The derivative gain simply adds damping to the system. if the system doesn't oscillate then it probably doesn't need the D gain. In mechanical system one can add mechanical dampers or friction but this wastes energy.
To really understand the derivative gain you need to try tuning a mass on a spring. When the mass is disturbed the mass will oscillate until friction slows it down but one can turn up the derivative gain and that will add damping too. A mass on a spring will never stop oscillating unless there is mechanical or electronic damping ( D gain ). The unqualified statement below about a well tuned PI control is better than a poorly tuned PID is simply BOGUS. In the case of the mass on the spring the PI control doesn't have a chance because there is no way of adding damping.
I realize the the gurus only meant this to be fore process systems but the editor didn't make that clear.
from the article said:
But let’s say the process in your case is a tank of liquid product that you’re trying to heat. If the process variable (temperature) goes above the setpoint, the product may be ruined or catch on fire. So, how do you get the process variable to move, but not overshoot the setpoint excessively? One answer is introducing a derivative factor.
Adding a derivative gain may be the answer. It depends on how many poles the system has. A SOPDT system will benefit from a derivative gain but a FOPDT system will not. I object to the blanket answer that a derivative gain will help without knowing what the system is. The there is a similar situation in motion control. Most conveyor/speed control applications work well with just a PI controller because the main time constant is the inertia of the system. BUT, when doing position control the velocity is integrated to become a position. These system usually require a derivative gain. A lot depends on where the closed loop poles are placed.
from the article said:
The more the controller tries to change the value, the more it counteracts the effort.
Not quite, it has nothing to do with how much the controller tries but rather on how high the error rate is changing. The derivative term opposes the rate of change in error.
from the article said:
It doesn’t move as quickly as the PI-only effort, but without the oscillations, the right amount of derivative action can stabilize the process variable at the setpoint sooner.
This assumes the P and I gains haven't been changed. If one adds the derivative gain and also increases the proportional and integrator gain one can actually get faster response. Especially if the system really needs a derivative gain to place the all the poles.
from the article said:
Looking at the situation in the top diagram of the sidebar, the operator has raised the setpoint. How does the controller respond?
When you raise a set point in a temperature system the integrator does wind up as stated. A motion controller's integrator would like wise wind up if the velocity set point is increased BUT, motion controllers have feed forwards. A properly tuned motion controller's integrator term should always be close to 0. This is a big difference. Another big difference is that in motion control the set points are rarely changed in a step. There is a target generator the smooth ramps the target position, velocity and acceleration through the whole profile until the set point is reached. Ideally the PID term is close to 0 and feed forwards do almost all of the work.
from the article said:
So if the main purpose is slowing the control effort of the other factors, what’s D good for? Fast acting loops, such as flow and pressure loops, don’t really need it.
Fast acting is not defined. A motion control system is fast acting compared to anything in the process industry and yet motion controller use the derivative term much of the time. Almost all the time in position control.
We make hydraulic motion controllers where oil flows and pressure on either side of the piston creates the force that makes the actuator move.
It should also be obvious that the article is talking about 10s of minutes and we are often talking about moves that take a few milliseconds to a couple of seconds.
A well tuned PI controller is going to beat a moderately tuned PID controller every time.
This is truly bogus. Anybody can make a PID controller look bad, but if the PID controller is right for the system then it will be easier to tune than the PI controller.
Adding the extra tuning parameter adds complexity, which can confuse a lot of people.
That is why auto tuning was created. I am very good at tuning PIDs on many different systems but I am pretty clueless when it comes to tuning the second derivative gain. I use math and other tools for that.
George Buckbee, P.E., vice president of marketing and product development for ExperTune warns that some traditional loop tuning beliefs should not be considered as universal. “That Ziegler-Nichols quarter amplitude damping thing is a bit of a fallacy for a lot of loops,” he advises
I agree and it shouldn't be used for motion control at all. One can do much better.
“The criteria for what is 'good performance’ really do change from one loop to the next, but usually it’s faster movement toward the setpoint with less risk of overshoot. You have to choose those criteria wisely, and define them for control performance on a per-loop basis”.
True, but the article could mention how the tuning is evaluated. There are methods such as the sum of squared errors(SSE), mean square error(MSE), integrated absolute error(IAE), integrated time absolute error (ITAE)
However, as Buckbee points out, each loop has to be approached individually using the right tuning for a given situation.
YES!!! This is why I cringe when I see people giving advice about how to tune a system when they don't know anything about the OP's system. Usually the OP doesn't even know.
Buckbee compares using derivative to learning to drive: “Derivative is like trying to drive your car with one foot on the gas and one foot on the brake. To my 16-year-old son who’s just learning, that was the first thing he wanted to do.”
What nonsense. If implemented properly the derivative action should be smooth but often it isn't because of sample jitter and non-linearities due to quantizing. You all have heard me rant on that before. The sad part is that PLC PID controllers suffer from slow analog feed backs with low resolution and lots of sample jitter but that is the PLCs fault. Not the PIDs. Our motion controller samples 8 times a millisecond, has 16 bit resolution and all sort of tricks to estimate the rate of change which most on this forum ignore.
Such an approach might work for a skilled race car driver, but most loops don’t need that kind of immediate and violent action.
These guys haven't met most of my customers. They want to go faster, smoother and more accurately all the time and are constantly pushing the hydraulic and mechanical limits.
Buckbee adds, “Derivative is looking at fast, short-term changes in the process variable, and that’s all that noise is. It goes up by 1%, and the next sample it’s down by 1%.
If our product wandered around by 1% it would get yanked out in a millisecond. I have customers grinding to 10 microns. Back in the 80s we had to hold to within 0.001 in. Now there are rods with 1 micro resolution.
Rice suggests that if the dampening action is too high, you have to turn up the P and I action to compensate, like trying to accelerate your car with a foot on the brake.
True but I hate the accelerate your car with the foot on the brake analogy.
At the bottom you can see that ZN was developed in the 40's. Control theory has come a long way since then.