Chilled Water Constant Flow Reverse-Return Bypass

[Btutiful]

Hey all,

I'm working on my first Cooling Tower Chilled Water system, and I was hoping to get some input on some solutions I've come up with for a reverse-return bypass. (Everything is pretty much self-taught since undergrad days and I don't have access to a mentor to get their opinion.)

The customer wants a constant flow system that'll feed a bunch of chillers they've got around the lab. Simple enough - I just put in the typical auto balancing valves at each drop and make the cooling towers closed loop with VFD fans to maintain temperature difference (also wanted by the customer). The catch comes when they say they want to add some chillers in the future. Luckily, I have information on these future chillers, and I've been able to size the pump, towers, etc. to account for these. My question comes when managing the variable amount of bypass I will need during the time they don't have these new chillers installed.

Can I simply put in an "Automatic Recirculation Valve" downstream of the cooling towers and have it's bypass go to the inlet of the pump, or do these valves not modulate but rather open and close? Or maybe put it at the end of my supply with the bypass going into the reverse-return? I've also seen "Flow Bypass Valves", but I'm not sure if these are just set at a single pressure difference and don't modulate.

Is there a different way you guys would tackle the bypass? The customer seems to want the system how I've described, or I'd just do a variable pump system.

Thanks for the help in advance.

 

[goutam_freelance]

Can you illustrate with a simplified flow diagram?

 

[LittleInch]

Why do you need a bypass?

This is only if your future flow is x 2 or x 3 your current flow?

So you're running an oversized pump, but pumps have quite a large acceptable flow range of probably 30-120% of your "rated" flow.

If all your cooling water outlets are flow controlled then you will only pump what is needed. Anything else is basically waste of energy pumping water around you don't need.

Needs a bit more data on pump capacity, current flow and how this is laid out.

 

[Btutiful]

Thanks for taking a look at this guys!

Attached is a rough draft of the P&ID that shows existing and future loads. Any criticism for this would be appreciated as well. This is a critical system for the customer, so it is setup for duty/standby as requested by the customer. I don't have the bypass roughed in yet.

To accommodate the existing load, 34.9 GPM is needed. The future load requires an additional 16.9 GPM (about 40% of the final GPM requirement of 41.8 GPM). The only flow control I have so far would be the valves at the pumps and the balancing valves at each drop. I figured that with this much flow variation, I would want some sort of bypass, but feel free to let me know if I'm wrong. It could be I didn't have a problem to begin with.

 

[LittleInch]

So some of the symbols I don't understand, but you say you have flow balancing / control at each location yes?

Your pump is good for 51.8 GPM, [was that a typo above to say 41.8?] presumably somewhere close to BEP ( A pump curve would really help here)
At 34.9 GPM, it is therefore running at ~ 65% of max duty.

I don't see a problem here that needs any fixing. Most centrifugal pumps will only start complaining when you get to about 30-35% of rated flow.

OK your pump might not be as efficient as it will be at 51.8GPM, but absorbed power will be lower as you're only pumping 35 GPM. This is almost certainly better than pumping all that water (52GPM) then just wasting the pressure by dropping the pressure back to your inlet pressure.

Either that or add a VFD and control on pressure at the end of the supply loop to maintain a min pressure for the unit furthest away. Note that VFDs use power so you might not save any money in this initial operating period.

Does that help?

But post the pump curve if you have it or advise make and model no / size.

 

[Btutiful]

Yes, there will be auto balancing valves on the return side of each drop.

Yes, there was a typo, but it was for the 34.9 GPM. It should be 24.9 GPM initial load and 16.9 GPM future load, totaling to 41.8 GPM.

And yes, that helps a lot. Thanks! Attached is the curve of the pump I had planned to use. It's a Taco 1919 in-line. It looks like I might be in the ~36% range of the pump efficiency before future loads are added, but it could still do it. Right?

 

[Rputvin]

Your towers don't look setup correctly to me. It might just be the schematic representation missing some detail but this appears to be setup to flow (treated) domestic water into the basin, and then pump the basin overflow and drain connections to facility drains.

The basins will hold water that will become concentrated over time. I would expect the basins to have the water treatment and monitoring instead of the supply line, and that control would trigger blowdown and run your drain pump as needed. Multiple ways to accomplish this task though.

You have a ton of strainers shown. If this system ever has issues with flow/pressure troubleshooting it will be a nightmare. I'd suggest going to a single device that can be maintained, stick a pressure alarm on it, and let it tell you when it needs to be serviced.

If you use a VFD for your main pumps - run a differential pressure between the suction and discharge headers. Your static or zero-point pressure is variable in a charged/hydronic loop. If you just use a pressure sensor at the discharge for control reference you'll have variable pump output based on that varying suction pressure.

 

[Btutiful]

That's a good point on the towers, thanks! The customer has water treatment staff that is determining their water condition and how they want to treat it, so I don't have anything solid down yet for that part of the design. I will definitely present that method to them. They have their preferred ways to handle controls as well, so I will need to coordinate that with them. Good information though!

The pump on the blowdown is just to give some extra feet to the line so that it can reach their drain.

I didn't even think about the troubleshooting part of things. Having strainers at each drop was another one of the customer's design parameters, but I'll see if I can convince them otherwise - I think they'll agree if I bring it up.

That's good to know how to control a pump with a VFD. Quick question on that. Would I need to read the differential pressure between the upstream side of the expansion tank and downstream of the pump, or does it matter on which side of the expansion tank you read from? I would imagine that downstream of the expansion tank wouldn't be as reliable because the expansion tank would keep the local pressure constant. But maybe that's a good thing?

 

[goutam_freelance]

Best plan of action is to do a techno-economic analysis considering the time when the additional flows will kick in. The options to be evaluated are as follows:

Option-1. Consider the additional cost of running the present pump(34.9 GPM) with higher power consumption(away from BEP at 51.8 GPM), provided the pump motor can take up the additional load.

Option-2. If a bypass is provided(with 51.8 GPM pump at present), consider the cost of loss of energy due to the operation of the bypass.

Option-3. Consider a VFD with additional cost and additional power consumption for the VFD with 51.8 GPM pump

Option-4. Consider the additional cost for 4x18 GPM pumps(3W+1S). 3 pumps(2W+1S) can be installed now and 1W installed in the future. Standby pumps can be 2 if required.

BTW, if the chiller loads vary significantly, you may need a minimum flow recirculation for the pumps.

 

[SAK9]

Cooling tower systems are generally designed as constant flow systems as tower characteristics change drastically when the flow drops below a certain limit. As you have a large number of chillers/heat exchangers receiving water from the system it is quite possible that some of the chillers may be off at some point of time.In order to account for the rise in pressure and variation in flow under such situations a modulating bypass valve between the supply and return lines is the best solution to hold the flow constant in the system.This valve needs to be carefully selected as it needs to operate over a wide variation of pressure.

 

[LittleInch]

Re reading this I'm not exactly sure what a "closed loop" cooling tower is?
A closed loop with a VFD fan sounds more like a fin fan heat exchanger to me....

Is this a set of tubes inside a tower and then you run water over it?

The only pumps I see seem to be circulating the cooling water which then goes to all those individual chiller packages??

Why not have one central chiller package and then distribute chilled water? This option seems very inefficient to me.

I've now seen the pump curve and the pump looks a bit oversized, but also at this level of power I wouldn't be worried about the energy efficiency here.

 

[Btutiful]

Thanks goutam, I will look into the cost differences.

That's a valid point SAK9. One thing I should clear up, though, is that flow through the chillers is planned to be constant. The only variability will be when they add the additional chillers to the blank drops that will be provided as part of the design. The P&ID shows those chillers already added, but after construction is complete, some of those drops will be capped and awaiting those future chillers.

Closed loop may have been an incorrect term to use. Yes, the cooling tower has a coil inside that the water is ran over. The coil is what the glycol/water runs through, allowing us to run this during the winter (-20°F) - with the basin drained and no water running over the coils anymore, of course.

A central chiller package was actually considered during the customer's conceptual design of this project, but the approach described is what they settled on. To be honest, I'm not sure why.

I was starting to second-guess my selection of pump as well. I think I have to agree that with the constraints given I might have to accept the temporary inefficiencies of the pump. I will see if I can find a pump slightly more optimized, though.

Thanks a million for everyone's responses!

 

[LittleInch]

Well it's all a bit odd as you now have two sets of pumps and water to deal with. One in the closed loop which is basically a water / water HX or in the winter a water/ air HX and the other in the water loop inside the cooling tower.

What SAK 9 was noting was that most cooling tower systems run an open loop picking up water in the basin, running it round your chillers taking heat away then dumping that warmer water in through the top of the cooling tower in a continuous cycle. That cooling water cycle tends to have a much more limited turndown.

But whatever, that's the system you have and my comments don't change - basically I can't see a reason to do anything as your current pump might be bit bigger than you need, but it will cope happily with either your initial flowrate or the larger flowrate with no other bypasses etc needed.

 

[Btutiful]

Oh, I see. My misunderstanding. There will be two sets of pumps as you say. The pumps on the P&ID are for the closed loop and the drain. I will make this more clear on the final version of the P&ID.

Thanks LittleInch. I appreciate the clarification and your input.

 

[SAK9]

If all chillers are going to be operating all the time a bypass valve is not required. You can select a pump for the flow anticipated for future chillers as well and run it at reduced speed for current flow. Please make sure that the pump selection will meet flow and head requirements under both the conditions.

 

[LittleInch]

Not necessarily operating, but as long as the water keeps flowing through all the chillers then I agree with Sak 9. It just needs to operate at lower flow, not necessarily reduced speed. It's only a 1kW pump....

 

[Btutiful]

I will verify flow and head requirements at both states of the system, thanks SAK9!

The system will continue to provide flow to each drop regardless of the state of the chillers, so it sounds like I should be good without a bypass.

Thanks all!