucial.
|Fig. 1. Low-emissivity and high-emissivity bowls|
Direct-contact measurement using a thermocouple is usually the most inexpensive and accurate temperature-measurement method. However, the challenges associated with using a thermocouple become almost insurmountable at very high temperatures. For example, graphite embedment at temperatures above 1300°C (2372°F) requires a platinum sheath – a very expensive proposition. Furthermore, the sheath composition may need to be changed, depending upon whether the atmosphere is oxidizing or reducing. High-vacuum applications present even more challenges. The result is substantial cost.
In response, several manufacturers offer noncontact optical pyrometers. These devices have dramatically simplified the challenges associated with high-temperature thermometry. Pyrometers have become the instrument of choice, even in applications where they are inappropriate or poorly suited. On one hand, these devices have revolutionized the furnace industry. They are capable of deriving a temperature in a variety of difficult applications where thermocouples could never be used. On the other hand, there is a fundamental misunderstanding about their limitations.
Pyrometers are used to derive the temperature of an object without touching it. They do this by using special detectors that measure the amount of infrared energy being radiated from an object. The intensity of this radiant energy is proportional to the object’s temperature. Pyrometers are designed to detect this emitted energy. They then compute an equivalent temperature based on blackbody curves, which are offset to accommodate emissivity. Emissivity is the ratio of infrared energy an object will emit at a given temperature when compared to a blackbody of the same temperature.
A blackbody is a perfect radiator of infrared energy. For a given temperature, a blackbody will emit 100% of its infrared energy. Thus, the emissivity of a blackbody is always 1.00. All other objects in nature will emit less infrared energy than a blackbody of the same temperature. When the emissivity setting of a pyrometer is adjusted to a value of 1.00, it will only be accurate when pointing at a blackbody. When it is pointed at anything else, the pyrometer will display a temperature that is lower than actual. Thus, in order to be accurate, the emissivity of the object must be known. This value must be programmed into the pyrometer.
One way to determine emissivity is to simultaneously measure the actual temperature of the object using a thermocouple. As the emissivity setting is adjusted downward, the temperature displayed by the pyrometer will rise. The emissivity of the object will be discovered when the temperature displayed on the pyrometer equals the actual temperature of the object. Unfortunately, determining emissivity by this method is not always feasible. As a result, emissivity is most often determined by using an emissivity table. These tables provide an emissivity value for many common materials, but only at a fixed wavelength. The problems associated with using a pyrometer usually begin with an emissivity table. These problems are then compounded by a basic misunderstanding about emissivity and reflection.
|Fig. 2. Low-temperature heating test|
Reflection vs. Absorption
A sheet of aluminum foil has an emissivity of 0.05 at a wavelength of 8-14µ, which means that it will reflect 95% of the infrared energy that strikes it. The foil itself will radiate only 5% in the infrared. If we measure the temperature of the foil using a pyrometer, we must take into account any infrared energy reflecting off its surface. Otherwise, the pyrometer will assume this energy is emanating from the foil, and the resulting display temperature will be much higher than actual. In fact, a pyrometer pointed at the surface of aluminum foil can detect the reflected body heat of a nearby person. Since the emissivity of aluminum foil is low, it will emit very little infrared energy, even when its temperature is very high. In other words, the foil may not seem to radiate much “heat” from a distance, but we could be burned if we touched it.
The emissivity setting of most pyrometers can be adjusted from 0.1 to 1.00. By reducing the emissivity setting from 1.00 to 0.05, the pyrometer will internally amplify the infrared signal and calculate an equivalent temperature based on a 5% infrared emission. It sounds simple enough. Determine the emissivity of the foil, then adjust the pyrometer to that emissivity and presto … accurate thermometry. Unfortunately, this approach will not work because a pyrometer cannot tell the difference between foil that radiates infrared energy and foil that reflects infrared energy.
Using what we’ve learned about aluminum foil, let’s place a block of metal at its center. Let’s assume the block of metal has an emissivity of 0.50 at room temperature. An emissivity of 0.50 means the block will be 50% reflective. When foil and block are exposed to a source of infrared energy, such as a heat lamp, the pyrometer will be unable to correctly derive the temperature of the block – even though we’ve adjusted the emissivity setting to the correct value of 0.50. The temperature derived by the pyrometer will be much higher than actual due to the added reflection from the foil. This thought experiment simulates the effect of placing a reflective object inside a furnace, where it may be surrounded by heating elements, crucibles or other objects that are hotter.
Effects of Reflection
In order to verify this hypothesis, an experiment was conducted using an electric radiant heater. The heater was used to raise the temperature of two identical objects. One object was placed at the center of a low-emissivity (ε = 0.05) bowl, the other at the center of a high-emissivity (ε = 0.80) bowl (Fig. 1). The objects at the center of both bowls had an emissivity of 0.80. Thus, it was anticipated that 20% of the infrared energy emanating from the radiant heater would bounce off the surface of the objects (the remaining 80% would be deposited as heat). At the outset of the test, the ambient temperature inside the oven was 70°F. The oven temperature setpoint was 180°F. The heating element was located above the bowls. After five minutes, the pyrometer was pointed at both objects. The pyrometer displayed a temperature of 204°F for the object in the low-emissivity bowl, whereas the object in the high-emissivity bowl was only 98°F (Fig. 2). When these measurements were made, the ambient temperature inside the oven was 120°F. Clearly, infrared energy from the heating element was reflecting off the surface of the low-emissivity bowl and interfering with the pyrometer. In fact, readings as high as 515°F were observed when the pyrometer was pointed at the interior surfaces of the low-emissivity bowl. Nevertheless, the actual surface temperature of both objects was only 98°F.
Unfortunately, a pyrometer cannot discern between what is being reflected off of an object and that which is being emitted by an object. Consequently, the resulting display temperature will be a composite of the radiant heat source and energy being emitted by the object. Moreover, the larger the temperature gradient (i.e. ∆T), the more inaccurate the pyrometer will become. In other words, if we increase heater power in an effort to rapidly heat an object, even more radiant energy emanating from the heat source will be detected by the pyrometer. As a result, the pyrometer will be more inaccurate than before – something intolerable if the pyrometer is connected to a PID (proportional integral derivative) controller.
This problem is really exacerbated if we are melting metal because the emissivity of most metals will dramatically drop once their melting temperature is reached (while the pyrometer emissivity setting remains fixed). For example, assume our metal block has an emissivity of 0.50 at room temperature. As the temperature of the block is raised, its emissivity remains constant, but it rapidly drops to around 0.28 at its melting point. Unfortunately, the programmed emissivity value is still 0.50. Depending on the amount of reflection, the display temperature could be hundreds of degrees hotter or colder than actual. So, what was the actual temperature of the metal?
|Fig. 4. Emissivity targets|
According to one pyrometer manufacturer, at a wavelength of 1.6µ, a 20% emissivity error should result in a display-temperature error that is somewhere between 30 and 123°C. In other words, if an instrument reads 1969°C (3576°F), the actual temperature should fall somewhere between 1846 and 2131°C (3355-3868°F).
In order to validate this possibility, another experiment was conducted – this time using a high-temperature, graphite sintering furnace. A flat, graphite target was placed inside the furnace. The furnace was slowly heated over a period of 24 hours. It was then allowed to stabilize for six hours. The temperature of the target was measured using a single-color Raytek 2MH optical pyrometer at a wavelength of 1.6µ. The temperature displayed by the pyrometer was recorded as its emissivity value was varied from 1.0-0.80. The result is shown in Figure 3. The temperatures displayed by the pyrometer varied 132°C, over an emissivity range that was varied by 20%. The results validated the theoretical range error that was predicted. The temperature variance that was predicted theoretically and then observed experimentally was 30-123°C.
The effects of reflection will be similar inside a metal-melting furnace. According to this same manufacturer, a 20% emissivity error at 1350°C will result in a display error of 110-470°C. In other words, the actual metal temperature would be somewhere between 1460 and 1820°C (2660-3308°F). If the pyrometer is connected to a PID controller, the controller would drive the furnace temperature hundreds of degrees hotter than necessary, resulting in a tremendous waste of energy and potentially affecting the quality of the casting.
In my own experience, precision (i.e. repeatability) is usually more important than absolute accuracy. Thus, quirks associated with using a pyrometer are an acceptable trade-off, as long as conditions (i.e. the types of materials being heated and the furnace profile) remain fairly consistent. In these applications, time versus temperature is usually taken into account. Thus, temperature data can be weighed against actual heating time in order to ensure relative accuracy, which is possible because furnace power and heating time result in a predictable heating rate.
There are applications, however, such as those relying on dielectric heating (i.e. microwave heating), where temperature cannot be reliably predicted as a function of power and time. In these applications, pyrometer accuracy is essential because the amount of heat produced for a given amount of power applied cannot be reliably predicted. Further, any miscalculation with respect to emissivity and reflection will be compounded by the inherent accuracy limitations of the pyrometer. In terms of instrument accuracy, one manufacturer quotes an accuracy tolerance of 0.5% +2°C for their device. If we add this accuracy tolerance to a 20% emissivity error, the resulting temperature readings would be substantially in error – unusable, in fact.
|Fig. 5. Target time vs. temperature|
Ultimately, a pyrometer really causes havoc in applications where precise temperature control is needed, but reliable thermometry cannot be obtained because of reflection or emissivity variance. In many of these applications, inaccuracy will affect not only product quality, but a major headache is also created for QA (quality assurance) in the process. In these situations, one must be creative. In order to eliminate reflection, temperature readings should be derived indirectly. For instance, the pyrometer can be pointed at a detection target that is specially designed to eliminate reflection and emissivity variance.
If the target is kept in intimate contact with the object being heated, the temperature of the object can be accurately measured – albeit indirectly. Using this method, the target would need to have a cylindrical pit machined into its center. To be effective, the diameter and depth of the pit would need to be carefully chosen based on the distance between the pyrometer and target (i.e. spot size). When the target is placed inside the furnace, heat radiating from the pit would be nearly indistinguishable from a true blackbody. By aiming the pyrometer at the base of the pit, external reflections would be blocked by the cylindrical walls of the target. Consequently, the pyrometer emissivity setting could be left at 1.00 (its calibration point), and highly accurate and repeatable temperature measurements could be made.
In order to validate this method, the electric radiant-heat experiment was repeated. This time, a large cylindrical object was placed at the center of each bowl. The material composition (carbon steel) and mass of both objects were identical. The surface of both objects was painted with high-emissivity (ε = 0.80) paint. One object was placed at the center of a low-emissivity (ε = 0.05) bowl, the other at the center of a high-emissivity (ε = 0.80) bowl (Fig. 4). It was assumed the interior surface of both objects would behave as a blackbody radiator. If the method was viable, the pyrometer would display the same temperature at the center of both objects. At the outset of the test, the ambient temperature inside the oven was 70°F. The heating elements were controlled using an ON/OFF control method and were located underneath the bowls. The oven setpoint was 250°F. During the test, the pyrometer was used to record the surface temperature of both objects and the interior surface of the bowls at two-minute intervals. The heating test was conducted for a period of 30 minutes.
The results of the heating test are displayed in Figs. 5 and 6. The test revealed that the exterior surface temperature of the object inside the low-emissivity bowl read much lower than actual. However, the interior surface temperature was very close to actual. The unusual exterior surface temperatures seemed to indicate that pyrometer accuracy was being affected by the heating elements – in this case, the ON/OFF action of the heating elements was reflecting off the surface of the bowl. This result was not anticipated, as the heating elements were underneath the bowl and the oven walls were not particularly shiny (so it was assumed that the effects of reflection would be minimal). The result demonstrates that temperature accuracy will be dramatically affected if a pyrometer is aimed at an object of high emissivity, but the target surface is not shielded from heating elements or objects of lower emissivity (e.g., shiny furnace walls, a hot crucible, etc.).
|Fig. 6. Time vs. temperature interior bowl surfaces|
The radiant-heating tests clearly show that pyrometer accuracy is dramatically affected by reflection, which is a natural property of any target material. If a target is surrounded by objects of lower emissivity or higher temperature, such as crucibles or heating elements, then infrared radiation from these objects will strike the target and add to the amount of infrared energy detected by the pyrometer. As a result, the pyrometer will measure a higher-than-actual surface temperature. The second test clearly shows that shielding the target surface from external reflection improved accuracy. Had the cylindrical object been a properly engineered detection target, pyrometer accuracy would have been much better.
If a detection target was used in all furnace applications, the effects of emissivity variance and reflection could be eliminated altogether. Ergo, we could heat any object, regardless of emissivity characteristics, and never worry about emissivity tables again. The lesson to be learned here is simple. Consider involving the pyrometer manufacturer. They are very adept at solving complex instrumentation issues. In fact, it was through collaboration with IRCON® and Raytek® that solutions presented in this article were made possible. IH
For more information: Contact Kevin R. Brooks, NSA-Nuclear Safety Associates, Inc., 100 Union Valley Road, Suite 101 Oak Ridge, TN 37830; tel: 865-576-3723; e-mail: email@example.com; web: www.nuclearassociates.com
SIDEBAR: TiN Mold Coating
A ceramic plate, fashioned from a mixture of alumina and silicon carbide, was brush-coated with titanium-nitride mold coating (A). A pyrometer was pointed at the surface of the plate (B), which was then radiantly heated under argon using microwave energy. The plate was heated until the pyrometer displayed 1500°C (2732°F). Note the surface condition of the plate following the test (C). Pyrometer inaccuracy resulted in significant overheating. Consequently, the mold coating was damaged (maximum-use temperature of the coating was 1900°C). Accurate temperature measurements of the ceramic would have been possible if a pyrometer target had been used. Accordingly, the plate could have been reliably heated to 1500°C without damage to the mold coating.
The use of a pyrometer target for ceramic heating applications is a good example of why collaboration with a pyrometer manufacturer is essential. In this instance, the manufacturer was able to provide the necessary product support to ensure the correct device was specified. Moreover, their suggested design for a pyrometer target ensured that the specified device would function as intended.
The heating of ceramics to high temperature is one example of an application where pyrometers are not particularly well suited. Other difficult applications include microwave, laser, plasma, salt baths and electron-beam heating.