However, there are applications where you can us PID loops but they are not the best method. For instance, if you have a tank and you are trying to swing a very long barrel around to a precise location, you want it to be as fast as possible as well as being accurate as possible. It may be worthwhile to spend the extra time and effort to study and mathematically model your process and go with an more advanced control technique than PID.
Yes, but also use feed forwards as well as a PID.
There are also several applications where it is impossible for PID loops to control. Functions were the process constantly changes (such is robot devices where the environment and tasks constantly change) cannot be controlled because the control loop criteria is constantly changing.
There is nothing to stop you from changing the PID gains on-the-fly.
Also, rocket guiding systems are something else that PID loops generally do not have the power to control.
Not by themselves but they are used in conjunction with other techniques.
A PID can be used to place the poles to get a desired response. Sometime the plant poles move during the process. Now the PID gains need to change to keep the closed loop poles in the same location so the response stays constant.
Below is a link to a simulation of level control for two tanks. The idea is to control the level in the second tank using a VFD controlled pump that pumps water in a first tank than has a hole or valve in the bottom so that water can flow to the second tank. I think this was the last pdf because I started off simple and worked up to this. I was helping a student on LinkedIn with his project. I did not provide this final pdf until after he handed in his work. The student did get a A but I told him I would not have passed him because he wasn't able to write the first two differential equations that explain how the water level in the tanks behave. Meanwhile there were a lot of other 'engineers' that were annoying me because they said it is a simple application and one can tweak gains until it worked. That is true but a student doesn't go to college and take a class in controls just to learn how to tweak gains. I showed the student how the problem should be approached starting with writing the differential equations for the water level in the two tanks. On page one there is a Cv or flow coefficient for flow as a function of level. There is also an area A for each tank. At first we assumed the Cv and A were constants but in reality they aren't. The flow goes through the valve proportional to the square root of the pressure across the valve which is proportional to the level. The area of the tank doesn't need to be constant either. The time constants tau1 and tau2 are not really constant in this case and change as the plant changes. Now look at this:
http://deltamotion.com/peter/Mathcad/TwoTanks/Mathcad - t0p2 2pi tc - Alin's two tanks Cascade.pdf
Page 1 is the simple differential equations with out which I would not provide a passing grade.
Page 2-3 is where I computed the symbolic formulas for the PI controller for each tank as a function of the tank gains and 'time constants'
Page 4 is where I assigned some values to the variables. Notice that I didn't assume the areas or flows are constant.
Page 5-6 is where the differential equations for the simulation are executed.
Page 8 is where the response is plotted out.
You can see the response is a nice critically damped response regardless of the level set point. This occurs because I keep the closed loop poles in the same location by calculating new controller gains as a function of the changing level.
Notice also that at 28 minutes I added an disturbance by dumping more water to the second tank.
I did this with two simple PI controllers. I could do it with the just a P controller for the first tank and a PI controller for the second tank.
http://deltamotion.com/peter/Mathcad/TwoTanks/Mathcad - t0p2 p pi Alin's two tanks Cascade.pdf
So Helliana, a P, PI or PID controller is really pretty powerful if you know how to use it.
BTW, the first tank was supposed to be only one meter tall. One can see that the water would have over flowed the first tank. A real designer should do a simulation like I have done. Then he would know how the water level change s in the first tank as a function of the level in the second tank. Those that simply want to tweak gain would not do this simulation and the real design would fail.
I took a lot of **** from the other so called 'engineers' that suggested how to tweak gains but I asked the student, what do you want to be, a designer or a user?
