PID stands for "Proportional Integral Derivative." PID control is the most popular form of control algorithm used in industry today. It is an appropriate control methodology for almost 95% of control loops in industry and over 99% of control loops found in HVAC. The proportional part of the control provides an immediate reaction to errors or disturbances in the system. The integral part provides a "memory" that adjust the control based on the error in the past, while the derivative part is an "anticipatory" mechanism that adjusts control based on expected future errors

While PID control is general enough to control just about any kind of (controllable) plant, some control loops might benefit from more advanced control techniques. Loops with extremely long dead times can be better controlled using methods that utilize some form of predictive or time-advance control algorithms, such as the Smith Predictor. However the complexity of such algorithms, and their tendency to degrade or become unstable if the plant changes, or is inaccurately modeled, has confined their use to a narrow range of processes. If the plant is non-linear or time varying, the math becomes so complicated that the topic is rarely breached except in extremely advanced discussions

The combination of P, I, and D produces a versatile and robust controller that reacts immediately to disturbances, has zero steady-state error, and starts backing off before the setpoint is overshot. Tuning PID controllers has generally been considered an art, and not a science. This is because the robustness and versatility of the PID make any sophisticated tuning methods unnecessary for controlling a plant adequately. A PID that is tuned to perform adequately, however, is not necessarily tuned to perform optimally.

The difference between the performance of a well-tuned loop and a badly tuned one is quite drastic. In some situations, poorly tuned loops may not be able to recover from sudden disturbances except after an unacceptable time delay. In other situations, they may experience too much overshoot, wasting energy in trying to bring the plant output back to the setpoint. For example, in a system where a space temperature is being controlled by cold air and heaters, it is highly undesirable to have to bring heaters on-line in order to recover from overshoot caused by bad control. If the gain is set too high, overshoot will occur. Setting the gain too low causes an unacceptable delay in reaching the setpoint, also wasting energy because the plant now has to be turned on many hours before the tenants arrive in the morning.