Thermocouples

Measurement and control of temperature is one of the most common requirements of industrial instrumentation and the thermocouple is by far the most widely used temperature sensor. Its characteristics include good inherent accuracy, suitability over a broad temperature range, fast thermal response, ruggedness, high reliability and low cost.

How does a thermocouple work?

T.J Seebeck discovered in the 1820s that an electric current flows in a closed circuit of two dissimilar metals when one of the two junctions is heated with respect to the other. In a thermocouple circuit the current continues to flow as long as the two junctions are at different temperatures. The magnitude and direction of the current depends on the temperature difference between the junctions and the properties of the metals used in the circuit. This is known as the Seebeck effect. Click here to see an example of the circuit.

If the circuit is broken at the center, the net open circuit voltage (the Seebeck voltage) is a function of the junction temperature and the composition of the two metals.

If the hot and cold junctions are reversed, current will flow in the opposite direction. Any two dissimilar metals can be used and the thermocouple circuit will generate a low voltage output that is almost (but not exactly) proportional to the temperature difference between the hot junction and the cold junction. The voltage output is between 15 and 40V per degree C, dependant on the thermocouple conductor metals used. The actual metals used in industrial thermocouples depend on the application and temperature measurement range required.

Thermocouple failure prediction

Like any other metal object, thermocouples are subject to metal fatigue wear and tear; they have a finite life. Many users of thermocouples are not aware of thermocouple deterioration until the sensor breaks, often causing an expensive interruption of a process. Removing a thermocouple from a furnace when at operating temperature can be difficult and dangerous. In fact the thermocouple, a simple and generally inexpensive sensor, can cause inaccurate readings for some time before any errors are detected. The errors usually cause low readings due to the thermocouple wires becoming thinner.

Impurities induced by any handling during manufacture or installation can accelerate chemical deterioration of the thermocouple. For base metal thermocouples, deterioration occurs slowly due to contact with the atmosphere, which in turn causes oxidation. As the surface of the thermocouple wires oxidises the current carrying cross sectional area is reduced. Nobel metal thermocouple deterioration is also well documented.

In "Principals and Method of Temperature Measurement", Thomas D McGee explains that the usual result of deterioration is the gradual reduction in the Seebeck voltage, often extended over several weeks and not frequently detected. If the Seebeck voltage is low, the measured temperature will also be low, so the actual process temperature will be increased to produce the required Seebeck voltage. The net result will be excessive temperature generation with resulting damage to material and processes. Those who use thermocouples should be aware of the possibilities of slow deterioration and its consequences.

A temperature controller, for example, would actually compensate for the thermocouple's loss of thermoelectric power by putting more heat into the process with all the energy, environmental and process plant costs that would be incurred. Fortunately, while Mr Thomas Johann Seebeck was experimenting with his wires in the 1820s, his contemporary and fellow countryman, Mr Georg Ohm, was also conducting his own experiments. Fortuitously because as the thermocouple conductors become thinner, their resistance changes as described in "Practical Temperature Measurement" by Peter R. N. Childs.

"The loop resistance of a thermocouple depends on its length, type and diameter of the thermocouple wire, the length type and diameter of extension wires, temperatures along the circuit and the contact resistance at any connections. If on installation, and at regular intervals in use, a measurement is made of this loop resistance, then a change in this value can be used to indicate wire thinning due to chemical attack, loose or corroded connections, contact resistance due to broken but touching wires or electrical shunting due to loss of insulation at some location along the wire."

Regular measurements of the thermocouple loop can indicate that the sensor should be replaced for reasons of accuracy and can also be used to predict its complete failure (sensor break). As thermocouple conductors oxidise they become brittle, making them more susceptible to breakage due to bending or vibration. Replacing thermocouples during a planned maintenance period is easier and more cost effecting than replacing thermocouples while the plant is running.