Multiple Units on common duct - air flow issues

[BronYrAur]

I'm trying to explain in technical terms what is happening with the attached system. There are 3 Liebert model VH380AUAAE1A546 upflow units dumping into a common duct system with provisions for a fourth. Each unit is rated for 15,200 CFM @ 0.5" ESP.

With only one unit running, I have no issues. As a second unit is brought on, however, I am seeing liquid slugging back to the compressor on the first unit. I assume that this is caused by a sudden loss of air flow across the first coil.

Just adding up the diffusers (12 @ 3,720 CFM) I don't have enough air outlets for the fourth unit, but that not my concern right now. I need to address the immediate problem of slugging and make a reommendation on how to correct the problem.

It too late to get units with high external static capabilities, so I am almost thinking the the solution is to leave one unit connected to the duct and the others to "free blow". The entire space is only about 50'x40' and it is a server room. I think the duct may be undersized for multiple units with only 0.5" esp, but I haven't done this type of system before.

I didn't create this problem (thank goodness), but I have been asked to write a letter explaining the problem and a recommendation for correction. Any thoughts?

 

[hoodguy]

What is the neck size on the diffusers?

 

[hollandhvac]

What is the type of supply fan in the unit, a backward curved or a forward curved fan? I think it is a forward curved fan.

 

[ChasBean1]

Good question by inventaguy... size should be either 24" or 30" for this application. Interpreting the drawing directly, I'd say they're meant to be 24". ALSO very important to note that the diffusers are manually adjustable. Start more CRACs, you probably have to manually open more diffusers. If you've already done that, please post and clarify more. CB

 

[KiwiMace]

The design intent for most computer rooms, and likely this one, is to have one unit as a redundant, thus the three units = diffuser capacity.

I think your main problem is the diffuser locations, particularly on CRAC-1 and 3 will dump cold air straight back into the RA intake on the front of the unit. Poor balancing would make it all the worse. This duct configuration would be v difficult to balance with the 2/3 duty cycle. Can't think of an easy way to fix that off hand. Maybe headers across 1,2 and 3,4, and H the other way between across the room, then branch all diffusers L-R from the H crossbar?

I can barely make out the rack layouts, but it looks like a hot/cold aisle layout, so you might be able to effect a reasonable solution by containing the ends of the cold aisles to prevent this short circuiting back to the CRACs.

 

[Waramanga]

I have been involved with a project with similar concepts involved, ie multiple id units feeding into common ductwork. Our head engineer was very careful to give each indoor unit a good length of duct before they joined common ductwork and they were certainly not pointing into one another. His explanation was that slight variations in esp resonated to cause big variations in air flow of individual FCUs.

 

[BronYrAur]

The diffusers have 16" diameter necks

 

[hollandhvac]

Yes good comment Waramanga, seen the same with 4 units that were combined in one plenum and supplied the air to an exchange building. Due to the forward curved fans and the low esp of the units, the last unit did not really help the system. In this you have to combine the 4 units and make the combined fan curve against the esp curve of the new system.

 

[DrRTU]

At 3720 CFM each, your grille selection is off the Titus chart. You will be over 0.5 TP just due to the grilles. I believe your design is grossly short of outlets. The VH380 units with the oversized drive is pushed to get 15.2K of flow. Most phone switches / data rooms want to have the 4" 85% filters in the 30 ton box and the filter area is the limiting factor. I do not know the ceiling height in this design or if this is a hot isle or cold isle design. The XC310 will have a throw over 30' even with the vanes at 45 degrees. The XC310 looks like an incorrect quantity and style of grille selection.

I would look into your back draft damper selection. I prefer a non-powered high pressure BDD with a viewing port window to make sure the damper is not stuck open. We have seen air leaking in the standby unit and fan trip on start due to fans spinning backwards.

I believe your operational issues are caused by low airflow. I have posted my design of a switch room with Four VH380 units in an N+1 design (2 units running and 2 standby). I listed this Lieberts at 15,200 CFM each. This is an all overhead, non hot isle design and front return. We did operate all 4 units w/o slugging issues. We did not use alternate unloading. The duct to the side is for smoke purge.

 

[BronYrAur]

Wow, thanks for all the info. I can clearly see how the diffusers are a major cause of the problem.

I didn't even mention the backdraft damper issue. They don't have any BDDs, but they do have isolation dampers with Belimo motors. They are slow to open which isn't doing the compressors any favors either. I'm going to take your recommendation on the HP BDD's with viewing ports.

Aside from being grossly shortly on diffusers, how does the ductwork look? Is is suitable for 3 to 4 units assuming adequate diffusers are added? I'm not a duct expert, but it seems to me that it is sized for no more than 2 units with 0.5" esp fans. Any thoughts on what I should recommend? Perhaps leaving the two on the same wall connected to the duct, and disconnecting the third and future units. I don't know ceiling height, but it's probably 12' to the deck. There is no suspended ceiling.

 

[ChasBean1]

Ductwork looks ok to me - highest velocity is only ~1550 fpm. One possible solution (see attached). Keep all your existing and just double your diffusers by installing 12 more of the same...

 

[SAK9]

I am not sure what is the distance between two opposite units.When two opposite units work,there is a strong possibility of "Bullheading" of the two air streams.The higher the velocity at which they meet the greater its effect and resulting reduction of air flow across the coils.I believe this is the reason for liquid flooding back to the compressor.The units could have been placed in such a way that the two streams flow parallely into the common duct

 

[ChasBean1]

A bullheaded tee is more of an issue in a water system than an air system, but agreed it's not ideal. There are plenty of ring duct systems around where bull heading concern could apply, but air generally provides additional pressure/flow versatility without dynamic issues you’d see with water.

You could ride up the curves toward the 'do not select' region with the added static of fans fighting each other, but enough air outlets (reduced backpressure) would help with that.

 

[SAK9]

Chas,

I believe you are referring to the duct loops used on VAV systems in office blocks.In such instances the duct splits into halves at the AHU outlet and take a big loop around the space and join at the farthest end from the AHU so the velcities are likely to be low at the point of joining.

 

[willard3]

To start, the fittings shown will create lots of turbulence and "fan effect" problem will effect fan discharge though I don't see a unit elevation. Read AMCA on fan effect. Liquid slugging is s sign of lack of airflow.

Forward curved fans can run at two pressure points for the same volume and do not operate well in parallel. Are the ducts drumming when more than one unit is on?

Control of forward curved fans in parallel is also a nasty control problem and you don't say how they are controlled.

 

[willard3]

"System effect" = "fan effect"

Disculpeme, disculpeme......

 

[ChasBean1]

SAK9, I was and I mostly agree - the velocities won't necessarily be low but the effect on the source fans is negligible because a majority of the volume's been dissipated prior to the streams colliding. The closeness of these CRAC units to each other and the bull-heading configuration won't help but it's not uncommon for multiple fans or AHUs to serve common outlet plenums.

I think if he has enough outlets, the heads of air that collide should have less ill effect - fans are lower on their curves and further from surge with less backpressure.

 

[sridhar1312]

Moment you introduce plenum where multiple units especially standard product the static gets affected. If you can make each unit connected to the duct without plenum and with one way gravity louver to stop air coming back in the unit which is not operated, the system will perform better.

Any fan manufacturer will vouch that by adding Plenum you get into problem area.