PID Loop Tuning

John Brown

I'm looking for a simple generic PID Loop Tuning Method with simulation of variables for testing controllers outcome. I seek software that will allow modeling to fine tune HVAC office building systems typically, VFDs, VAV motor operators, and hot/cold water valves, pumps...etc.

I found a manufacture at www.expertune.com and they produce realtime I/O test equipment software packages starting at $5,000 to run tests, a sort of plug n'play setup. This is all fine and well if you are maintaining an oil refinery, but I'm looking for something that will tune a typical office building at a fraction of the cost or no cost. Generally, medium office building budgets cannot afford this high cost with an ROI beyond two years. The present economy will not allow such expenditures. Has someone out there created a program which the user estabishes the PID perameters base on site specific I/O system design? Is MS Access and Excel a possibility to be used for data base and graphic predictions.

Help, I'm "Hunting", and so is my building HVAC System!

John Brown

Jerry Boulanger

That's a serious can of worms

you're looking at opening there. I'm not aware of any such software, but I can tell you that the most common cause of hunting in HVAC systems is oversized dampers and control valves. You will likely find that all of the temperature control valves in your system were selected using 'rules-of-thumb' of 3-5 psi valve pressure differential, when in fact they should be selected for proper authority in each application. The best controller available cannot overcome the deficiency created by an over-sized control device.

John Williams, Jr.

John?????

Exactly what DDC system is controlling your building at this time? There is no generic system that can simply talk to any major brand on the market. All of them have their own program and means of access. That means of access is not just simply having the program, thats easy to obtain, but rather having all of the keys and proper cabels to talk to such equipment.

Is it 100% pneumatic? if so than there may be problems with the pressure and control ratings.

As for the $5000.00, thats cheap for complete DDC program, but that's not going to simply talk to any brand of control system. All of the name brands are proprietary, you can purchase the programs and even receive classes on how to operate it, but not at any cheap price.


John

John Brown_2

PID Loop Tuning

Jerry,

OK, The issues I have are not pneumatic. That would be easy for me to solve from the ground up including performance diagrams. You are correct about proper valve sizing for applications with controllers and adapters that should be engineered in range of 1 and 2 stage PB, TR, S-Span to maintain SP. Thanks for the feed back.

John

John Brown_2

PID Loop Tuning

John W,

OK, I'll get right to the point. Below is one example.

Carrier DDC system with ComfortWorks (w/Keys)at the front end. I have a 100% OSA air handler with HHW Belimo two position valve controlling the HW coil which is hunting from 95 degress to OSA temperature when calling for heat to maintain SP. A PI is used according to the data base and is adjustable. D is advailable but was not originally programmed to hold a value in the data base. I hear two position valves are used for PIDs, but the operator has to be good to hold the occillation down at setpoint without the use of D. The drawback is, the motor operator gets a workout(shorter life cycle)trying to predict and hold SP. Anyway, I have yet to contact the manufacture of the controller. I plan to investigate the sensors and valve specifications. I think the valve range is 10 Vdc and the RTDs are 250 degrees. Basically, I am evaluating the system engineering approach prior to correction. I thought to hit the WEB for a prediction program. I hoping a super sharp guy that sleeps and breaths PID open/closed loops out there and has developed a simple prediction program where the user enters raw variables to a table or simulation model. Please remember, that Loop Tuning I seek is for office building applications with medium size DDC systems applications. Thanks for the input.

John Brown

Aidan (UK)

PID tuning

I’ve always used Trend, but I don’t think it’s marketed in the US. There is a loop-tuning program available in the Trend software suite, but this also assumes you’re using a Trend supervisor & outstations. I’d think competitors would offer something similar, so I’d try Carrier first. You’d otherwise need to find a DDC/BMS forum. Let me know if you do.

I’ve always understood that a D(erivative) component wasn’t usually necessary for building services applications and was usually omitted because of increased valve wear. I’m not a DDC/BMS specialist, so my grasp of the theory is a bit fuzzy.

The cost of the tuning software is usually excessive for a client, but it would be available to the DDC Contractor. Loop tuning is often included in the specification, but the Engineer would need to stand behind the installation Technician to ensure it was done. Engineers often don’t understand DDC, so it often doesn’t happen. I find most systems are installed with a guesstimated round number in the P(roportional) and I(ntegral) components, e.g., 5 or 10, and these will usually give adequate control of the system.

It sounds like there’s a fairly fundamental problem, causing the system to hunt from 0 to 100% output. PID loop tuning would fine-tune the system to give values of PID components to 2 decimal places. It doesn’t sound like what you need for this problem.

I’m not certain what you mean by “two position valve”; is this what I’d call a 2-port valve or is it a 3-wire actuator? What’s an ‘RTD’; return damper?

I’ve found cascaded control to be most effective with air supply AHUs. The room temperature controls the supply air set-point. The AHU just has to maintain the set-point. A change in outdoor conditions will cause an immediate change in the control signal, rather than waiting for the subsequent error in room or exhaust temperature to cause a correction to the valve driver output signal.

brent

pid

Having a 2 position valve means that the valve is either open or closed. Hopefully you do not have that. If you do tuning the loop won't do much. Hopefully you have a 2 way valve that modulates. Is the set point fixed at say 70° or does it vary?
What settings do you have to change? I know some carrier software has proportional gain and master gain and some other settings. Usually you need to slow the response down. The outside air temp doesn't change very fast and hopefully your leaving air temp set point doesn't either.

brent

John Brown

Loop Tuning

The 100% OSA AH has a two position actuator controlling a three way diverting valve within a two pipe HHW system. HHW Pump is non VFD controlled. Presently the settings are: SP=71; P=8; I=.3; D is not used. The two postion valve appears to be not modulating, I have to follow up on the manufacture. When the valve is operating I only see it move open or bypass 100% in either direction, but that could be because of the PI setup. Obiously, I have more investigation to see if is proportional. I'm going to try and increase the "I" by 50% and time the modulation/temperature swing/change. I had an appointment with the local Carrier Systems Specialist this week to go over their submittle package and approach, but he was activated for military duty this past week. I'm trying to get the jump on this system and correct the shortcomings.

Dan Law

Valve hunting

Sounds to me that you have two problems. One is the two position actuation, the
other is just a general problem with digital actuators. The problem is the inherent
slow response time. Just think about it for a minute, and it becomes obvious. The
air in your duct is moving past your sensor at something like 1500 fpm or roughly
25 fps. The Belimo actuator has a stroke timing of what? 60 - 90 seconds?
Contrast this with a 3 second valve timing on a pneumaticly actuated valve or
damper. See
where I'm going? By the time your system senses the change, and positions the valve, your way out of the original calculation. Add in 2 position control, & it can't
possibly find a setpont. As the hunting continued,
the control point variation increases and gets worse and worse.

Nobody in digital controls will ever
point this simple fact out, but actuator response timing is THE reason you need derivative control. The nessesary
control signal and subsequent valve position need to be anticipated PRIOR to
reaching a fixed setpoint. It's like taking your foot off the accelerator to remove some of
the cars momentum in anticipation of stopping. To continue the metaphor, two position would be like slamming on full brakes when you sensed you were entering the cross walk. Two position just makes
matters worse.

You need to get your controller to put out a floating control signal.
This is the design intent for the Belimo actuator in the application you describe - 3 wire, floating
control, (if not in fact full analog modulation). If it is not 3 wire floating (or fully modulating), but rather 2 wire spring return, your sunk. Got to
change the actuator first, then get into the PID loop second if this is the case.
Hope this helps.

[Deleted User]

Give these guys

a call. If they can't help, they might be able to lead you to who can. They're in Baltimore.

1-800-333-4822. Ask for Paul or Pat.

brent

pid

In my opinion you want a slow response time. It's been my experience that most control problems in hvac applications are because the device is trying to react too quickly.
In your application I'm assuming you have a constant volume fan that is bringing in 100% outside air. You said the setpoint is fixed at 71 (which sounds good). The temperature of the outside air doesn't change very fast. The valve position therefore doesn't need to change very fast.

All this doesn't matter because you can adjust your settings all day and it won't do any good. A 2 position valve won't work. If it's 10° outside and the valve is open you may get 90° (what do you get?). If the valve is closed you get 10°. Those are your 2 choices.

imho
brent

Eric

Ziegler and Nichols

These are the guys who invented PID control.

There are two basic types of loop tuning. Open loop and closed loop tuning.

I have the formulas and procedures for both, closed loop is easiest. I can fax to you. You will also need know what gains your control system adds to your P,I,D factors.
You usually have to dig deep in text to find that answer.

Aidan (UK)

Actuators, PID & 3-port valves

IF it is a 2 position actuator, then that’s the major problem. You could check the actuator model number on the Belimo website. All the PID settings are irrelevant until you get a modulating/proportional actuator. The system seems to be acting as if controlled by a thermostat, either on, off or in motion between the two extremes.


I don’t agree about pneumatic control being a necessity. The outside air temperature will change gradually, taking 12 hours to go from the minimum to the maximum or vice versa. An actuator speed of 60 or 90 seconds seems rapid in comparison. I’ve had experience with both pneumatic and electric actuators. Whilst the speed of the pneumatics can give the most accurate control, electric actuators are more than adequate for most building control systems.


I agree with Brent, most of this was typed off-line before I'd read Brent's post above. Don't mess about with the PID settings until you're certain you've got a proportional actuator working properly.


3-way diverting valves, used with constant speed pumps, was the standard control set-up on most UK systems until 5 or 10 years ago. 2-port control valves are only now becoming more common, used in conjunction with the increasingly common variable speed pumps. The flow rate through the pumps should be constant. All the valves’ by-pass branches should be fitted with regulating valves, adjusted so that the hydraulic resistance of the by-pass is the same as the resistance of the coil. The water flow-rate will then remain constant whether the water passes through the coil or by-passes it. If these regulating valves aren’t set correctly, a movement of one control valve in the circuit will cause a change in all the other coils’ heat outputs.

John Brown

PID Loop Tuning

Update:

I got lucky after two hours of fighting this fire for now and "put out the fire" so to speak. I managed to get the air handler under control by intepolating the known settings to a new approach, then spliting the resuts in half until the loop drifted open or hunted. After fair stabilization, I tweeked the Integral up a bit narrowing the offset. I ended up with 16=P and .9=I, with +-2 degrees F from set point and a 20 minute cycle time. I was about to adjust D to tighten up the error but decided to leave it alone, since I do not have all the facts about system componets. I suspect the valve operator is two position because the feedback I get from the BMS system is either it is 100% open or 0% closed. Anyway time for me to go to work and see what exactly I have.

Dan Law

Gradual Change / slow response

As all have stated, the big problem you have here is the 2 position valve operator.
As for slow response being better than quick - in my opinion it's a relative
measurement. Never been to the UK, but my understanding is it's a relatively
(there's that term again) moderate climate. As an example of the contrast, here in
Northern Michigan the last 24 hr. have produced the following readings. Noon
Sunday = +41F, 6PM Sunday = +19F, 6AM Monday morning = -15F Current =
+10F, predicted overnight lows = -20F. There's a world of difference in adding
heat to a fast moving air stream entering a coil
+40F. (particularly in terms of freezestat trips) Keep in mind the coil sizing must
accommodate all entering air temps.
The valve and it's attendant capacity modulation and response time
becomes critical. I disagree with slower being better. In my experience, it is more
a case of the valve sizing being incorrect, rather than the valve timing being too
fast.

When we have the option, we always prefer digital control of pneumatic
operators. In this configuration, you have the best of both worlds. All the flexibility
and reporting attributes of digital measurement and scheduling, along with the
power, speed and reliability of pneumatic actuation. Just our preference and one
man's opinion.

Aidan (UK)

actuators

“Just our preference and one man's opinion.”
The opinion is respected and read with interest because it’s been acquired by a different route to that which I followed. You probably have different insights into the same problems.


“I disagree with slower being better.”
I don’t think that slower is necessarily better; I don’t think that you need to pay the premium necessary to achieve rapid response in most building control systems. How fast is fast enough? You don’t need a Formula 1 engine in a suburban taxi.


IF you could achieve adequate control with electric actuators, THEN you could eliminate the compressed air. If it’s essential for you to have the fast response of pneumatic actuators, then you’ve got to pay the extra.


Pneumatic systems are unusual here. We fitted electric/pneumatic transducers to the existing pneumatic actuators on the MTHW and CHW valves on a 30-year-old close-control system, about 7 years back. The system ran at 20 AC/hr with 100% fresh air. We got adequate control of the temperatures initially, probably about +/- 0.5 degC variation around the setpoints.


Later on, I tuned the control loops, first using the Ziegler and Nichols manual step method in the Trend manuals, and later using the Trend loop tuning program. After that, there was generally no deviation from the setpoints. The Trend temperature sensors measure to an accuracy of +/- 0.1 degC and it was unusual to see even a 0.1 degC spike in the temperature graphs. It was the most accurate control of temperature that I’ve seen. The system had a 0.5 degC deadband, so I doubt that control to 0.1 degC was essential. I couldn’t have justified the additional cost of pneumatic actuators on a new control system.


When the old Satchwell MTHW valves started to expire, I had some difficulty in finding replacement pneumatic valves. I bought Siebe ones, but I don't think there were any other makes advertised in the UK. Some of those valves have since failed and have been replaced with electric ones. The flat-line graphs are no longer achieved, but no one has complained. In fact, no one has noticed, except me. I’m no longer involved with that system so I’ve said nothing.


“In my experience, it is more a case of the valve sizing being incorrect, rather than the valve timing being too fast.”


That’s my experience too; or it’s the regulating valves/circuit setters not being correctly set, or being altered by well-intentioned, mis-guided maintenance; or a clogged strainer; or the flow connected to the return port of the heating coil. If you’ve got all that right, the valve’s speed probably won’t be critical.


I haven’t been to Northern Michigan, but I’ve heard tell of it. I think you’d find electric actuators adequate, even there. You’d need a spring return actuator on the inlet damper, hard-wired to the frost stat, because, if the heating fails, you'd need it to shut faster than an electric motor could move it. I don’t think I’ll ever get to install an AHU in Michigan, so I’ll never have to put my theory to the test. Which is probably just as well.


Regards,

Aidan (moderate UK)

John Williams, Jr.

Call me....

maybe I can help you. I work for Carrier directly. I need to know what you have, in controller type that you are trying to talk to. In many of our applications we do not use fixed, or typed in, PID settings we use other ways of getting the same results. Many of those ways are old DOS ways of achieving better results without "loop tuning".
610-567-5038


John

Dan Law

Hands across the water

Hey, wasn't that a McCartney lyric? Just another example of Anglo - American
cooperation. Now if you could get Tony Blair to invade the French, we’d back you
up and..................never mind.

Pneumatic systems are still plentiful here, but rarely installed as new construction
any more, so we get involved in a lot of additions and remodels. No doubt the
opportunities for digital / pneumatic hybrid systems present themselves here
more often. Being an old Pneurotic guy, I guess I'm showing my age, clinging to
the past a bit. But what are we Americans known for if not our stubborn attitudes?
It's a strength and a curse all at once.

Anyway, thanks for the response. What a great thing this site and the Internet
has become, ehh? Can you imagine our fathers having a casual intercontinental
conversation over heating trade matters?!! It boggles my mind sometimes.

Regards,

Dan Law

“God save the Queen”

PS,

Not that this has anything to do with heating or controls, but the international
aspect of our postings put me in mind of it.

In response to the world’s ostriches who will not stand and be counted with the
US & Britain, but rather would mollify and extend time limits to Saddam, I offer
this quote from what is probably our most eloquent military man, Brigadier
General Anthony Maculifee.

During the Battle of the Bulge upon receiving an ultimatum from the German
commander of Panzerhops to surrender Bastogne post haste, Macauliffe sent the
Nazi his response: ................................................”Nuts!”

Remember the first rule of international warfare re: Franco military prowess: Going to war without France is like going deer
hunting without your accordion................your better off without it.