Some Myths and Facts About Temperature Sensing: Part I
Contact temperature-sensing devices, such a thermocouples, RTDs (resistance temperature detectors), thermistors, bimetallic thermometers and liquid-filled sensors, measure the temperature of their sensing tips; that is, they do not measure the temperature of gases, liquids or solids surrounding the sensing tip unless certain requirements are met. Differences in the thermal properties of the sensor and the media surrounding the sensor (e.g., thermal conductivity, thermal diffusivity and emissivity) can produce large differences in the observed temperature readings. Additionally, isotherms (areas of constant temperature) within the gas, liquid or solid material can produce large differences between observed and actual temperatures. The test engineer (user) must consider these factors before choosing the particular device design. This article describes various situations where temperature measurements are made, and studies temperature-sensor characteristics under dynamic and static conditions. Where appropriate, typical test data are presented to illustrate the performance of temperature sensors under actual tests.
Selecting the correct temperature sensor requires a basic knowledge of the temperature profile and the factors that influence the profile within the wall and adjacent gas or liquid. The temperature sensor must measure the desired temperature without disturbing this temperature by its presence, and also must have sufficiently fast response time to follow temperature changes accurately.
Consider a hot gas or liquid on one side of a wall and ambient temperature on the other side. The wall could be a chamber such as a furnace or pipe through which a gas or liquid is flowing. Figure 1 shows the temperature profile across the wall at a given time for a case where the wall is made of one homogenous material such as steel or ceramic. For a wall consisting of several materials, such as steel, insulation and graphite or ceramic, each homogenous material has its own temperature profile across its cross section. The following observations are made from Fig. 1:
- Heat always flows from the hotter medium to the cooler medium
- Heat energy is continuously absorbed by the wall at its hot side and released at the cold side to the cooling medium (e.g., air, water, etc.); under steady-state conditions, heat absorbed must be equal to the heat released or the wall will begin to melt.
- The temperature profile at each interface is asymptotic.
- There is no region where temperature is constant over a given cross section. Constant temperature zones are called isotherms, which occur at right angles (normal) to the plane of heat flow. Adjacent to the hot wall there is a laminar layer of gases, which tend to inhibit the heat flow and serves as an insulator. One could imagine the laminar layer as a layer of heat insulating tape that functions like another material having its own inherent thermal properties, and, therefore, having its own temperature profile across it.
Five distinct areas for temperature measurement in the above example are:
- Exterior wall surface temperature (cool side)
- Interior wall temperature at a specific point
- Wall surface temperature (hot side)
- Local gas/liquid temperature at the interior surface of the wall (hot side)
- Gas temperatures on the hot side of the laminar layer
These areas together with sensor requirements are discussed below.
Exterior wall surface temperature
The temperature sensor must be in intimate thermal contact with the wall. Also, the sensor thickness must be small relative to the wall thickness. Ideally, a flat, two-dimensional (2-D) sensor is the best for this application. If the wall is nonmetallic, a small-diameter sheathed thermocouple can be attached by bonding it to the wall. The diameter of the thermocouple must be small compared with the wall thickness. The probe must also be installed so it is in intimate thermal contact with the wall for a distance of 30 times its diameter. For a metal wall, an intrinsic-type thermocouple can be used. Each leg of the two elements should be flattened and welded to the metal wall in close proximity to each other. A schematic of a typical installation is shown in Fig. 2. Typical response curves obtained from various flattened and round wire sensors are shown in Fig. 3. More details on this technique are contained in .
Interior metal wall temperature
The temperature sensor must have the same thermal properties (i.e., thermal conductivity and thermal diffusivity) as those of the wall, and it must be accurately located within the wall and make good thermal contact with the bottom of the blind hole. Spring-loaded thermocouples are used for this type of application. Also, the sensor hardware, such as the protection sheath, mounting hardware, etc., must not introduce large conductive paths for heat transfer to occur.
The type of thermocouple assembly shown in Fig. 4 with in pocket depth installations up to 1.5 in. (38 mm) has inherent errors caused by the depth of the pocket, nipple cap, air currents in the vicinity of the heated cylinder and the position of the thermocouple relative to the top, bottom or side of the heated cylinder. Bayonet-type thermocouples show the following errors in controlled tests :
- Air drafts from open doors and windows, air circulating around heated cylinders and cooling air caused errors of up to 90 F (~50 C) in thermocouples installed in pocket depths of 0.5 in. (13 mm) using conventional-style twist-lock nipples.
- Errors from 7 to 36 F (~4 to 20 C) were observed simply by positioning the thermocouple in the top, bottom or side of a heated cylinder. An error in measurement was also noted with respect to the type of nipple used to install the thermocouple in the wall. All thermocouples in this test were also at pocket depths of 0.5 in.
- Errors of 18 F (~10 C) were measured in pocket depths of 0.5 in. In pocket depths of 1.5 in., the error was reduced to about 4 F (~2 C).
- The additive effects of these errors can easily create an overall error of 90 F or more in shallow-depth applications of conventional bayonet-type thermocouples. Therefore, a more precise temperature sensor is needed, particularly in applications where the pocket depth is 1.5 in. or less.
New designs have been developed for shallow pocket depths. A right-angle ribbon type (one of the newer designs) was specifically developed to eliminate conduction errors caused by temperature gradients . The design is a thermally isolated, or adiabatic, probe, which measures the actual local temperature independent of conduction effects. Controlled-gradient tests show no measurable errors in temperature, and response times are a few milliseconds. Figure 5 illustrates a modified bayonet thermocouple with ribbon elements. Typical applications of the spring-loaded thermocouples include thick-walled chambers and cylinders such as gun barrels, injection molding machines, extruders, etc.
Interior nonmetallic-wall temperatures
To determine thermal properties and heat flux in certain experiments, the temperature history of an interior point in low-conductivity materials is frequently obtained from a thermocouple appropriately installed in the material. Large errors can be produced by the presence of the thermocouple itself if certain precautions are not observed. The errors are pronounced when the thermocouple is installed in materials such as Teflon‚, nylon, phenolic, silica phenolic and similar low-conductivity materials.
The parameter to consider is the ratio of the thermal conductivity of the base material to that of the thermocouple assembly. If the ratio is significantly less than 1.0, temperature errors as large as 700 F (385 C) can be created . In transient cases, even larger errors are produced. Methods to reduce the error include:
- Designing a thermocouple assembly having a thermal conductivity matching that of the measured material
- Reducing the radius of the thermocouple wire, or increasing the surface area/cross-sectional area ratio
- Placing the thermocouple and adjacent lead wire parallel to the plane of heat flow
Figure 6 shows the essential features of a thermocouple specifically designed to meet these requirements. The thermowell material is the same as that of the test wall. The thermal junction and extension leads in the immediate vicinity of the junction are of ribbon form, which yields the largest possible surface area/cross-sectional area ratio for a given wire size. Both the junction and the lead wires are in the same plane.
The thermocouple, when installed so the probe is normal to the plane of heat flow, produces the most accurate in-wall temperature histories in low-conductivity materials. The sensor can be made in any standard pair of elements with fiberglass insulation, from No. 24 gage down to No. 36 gage wires. The probe length and diameter are optional, limited only by the machinability of the material used; a 0.1875-in. (~5 mm) diameter wire is convenient for most materials. All probes should be potted into the mating hole with suitable bonding agents.
Wall surface temperature (hot side)
A surface thermocouple must match the thermal properties of the wall, must not disturb the surface contour of the wall, must be two-dimensional, must have low millisecond response time and the thermal junction must be self-renewing if ablation and/or erosion are present.
The self-renewing thermocouple shown in Fig. 7 meets these requirements . This unique thermocouple design  uses flat ribbons, mica insulation and a sensing junction formed by abrasion. The design has a two-dimensional surface measuring junction with microsecond response times to surface temperature fluctuations. The sensor thermowell can be made of any machinable material, thus matching the thermal properties of the wall precisely.
The thermal junction is formed by performing a simple abrasive action across the sensing surface using a medium-grit size abrasive paper. The sanding and polishing action produces thousands of microscopic hot-weld junctions that join one ribbon to the other ribbon. Since the thermal junction is formed via an abrasive action, any additional erosion from use simply removes the old junctions while simultaneously forming new junctions. Thus, the thermal-sensor design has a self-renewing feature, which is useful in applications where the wall is subject to wear.
Figure 8 shows temperature and pressure versus time of a shock wave as it passes the location of the two sensors. A shock wave is generated by evacuating the shock tube, filling the tube with a combustible gas and igniting the gas electrically at the closed end of the tube. The measured response time of the self-renewing thermocouple in this test was 8 microseconds, which is equivalent to a rate of 28 x 10(2) F/sec (15.5 x 10(6) C/s). This unique thermocouple has been used in many transient applications. Typical examples of temperatures recorded by this thermocouple are shown in Figs. 9 and 10.
Gas or liquid temperatures inside the chamber
The gas region begins at the boundary layer at the surface of the hot wall and extends to the opposite wall. In steady-state flow, gas temperatures follow a bell-shaped profile; that is, it is coolest at the wall surfaces and reaches a maximum at the center of the flowing gas. (In cryogenic flow, the opposite occurs; that is, the cryogenic gas or liquid is coolest at the center of the flow and warmest at the wall surfaces.) In furnaces or closed chambers, the gas is not moving, but there are isotherms within the gas (this will be discussed in Part 2 of this article).
To accurately measure the gas temperatures, the sensor must meet two requirements:
- The sensing tip of the sensor must not disturb the local temperature of the gas by its presence.
- The sensor must lie in the plane of an isotherm.
Special types of ribbon thermocouples meet these requirements as shown in Fig. 11. Note that the ribbons and junction are heated simultaneously, and conduction along the stem is eliminated. If round wires or round metal sheaths are used in the sensor design, the tip of the sensor must be bent into an "L" shape for a distance equal to at least 20 diameters of the sheath.
Look for Part 2 of this article in the September 2004 issue of IH.
- The sensing tip of the sensor must not disturb the local temperature of the gas by its presence.