Question #7 - Scan time of control loops.
December 26, 2003

Question:

How fast should a process control loop be scanned? If a loop has a time constant of 10 seconds, shouldn't the scan rate be closer to 10 seconds rather than the more typical (for DCS control) 1 second or faster scan rate?

Answer:

The quick answer is "as fast as possible". There is no harm in scanning a loop very rapidly, and no advantage in scanning it slowly. Usually in a Distributed Control System (DCS) you have a choice of several scan rates, sometime as fast as 1/4 second or faster. However, only a certain number of loops can be scanned at the high speed, others must be scanned as slower speeds, such as 1 second or even 2 seconds.

There is a disadvantage to scanning a PID algorithm. As described in more detail in my tutorial, the controllability of a process is governed by the ratio of the largest lag in the system to the dead time in the system. The higher this ratio, the more controllable the process. The gain can be higher, providing better setpoint response and disturbance rejection.

There is a dead time in any digital controller equal to 1/2 the scan rate. So if a loop has a scan rate of 10 seconds, there is a dead time of 5 seconds, added to any other dead time in the process. This reduces the controllability ratio, and therefore the stability of the control and the amount of gain and reset that can be used. For this reason, there is an  advantage of a scan rate much faster than the primary lag of the process;.