How to Tune a PID Controller

▶ Check out the full blog post over at

=============================

⌚Timestamps:
00:00 - Intro
01:06 - Proportional term
02:04 - Integral term
03:06 - Derivative term
03:54 - Algorithms and parameters
04:44 - PID tuning methods
05:59 - Tune a PI controller

=============================

let’s discuss what the PID parameters are and how they are used.

In the most simplistic terms, the controller calculates the P, I, and D actions and multiplies each parameter by the error or E, which is equal to SP-PV indirect-acting, as discussed previously.

Then, all parameter calculations are added up to produce the Control Variable.

1) The proportional term, often called P Constant, can be referred to as Proportional Gain or just Gain, which is not a unit but instead a ratio.
For controllers that use the term Gain, adjusting this tuning parameter higher may cause more sensitive, less stable loops.

Conversely, on controllers with proportional band units, decreasing this tuning parameter affects the loop in the same manner.

2) The Integral term or I Constant, often called Reset, can be expressed differently as well such as repeats per second, seconds per repeat, repeats per minute, and minutes per repeat.

Essentially, regardless of the measurement type, the integral is the sum of all of the values reported from the signal, captured from when you started counting to when you completed counting or the area under a plotted curve.
This parameter can be called Ki, Ti, or others. This parameter determines how fast the steady-state error is removed.

3) Derivative or D Constant units are typically seconds or minutes.
The purpose of the Derivative constant is for predicting change. The Derivative action acts of the rate of change measured in the Process Variable.

The value of this parameter basically means how far in the future you want to predict the rate of change. This parameter can help to create a faster response in your loop and a better performing loop as well.

The most commonly used controller is the PI. Most processes can be well served with this type of control. P and PID controllers are occasionally used while PD controllers are rarely used.

PID controllers are very sophisticated devices with likely many adjustable parameters. The process and algorithm types can also vary.

Other parameters to research for PID controllers are Series, Ideal, and Parallel algorithms, filtering, scan times, anti-windup, self-regulating versus integrating processes, reverse acting, dead time, lag, Derivative on E or Derivative on PV, just to mention a few.

There is a science to tuning a PID loop but the most widely used tuning method is trial and error. There are other methods that require a multistep process to determine where your numbers should be. The goal of tuning is to ensure minimal process oscillation around the setpoint after a disturbance has occurred.

The first step in tuning your controller is to determine just how much adjustment you can make without serious implications to the process.

Talk to the plant personnel, if adjusting the parameters of the PID controller will not have an adverse reaction then you can begin your adjustments. If the effects will be detrimental, you must take a more measured approach.

=============================

You might want to review our previous articles:

=============================

Missed our most recent videos? Watch them here:

=============================

To stay up to date with our last videos and more lessons, make sure to subscribe to this YouTube channel:

=============================

=============================

#RealPars #PID #IndustrialAutomation

[Stripping_Bolts]

I'm a chemical engineering senior and I took Process Control last semester. My "Professor" read a PowerPoint in broken English and muted all chat and turned off cameras. I learned more from these 2 videos than I did that entire semester

[TheWolfmanG]

You guys have just cut my anxiety in half... all your videos. Thank-you so much!

[warrenosborne]

I started tuning controllers in the late 70's vacuum tubes no less, went on as a engineering tech for KC. Did 30 years til retirement. Tuning a pid loop is more art than science. I never used derivative. derivative is like pouring molasses on the loop.

[muchbettethanthis]

I've studied PID-controllers in university but found that my knowledge is really chaotic and unstructured. This video really helped me to structure the data. Looking forward to a new one on this topic!

[danielesauhs]

This is the best short video I've seen about PID, really like your content, understandable, interesting and straight up to the main point, this channel is awsome!!
¡Saludos desde México!

[farhad]

Most PI setups, fill void of missing D with feed forward, specially in motion applications. It's worth to include this parameter in next videos.
Thanks guys and keep up the great work

[trippermostafiz]

Thanks so much for this clarification!! Respect from a Bangladeshi Automation Engineer.

[davebennett]

The analogy to a person holding their hand on a valve and opening or closing it while monitoring a pressure or temperature increase/decrease is a good one. Potentially worth incorporating into future explanations. You can intuitively understand that opening a valve rapidly would result in a big change, because most people know what happens in their shower when they turn towards the hot too fast!

[smokes]

I usually set the proportional band (gain) to zero, and set the setpoint a little below what I need so when it overshoots it wont hurt anything. The temperature is going to oscillate at a specific frequency that is natural to the process. If you know that frequency (time period for 1 cycle) and the temp extremes you have everything you need to enter PI and D values. Get a stop watch, start it at the precise moment the heat turns off, on rise, just after PV passes the set point. PV will overshoot and start to cool. Note the max temperature. Its going to cool and just after crossing the set point the heat will come back on. The temperature will continue to drop, then bottom out at a certain value. Note that min temperature. The temperature will rise and just after it passes the set point the heat will turn off, stop the stopwatch at that precise moment. You now have everything you need to dial it in on the first try. Note the max and min temps (subtract the max from the min), what percentage is that value of the set point? Enter that percentage for your gain (P). The total time it took in seconds is the reset number (I or Ti). Take that Ti number and divide it by 6 and that is the Td (D) value (a good starting point for an oven). That is usually enough for everything to settle down to a nice flat line. If it is still oscillating make the proportional band larger. If it is too twitchy (reacts to quickly to changing input) or never quite recovers from an overshoot, lower the derivative. If you use this method rarely is the I off because you can very accurately measure it. Don't play with this number, especially if you got it after the system was all warmed up.

[elliotlaw]

Currently writting a report on PID controllers as part of my control and mechatronics unit for engineering, very useful video and well explained, I liked the added information around noise in the derivitive gain, and the different layouts of a PID controller (Parallel and series ect)

[livesimply_]

Most processes will benefit from adding the D coefficient. Start by slightly decreasing the Pb and setting a small D coefficient. This is especially helpful when you want to tune to be more aggressive (smaller Pb) but to minimize overshoot. Use higher D coef for systems that have higher momentum (which tend to be slow processes with undesirable overshoot).

[ahmedb]

A very good demonstration, thanks for your great efforts. From Systems Engineer. I have tried auto tuning several times and it does not work. So the best tuning method what you have explained in the last section, adjust P first, then adjust T, then fine tuning.. Thanks again.

[FirstnameLastname-fnik]

PLC programming software like Emerson's Proficy Machine Edition and Schneider Electric's Control Expert have PID blocks. It would be really awesome if you guys made a longer video or video series where you explore the different parameters in PLC programming PID blocks, the math behind the parameters, and then maybe tune something like a valve opening/closing.

I know that's asking for a ton, just throwing stuff out there that I would personally really love to see! Thanks again for the great informative videos!

[aldob]

at school the trial&error process is not explained, but the model of the controlled system is rarely know so to me this is the only option. thank you

[NavySturmGewehr]

This video is fantastic! I've got to adjust the pid loop for my turbocharger controller, this really helps!

[CaminoYVerdad]

I would LOVE to see the scientific approached explained. Looking forward to that video.

[praveenr]

So much of respect for RealPars effort...

[digi]

I cant wait for scientific approach explained. Thank you alot, that help me clearly about PID controller

[AndreGobin]

This was such a great review and tutorial. Love it.

[carmelpule]

Congratulations on a well-presented video.
When our daughter was born, I decided that through playing, she would learn analysis, logic and reason, and rational thinking, rather than develop her memory to be able to recite what she read. I played with her by making her many wooden toys relevant to dynamic behavior rather than her playing with static toys. The use of mediums that do not keep their shape as, water and sand and balls and rubber bands, which could be used in both static and dynamic behavior was very helpful, in coaxing my daughter to see what matters most, but cannot be seen nor said, so easily to young children. The idea that children should be conscious of how their toys can have a Position, Velocity, Acceleration, or RATE OF CHANGES interested me, while how a child could be conscious of what needs to be "integrated" while they play, was another issue that racked my mind.
* Teaching how a toy could be placed in a location is easy for any child as it is only a static situation with no rate of change.
* Catching a ball in the still air, becomes more challenging.
* Catching a ball while the wind is blowing at a constant rate of change or when gusting becomes an interesting mind process.

* POSITIONING a toy in a particular location when the toy is suspended from a solid rod, or while the toy is suspended from an elastic rubber band, has its difficulties, and it is interesting how a child learns how to tackle the issue of positioning a toy in its place when this hangs by a non-stretchable or a stretchable string/ elastic band when there is no wind, or with constant side wind and with bursts of side wind from a fan. It is interesting to see the oscillatory motion that takes place as the child operates her hand according to the error angle when the toy reaches its final point and its inertia takes it to overshoot the target.

* It is interesting how a child learns to slow down the action by introducing a slower RATE OF CHANGE by using her eyes to detect how fast the toy is approaching the target position. Through his eyes, a child detects the velocity of the toy and this is why it is imperative for children to play with dynamic toys rather than being told to just read books to study hard!! These proceedings in detecting the RATE OF CHANGE or the Derivative of position one can do by giving a moving object at least two glances where the difference of distance will give an idea of the velocity. In the industrial world, one can use a tachometer that operates on the rate of change of its coils with respect to the magnetic field but a human can do it with two consecutive glances. It is this issue of being conscious of how fast the toy is moving which would help a child slow down the final movement of approach as he comes close to the target to hit. The Derivative control signals can be learned very quickly by a child who plays with toys that have motion as bouncing balls and placing toys in their place at a fast rate. Writing the alphabet is a perfect manner to teach all the control processes that exist in an automatic control system and I do believe that people who cannot write neatly have some issues with their combination of body/mind control which is the motoring and signal processing system.

* If a child is asked to place a toy that is suspended by a string on a particular location on the ground without the toy being placed on the ground, the child would do it and he would notice that the string suspending the toy is perfectly vertical with the suspension point right above the toy at the final positioning, hence no positional error exists. If the child is made to suspend the toy on its target point while a fan is placed to blow the wind horizontally, then the child at first pauses and tries without thinking too much but he would soon realize that he need to pull the hand towards the fan to ensure that the inclination of the string will pull the toy on its target while the fixed inclination on the string towards the fan will generate a force that will overcome the side drag force generated by the side wind. The child soon learns about using the first integral to get a zero error when the toy is loaded with a side wind. That is INTEGRAL CONTROL.

I used to find it interesting with our daughter to set up a windy condition which is bursting effect n pulses and she used her eyes to predict what I am going to do it and she used to react beforehand, which is a prediction capacity that no industrial machine can have unless its environment is carefully monitored. When I did not permit our daughter to predict the bursts of wind then she was rather aggressive on her hand motion and I would attribute this action to a double integral action to account and react for the burst of wind faster than she would have done with a constant side wind blowing.
I am fortunate to say that my daughter learned all this through playing with dynamic toys and she had no idea that she was handling PID control, but also acceleration feedback or double derivative or double integral, which could be said to be useful in controlling burst of the use of acceleration and double integrals could be exciting due to the integrals introducing delays hence instability.

Learning balancing and roller skating, ice skating, skiing on water and on snow, and balancing acts and all sport have PID control in the human having to catch or aim at a target. Unfortunately during WW2 which I remember, many young pilots died because they could not handle oblique shots at the aircraft in front of them, which not only requires a PID and detecting the rate of change at higher levels to be able to predict where the target is going to be. Getting drunk is a good way of analyzing the delay effect of applying the PID when one is sober and there one begins to understand why a drunk person oscillates to a point of instability.
Well, when our daughter was of age, I took her flying and I had told her that if she wants to learn how to drive any vehicle including bicycles, motorcycles, cars, she should apply continuous movements as that would be applying a persistently exciting input which would cause the craft to undergo all movements and hence the craft would PROJECT ITS OWN CHARACTERISTIC very quickly to the operator. Well, when our daughter flew for the first time I had to take a back seat as she sat near the pilot who gave her the controls when it was safe to do so. As she took the stick in her hand, she gently applied a continuous repetitive oscillatory pitch control to which the aircraft reacted and after that, she did the same with the rolling control which she gently repeated in an oscillatory manner. Then she went for the rudder control. With her persistently excitation signals the aircraft was made to give all its secrets and information that was required about the delays in its inertia. As it was a slightly windy day, I asked her to take a route sideways to the wind and she quickly reacted to the use of the rudder to keep the aircraft straight. I had prepared the pilot as to what she was going to do and I saw him smiling, and after the flight, he said, " Does your daughter know about PID control?" to which I replied, " She does not know them by those letters or that language written or spoken symbols/words, but she knows when and how to apply them, while she plays and follows any sport and about her own stability in all circumstances of her life, even while applying lipstick and makeup which need accurate hand control.

Basically, our life motion ranges from situations involving RATE OF CHANGE OF ACCELERATION, ACCELERATION, (A) VELOCITY (D), POSITION (P), FIRST INTEGRAL (I forCONSTANT WIND LOADING), AND (SECOND INTEGRAL POSSIBLY USED WITH GREAT CARE IN BURSTING WINDY CONTITON /LOADING) to replace that which cannot be predicted and could affect the motion of any entity.