## Proportional PlusIntegral Control PI

An integral function is added to a proportional-only controller to eliminate the residual error. This control action adjusts the gain to a stronger and stronger value until the error is eliminated. In theory, the integral controller will not rest as long as any error exists, however it is common to allow a small acceptable error band around the nominal set point to prevent incessant low-level hunting as the controller seeks the perfect "zero" error condition. In practice this has the appearance of slowly but surely building up the output to taper off the error. Since the "wind up" effect is slow, but also relentless, integral control problems can occur in processes that change rapidly. Also, if the controller is left active for long periods while the controlled equipment has been turned off, the integral controller will "wind up" to a 100% output. Upon startup after a long period of wind-up, the integral function can be strong enough to "stick" at full output for long periods of time with complete loss of control. Therefore, whenever integral control is used, some form of hardware or software interlock

Figure 22.4 Proportional Control Mode Diagram
Figure 22.5 Proportional-integral Control Mode Diagram

magnitude of the error (proportional) or the duration of the error (integral). In reviewing the characteristic response curves, PID control looks like the absolute best and, in fact, does provide the tightest control of all the control modes. However, the derivative gain is very touchy to set up properly and can easily cause instability of control, especially at the beginning of a batch process or after a large step change. In HVAC work, the derivative term is seldom used to avoid the potential for instability, and since most HVAC processes are relatively slow acting and are tolerant of temporary overshoot of PI control. In many process control applications, PID control is essential since close control is often tied to product quality.

• Normal offset from proportional control.

• Normal time lag from Integral control.

• Independent, but sequenced, modulating controllers.

• Adjacent processes, such as two comfort zones in an open bay with widely different user set points.

Where the process does not require absolute tight control, building in some blank spaces or dead band between sequenced elements is a simple way to achieving and sustaining energy savings.

0 0