Cooling Water Supply Pump Challenge! 1

[francis_mechanical]

I was sizing a pump that will supply cooling water to 26 diesel engines.

The capacity of the pump is near 2600 GPM

(Im new in sizing pumps)

I was searching the internet to get a physical feel on how big this pump is. And 2600 GPM pumps are quite big (for me)

The estimated size of the discharge is 12in! Maybe this figures are realistic but I don't want pump discharge this big.

I'm planning to install 2 set of pumps instead of one to reduce the size of the pump. so 1 set of pumps for 13 engines, and other set of pumps to the remaining 13 engines.

Is this a good design?

 

[LittleInch]

12" discharge sounds about right for a pump that big.

A pump half the size will still be say an 8" nozzle and probably about 2/3 the size as much is taken up with bearings and structural elements.

Bigger pumps are inherently more efficient than smaller units, but a lot depends on how you plan to use the system. Is it 26 engines or no engines or is there some flexibility required? If so how much? If it's often only 13 engines or less then 2 smaller pumps would be more efficient.

How often does it run? two pumps cost more than one once you add in extra pipes, valves, cables, foundations etc, so if you're only running 10% of the time then OPEX costs will not be as big an issue as CAPEX.

There's a lot to consider to get the optimum for your system.

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

 

[Artisi]

100 gpm per engine, what size engines are they?

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[francis_mechanical]

@Artisi
I have 26 engines, all with 550 KW rated power each.
My calculated cooling water requirements is 88 GPM(US)/Engine
The Power Plant will use this engines have a estimated capacity of 9 to 10 MW.

@LittleInch,
Yes, you are right, I tried the 13 engines configuration, and my pipe size was just around 10in, so its not really a big difference. I was hoping around 6. I guess this is because of my experience, I was always involved in designing systems with 6in pumps, that is why I was shocked when I arrive on a 12in pump.
I tried to narrow down the division to 3 circuits, that's 9 for 1st cooling circuit, 9 for the other and another 8. But still I was stock at 8in Pump!

Also, what default schedule of pipe should I use? I was using the Sch. 40 pipe. Maybe I could check this later if I already have a more precise estimation of the required head.

I'm sizing the pipe system using this method.

1. Determine 1st the capacity of cooling water on engines.
2. Set and allowable water velocity of 2 m/s.
3. Calculate the estimated area, and select pipe size with the default Schedule of Sch. 40.

Please Criticize if I'm off the track.

@littleinch,
I was thinking if you could spend some emails with me, I see you are very active here. Can I directly connect with you? Well, if your not managing a huge audience, :-D

 

[georgeverghese]

Arrange the supply mains to allow for about equal flow to each machine. Watch out for poor quality cooling water, high TDS and high dissolved O2 that could ruin the cooling exchangers (radiators?) performance and increased pressure drop over time. If the radiators are finned, choose marine grade aluminum if this is a coastal facility.

 

[LittleInch]

Ok,

Thanks for the compliment - always goes down well

You need to be careful here. Working out a header and return system with a number of offtakes is not so easy to get it right. You will probably need to install a flow control on each engine offtake to make sure you don't starve one engine and have a different one with twice the flow.

Depending on your layout you're probably best off using what I think they call an inverse return, i.e. try to make the total length of pipe from pump to return point as eequal as you can for each engine.

At those size I wouldn't go less than sch 40 - sch 20 is possible, but gets rather thin and isn't really man enough for the job, especially as you'll have some thermal stress if you're operating at 90-100C for the cooling water return.

2m/sec is about the highest you want to go for such a system, but you could go to 2.5 if you're at the limit.

sounds like you're building a small power plant.

Prefer to operate like this - anything else needs to be paid for.....

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

 

[Artisi]

I would suggest thinking along the lines of 2 pump systems -if using 1 pump system any major pump or system problem puts all engines out of service unless you are considering a full backup system.
If it were me I would initially look at 3 × 50% units, this gives good back up and easy maintenance .

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[georgeverghese]

No details here - check if at least one of these pumps needs to be connected to a critical mains power bus with control power from UPS.

 

[MikeHalloran]

Where does the cool-ish water come from?
Where does the heated water go?
Why introduce a single point of failure, or two, or three, into a large system like that?
I.e., why not one pump per engine, since they're typically equipped to drive a properly sized cooling water pump or two per engine anyway?



Mike Halloran
Pembroke Pines, FL, USA

 

[Artisi]

Mike: Good question but as usual the OP is always short of real meaningful detail - my understanding is the water supply is to each engine already equipped with their own engine driven waterpumps and being supplied to negate the need for radiator cooling.



It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[francis_mechanical]

@LittleInch,

I never though about the unequal flow. I was just assuming as long as the areas are equal then fluid will flow as what Q=AV is talking about. (chuckle).

I have an idea on how to solve the possible unequal flow of water (Figure 1 of the attachment). Although I cant (for now) back up this with calculations, I think it would somehow make the flow even on all of the engines.

As for the inverse return, I saw this video on youtube, (screenshot on Figure 2 of the attachment). Is this what you are talking about?

I tried to analyze, I think it was trying to normalize the flow and pressures on the headers that are coming from and to the branches.

Please consider my grammar, I'm no native English speaker.

 

[francis_mechanical]

@Artisi

Yes, I also have the same idea with you, but my concern with the 2 supply pump system is economy. I'm also aware that using 1 system pump would put the entire plant at risk of failure if major damage of the supply pump is to occur.

But for now, I'm gonna go with 1 supply pump,. But still prepare a calculation for the 2 system supply pump and present it to my boss.

 

[francis_mechanical]

@MikeHalloran,

I'd like to post the diagram here but I don't want the topic to be diverted to other issues that people might see on the diagram. For now, is for the supply system of the cooling water, its sizing and arrangement.

"Why introduce a single point of failure, or two, or three, into a large system like that?
I.e., why not one pump per engine, since they're typically equipped to drive a properly sized cooling water pump or two per engine anyway?"

Yah I agree with the single point of failure. I can't give 1 pump per engine (maybe 1 supply for 5-6 engines - possible). maybe as we progress to this thread, I could finally decide on the configuration of the cooling system. But for now, I'll go with 1 supply pump.

Thanks for the response.

 

[Artisi]

For the total cost of a 10mw power station,the costs associated with 2 or 3 pump systems would be insignificant, not withstanding the costs and drama of a complete outage because 1 pump failed.
But guess it's not your decision at the end of the day, however I would cover my arse with a properly prepared engineered study of the alternate suggestions.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[curtis2004]

francis_mechanical,

You are in right path.

1. You have to find out what is cooling water requirements for each engine. This give you water flow, inlet and outlet temperature ranges. By multiplying by 26 you have got total cooling water flow, which is you calculated to be 2,600 gpm, I assume. We probably will have constant flow closed circuit cooling water system.

2. From Crane reference table I can find out that for 2500 gpm: 7.17 fps for 12" X SCH40, and 5.93 fps for 14" X SCH40. I think, for this type of application 12" pipe is minimum you can use. Please note that 12" pipe STD (0.375") and SCH40 (0.406") doesn't much, so be careful! Corrosion Allowance for this type of application will be around 1/16"-1/8".

3. So you probably will have 5 psi pressure drop across each engine (please check this number from engine manufacturer), so you will have about 10-15 psi differential pressure between supply and return headers. You will have another 10 psi allowance for your Air Cooler (I think you can use Air Cooler to cool water), and 10 psi pressure drop through all piping. So we will need 35 psi total DP for your pump. Probably three (3) pumps at 50% flow is a good idea as mentioned above (two pumps running, one stand-by), so your pump shall be able to deliver 1300 gpm @ 35 psi. Your system will have 50 psig design @ 100C. Piping will be uninsulated (except areas where human can hurt themselves by touching high temp surface).

4. You will have supply and return header 12"xSTD and 26 branches going to and coming from engines. For 88 gpm flow for each engine 2" pipe may be sufficient (Crane says 8.6 fps for 90 gpm). I think you can use STD wall pipe, but many specs use "extra strong" XS wall for small bore piping 2" and below for corrosion allowance.

5. I recommend to use manual butterfly valves to adjust flow on each branch going to each engine. These come flangeless body and installed between two flanges. During start-up these will be adjusted to have 88 gpm through them and locked. It is simple and not expensive solution for constant flow application. Of course you can install control valves with flow meters on each line and control flow through it (I don't think it will add any value, except cost).

6. You will have in the end supply and return header 3"-4" by-pass line with by-pass valve. This line needed to control differential pressure between supply and return header, and will be helpful during start-up and adjusting flows to each of 26 engines.

7.Now, you can size Air Cooler, based on supply and return temperature ranges, we can determine how big cooler do you need. Use summer hottest air temperature for Air Cooler sizing. 10 psi pressure drop should be maximum, but do not go below 4 psi because you may have problem with flow distribution. I think stainless steel tubes with extruded AL fins are most cost efficient Air Cooler designs. You can choose from number of suppliers who can built and supply complete package with fans.

8. Once you have sizing of Air Cooler, and preliminary layout, you can calculate internal volume of your system. You need this to calculate an expansion tank, because your system is closed system, and you need a space for expanding volume of water. So you will need an expansion tank, which should have internal volume of 200% - 300% of volume calculated expansion volume of water. You expansion tank will be installed close to suction side of the pumps.

9. If you have gone this far, now you can draw P&ID of the system, have Line List, Equipment List, Valve List, PRV List (you will need at least two, one for Air Cooler, one for Expansion Tank), Instrument List (Pressure and Temperature gauges), and start cost estimate of the system.

10. Anything else? You can try to get built a Pump Skid and install all three pumps in a steel skid with all piping, instrumentation, valves, expansion joints. This will reduce site costs, because all of this can be built at fabrication shop for a fraction of the cost of the on site costs. Of course you have to have a space to bring in this skid into you location. You have to have close attention to structural and foundation design for this pump skid because of vibration issues you may have. You may consider using glycol, if you need freeze protection in case you system will be down (emergency, maintenance, whatever reason) in winter. Please keep in mind there is slight difference between properties of water and glycol-water solution.

Regards,
Curtis

 

[francis_mechanical]

@Artisi
"For the total cost of a 10mw power station,the costs associated with 2 or 3 pump systems would be insignificant, not withstanding the costs and drama of a complete outage because 1 pump failed. "

This is eye opener, thanks. I think I have more study to do.

I'm always thinking about this phrase "Point of Failure" (chuckle) I guess its already in my mind I just don't have the right words to describe it.

@curtis2004

I can't help but thank you in advance, I'll be studying your book long comment. This will give me a lot of technical ideas on my design! Glad to connect with you here.

 

[MikeHalloran]

What drives these monster cooling pumps, whether they number 1,2 or 3?
The plant's output power? Not available in a Black Start condition.
How many engines have to run without cooling water flow before you have enough power available to start the cooling water pump to cool them? Will you provide three independent power supplies to the cooling water pumps skid so there won't be a single point of failure on the electrical supply side?

Butterfly valves and ball valves are highly nonlinear WRT Cv vs. handle position.
How will you adjust 26 such to a particular GPM without 26 flowmeters?
You will need 26 people doing the adjusting all at once, or you will go crazy trying to do it with a small staff because the valves will all interact.
What happens when some damn MBA comes along and decides the installation would 'look better' if all 26 valve handles were perfectly aligned and parallel?
Flow or temperature controlled valves might work better.

So far, OP, it sounds like you've got a huge mess on your hands and/or are in way over your head, no offense intended.



Mike Halloran
Pembroke Pines, FL, USA

 

[LittleInch]

Franscis,

After Curtis reply I don't think there's muchmore to say.

A lot depend son how you're physical arraging these 26 engines.

In a single row?
blocks of 4 or 6 or 8?

Yes I meant option 2.

If you're going to go for option 1, then you really need to make the two header pipes over sized by 3 or 4 pipe sizes, i.e. stick something like a 24" pipe in. The extra cost is nothing, but the operability and simplification of design is much better than trying to optimize something which costs 0.05% of your installation, but has a huge impact on its operability. Availability of cooling water is much better than starving an engine and have it over heat.

This way there is virtually no pressure drop difference in the header from the engine closest to the pump connection and the engine furthest away.

Then all you need to do is adjust the flow on each engine one by one as the header and return header pressure won't really change. either rthat or each engine will have a thermostatic valve which opens at the right temperature once the engine has warmed up and cuts in if too much cooling is being done. Pretty simple standard valves, but understand how your system is going to work.

One thing with your pump. circulation systems often need low head, high flow pumps. These can be difficult to get and you may need an axial flow type unit.

Also don't forget your pressurization tank / accumulator to keep the pressure up and allow compensate for the changes in temperature from start-up to operating.

Good luck and let us know what you decide to do or come back and check if you're on the right track.


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

 

[francis_mechanical]

@MikeHalloran

yah. But I guess I made up my mind and finish this challenge. I just think that it would be very rewarding if I got this right. Thanks for the advice!



@LittleInch

"Also don't forget your pressurization tank / accumulator to keep the pressure up and allow compensate for the changes in temperature from start-up to operating."
Yes, I'm referring this as the expansion tank.

One last question before I focus, can you a give me an idea on how often and how fast does the fluctuation of the cooling water temperature of an engine?

Like, during start-up, when the water is at low temp say 27°C, how long (or how many loops) does it take for the water to reach around 85°C.

 

[Artisi]

There must be other large diesel engine installations around the world, have you bothered checking with the engine company for advice, or are you trying to re-invent the wheel?

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)