Chilled water buffer Tanks

[Bos88]

Hi All,

I am working on a 700 TR process cooling application having 2 water-cooled screw chillers. The chillers are piped to several plate and frame heat exchangers and primary pumps are constant flow. As the primary loop volume is very small, we have added buffer tanks on the return piping (before the pumps) to prevent chillers from short cycling. Due to some local availability constraints, we have to use 4 buffer tanks (1000 gal. each) piped in series instead of a large one.

One of the supplier has offered a 4-port buffer which to me is more appropriate for a primary-secondary application which is not the case in my system. He mentioned we can close 2 of the ports and use it as if it was a 2-port system. When I looked at the design of this buffer tank, it does not have any internal baffle plate (as it is design primarily for P/S applications). My colleague handling the layout of the project was planning to pipe the buffer tanks in series with either
- Solution 1: tank 1 having top inlet and bottom outlet piped to tank 2 bottom inlet, then from tank 2 top outlet to tank 3 top inlet and from tank 3 bottom outlet to tank 4 bottom inlet
- Solution 2: Tank 1 bottom outlet piped to tank 2 top inlet, tank 2 bottom outlet to tank 3 top inlet and finally tank 3 bottom outlet to tank 4 top inlet

My questions:
1) Do you think that any of these configuration will work properly (adding volume to the network to prevent chillers short cycling). My main concern is that there are no baffles in these tanks and from what I read on different suppliers web site, baffles in buffer tanks are required to prevent stratification and ensure adequate mixing.
2) Which piping configuration is preferable: solution 1 or 2?
3) As it is a process cooling, do you think it is more appropriate to locate the tanks downstream of the chillers (supply side) rather than upstream (as we have currently considered).

Thanks in advance for your comments/suggestions.

 

[itsmoked]

Delaying the chiller from cycling off would likely work better using supply side control. Tanks in series and without baffles then completely homogenize them. Add circulation from the last tank back to the first tank. This forces the entire 4,000 gallons into thermal mass participation and having to be cooled to cycle-off. I'd add a circulation pump to do this. Use eductors for all tank inlets to rapidly, actively and throughly stir the tank volumes. The circulation pump will keep the volume stirred during off cycles - unless the system always flows during the off cycle.

Eductor

Eductors translate pressure into flow multiplying the source flow by 3 to 8 times. One properly placed in a cylindrical tank can stir it well.

Example Eductor guide in PDF

Keith Cress
kcress -

http://www.flaminsystems.com

 

[LittleInch]

No 1 - No real idea, but seems a good solution to give you a minimum volume for your chillers, but how big a volume you need and how variable your secondary flow / heat load is will need to be looked at to give you an idea. How many times you can have your chillers cycle, the options for turndown etc etc all bring it into the mix

Where did 4000 gallons come from?

Can't see any real difference in solution 1 or 2, but agree you either need baffles or some sort of mixer in the vessels to prevent stratification.

I would go for locating on upstream myself but maybe for some flexibility you could pipe in parallel or two tanks in series in parallel and you can isolate one or more streams as the duty changes?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Bos88]

Thanks all for your comments and suggestions

Regarding the 4000 gallons, this is based on chillers' supplier recommmendation which is typically 3 gal/TR for comfort cooling and up to 6 gal/TR for a process cooling. So for the 700 TR total load we ended up with 4000 Gallons. We have a small flow (about 600 GPM Total) which this will give us a 6+ minutes run time. If return temp to chillers stay constant we probably won't have much issues with cycling of chillers but in this application it is difficult to predict. So we prefer to add the buffer just in case and we could also build in a bypass around the tanks if needed. Best is that I foind a buffer with baffle or use the eductor as suggested,

 

[Compositepro]

Thermodynamically, mixing warm and cold water in a tank is is not efficient thermal storage. Stratification is a good thing in thermal storage tanks. You chiller should draw warm water from the top of the tank and return cold water from the bottom of the tank. Your process should draw cold water from the bottom of the tank and return to the top. The amount of cold you can store in a tank depends not only on the volume but also directly on the difference between the maximum and minimum supply temperature your process can tolerate.

If your chiller capacity is always greater than your process load, you would start the chiller when the bottom of the tank get too warm and stop the chiller when the top of the tank gets too cold. This will maximize your cycle time with a minimum of tank volume.

If your process load sometimes exceeds the chiller capacity, thermal storage can be used to provide a peak chilling capacity that is many times greater than average.

 

[Bos88]

Thanks Compositepro for your input. Just wanted to clarify that in my case, I was not refering to thermal storage tank (TES) but to a buffer tank aimed at increase the overall volume of water in the network to prevent frequent cycling on/off of chillers. Something similar to this type of product:

https://westank.com/buffer-tanks/chilled-water-buffer-tanks/

What I find confusing is that most of the suppliers for buffer tanks state that internal baffles are required to prevent stratification. One Supplier states the opposite or more precisely indicates that adding baffle + sparge pipe can transform the buffer tank into a TES tank (see

http://www.masterflow.net.au/buffer-tanks).

I can understand the sparge pipe to make it a TES but not the baffle. I will ask them but I never thought of using a Buffer tanks and transform it in a TES Tank with some accessories.

One of the chiller catalog provides guidelines on connection of buffer tanks. In my case the current tanks that have been proposed do not have internal baffles so they look like the one I circle in red which as per the chiller supplier is not recomended. On the right side of the picture is a sketch of the system I am looking at with 4 tanks in series. In my opinion these tanks should have a baffle to prevent stratification while my colleague believes this will work fine without baffle. Hence this debate.I think the first 2 posts I got seem to give me reason.

 

[LittleInch]

Bos88,

From where I'm sitting I think your system is a variable one, i.e. your load can vary up and down and hence temperature of the water entering your 4 tank system can sometimes be hotter and sometimes colder than the average temperature of the water in the tank. correct?

Hence from your diagram above, whilst I agree the top left one is bad, the bottom left is maybe just not as good as the others. At 600GPM in a 1000 gallon tank, your average residence time is less than 2 minutes. Doesn't give a lot of time for stratification, especially if you enter at a decent velocity or even induce some swirl.

But the hot / cold thing to me means that in some cases enter at bottom, leave at top is good (when water is colder as it will then sweep up the vessel, but entering hotter could result in that water climbing to the top faster. Hence your original solution 1 would cover both and get some sort of uniformity of temperature over time and stop the vessel short circuiting.

Vendors of chilling plant have no real idea how there system is used or whether heat loads very slowly or quickly or by how much. Hence why the guidance seems to be relatively slowly for "comfort" cooling, but quicker for process cooling hence why you need more thermal mass to even out the variances. Only you know what your cooling load demand looks like so make some sensible decisions from there.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Bos88]

Thanks LittleInch for your comments.
Actually the pumping system is constant flow in this case but load (return water temp from process will vary). I still included VFD on the pumps for flexibility but the plant is designed to operate at constant flow (client requirement)

Good point about the flow vs tank capacity not giving enough time for stratification.

Thanks for that

 

[LittleInch]

No problem. My point about the varying load is that depending on the rate of change you might be able to reduce your thermal mass, but otherwise a smaller vessel is better in terms of sweep through, but also means that a sudden loss of cooling load will result in a fairly sudden drop in temperature, just that it is delayed by a few minutes.

If you're getting large sudden drops in load then you might want to look at piping these in parallel to get you a bit more mixing into a larger volume to avoid the sudden temp drop and hence short cycling.

Only you can decide based on your system demands.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[Compositepro]

I just want to point out that buffer tanks work due to thermal storage. They supply the cold water while your chiller is off. They are just much less efficient than a well designed thermal storage system and therefore you require much larger tanks. While the controls for thermal storage are more complicated and expensive, you may save money due to the smaller size and foot print of the tanks and the higher efficiency of the chiller.

 

[BronYrAur]

If all you are doing is adding volume to the system with these tanks, what would be the harm in piping them in parallel? That way you could valve one off should it need servicing without shutting down the entire system.

 

[Bos88]

Thanks BronYrAur for the suggestions on the parallel piping. It is a good point.

Yes Compositepro, I always buffer tanks to be purely a volume solving equipment. Looking at some suppliers, I realize that some of them have as well modified their buffer tanks to add a sparge pipe to make them small compact TES tank as well.

Thanks to all.

 

[quark]

You will have problems with filling of tanks in a parallel set up. How do you fill and empty all the tanks simultaneously? If you don't do that, your buffer reduces, or you may need to go with complex control system. 700TR chiller and 4000 gallon tanks provide you buffer for 2 minutes at full load or 10 minutes at 20% load. Compositepro's suggestion is better and trouble free.

If you still decide to go with the set up, see that the final tank in series supplies water to the circulating pump from bottom, to avoid air entrapment. Your big challenge is to avoid air in the buffer tanks.

 

[Bos88]

Yes Quark you have a valid point.
My total flow is 600 GPM so I calculated 6+ minutes buffer at full load based on 4000 gals. How did you arrive at 2 minutes

 

[quark]

Are you working on 15C ΔT?

 

[Bos88]

Yes quark,

 

[quark]

That is high for a primary only system. Have you checked with Chiller Manufacturer?

 

[BronYrAur]

I'm not in a position where I can make a sketch and send it but just pipe the four tanks in parallel in a reverse return setup. That way the amount of piping to and from each tank is essentially equal and the pressures will be close enough that you will get the same flow through each one. Again, as I understand it this is strictly to put volume in the system so that you're not cycling the chiller so often.

 

[Bos88]

Yes I checked with manufacturers. No issue with the DT.
Yes BronYAur, only volume related

 

[quark]

Ok. BronYrAur's suggestion seems to be less complicated. Connect the supply to tanks from bottom (in parallel) and do reversed return arrangement from top of the tanks and connect the header to your pump. Do proper venting arrangement and your are done.