Chiller Plant Sequencing

[joey791]

I'm doing a chiller plant with 3 chillers and 2 chillers have isolation valves. My question is this, how do you go about sequencing your chillers such as resettting temps and when, when lead chiller fails how do you prove it, temp vs. time? How long do you wait after opening valves to start the other chillers?

I've done all this before but here in La, when the temps and humidity creep up 5 minutes could mean you dont catch up for the day, on the other hand properly protecting equipment so a chiller isn't cutting off and on every 5 minutes is paramount. Just trying to get a general concensus on how to make my sequencing more efficient.

[MatrixTransform]

Chiller Isolating?

Ive seen valves on condenser water for some screw chillers that need to get the pressure up quickly to move oil...but never really seen isolating on chw.

Youre controlling leaving water right? if LVG-T isnt down to setpoint, then start another chiller. wait-open valves-wait-start pump-wait-start chiller.

I suggest staging chillers off based on RET-T (inside the bypass/balance loop) as the bypass opens, the return water temp starts to drop (or flow in the bypass/balance line increases) or when the flow equals one CHW Pump worth, then you dont need the chiller....shutdown. Stop chiller-Stop Pump-close valves.

[joey791]

This is a 30 year old campus we are doing some upgrades on. It has a constant flow pump in each building(2) for this plant. Isolation valves are installed in the 2 smaller chillers because water flows through all chiller barrels all the time, now we can cut flow to the chillers that are not needed. Matrix how long do you usually wait to make the call for another chiller 5 minutes or 30 minutes, I've seen it spec'd both ways for failure on startup and for regular staging.

[Southern Mech]

most my chiller plants are on 10-15 min b-4 it brings on the next chiller.

[mechdorn]

Hey Joey,

By the short description, I'll assume a "variable flow primary pump" meaning a single VFD pump does all. In this application chillers are normally isolated out when not online. Usually, with all multipliable chiller applications, the question that comes to the top is " when to add or subtract a chiller?" The answers are as many as there are licensed PE's in your state. Here is what look for in a good sequence. Two or more arguments must agree to start and two have to agree to stop. Example, common leaving temp setpoint is 45 deg, above 47 deg add another chiller? OK second chiller starts, temp now goes to 45 deg in 20 minutes, second chiller Off? Better, common leaving temp setpoint is 45 deg, above 47 deg and chiller1 KW is greater than 95%, start chiller 2. To remove a chiller, common temp needs to be lower than 45 deg and both chillers KW is less than 65%. The same principled can be applied with several different inputs, flow rates, valve positions, temp sensors, KW readings, all kinds of combinations. The " latch logic" I described can be deployed with most systems. If it is the system type I described, flow rates will usually have a major say-so in the sequence of operation.

[joey791]

Hey Joey,

By the short description, I'll assume a "variable flow primary pump" meaning a single VFD pump does all. In this application chillers are normally isolated out when not online. Usually, with all multipliable chiller applications, the question that comes to the top is " when to add or subtract a chiller?" The answers are as many as there are licensed PE's in your state. Here is what look for in a good sequence. Two or more arguments must agree to start and two have to agree to stop. Example, common leaving temp setpoint is 45 deg, above 47 deg add another chiller? OK second chiller starts, temp now goes to 45 deg in 20 minutes, second chiller Off? Better, common leaving temp setpoint is 45 deg, above 47 deg and chiller1 KW is greater than 95%, start chiller 2. To remove a chiller, common temp needs to be lower than 45 deg and both chillers KW is less than 65%. The same principled can be applied with several different inputs, flow rates, valve positions, temp sensors, KW readings, all kinds of combinations. The " latch logic" I described can be deployed with most systems. If it is the system type I described, flow rates will usually have a major say-so in the sequence of operation.

No variable flow pumps, this plant serves 2 buildings with a constant flow pump in each building. I am not taking KW readings and have used that approach before but with results I wasnt too happy with because one chiller was a giant piece of crap and the other one was new and up to the task so depending on which chiller was lead the other one would cycle on and off every 30 minutes

[mechdorn]

No variable flow pumps, this plant serves 2 buildings with a constant flow pump in each building. I am not taking KW readings and have used that approach before but with results I wasnt too happy with because one chiller was a giant piece of crap and the other one was new and up to the task so depending on which chiller was lead the other one would cycle on and off every 30 minutes

So the same flow is cut in halve when another chiller is brought on-line? The place where your working should consider some professorial engineering consultation. Run, joey run.

[emcontrols]

Sounds challenging! May I ask a few questions.
1. What are the 3 chillers tonages and type? Centrifigul, screw etc.. water cooled or air cooled. and which chillers have the iso valves(need to know if the valves are 2 pos or modulating and where located chilled water or condenser side or both.
2. Constant volume Pump in each building (I assume they run 24/7 or when buildings are occupied and at the same times)
3. Are there any dedicated pumps for the chillers, or pumps that serve the chiller plant(What im getting at is primary/secondary system, series or parrallel and or decoupler/bypass)
4. What temp sensors are in place and flow meters if any? (Is this a design build or existing system you are fixing)

Need this info to start.

[supertek65]

obviously, you can no run chilled water through any stopped chillers!!!!!!!!!!!!!!!!!!!!!!!

I imagine you have these chillers piped in parallel and 3 separate chw pumps????????

I have many accounts with 3 chillers.

44 degree leaving water, 45 degrees for 15 minutes and chiller 2 comes on 46 degrees for 15 minutes and chiller 3 comes on.

I havd ddc controls closing or isolating the chw on chillers not calling!

ofcourse lead lag all 3 chillers.

Mostly what I have is centravacs, but I also have a few 19d and YK!!

The carrier chillers do their own lead lag and the YK I lead lag them every start!

just going on bla bla bla!!

Frank

[berg2666]

I'm doing a chiller plant with 3 chillers and 2 chillers have isolation valves. My question is this, how do you go about sequencing your chillers such as resettting temps and when, when lead chiller fails how do you prove it, temp vs. time? How long do you wait after opening valves to start the other chillers?

I've done all this before but here in La, when the temps and humidity creep up 5 minutes could mean you dont catch up for the day, on the other hand properly protecting equipment so a chiller isn't cutting off and on every 5 minutes is paramount. Just trying to get a general concensus on how to make my sequencing more efficient.

Can you make inputs to a controller from each chiller for chiller status and alarm points from the individual chillers. Are the chillers equal tonnage? Do you have ways to load and unload chillers say water setpoints or current limits setpoint control?

[amigo]

this may be useful for you:

http://www.trane.com/commercial/libr...31_4/index.asp

[joey791]

Sounds challenging! May I ask a few questions.
1. What are the 3 chillers tonages and type? Centrifigul, screw etc.. water cooled or air cooled. and which chillers have the iso valves(need to know if the valves are 2 pos or modulating and where located chilled water or condenser side or both.
2. Constant volume Pump in each building (I assume they run 24/7 or when buildings are occupied and at the same times)
3. Are there any dedicated pumps for the chillers, or pumps that serve the chiller plant(What im getting at is primary/secondary system, series or parrallel and or decoupler/bypass)
4. What temp sensors are in place and flow meters if any? (Is this a design build or existing system you are fixing)

Need this info to start.

1. I do not have tonnages right now. 2 Trane air cooled screw chillers and 1 Carrier scroll chiller air cooled. 1 Trane chiller is about a year old and will be lead, 1 RTAA104 has an iso valve on the chill water on/off, Carrier has an iso valve on the chill water on/off.
2. Constant volume pump in each building seem to run at the same time, I haven't seen one drop yet and the other run, both run on occupancy.
3. No dedicated pumps for chiller just these 2 loop pumps, 1 in each building.
4. Chill water supply out of each chiller and common return for both buildings. No flow meters, there never were any. One building is 30 years old the other is 40 years old. The original plant had 4 chillers but since all chillers have been changed over the past 5 years has been widdled down to 3. Controls were originally JCI pnuematics then start/stop control was changed to Robertshaw controls, now I'm pulling the Robertshaw controls out and putting our stuff in.

[joey791]

Can you make inputs to a controller from each chiller for chiller status and alarm points from the individual chillers. Are the chillers equal tonnage? Do you have ways to load and unload chillers say water setpoints or current limits setpoint control?

Wire has been pulled and alarm points were not included. Chillers are not equal tonnage(at least the Carrier is a smaller tonnage for sure). We have wire ran for chill water reset on all chillers, Trane came with it, the Carrier does not have an EMM module to reset chill water with.

[emcontrols]

Ok, lets begin. We know the trane with no iso valves will have to be lead. We also no the reason for the iso valves on the other 2 chillers is to prevent chwr mixing in to supply and thats all they are for. We are also assuming the engineer is ridding the pump curve on GPM when only 1 chiller is on line (Lead w/out iso valves) only way you could change the lead would be add iso valves to chiller without them and add a decoupler line. (Might not be an option but would be recommended) The next feasible solution is to determine if the remaining 2 chillers are equal in tonage(If not, rotate is not an option) You will then bring on the smaller of the 2 and or the chiller that unloads the most and finally chiller 3. (It would be nice if you had percent chiller loaded/unloaded-add current transformers etc.., Then you could wait to bring a chiller on after chiller is fully loaded, then unload chiller 1 and bring on chiller 2 and so on as well as rverse order. Controlled by resetting your chilled water supply temp set point based on return water temp (Building Load).

You mentioned timing on valves etc.. Pumps are already running, Im assuming there are chw flow switches dedicated to chillers so logically you want to open valve confirm how long it takes to prove flow (record and put into program) and initiate chiller start. And stop chiller then close iso valve to avoid nussance flow loss alarms.

As far as chiller on/off minimums they should be already factory set, if not get M# & S#'s and set for recommended settings and incorporate in to programming to eliminate control fighting. I know us control guys like tight accurate control. I would use and abuse the DDC loop, tweek the PI and stay away from the D quickining the anticipated cycle times would make things hectick on the staging and the min on/off times.

[supertek65]

your other option, which is one I use at a good account !
put drives on all the pumps and run them on pressure drop across each chiller!!!!!!!!!!
but you still need an iso on that third chiller!
Unless of course you wish to rust the chiller out?

Ok, lets begin. We know the trane with no iso valves will have to be lead. We also no the reason for the iso valves on the other 2 chillers is to prevent chwr mixing in to supply and thats all they are for. We are also assuming the engineer is ridding the pump curve on GPM when only 1 chiller is on line (Lead w/out iso valves) only way you could change the lead would be add iso valves to chiller without them and add a decoupler line. (Might not be an option but would be recommended) The next feasible solution is to determine if the remaining 2 chillers are equal in tonage(If not, rotate is not an option) You will then bring on the smaller of the 2 and or the chiller that unloads the most and finally chiller 3. (It would be nice if you had percent chiller loaded/unloaded-add current transformers etc.., Then you could wait to bring a chiller on after chiller is fully loaded, then unload chiller 1 and bring on chiller 2 and so on as well as rverse order. Controlled by resetting your chilled water supply temp set point based on return water temp (Building Load).

You mentioned timing on valves etc.. Pumps are already running, Im assuming there are chw flow switches dedicated to chillers so logically you want to open valve confirm how long it takes to prove flow (record and put into program) and initiate chiller start. And stop chiller then close iso valve to avoid nussance flow loss alarms.

As far as chiller on/off minimums they should be already factory set, if not get M# & S#'s and set for recommended settings and incorporate in to programming to eliminate control fighting. I know us control guys like tight accurate control. I would use and abuse the DDC loop, tweek the PI and stay away from the D quickining the anticipated cycle times would make things hectick on the staging and the min on/off times.

[joey791]

No variable speed has to work as is

[supertek65]

oh darn!!!!!!!!!!!!!!!!!!!!!!!!

No variable speed has to work as is

[hvac ddc]

Could calculate approximate tonnage based on temp drop across chiller. If design TD is 14 deg, 7 deg = 50%. We do something similair with a chiller plant w/ 3-950 ton Yorks. There we calculate plant tonnage based on main loop TD and flow. We run a minimum of 1 chiller and then if the calculated main loop load is >80% for 10 min, another chiller stages on. This next chiller will stay on until loop load is <80% for 10 min.

[MatrixTransform]

How long? ... I dunno, A few minutes I s'pose ... if you need chilled water, then why wait at all????? ... 'cos the spec says so?

I guess I would tailor the start delays to meet 2 critereon:

1: not exceed recommended starts per hour for the chiller(s)
2: not wait so long that Lvg water drifted too far from desired setpoint.

LVG CHW not at setpoint ? ....wait (?) start another chiller.

...as return water drops, the chillers will unload themselves

...when flow volume thru balance/bypass equals the flow a of Prom CHWP, then shut down that chiller.
OR
when return water temp (inside the bypass loop) drops below xx.xx °C , then shut down that chiller

Shutdown takes priority over start up ... unless extraordinary conditions such as CHW LVG T > xx.xx

etc