PID theory: Fixed delays
In previous sections we have explored the delay mechanism generated by “time constants”. You may recall that with at least a simple time constant, the moment the input is applied the outputs starts to change. With multiple time constants the outputs seems to hesitate, but in fact it does still start to change immediately, albeit very slowly.
What about true delays? This is the sort of delay you would get if you had a water heater with a long outlet pipe and a temperature sensor at the end of the pipe. Chances are you experience exactly that kind of delay each time you take a shower. For a start, the water must travel from the heater to the shower head. And even after that delay, once the hot water has come through there is a perceptible delay between adjusting one of the taps and the warmer or cooler water getting to the shower head.
Fixed delays like this are called Dead Time or Transport Delay (take your pick). Sometimes the term Dead Time is incorrectly used to describe the apparent initial delay we see when there are multiple time constants. Transport delays can add further to the challenge of getting a PID control system tuned up. They are not usually impossible to handle, especially if significantly less than the dominant time constant, but they can be a challenge. The main thing is that you are aware of them, and also that some types of “fixed” delay can actually vary. For example, the more water flow you have in your shower the shorter the delay (can you figure out why?)
The PIDassist program allows you to simulate a fixed delay. When you simulate your system, you should always test it over the expected range of delays, as response of PID controllers is strongly influenced by delay. More on this later.