Iterative Control Algorithm SAT control

[sysint]

Anyone using Iterative control methods for VAV SA temp and staging control?

[control_noob]

Originally posted by sysint
Anyone using Iterative control methods for VAV SA temp and staging control?

GAAAAAAAK Brain Overload!!!

[nikko]

You typically don't want to adjust SAT in a VAV system. You may get away with two to four degrees using terminal load as an indicator of the system, but this is a summer to winter range, not a range to swing through during an average day In general, it's not a good idea as it will unbalance the system (temperature speaking), could cause oscilation in space temps and will take time to reach equilibrium if knocked too far off balance..

For seasonal changes, monitor your return air temp. You want approx 77° with 1 or 2° fluctuation. If it's higher, drop your SAT by the difference and leave it for a day or two, watching it closely (or set alarms accordingly). If it's lower, raise the SAT and watch it. If your RAT normalizes at 77° again, you're on the right track. Keep in mind a properly engineered VAV system is designed around a specific (and constant) SAT but also assumes a specific load. If the load is altered, the SAT needs to follow.

Instead, try adjusting your SAP(ressure). Allow it to swing from 1" WC to 2.5" WC on 0.5" WC incriments based on box damper position.

Poll all your damper positions and ignore any box reading greater than 99% as it could indicated a problem zone.

On all other boxes, monitor max damper position and if the single largest damper position is less than 95% open, decrease SAP by 0.05"WC. Wait five minutes and check again. If any box has a damper position greater that 98%, increase SAP by 0.05". This loop runs during all occupied hours.

Energy savings realized through riding your fan curve will far outweigh energy savings gained through adjusting SAT and your system will still heat/cool effectively and be able to respond to solar /wind loading throughout the day.

Good luck
Nikko

[osiyo]

Originally posted by nikko
You typically don't want to adjust SAT in a VAV system. You may get away with two to four degrees using terminal load as an indicator of the system, but this is a summer to winter range, not a range to swing through during an average day In general, it's not a good idea as it will unbalance the system (temperature speaking), could cause oscilation in space temps and will take time to reach equilibrium if knocked too far off balance..

Hmmmm. We reset the SAT, I presume you mean the AHU discharge air temperature?, routinely. Depending on the installation, we'll use a couple methodologies.

One is the reset off the return air temperature which you mention. Tho, we sure as heck don't wait a day or so before checking that again and doing further adjustment. It's done dynamically, all the time, tho the PID loop for that control is set up to respond slowly. Typically slower that the rate of change. With, of course, max limits on the change. Max limits we use being greater than those you mention.

The other method being to poll all the VAV's served by a particular air handler to fetch all the space temps. In summer mode AHU DAT is reset on a slow sliding scale based upon highest room themp. In winter, reset is done by lowest room temp. Again, reset has limits (greater than those you indicate), and the contolling PID loop is deliberatly set to be slow and gradual.

One does not want fast swings. It's counter productive. Particularly as one may have rogue, temporary fast swings caused by unusual conditions. ie One installation comes to mind. A school, music room. Which had an direct outside exit door sometimes used by the students. ie When band class was going outside to practice. So there were short periods when that door would be open continuously while 30 or more students filed out or back inside. During which outside ambient air washed over space sensor and it'd jump or lower it's reported temp quite a lot. In this case, I observed over time, via keeping a data log I set system to keeping temporarily. And as it turned out, a slow response PID loop for the AHU resetting of DAT was adequate. It compensated well for what was largely a false high or low temp. That inrush of OA was not enough to warm or cool the actual structure materials of the room, so once the door was shut, the room fairly quickly went back to, or near, desired set point temp.

I have seen similar cases where the only adequate solution was for us to program the system to exclude a particular space for polling when deciding upon reset. ie A Maintenance office for a building which had both inside and outside access doors and at particular times of days, for particular activities, outside door was opened for significant periods. To have allowed a reset of AHU DAT off that space would've mucked up the balance of the whole system. We just explained situation to maintenance supervisor, he understood problem.

The thing is, if the resets are set up properly. With suitable limits on change, and the PID settings are good, I've not seen it to be a problem to vary the DAT. Given that all components are working correctly, and other control loops are adequate. And that one accounts for extreme and unusal cases. ie Our decision to exclude that maintenace office I mentioned from polling to determine higest and lowest space temp to be used as a reset.

[QUOTE]Keep in mind a properly engineered VAV system is designed around a specific (and constant) SAT but also assumes a specific load. If the load is altered, the SAT needs to follow.[QUOTE]

Depends on who engineered it. I routinely see plans and sequences of operation drafted by design engineer who took into account the need to reset SAT/DAT based upon conditions and who has specified max range of reset. Sometimes they even get more detailed. ie One project last summer. HVAC engineer for architectural firm which designed new building not only specifed the limits, his plans called for method of determining reset. In this case he called for polling and using the average of space temps served by air handler. Didn't work worth a durned using the average, tho. At least not in the case of one area served by a particular air handler. Area had two rooms that were corner rooms, each with two walls having quite a lot of window exposure. In the end we proved to him it wasn't gonna work. And we resorted to to using high and low appropriately, with a tweaked PID loop, and adjusted max reset. Still wasn't perfect and personally I think he screwed the pup by failing to properly design more air flow for those rooms, better placement of the diffusers, bigger reheat coils, and probably should have upgraded the windows to ones with better thermal characteristics. But, of course, they were trying to save a buck here and there.

Instead, try adjusting your SAP(ressure). Allow it to swing from 1" WC to 2.5" WC on 0.5" WC incriments based on box damper position.

I'm not going to repeat all of your comments on resetting supply duct pressure. Gotta think about this one. First time I've heard of this tactic.

Not saying it's wrong or anything. Just never heard of it and have to think about it. While I've experience with a lot of things, that's not to say I've experienced everything or know everything. This thought of yours is unknown territory for me.

Anyone else ever see this? What's the strategy? Pros? Cons?

Right off the bat I wonder whether or not this might cause fan motor issues in some cases. I also wonder about noise increase and resulting complaints.

Hmmm. Well, this morning I'm going to a job site where we'll have a balancer on site, and a 30 year commercial HVAC systems pipefitter on site, and I've got to quiz them about this idea.

Not saying you're wrong, Nikko, not at all. I'm saying I just don't know enough about this tactic one way or another. Gotta get myself some educating on the subject. It's outside my experience, and I've not seen it in any of the texts and references I've read.

Right off the bat I wonder about excess hunting and system pressure changes. Since VAV damper positions tend to change at a faster rate, and probably more often, than -significant- changes in room temperature.

Next thought, if in fact a VAV -IS- maintaining set point within the space, and air flow is properly at or between min and max settings, why the heck do I want to reset anything? Why monkey with what works?

Like I've said, I'm here in this group to learn. So anybody, please feel free to educate my dumb ass. I'm not being a smart ass with that last statement. I'm serious. I try, everyday, to learn something new. And this is new to me. And it won't surprise me or upset me in the least to find out I'm ignorant in this matter.

[sysint]

Nikko's not from the Midwest..... so he's got some programming liability for the Midwest... (small ordinance shot - take it and pass it along) we have "seasonal changes" in the span of 8 hours. Today may be a classic. Start the day at 55F, watch it move to 80F and down to 45F by 5pm.

One does not want fast swings. It's counter productive. Particularly as one may have rogue, temporary fast swings caused by unusual conditions..... Next thought, if in fact a VAV -IS- maintaining set point within the space, and air flow is properly at or between min and max settings, why the heck do I want to reset anything?

...which Iterative control is supposed to prevent, not solely being reactionary. And if everything is good, it stops. The best part is that it's smarter response may mean you don't have to spend time figuring out good PID settings.

Agreed however on the pressure reset. I'll look into that. Almost like the lower SAT type delivery w/low volume.

EDIT: - Then the question for Nikko is have you used any Iterative schemes for duct pressure reset?

Maybe?:

New Pressure = Current Pressure + (Cool Capacity/VFD RPM)(SA Temp/SA Temp SP)(Slope of Fan characteristics)

I would think you need a factor included in the vfd area and you would have to establish some base of fan curve. Then you need another equation over time that you would use. Then temperature probably would end up being the offset, not the ratio. Also, cool capacity would have to be determined for what units and fan rpm adjusted accordingly in the equation.

[Edited by sysint on 04-19-2005 at 07:57 AM]

[smokies]

We do program resets on most of our jobs. Maybe our strategy is for our area - central florida. Never worked in the northern climates so we have it pretty easy.

We start the system in the morning with default design setpoints( example 55 dat setpt with 1.5" static setpt pretty standard for cooling).After an adjustable delay 30 mins - 2 hours we begin our reset.

We used to look at max VAV position and reset from there. However, it seems that now VAVs are being over sized on many jobs, so max airflow may be at 60%. So now we look at max deviation from setpoint from every box along with max damper position. This will give us a real look at the areas to see what needs what.

A very important point to remember the energy savings for an AHU is in the horspower of the fan. If you reset the SAT before the Static you are wasting energy.We start by resetting our static pressure by an adjustable increment- usually .1. We run the program every 15 minutes. If all VAVs are within .5 degrees of setpoint we will subtract .1 from active static setpoint.The process continues until the static setpt is at an adjustable low limit. If any box gets 1.5 above setpt we add .1 to the static. If the static setpt is at it low limit and all boxes are within setpt only then will we reset SAT or DAT setpts. We will add 1 degree.Usual Range is 52-62 in summer. The SAT setpt will continue to be reset as long as the VAVs are within setpt. If a VAV gets 1.5 above setpt we subtract from the SAT setpt first. We do not reset the Static setpt up until the SAT setpt has reached its low limit. This program works very well here. The engineers we've dealt with all like it. Never seen return temp reset work here.

We use Trane controls. Wondering how long it ould take to program this in a LON system. Just wondering in comparison.

Wondering what working for resets in other places.

[sysint]

We use Trane controls. Wondering how long it ould take to program this in a LON system. Just wondering in comparison.

If Trane made plugins (like everyone else) then it would take exactly the same time for you as they use Lonworks. (Comm5) Anyway, it goes fast with other controllers also. Download the program and make a few binds.

[sysint]

Anyone find a way to adjust this equation and take it to the next step?

New Pressure = Current Pressure + (Cool Capacity/VFD RPM)(SA Temp/SA Temp SP)(Slope of Fan characteristics)

Many are talking the demise of the PID loop - I'd like to see if this can be resolved. I just took the variables I thought would be involved. However I have no good definition of "Cool Capacity" or "slope". Or, even if the VFD rpm needs a factor applied.

[davem]

Sysint,
You brought out a couple of posts ago that the whole point of iterative control is that when conditions are good, it stops. Applying a trade-school-mathematics view of this formula in question, I'd expect "good conditions" to mean that New Pressure = Current Pressure + ZERO; leave it where it is. What that means is that in this formula, at least one of the terms (cool capacity, vfd rpm, sa temp, etc., etc.) would have to equal zero to get the formula to calculate out to zero and keep the pressure where it is. The way it looks now, the formula's result will always be in motion, ala PID without a deadband.
Am I oversimplifying how it works?
Or many I've naively stumbled into the "adjustments" you were inquiring about?

[sysint]

Dave, maybe you are simply asking "how close is close?" The answer is "n" close. So close to the same it is the same. Or, smart enough (in this case) to go where it needs to be next.

The PID loop by contrast is reactionary.

"stop" may not have been the best choice of words.

[davem]

Wow, this is quite a schooling in control theory - and I thought learning how to tune PIDs was complicated enough...I had trouble imagining how that formula (or one like it) could be predictive, as opposed to reactionary, because I thought that the only data available was real-time temperature/pressure data, to which the control loop would have to react. After reading this article: http://www.hpac.com/microsites/netwo...man_0307cf.htm, I can see how the application of certain "constants" (the determination of which seem analogous to tuning a PID loop) yields a predictive effect. The constants mentioned in this article may very well be the adjustments or tweaks you were referring to, sysint.

Of course, you might have already seen this article, but it proved quite informative to me - now if I could figure out a way to test it...

[sysint]

Actually, when you think about it I'd rather have the equation. No adjustments.

[osiyo]

Originally posted by sysint
Anyone using Iterative control methods for VAV SA temp and staging control?

Thought I'd answer you're original question.

I was not familiar with the concept of iterative control as applied to HVAC systems. Tho I pretty much immediately figured out what you meant by your question. As iterative methods of determing an answer to a math problem would be the same whether talking about HVAC systems or anything else.

In any event, the answer is no. Never done it, never considered it.

Hadn't even realized anyone was considering the method for a dynamic operating condition such as a HVAC system, which is a rather constant state of change. Particularly since many of the indivdual components do not have a direct, fixed, and linear relationship to each other. Add, the consideration of individual component limitations. ie, maximum allowable rates of change, to remain within the abilities of the individual components and to adhere to manufacture's suggestions. Then toss in, just to make it fun, suggested or required minimums and maximums of temp, speed, pressure, etc ... required by system designer or individual component maker.

Chuckle, talk about a possible can of worms !!!

In any event, I got curious and looked up Hartman's papers and writings about this.

Interesting. Never read his stuff before. But he does make some interesting and valid points. Not that he's the first, or the last, or the only one to cover the points he does. Rephrased, and using other techniques and solutions, I've seen various elements of what he discusses used in practice. Since before his first article on the subject in 2001.

Off the bat, a few thoughts.

Developing the iterative control algorithms could be very time consuming.

Each would have to be at least somewhat customized to each individual site and each different component. As one would have to take into consideration equipment limitations, operational restrictions, and so forth.

One is gonna need to include overrides, or exceptions, to meet unusual or emergency conditions.

Without PID loops, with their associated resets if such are necessary, what is each end component gonna do upon network failure? ie We frequently see it required by a customer or their A/E firm that systems or subsystems be able to operate in a standalone condition upon failure of a network for any reason. Certainly a control scheme for each controller is possible using the iterative method based upon information gathered from the network. But one is definitely going to have to ensure the resident program can run alone, continuing to use last received info, without failing or exceeding inherent machine limits imposed by maker of whatever physical equipment.

I'm not saying iterative control is not doable. Certainly it is possible. But I don't think the matter is quite so simple as Hartman hints it to be. And could be a real bear if one is working on a project where yah only have the controls portion, and others have the mechanical and electrical.

As concerns his comments about PID loops. <Shrug> We don't have the issues he does with them. Nor do many others of my acquaintance. PID loops can be very stable, properly adjusted they can avoid the too frequent adjustments he mentions. Etc. And they can employ interaction with other components of the whole system on the network by using changeable setpoints or resets that're changed based up whole system demand or factors from some other particular part of the system. ie We have PID loops where resets are changed on the fly during normal operation, so that the new reset matches new operating conditions as loading changes.

Just some thoughts. But a very interesting idea. His iterative control concept. In a number of installations we have we accomplish much, if not all, of the objectives he mentions, using other means. But I'm enjoying his thoughts and am interested in seeing some real, working examples.

Chuckle, tho I suppose he's keeping the details to himself. "Wanna know how? The precise algorithms I use? Hire me !"

I suppose that goes along with this business, and most others. The devil is in the details, and the inventor or developer of said details wants paid for them.

[sysint]

I understand about the network dependancy. However, you could generate some defaults fairly easily I think.

This relies on the communication with the VFD and the controller. I don't think the equation is overly difficult to start... what I mean is that you take in all the pertainent variables. However, looking at Hartman I see he has RPM 100% = 1 and I don't know where he gets his factors...

In the equation Hartman uses for Valve position you have
FC1CCVnew = FC1CCV + ( C2 / ( 1.0 + C1 x (PumpRPM - 0.3 ))) x (FC1DAT – FC1DATSP) x (C3 + FC1CCV )

Anyway, the majority of the equation has to cancel itself out leaving just a percentage. Seems like there is a BTUH equation in there somewhere.

[the_dude]

It appears to me that the resetting of SAP could be used in any climate zone really. Am I right?
This does seem to be a logical way to save energy prior to resetting the SAT.
How would resetting the SAP affect the required amount of outside air for the zones? Would we need to reset the mixing dampers open (if there were no dedicated outside air damper system)?

Cheers.