chilled water storage

[jds1901]

I have installed large thermocline type chilled water storage tanks for energy cost reduction. The tanks I installed were several million gallons each and were 30 to 40 feet tall.

I am interested in applying a much smaller storage tank that would provide a short span backup of a critical load. If the chillers serving the load trip, I need to be able to continue serving the load for 10 to 15 minutes while the generators start, automatic transfer switches close, and the chillers come back up to load.

Thread403-153938 by Orah back in May 2006 talked about an application just like this but there were no useful responses.

Does anyone know of a firm that makes small chilled water storage system in the 50,000 gallon range? Has anyone ever applied a chilled water storage tank for this purpose?

 

[cry22]

I've seen a design (IBM buuilding I recall) that had a chilled water storage tank in series with a centrifugal chiller plant for the same purpose you're describing. The tank was part of the piping system, located downstream of the chillers, that simple.

 

[jds1901]

I have put tanks on the outlet of a chiller before but generally that is to keep the chiller from short cycling. If the amount of water in the system piping and coils is too small, the chiller will load and unload rapidly.

A simple tank on the outlet of the chiller will not protect the load during an outage since the warm water pumped into the tank during the chiller outage will still end up going through the cooling coils.

I need a separate thermal storage tank that is normally offline but which can be quickly put into operation if the chiller shuts down. The tank can be recharged after the chiller is back up. I will probably need a small trickle charge to keep the tank cool when it is not being used since it will not be discharged regularly like most thermal storage tanks.

Also, this load is a small part of the load served by a very large chiller plant. If we could justify the economics of a large thermal storage system for the campus that would be great but it is not economical in this case.

 

[Wareagl487]

WoW!!! - Never ran across or thought of this strategy - I'm very interested to learn more!!

Andy W.

 

[hollandhvac]

I have seen once a design where there was a buffer vessel in between the supply and return from the chiller. This line fitted with a pump would fill with cold water. The pump, in this line, would be switched on incase the chiller stopped. I think this is what you want?
Also Trane have got a design in there catalog where there is one tank in the supply and return from the chiller. The supply (cold) is connected to the buttom and the return (warm) is connected to the top.
just my one cent contribution.

 

[KiwiMace]

This strategy is becoming more common in critical facilities. An area in which I have recently started working, so don't hold me to specifics.

The issue is the time it takes for the chillers to come back on line after a trip (order of minutes). In high density datacenters and the like, the heat load can cook the servers in as little as 30 secs. Pumps can respond quickly enough.

The chilled water tank is plumbed in pretty much how you would configure your other TES tanks. This is not the same as a buffer tank mentioned in some of the responses above. The tank could be located in a bypass loop that you can bring in and out to charge during low load times.

A 50,000gal size would probably be a one-off concrete, rather than some arrangement of modular tanks. Ice TES is much more compact.

 

[Wareagl487]

The two large data centers I have done have been DX - the facilities started small and have expanded 5 fold over the last few years. Original design did not support a chilled water system - but I bet if the owners looked back - they would probably have wished it had went chilled water.

Anyway, its easy enough to figure the chilled water demand at peak load over a given time frame - so - what is the given time frame to have adequate chiled water in reserve?

Inquiring minds want to know.....

Andy W.

 

[SAK9]

Why not install a DX coil in the air handler which can come on when the chilled water supply fails for any reason?This will certainly raise system reliability.

Maintaining a fully charged chilled water tank may not be energy efficient if it is not used regularly

 

[zerosum]

If 50,000 gal. of chilled water handles your critical load for the chiller recovery time, why not repipe your distribution system so that as you say the new storage tank (say, 30 ft tall x 15 ft dia) is in series downsream of the chiller and that the tank outlet piping leg serves only the critical load and then returns back to hot-side header system. Thus the new chilled water tank (dedicated to the critical load) is always having circulated cold-side water going thru it, and controls are greatly simplified or eliminated. I would go in the top and out the bottom, forget stratification in this case.

Many fabricators can build a carbon steel tank like that. Use API 650 as a standard. Good painting and vapor-sealed insulation is essential to avoid exterior corrosion.

You might explore a 60 ft hi x 12 ft dia tank which possibly would allow it to be shop fabricated.

A good reference is Erwin G. Hansen, "Hydronic System Design and Operation" McGraw Hill.

 

[zerosum]

Forgot to say, you would want to minimize mixing by slow entry and exit velocities.

 

[HEC]

I like SAK9's solution. The DX coil could be fed from a simple liquid reservior of sufficient size to allow cooling for the required time. The liquid reciever would then be recharged when the power comes back on.

Just a thought out of the box!

Mark Hutton

 

[jds1901]

DX is a reliable option but is not efficient for a baseload of approximately 1200 tons. A backup DX of that size is also very expensive and adds inefficiency due to the pressure drop across the DX coil which is normally not being used.

I found one ice storage application for a similar application but that requires an external melt ice system in order to be able to melt fast enough. External melt ice tends to bridge if it is not discharged regularly and could result in channeling during an emergency with most of the ice not melting quickly. Internal melt systems such as the Calmac tanks are not suitable for quick discharge.

I hoped some manufacturer was offering a chilled water storage tank just for this service and already had the kinks worked out. Apparently not.

I will probably do as zerosum suggests and just use a standard tank. If so, I will design my own diffuser for the bottom and top to minimize mixing and get the most out of the tank. It may not be perfect but I bet I can recover 90+% before I lose water temperature. I also like the idea of a 12 foot tank that can be factory fabricated, maybe two tanks 30 foot tall.

Thanks to everyone for the suggestions.

 

[zerosum]

jds1901

Give us feedback when you get-r-done.

In a previous life I had seven exotherming batch reactors with 3000T chilled water plus CTW to keep them cool. My interest now is just academic.

Good luck

 

[cry22]

My two cents. Keep it simple.
Nothing beats a storage tank full of coolimng medium ready to use with a dual set of pumps on emergency circuit (zerosum description).
Adding a DX system as suggested here will require a large emergency generator AND it will not be as reliable as a storage tank.

Two engineers, two opinions. well, if two people got along 98% of the time, one of them is not necessary.

Good luck jds.

 

[KiwiMace]

The extra DX coil will also cost good money in fan power over the life of the equipment.