Modular Scroll Chillers

[PowerPlay]

Recently installed three dual-circuit modular water chillers with scroll compressors, total capacity 170 tons. Unless load is a good match to a given number of compressors (which is rare), frequent compressor cycling is the norm. Compressor cycling adds & subtracts capacity in fairly large increments. The chilled water system is constant flow. Chiller mfg. recommends putting VFDs on the chilled water pumps and making it a variable flow system. How would that be better with respect to compressor cycling?

[jayguy]

brand? model? if it what i am thinking...do you have constant water flow through each module even when the module is off or do you have an actuator that closes and prevents water from going through the non-running module?

[motoguy128]

How large is you chilled water storage tank and how much water is in the system?

[jayguy]

How large is you chilled water storage tank and how much water is in the system?

i haven't seen a single one of these units that had a stoarge tank.

i haven't seen a single one of these units that didn't need a storage tank either.

[CCSPIERCE]

Jayguy you are correct. If system'sdesign were held to the minimum 3 gallons per ton loop volume for TOTAL MACHINE TONNAGE there would be a lot less problems. For process chilling 6 gallons per ton. I haven't seen a control scheme that lead lags machines within the chillers controls, that actually works in parallel. Although if isolation valves are placed and controlled properly things improve. It has been my experience that a third party control scheme controlling the chillers and isolation works best. In the end, loop volume is the key to success. With the scroll staging capacities you get less variable control, however the scroll unloading scheme Carrier is using with Danfoss works pretty well. It's just annoying to listen to it unload/ load within the seconds of parameter timing.

[PowerPlay]

They're ArctiChill machines. They're controlled by the master module internal controller, based on chilled water leaving temp setpoint.

Each module has isolation valves on chilled water side but the valves are open all the time. Mfg. rep said it needs to be that way for a constant flow system. Loop temperature fluctuated more when the isolation valves were active.

There is no storage tank on the chilled water loop. I do not know system volume but it is not a very large system -- serves two main air handlers, a couple fancoils and three chilled water CRACs. The benefit of a tank would be to act as a shock absorber for temperature transients? Due to space limitations we probably couldn't install a decent sized tank if we wanted to.

I'd still like to know how VFDs on the pumps would make things better.

[jayguy]

...I'd still like to know how VFDs on the pumps would make things better.

right now: when there isn't a compressor running on a module, there isn't any cooling on that module, so you pump un-cooled water right back into the loop. this causes the water temperature to rise and you slam compressors on and off.

with drives on the pumps: you would shut down the inactive cooling modules. water would only flow through the running modules...less cycling.

with constant volume, you can not do much which is why most technicians think these chillers (by type, not necessarily by manufacturer) suck. they are usually hyped waaaaaaay beyond their capability.

[19characterusername]

Do you have constant primary/variable secondary or constant volume with 3 port valves everywhere. A constant flow system with three port valves everywhere would move the entire mass of the circulating loop through the chiller at all times, but would be less efficient in terms of pumping energy.

You could put a (or multiple) 3 way valves on equipment far out on the loop with the sum of all bypass legs balanced to the evap flow of one module. That will get the mass of the entire circulation loop moving through a single module of the chiller. Then dP control on the chilled water pumps.

If you have primary/secondary pumping and if your primary loop is small, you're going to get short cycling. Full stop. There should be some anti short cycle timing in the controller. However, that may be getting overridden by some evap freeze protection control.

[PowerPlay]

Do you have constant primary/variable secondary or constant volume with 3 port valves everywhere. A constant flow system with three port valves everywhere would move the entire mass of the circulating loop through the chiller at all times, but would be less efficient in terms of pumping energy.

You could put a (or multiple) 3 way valves on equipment far out on the loop with the sum of all bypass legs balanced to the evap flow of one module. That will get the mass of the entire circulation loop moving through a single module of the chiller. Then dP control on the chilled water pumps.

If you have primary/secondary pumping and if your primary loop is small, you're going to get short cycling. Full stop. There should be some anti short cycle timing in the controller. However, that may be getting overridden by some evap freeze protection control.

System has 3-way valves at the major loads (air handlers and CRACs).

The master controller has minimum run time and minimum off time for compressors.

[19characterusername]

System has 3-way valves at the major loads (air handlers and CRACs).

The master controller has minimum run time and minimum off time for compressors.

Are you getting multiple modules short cycling or just one? How far are the AHUs and CRACs?

What might help is cranking down the balance valves on the bypass legs of the three port valves so that the sum of the flow through all bypass legs is equal to the flow of one module. Along with that I'd add two position two port valves on each of the evaps and add dP control to the CHW pumps. You gotta do all three. This is actually going to reduce the mass flow rate of water through the chiller plant, but it is likely to increase the mass flow rate per active module. More importantly, it will improve the dT across the active module which may encourage the controller to leave compressors on longer. The controller behavior discussion should go to your ArctiChill rep as I'm just guessing, although I suspect this is part of the reason they suggested variable flow.

Good luck!

[PowerPlay]

Are you getting multiple modules short cycling or just one? How far are the AHUs and CRACs?

What might help is cranking down the balance valves on the bypass legs of the three port valves so that the sum of the flow through all bypass legs is equal to the flow of one module. Along with that I'd add two position two port valves on each of the evaps and add dP control to the CHW pumps. You gotta do all three. This is actually going to reduce the mass flow rate of water through the chiller plant, but it is likely to increase the mass flow rate per active module. More importantly, it will improve the dT across the active module which may encourage the controller to leave compressors on longer. The controller behavior discussion should go to your ArctiChill rep as I'm just guessing, although I suspect this is part of the reason they suggested variable flow.

Good luck!

Cycling happens when building load is not a close match to the capacity of a given number of compressors. Usually two compressors isn't quite enough and three is too much, so the third one cycles. Load varies depending on outside weather conditions (outside air is brought into the main AHUs). Sometimes load gets in a sweet spot and there is minimal or no cycling for a half-day or more.

"How far are the AHUs and CRACs?" The main AHUs are in the same mechanical room as the chillers. CRACs and a couple fancoils are one floor below.

[PowerPlay]

Jayguy you are correct. If system'sdesign were held to the minimum 3 gallons per ton loop volume for TOTAL MACHINE TONNAGE there would be a lot less problems. For process chilling 6 gallons per ton. I haven't seen a control scheme that lead lags machines within the chillers controls, that actually works in parallel. Although if isolation valves are placed and controlled properly things improve. It has been my experience that a third party control scheme controlling the chillers and isolation works best. In the end, loop volume is the key to success. With the scroll staging capacities you get less variable control, however the scroll unloading scheme Carrier is using with Danfoss works pretty well. It's just annoying to listen to it unload/ load within the seconds of parameter timing.

"Loop volume is the key to success." Does that mean more is better?

Using third party control of the modules means taking control away from the Master module controller. This could void the mfg. warranty since these are new machines.

[CCSPIERCE]

Loop volume determines the amount of time that water flows from the cooler thru the piping and back to the cooler. In the case you describe, envision that the water has left the chiller, traveled out into the mechanical room piping, thru a single AHU 3 way valve which is in bypass. That water has essentially picked up very little heat. In the case of 100 tons, at a flow rate @ 1.8 gpt/m you would have theoretically ( with no losses) 180 gallons per minute flow rate. If the loop volume is 180 gals. then your chilled leaving water returns back in one minute to the cooler. Increasing that loop volume to 3 gals per ton achieving 300 gallons per minute nearly doubles your returned water timing. The chiller's response time to react to changes in 3 way valve(s) bypassing ends up with unit off, water warms up when 3 way valve quits bypassing, chiller restarts, chills water and bypass starts the whole thing over. Utilizing the chiller's return water reset program to raise leaving chill water setpoint sometimes can get you above the supply air setpoint which is controlling the 3 way valve. If return water warms up the chiller should lower the LWT back down to 44.0...
Using a control scheme outside of the multiple chillers to determine the AHU requirements and calculating BTU required would minimize multiple chiller staging, thusly the chillers controls would stage the single chiller running.

Since you have CRAC units you most likely need to maintain 44.0 without resetting the leaving chilled water setpoint. More than likely you need 6 gallons per ton volume. Add up chiller's vessel volume, piping volume and ahu coils piping as best as you can to determine your actual loop volume. ASHRAE handbook has some charts to help determine pipe volume. The chiller's data sheet should tell you the vessel's volume. If AHU coils volume is undeterminable, count coil tubes and use handbook to determine tube volume. Then add it all up to see what the loop volumes are per each stage of chillers.
Keeping in mind that when the AHU control valve bypasses, that AHU's coil loop volume is removed from the loop. If a chiller is off and valve isolated then that cooler vessel loop volume is also negligible at that time.

[heatingman]

VFD'd on the house circuit will do you no good, especially since there are 3-ways at the load points.

The 3-ways make for constant flow in the loop. You could utilize the vfd as a means to set the flow rate instead of using discharge valves.

Did anyone ever set the flow rates? Through the loop, or the chiller bundles?

To utilize a VFD properly, the loop volume needs to be variable, as it would be if the control valves at the loads were 2-way.

In either case, vfd's won't help with chiller cycling. For that, timers, buffer tanks, HGBP, or an APR valve would be needed.

Personally, I'd be looking at an APR valve added to the 1st stage compressor on each chiller module.

[CraziFuzzy]

Did you ever get this sorted? I'm not sure it was every really explained what you need the extra volume for. The idea is, the cycling is due to a difference in chiller capacity, and system load. This results in total heat being either removed, or added to the system. This heat results in a temperature change in the system, which leads to the cycling, where the temp starts swinging the other way. There is no 'fix' to remove the cycling - and a variable flow system is going to have the same problem. All you can hope to do is slow down the cycling. This is where loop volume comes in (A chilled water buffer tank). By increasing the overall system volume, the rate of temp change for a given imbalance is greatly reduced. Slowing the rate of temperature change will slow down the staging up and down rate.

That said, there are other advantages (on the energy saving front) from using a primary/secondary system, and getting rid of the 3-way valves. Currently, you are using extra electricity most the year just to move water in a circle, for no real reason, other than the system was easier to design. You can fix two problems in one by using a chilled water buffer as the decoupler between a primary and secondary system. Put a fixed speed primary pump (they don't have to be very big) for each of the chiller modules, that only turns on when that chiller is needed. Get rid of the 3-way valves, and use a VFD on the system pumps (which would now be secondary pumps).

[PowerPlay]

Did you ever get this sorted? I'm not sure it was every really explained what you need the extra volume for. The idea is, the cycling is due to a difference in chiller capacity, and system load. This results in total heat being either removed, or added to the system. This heat results in a temperature change in the system, which leads to the cycling, where the temp starts swinging the other way. There is no 'fix' to remove the cycling - and a variable flow system is going to have the same problem. All you can hope to do is slow down the cycling. This is where loop volume comes in (A chilled water buffer tank). By increasing the overall system volume, the rate of temp change for a given imbalance is greatly reduced. Slowing the rate of temperature change will slow down the staging up and down rate.

That said, there are other advantages (on the energy saving front) from using a primary/secondary system, and getting rid of the 3-way valves. Currently, you are using extra electricity most the year just to move water in a circle, for no real reason, other than the system was easier to design. You can fix two problems in one by using a chilled water buffer as the decoupler between a primary and secondary system. Put a fixed speed primary pump (they don't have to be very big) for each of the chiller modules, that only turns on when that chiller is needed. Get rid of the 3-way valves, and use a VFD on the system pumps (which would now be secondary pumps).

No, this has not been fixed yet. We have someone doing evaluations and hopefully they'll come up with a solution which can reduce cycling and not require major hardware changes.

A buffer tank would probably help, but it would be quite difficult to add due to space limitations. Putting it outside the building is not an option.

[PowerPlay]

Did you ever get this sorted? I'm not sure it was every really explained what you need the extra volume for. The idea is, the cycling is due to a difference in chiller capacity, and system load. This results in total heat being either removed, or added to the system. This heat results in a temperature change in the system, which leads to the cycling, where the temp starts swinging the other way. There is no 'fix' to remove the cycling - and a variable flow system is going to have the same problem. All you can hope to do is slow down the cycling. This is where loop volume comes in (A chilled water buffer tank). By increasing the overall system volume, the rate of temp change for a given imbalance is greatly reduced. Slowing the rate of temperature change will slow down the staging up and down rate.

That said, there are other advantages (on the energy saving front) from using a primary/secondary system, and getting rid of the 3-way valves. Currently, you are using extra electricity most the year just to move water in a circle, for no real reason, other than the system was easier to design. You can fix two problems in one by using a chilled water buffer as the decoupler between a primary and secondary system. Put a fixed speed primary pump (they don't have to be very big) for each of the chiller modules, that only turns on when that chiller is needed. Get rid of the 3-way valves, and use a VFD on the system pumps (which would now be secondary pumps).

I know cycling occurs because of mismatch between capacity and demand. The modular chillers add & subtract capacity in 25 or 30-ton increments. We need another small chiller in the loop that has variable capacity. Use the modular chillers for base load and the other chiller for load-following.

[CraziFuzzy]

Without knowing the system volume, it's impossible to determine the size tank that would be required. Still, it is likely to be much cheaper than a variable chiller. Keep in mind, cycling isn't necessarily a bad thing - but excessive cycling is. A buffer is not going to remove the cycling, just stretch it out - which should be enough. Truly variable chillers in the 30-ton range are uncommon, which usually means expensive. Most likely option would be a smaller multi-scroll - something with 5 or 6 5-ton copelands in it... even that would still cycle as needed though.

[19characterusername]

Scroll modulars typically have 2x10, 12.5 or 15 hp compressors and should be able to cycle one or both on. On the multistacks with the brazed plate HX the evaps are arranged CKT1|CHW|CKT2|CHW etc and the condensers are arranged CKT1|COND W|CKT2|COND W etc. So each module should be able to get down to half its capacity. Short cycling is an issue with low effective system mass as discussed or could be controls trying too hard to accurately maintain setpoint. Have you trended CHWS against stages? Maybe the chiller plant controller is being too aggressive in staging things on and off to tightly maintain setpoint. You might be able to fix this with software.