Successful control of your cooling process and selection of a suitable control require a good understanding of the flow of heat in that process. No matter how intelligent or elaborate a control is, it cannot overcome the limits of basic thermodynamics in a system or process. Heat applied to a process or generated within it by exothermic reaction will cause a steady rise of temperature — regardless of the control system — unless a path exists for removal of the heat energy.

Usually, provision for a suitable flow of heat out of a process is not a difficult task. It requires little more than balancing the heat applied (in BTU/hr or similar units) with the heat removed (by passive conduction, convection, radiation or active cooling) over a suitable period of time.

Only after this balance is designed should a control system (electronic or mechanical) be considered to keep process temperature within desired limits during operation. The selection of a suitable control system then is subject to another set of limitations. Accuracy, repeatability and speed of temperature control cannot be arbitrary. These parameters also must respect thermodynamic and practical limitations as well as cost.


Basic Heat Flow

How fast can heat accumulate in your process? A common measure is BTU/hr (or equivalent units). This value is available from heater specifications or via knowledge of exothermic heat generation in the process.

How fast can heat be removed from your process? This may be a more difficult value to procure but no less important. It can be obtained from conduction, convection or radiation charts, manufacturer specifications or calculated with knowledge of system dimensions, materials and environment.

Is the rate of heat removal greater than its generation? If so, then it will be possible to control the process temperature; otherwise, even the most elaborate control system available will be unable to help.


Selecting a Sensor

Sensor type will have a direct bearing on cost and performance. Select a sensor that meets your expectations for performance rather than specifying a particular type (thermistor, RTD or thermocouple) simply because it is familiar.

The most energy-efficient types are thermistors and semiconductors because they offer the best system stability at the lowest cost. If these types do not offer sufficient accuracy or range, then lower sensitivity but higher accuracy RTDs might be appropriate. Thermocouples offer extended temperature ranges and a fast response time, but they have the lowest stability and accuracy.


PH 1121 Thermalogic Desired Temperature Control Temp Sensors

Sensor type has a direct bearing on cost and performance. Select a sensor that meets your expectations for performance rather than specifying a particular type (thermistor, RTD or thermocouple) simply because it is familiar. Photos credit: Thermalogic Corp.


Sensor Placement

The location of the control sensor is critical. There are two considerations in placing the sensor: temperature gradients and thermal lag. These effects, if not addressed, can cause temperature control to be erratic or occur at a significant difference between setpoint and important regions of a process.

Gradients arise because of the thermal resistance of materials. They can be minimized by careful selection of materials and attention to the path of heat flow; however, a control sensor should always be placed at the point that is most important to your process.

A temperature controller alone, no matter how sophisticated, cannot ensure an even, gradient-free temperature throughout a process. Gradients should be minimized by careful design of materials and geometry.


Thermal Lag

Thermal lag is the time required for heat to flow through the process and from the process to the sensor element. It can cause erratic control if the lag occurs between the sensor and the system elements controlling heat flow.

Thermal lag can be partially offset by proportional control with integral and derivative (PID), but this approach will result in a slower overall system response time as well as a higher controller cost. Addressing thermal lag in system design is a more effective way to achieve process temperature stability.

Design strategies to reduce lag time include close-coupling between the process media and the cooling device(s), using materials with high thermal conductivity and minimizing the physical size of the system.


PH 1121 Thermalogic Desired Temperature Control PID Schematic

Adding derivative action will, in most cases, allow adjustment so that temperature fluctuation and average value are arbitrarily close to the setpoint with little or no overshoot. This comes, however, at the cost of slow rise and fall time, which cannot be arbitrarily adjusted on startup or change of setpoint. Photos credit: Thermalogic Corp.


Understanding the Difference between Accuracy and Precision

Accuracy and precision have different meanings; however, these terms often are confused when specifying process control. The difference between cost and performance is important.

Precision is how closely temperature can be read on a display or dialed for a setpoint. For example, a display of 102.3°F (39°C) is precise to 0.1 degree Fahrenheit while a display of 123°F (50.5°C) is precise to only 1 degree Fahrenheit.

Accuracy is how closely a reading conforms to the same reading calibrated and traceable to international standards measured at exactly the same point. For example, a process reading of 102.3°F (39°C) that is accurate to ±0.5°F could be showing an actual temperature anywhere from 101.8 to 102.8°F (38.7 to 39.3°C). A more precise temperature reading is not necessarily more accurate.

In general, high precision can be obtained more easily and at a lower cost than high accuracy. However, neither precision nor accuracy should be specified beyond what is necessary for a process.



Temperature Control Specifications

Expectations for cooling rate and process temperature must be realistic and within the capability of obtainable sensors and system construction. Arbitrary specifications of temperature stability and limits can result in unobtainable results or unnecessary costs. Consider the importance of temperature precision and accuracy in your system (see sidebar, “Understanding the Difference between Accuracy and Precision”). Relax these specifications as far as possible without affecting process results.

Temperature accuracy is important, but how much accuracy is really needed? Unless you are dealing with a phase change, distillation threshold or similar effect that has an inherently precise temperature, it is unlikely that accuracy of one or even several degrees will affect process outcome.

Specifying accuracy better than 0.1 degree could drive the control or system design cost and schedule well beyond acceptable limits. Be realistic about the accuracy and stability that your process must hold at the setpoint.


Temperature Control Selection – Algorithm

The most common methods of temperature control are on-off control (sometimes called “bang-bang”) and proportional with optional integral and derivative (PID). If a process has a low thermal lag time between the source and sink of heat, and if the sensor is closely coupled to this path, then on-off control will always yield the most stable and close-to-setpoint temperature. It will probably also be the lowest cost.

When either or both of these conditions (thermal lag and sensor coupling) are not adequately met, some form of PID control is appropriate. This selection should not be made without careful consideration: It will always result in a higher cost and additional effort to obtain satisfactory adjustment of control parameters.

Proportional control can be adjusted to achieve a smaller overall fluctuation of temperature from the setpoint than on-off control. The average value of the process temperature reached, however, will still be somewhere within the maximum fluctuation obtained with the on-off control. This average value will depend on system parameters and cannot be significantly affected by the control.

Adding integral action (PI) will allow adjustment so the temperature fluctuation is small, and the average temperature value reached will be close to the setpoint. Upon startup or change of setpoint, however, there will be overshoot, which can be reduced (but not made arbitrarily small) by parameter adjustment.

Finally, adding derivative action (PID) will, in most cases, allow adjustment so that temperature fluctuation and average value are arbitrarily close to the setpoint with little or no overshoot. Unfortunately, this will come at the cost of slow rise and fall time on startup or change of setpoint. This cannot be arbitrarily adjusted. With PID, the parameter adjustments to achieve the desired temperature response may be lengthy, tedious and sensitive to system characteristics.


PH 1121 Thermalogic Desired Temperature Control PID Graph

The most common methods of temperature control are on-off (sometimes called “bang-bang”) and proportional with optional integral and derivative (PID). Photos credit: Thermalogic Corp.


Temperature Control Selection – Display and Output

Consider carefully how process temperature and setpoint will be displayed to the user. Even a high quality controller and carefully designed system can appear to be unstable if the temperature readout is presented in 0.001 or even 0.01 degrees. Always specify readout precision in the largest unit that is meaningful to the process. One degree is usually adequate in non-scientific applications.

Control output to actuate active cooling devices will be specified according to the requirements of those devices. Common types include mechanical or solid-state relay contacts and analog voltage or current transmitters. The type of control of actuation usually has little or no effect on the quality of temperature control.

In conclusion, to design or troubleshoot satisfactory control of process cooling applications, start with system design, not control selection:

  • Is there adequate cooling for the heat generated?
  • Does heat have a low thermal resistance path to leave the process?
  • Are system temperature gradients understood and minimized?
  • Are precision and accuracy specified appropriately for the practical process requirements?
  • Is the type of sensor appropriate for these specifications?
  • Is the sensor placed at a location most important to the process?
  • Is on-off control satisfactory for the process and user? (Only proceed to P, PI or PID if absolutely necessary.)
  • Is the display precision appropriate to process and control stability?

Successful control of the cooling process and selection of a suitable control requires understanding the flow of heat in that process.