Getting Control of Cooling Processes
Many engineers and designers believe that specifying an appropriate temperature control for a cooling application is not the critical design decision it is when specifying a temperature control for a heating application. In some applications, this may be true due to the unique process dynamics associated with those cooling processes. In others, however, considerations remain that make temperature control selection an important factor in the eventual success -- or lack thereof -- of the cooling or refrigeration process. When specifying a temperature controller for cooling applications, the control engineer has several basic control options from which to choose:
- On/off control.
- PID control.
- Fuzzy logic control.
On/off control is the simplest and most often specified temperature control utilized in cooling applications. This type of control is just what it sounds like: The output will turn on at a certain temperature and will turn off at another temperature. The difference between these two temperatures commonly is known as the deadband, but it sometimes is referred to as the on/off differential.
The deadband can be preset from the factory as a function of temperature range such as in a bimetallic thermostat. Alternatively, the deadband can be user adjustable such as it is a digital temperature switch. Consideration regarding deadband size must be made relative to where the temperature sensing element is located. Typically, when the element is on the product and controls it directly, the deadband is relatively small, say 2 to 4oF (1.1 to 2.2oC). However, if the element senses space temperature, the deadband usually is larger and values of 6 to 8oF (3.3 to 4.4oC) are more typical. Because many cooling applications are, on average, well insulated or otherwise static in nature and temperature repeatability is not critical, on/off control can be utilized effectively in most cooling applications. In fact, a good example of on/off control is the temperature switch in your household refrigerator.
Used in conjunction, all three control terms will allow for the most precise and stable process control. However, in many cooling applications, only the proportional term is used to prevent excessive output cycling and to increase process responsiveness. To make life a little easier, most temperature controls have self-tuning features that will calculate all or some of the PID terms automatically. However, use caution -- many of these self-tuning algorithms do not work in the direct-acting (output increases when input increases) or cooling mode. If you want to use a self-tuning algorithm, ask the temperature control manufacturer if its autotuning algorithms work in the direct-acting or cooling control mode.
Fuzzy logic control has one distinct advantage over standard PID control: Fuzzy logic can account for process nonlinearities or other uncertainties such as intermittent opening of refrigerator doors or seasonal changes in cooling water temperature. PID control is based on purely mathematical linear process modeling. Many real-world applications do not act in a completely linear manner because the variables are too numerous to model. In these situations, fuzzy logic can offer better control. In addition, cooling output changes are minimized with fuzzy logic control, which thus reduces component wear while at the same time increasing energy efficiency.
6 Questions to AskOnce the designer has determined what type of temperature control is best for the application, other factors must be considered before the final temperature control selection can be made. The designer should always ask the following six questions when specifying a temperature control:
- What Degree of Stability Is Required in the Process? The desired degree of stability must be determined. For example, most laser manufacturers require cooling stability within 0.1 to 0.2oF of setpoint while many injection-molding machines typically require stability within 1 to 2oF (0.6 to 1.1oC). In contrast, a welding application may only require 2 to 3oF (1.1 to 1.7oC) stability. All of these applications may be better served with a PID or fuzzy logic control rather than a simple on/off control.
- What Are the Unique Process Dynamics? Determine unique characteristics of your processes. For example, ambient temperature may affect the process in such a manner that the control setpoint must be changed. Some manufacturers have temperature controls that allow the control setpoint to be automatically changed based on ambient or other conditions. Also, the designer must determine if the process is thermally insulated or to what extent external variations will influence the process behavior.
- What Outputs Are Required on the Temperature Control? Unlike most heating applications, many times the temperature control in cooling applications switches an inductive load (solenoid valve, relay, motor, etc.). In these types of applications, the control engineer must make sure that the temperature control is protected from inductive noise spikes by using an R/C snubber. In addition, if the control is expected to cycle constantly, a logic or solid-state relay output may be preferred over a traditional electromechanical relay output.
- What Type of Temperature Sensor Will Be Used? An entire article or book could be written about proper temperature sensor selection. However, one should consider the mounting requirements, desired response time and required accuracy of the temperature sensor when making this selection. It also should be determined if the control will accept the sensor input or if the control requires a specific input type. Many on/off or thermostat controls require manufacturer-specific thermistor inputs while many microprocessor-based PID controls can accept multiple standard temperature sensor inputs, including thermistors, thermocouples, thermal diodes and RTDs.
- What Type of Equipment Longevity Is Required? If excessive or continuous cycling of equipment is a concern to the designer, perhaps on/off or proportional-only control would be the ideal control type. Although PID or fuzzy logic control may provide for more stable process control, the constant output cycling might cause compressors or other cooling devices to wear out and fail prematurely.
- Will Other Control Options Be Required? Many control applications require additional control options such as serial communications, analog process value retransmission, alternating outputs (lead/lag) and alarms. Many industrial applications also require specific serial communication protocols such as SPI or Modbus-RTU; others may require digital inputs to indicate such items as low flow, low pressure, etc. One also may desire to have the control outputs turn off if certain conditions are met or not met.
Of course, even if all of these considerations are made and the right temperature control is selected, the refrigeration or cooling system must still be properly sized. Even the best temperature control will not be able to maintain or perhaps even reach setpoint in an undersized system.