PID Loop Tuning Before and After

Added by Abel B Ramirez II on June 22, 2013. · 2 Comments · Share this Post

Filed under Training

Following are some screen shots from two Air Handler Units (AHUs) that needed some major tweaking on the PID Loop for better Supply Air Temperature Control. I am actually using a PI Loop without the use of the derivative portion. As you will see there are always better ways to do something and providing a quality end product that is accurate will help either in comfort or energy savings to the end user.

This is AHU2 and as you can see the SAT(Red) sort of moves with the Setpoint(Blue), but you shouldn't dismiss all the spikey-ness especially when this trend is for 3.5 days worth of time. Those spikes now are looking more like 30 and 60 minutes from top to bottom. You also need to take into account that the tops and bottoms really aren't that far away from setpoint, typically only 0.5°F. All and all, not too bad...but I am itching to mess with this AHU later. ;)

This is AHU1 and still looking at a 3.5 day view we show each trend sample which I think was every 15 or 20 minutes (feel free to count). Here we see a much more distressing problem. We never seem to follow the setpoint and in some cases are 3°F away for an entire 24 hour DAY! Looking at the PID loop setting I found there was a really large deadband of 5 degrees, which was causing most of the issues.

This is the same AHU1 with the Red/Blue, but I changed the colors to Dark/Light Green. this graph starts where the one above stops and after I started jacking with the PID Loop settings. With one quick swoop I made a few changes at the beginning(details further down) and saw a slow improvement. But to be sure you got it perfect you have to make adjustments and see how the PID reacts. In the middle we see a perfect example of the over-under-tighter PID reaction. I wanted something better than that so... some more adjustments and now we have a really nice fast reacting PID Loop that hones in on the setpoint without alot of overshooting.

This is a screenshot of the CO (PID Loop) object inside a Delta Controls DDC Controller. Pretty simple, you plug in your Proportional Band, Deadband, Reset Rate and Reset Band which are both used for the Integral portion of the PID. Below I will discuss some simple base numbers to use, but even this one deviated slightly from the base.

Share This Post

Don’t be Satisfied with Something that ‘Kinda’ Works

The best thing about controls and automation that appeals to me is that I get to try and build a better mouse trap. Now there is a point to where you just have to step back and realize that you need to balance the time you invest in getting it perfect and the value of the work once you are done with it. In other words, just don’t get too carried away with it.

In the case of AHU2 (first screenshot) I knew this was pretty good even though I didn’t like the spikes; at the time I was just reviewing a project that was already installed, not something I was actually working on at the time. So I set that one aside and moved on to finding AHU1 which had a far worse issue that needed to be dealt with.

Good Base PID Loop Settings

I use Delta Controls the most. I have gotten very accustomed to how they adjust their CO (Controller Object) that spits out a PID Loop result percentage. Over the years I have used a pretty simple way of getting some base settings for some typical equipment and then I just adjust a little from there depending on the application.

Proportional Band
Chilled Water Temperatures15°F
Hot Water Temperatures25°-30°F
Supply Static Pressure3″-5″wc


After I choose a Proportional Band based on the type of media and range I will typically see in a good working system then I usually will skip the Deadband and leave it at zero for the moment. I always use a Reset Rate of 4.0 and a Reset Band equal to 1/2 the Proportional Band. If my Proportional Band on a CHWV is 15 then my Reset Band will be 7.5.

Stay Flexible or Else it Won’t be a Custom Fit

There will always be exceptions to the rule and various job specific issues like dampers that bind, valve actuators older than you are, or long and short pipes that add/subtract time from the moment you move a valve to the time the sensor actually sees a change. Things that usually call for changes to the base numbers are…

  • The reaction speed of the media being controlled
  • Multiple outputs or devices being controlled by the single output percentage
  • The stability of the input sensor reading

In the case of AHU2 there were actually 2 Chilled Water Valves being controlled off this one Air Handler Unit and PID Loop. They were setup in a 1/3 and 2/3 configuration and in the trends above we were able to use just the  1/3 CHWV to control the Supply Air Temperature.

Controls and Automation Equipment Manufacturers are All Different

Looking at the Delta Controls CO object above you would be hard pressed to use these same numbers in a different controls system because they may not even have fields called Reset Rate and/or Reset Band. They may also use a completely different algorithmic calculation behind the scenes that might require bigger or smaller numbers in the Proportional Band. So don’t take the above base values or my working settings above to mean that is what you should use…unless you are working with a Delta Controls controller. Then… yes, you will greatly improve the results of your PID Loops if you start with these base numbers.

IMPORTANT: Change the Setpoint and Watch the Reaction

The final and MOST IMPORTANT part of tuning a PID Loop is to throw it out of whack and see how it responds and reacts to the change. PID loops inherently will adjust the output percentage to try and make the Input equal the Setpoint. That is its job!

So if you put crappy settings that are super slow and wait for it to hit setpoint and call it done… you aren’t done. Period. You need to then make an adjustment to the setpoint like I did in the 3rd screenshot above with the Green lines. I changed the SAT Setpoint from 55°F to 52°F and saw the PID loop actually move towards the setpoint instead of staying several degrees away from it, but it did the ol’ over-under-tighter movement. Which isn’t bad, but I know I can do better.

At that point it gives you more opportunity to see where your settings need to be changed to help compensate. Once again, after making those adjustments you make another change… I put the setpoint back to 55°F and watched again. Sure enough my settings were even better and within 7 minutes I was only 0.1°F away from setpoint with a maximum overshoot, one time of 1°F.

Engineers, Commissioning Agents and Your Customers Will Love You

If you take this little bit of extra time to hone your settings on one unit, you can then replicate your results to other similar units on the project and build up your comfort with how the settings react so that it will help you set things up faster on the next job. In doing this you will surely be left alone by the overseeing eyes of engineers, commissioning agents and your customers when they realize how smoothly your controls handle the equipment and not only maintain setpoint by 5-10%, but NAIL setpoint and get there in a very short time.

Looking back at the green trend lines you will see it took me an hour and a half from start to finish. But understand I was at my desk working on other things and talk to other people etc. Plenty of time for you to commission and check out other points, make a few phone calls, talk to someone about an issue and all those other things you typically will have to do while out on-site.

There’s no excuse to NOT do a good job unless you just aren’t trained to know what a good job looks like. Hopefully this article has helped you realize how good you can be if you invest a little more time in tuning up your PID Loops.

2 Responses to PID Loop Tuning Before and After

  1. Jon J Darcy June 25, 2013 at 18:08

    It amazes me that we keep trying to reinvent the wheel. The standard for temp control loop for 50 years was start with 10% PB and 1 repeat per minute. There is no place for derivative in temperature control since there is almost never any acceleration of the process. Techs try to solve valve or damper sizing issues using derivative. 45 years tuning control loops has taught there is no set of tuning constants that will solve application mistakes for all loads. Bumping the process and tuning to 1/4 amplitude decay (look it up) will get you there. Tuning Cooling coils that fall into laminar flow is impossible and has to be solved in other ways. Being a control engineer is not about writing software routines blindly trying to solve unresolvable mathematical equations but rather about being an engineer and making sure all parts of the process is sized right and working correctly. Once the process works right the tuning is easy.

  2. Tim Wilson January 23, 2014 at 08:35

    when I look at the PID loops in the programming that we use, I don’t see the reset rate and the range for the P that you talk about. these are just P and I and D and are small numbers (usualy with a decimal place like .2 or .4) I don’t understand where the numbers for the bands come from in the videos. I was wondering if you can tell me where I can find the bands that you show in the video? I am using Allen Bradly RS logics and I am looking to tune a lupe for nitrogen to air so that we can control the O2 level in a kiln. Any help would be great. I also understand that there are inde[pendent and dependant PID lupes. Could you explain them? Thanks Tim


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

A Great Value in Online Advertising

The Controls Freak has slowly been building up a dedicated audience of Building Automation/Controls and Facilities Professionals in the United States and around the globe. As we bring aboard more contributors our viewership will continue to grow exponentially.

Full analytics reports can be requested on the Contact Form along with a request for current rates and available advertising spaces.

Now is a great time to support The Controls Freak and introduce your brand of products or services to even more potential customers for an extremely affordable rate.

Send a request today and let’s start working together.