A PVC pipe manufacturer interested in energy consumption invited his local electric utility into his plant to make some measurements. As extruder heaters accounted for a large portion of the energy bill, the company logged the kilowatt-hours consumed over 24 hours on one typical extruder's barrel zone. Then, a different brand controller with a few extra control parameters was substituted, and the test was repeated under the same conditions. As table 1 shows, the test using the enhanced control consumed significantly less energy.
The plant had 30 such zones on five extruders. If the plant operated 330 days per year, replacing the controllers could save the processor approximately $90,000 per year. In addition, the test did not include the costs of chilling cooling water. Five 30 to 35 hp compressors supply chilled water to the five lines, mostly for barrel zone and cooling trough water.
The results call for a closer look at how the control process works.
On Controller 1, the heat and cool output lights were pulsing every few seconds, indicating that the controller constantly alternated between heating the zone and dumping heat into the cooling water. Because it held the temperature within tolerable limits and the product was acceptable, in most plants the controller would not attract much attention. By contrast, Controller 2 spent most of its time doing nothing. Occasionally, it sent a short pulse of heat or cooling water to the zone to hold temperature within 1oF.
Let's track all this heat. It goes into:
- Bringing the zone up to processing temperature at startup.
- Continuously melting the incoming polymer.
- Supplying radiation, conduction and convection losses from the machine to the environment. Some small energy savings could be realized by using thermal insulation on the barrel heaters.
- With Controller 1, a fight between the heater and the cooling process.
Also, consider how controller adjustments influence heat/cool cycling. For example, if the proportional band is small, a small decrease in zone temperature will give an overdose of heat. The controller then will overcorrect the temperature and bring an overdose of cooling. Control loop instability will continue and can cause large temperature swings. Increasing the proportional band will bring a more gentle response to temperature deviations and is the first step to achieving stability.
After finding the heat proportional band that provides control stability when correcting for temperature drops, it is important to ensure that a temperature increase brings the same-strength corrective action from the cooling system yet maintains control stability. The controller's relative cool parameter serves this function. It is put into controllers because zone cooling (particularly water) often is much stronger than zone heating. Also, zone cooling and water shut-off valves sometimes are left in unknown positions, so the zone's full cooling capacity is not readily known. For a high capacity cooling process, set the relative cool parameter well below 1.0. (A 1.0 setting is valid when the heater and cooling process have equal capacity.)
The job of integral action is to watch for and slowly correct deviations from set temperature. A too low integral-time setting makes this action too eager and brings control instability. Too high a setting will make the correction of persisting deviations somewhat slow but does not harm stability.
Sometimes called rate action, derivative action watches for temperature changes and produces heat (or cooling) in proportion to the rate of temperature change, in the direction to oppose that change. Its contribution ceases as soon as the temperature stops changing. A larger value setting provides a stronger response but can lead to large power and temperature swings and excessive energy consumption if it is set too large.
Finally, most controllers have a self-tune feature that allows it to analyze the controlled process at startup or on-demand and adjust the controller parameters for stable and optimum performance. This can be a great time-saver, but don't let it override your knowledge of the process and the control principles involved.
The brand of controllers used in the test is less important than the availability of useful control parameters. Keep process behavior in mind when adjusting them.