A plastics recycling plant used ducted hot air to dry a stream of shredded and washed post-consumer polyethylene bags. Heat was provided by fast-acting, SCR-controlled electric heaters. It was important to keep the temperature high enough for efficient drying, yet not so high that the plastic melted.
The operator commented that the temperature controller was not doing its job, and no combination of PID or other control parameters would stop the severe air temperature cycling. An investigation showed that the control thermocouple in the duct was well made, but the hot junction was enclosed in a strong, 0.5" dia. outer tube.
When the thermocouple was replaced by one with an exposed, low-mass hot junction, response time went from minutes to seconds. The controller, now acting on up-to-date information, provided stable control with a fast response to process stops, starts and disturbances.
In this case, another contribution to good process design was the short time lag between a power change and the sensing of temperature change downstream. Thermal lag has two components. The first, called transport lag, is equal to the air velocity (ft/min) times the distance (ft) between heater and sensor. It is a pure time delay (dead time).
The second component of thermal lag is an exponential lag: the time required for the heater to warm up and start delivering heat. These principles apply equally to fluids and solids in motion. In fact, PID temperature controllers routinely handle exponential lags, but it takes a more advanced model to compensate for dead time and avoid cycling.
For fast temperature sensing, thermocouples generally beat RTDs because of their low mass, simple construction, point sensing and ease of exposure of the sensing element. With RTDs, their construction demands protection tubes for most industrial locations. This protection and the distributed construction of the platinum wire contribute to slow response.
A nonhostile location may permit use of small and fast thin-film wafer construction RTDs. You can protect exposed RTDs and thermocouples from abrasion with a cage over the tip. Small diameter MgO thermocouples with a grounded hot junction are nearly as fast as exposed junctions and provide a choice of atmosphere-compatible sheath material.
When you cannot make contact -- for example, when sensing the temperature of a roller, web or moving material -- consider an optical pyrometer. At a fraction of a second, response time is rarely a problem. Observe the manufacturer's specification on target size (expressed in degrees of angle), ambient temperature limit, optical cleanliness and, especially, compensation for target emissivity. Rather than depend on some listed figure for the emissivity of your material, it is better to measure the target temperature when it is stationary using a thermocouple and indicator. Then, adjust the emissivity setting to bring the optical pyrometer reading into agreement.
What is the right location for your sensor? Close to or in contact with the material you are processing. For example, in an oven, if you cannot make the whole workspace uniform, then one sensor cannot report what all of the work is receiving. Use several zone heaters, each with its own sensor and controller, and trim each zone to get the profile needed on the work.
On the sealer bars of plastic bag-making machines, the thermocouple often is attached under a screw and washer on the top or side of the bar. In this position, it senses more of the heater temperature than the sealing edge temperature. Instead, slide a small, long MgO thermocouple deep into the bar, close to the sealing edge. It now senses the variations in sealing temperature as the material wipes heat away each time a seal is made. It will also "see" the temperature wanting to rise as soon as the machine pauses. With this arrangement, the controller can deal with disturbances as soon as they occur.
Your attention to process design will improve control and lower control system costs.