EFFECT OF FLOWRATE AT PUMP "SUCTION" INLET ON NPSHa

[davincigee]

Hi guys,
I've been pondering over this for a while now. We have 4 pumps connected in parallel. Each pump is connected to a 12" suction manifold by 4" pipelines. 12" pipelines from the tanks are connected to the 12" manifold to supply gasoline to the pumps. Each pump has an NPSHr of 4m yet even at a tank head of 15m (that is the static head when measured from the centre-line of the pump suction inlet to the surface of liquid in the tank) we get these grinding noises or cavitation occurring in our pumps. This is surprising because at a static head of 15m in the tank, it should be enough to overcome all the suction losses and NPSHr and provide sufficient flow to the pumps. Is this a case of some of the pumps getting starved as a result of low flow in their suction pipelines? And how do u determine sufficient flowrate aside the NPSHa calculations to ensure that your pumps are getting enough flow?

Thanks

 

[LittleInch]

Try having a look at the FAQ on this forum on npsh. You don't say what the flow rate or length of the lines are so we can't see how this could occur. Be aware that gasoline has a notorious high vapour pressure and also pump npshr is not the same as the cavitation limit which can be several metres higher. A sketch of your system would also help with lengths, no of bends etc.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[BigInch]

I don't know flowrate or lengths, but something rings a little "not quite right" with those relatively small 4" diameter lines coming off a 12" header. I think there may be suction piping configuration issues.

Independent events are seldomly independent.

 

[davincigee]

Dear Big and LittleInch,
I have attached a schematic drawing of the suction system. I have also included the lengths of the various pipelines.
Here is the question again:
We have 4 pumps connected in parallel. Each pump is connected to a 12" suction manifold by 4" pipelines. 12" pipelines from the tanks are connected to the 12" manifold to supply gasoline to the pumps. Each pump has an NPSHr of 4m yet even at a tank head of 15m (that is the static head when measured from the centre-line of the pump suction inlet to the surface of liquid in the tank) we get these grinding noises or cavitation occurring in our pumps. This is surprising because at a static head of 15m in the tank, it should be enough to overcome all the suction losses and NPSHr and provide sufficient flow to the pumps. Is this a case of some of the pumps getting starved as a result of low flow in their suction pipelines? And how do u determine sufficient flowrate aside the NPSHa calculations to ensure that your pumps are getting enough flow?

 

[Artisi]

you will need to be more specific on the pipework / valve, strainer, concentric reducer locations in the 134cm distance from the manifold also,where does the 12" pipe reduce to 4" - It could well be you have very poor flow conditions into the impeller resulting from poorly positioned components or pipe reduction.

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.)

 

[LittleInch]

As you haven't given us the flow, I asusme that is what you're looking for. with these types of questions, a few metres here and there make quite a lot of difference, but from my working out and guessing a bit your gasoline vapour pressure (what temp is is stored at?), I think you have about 17m head to drive the fluid to your inlet flange before you hit cavitation. Your equivalent length with the bends, tees etc is about 170m. The key unknown is those filters. what mesh size / what is the state of them?

If you assume they take 6 or 7 m worth then that only leaves about 10 to 11 m for the flow friction. That equates to about 300 to 360 m3/hr total.

However a few degrees C up or a few more metres loss through the strainer and it could be quite a lot less.

Basically start with atmospheric pressure converted to m head (about 13.5), add your positive head (19), minus your true vapour pressure (10m?) mius your cavitation limit (say 6m), subtract any other known losses at your flow rate (filter 6 m?0 and what is left has to acocunt for friction flow losses. Plug that into a flow calulation program or web page and put in viscosity etc and flow rate arrives based on an equvalent length for your system including all the bends and tees etc (170m for the 12").

You soon run out of metres.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[tr1ntx]

Are those gasoline dispensing pumps, like for fueling vehicles?

I was expecting transfer pumps and for the diagram to show something of downstream piping. What is the reducer dia?

 

[Artisi]

One thing not raised or discussed is there are the pumps operating on their performance curve?

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.)

 

[Artisi]

Correction to earlier post -- Read --- where are the pumps ---

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.)

 

[chicopee]

What I don't see in the diagram is the return vapor line if these pumps are for tanker truck refilling.

 

[Krausen]

My first question with this layout is why would you neck down to 4" to feed these pumps? Even if flow rates are minimal, it'd be to your benefit to keep the suction lines at 12" minimum. That is just good standard practice when feeding pumps. Also, why would you install suction-side strainers when pumping product out of tankage? As mentioned above, the strainers are the x-factor that may be contributing to a significant pressure drop.

 

[davincigee]

Hello Guys,
Sorry for the long delay. I lost internet connection. Thanks for the inputs. I really appreciate it.

Mr. tr1ntx yes those are gasoline fuelling pumps for loading Bulk Receipt Vehicles (tankers). The reducer diameter is 4".

Mr Artisi, the pumps have a head of 62m at a flowrate of 142m3/hr. Unfortunately, since the plant started running 2 years ago there haven't been any flowmeter available on each pump. It thus makes reading the pumps performance virtually impossible.

Chicopee, we don't have a vapour return line. All our tanks are not pressurised but have contact with the atmosphere. But we do have a 1" pressure relief valve that relieves pressures as high as 75psi back to the suction pipeline.

Krausen, you most definitely need to keep your pump clean from debris or particulate matter within the tank and pipelines from entering into your pump suction else you will destroy your seals in no time. Perhaps the most appropriate option is to keep the strainer at a minimum distance (about 10 pipe diameters I think)from the pump.

LittleInch, thanks for the info. The gasoline is stored at 28 degrees celcius. For the fittings, you have a 4" Y-strainer (don't exactly know the exact size of the mesh. Lets say 12-15 microns.), there is a 4" gate valve always at fully opened state when pumps are operational and then an eccentric reducer 3" that is attached to the inlet of the pump. All these fittings are present in each of the 4" pipeline that is attached to the 12" header and serve as the suction feed of the pumps.

All in all, I want a rough estimation of what the flowrate will be like at the suction end of the pump. Is this flowrate sufficient for the minimum velocity of 2.5m/s stipulated in suction piping.

 

[LittleInch]

Ok, a little bit more info required -

1) When do you hit problems? with 1,2,3 or only when you have all 4 running
2) Do you have any differential guages across your filters?
3)12 - 15 microns (mm x 10-3) is a fine filter. This will have a significant pressure drop clean, never mind dirty. Are you sure? Why are they there as I think they are the root coause of your issues and the only thing you can easily do to make things better is remove them.

As noted above, my current estialte is that beyond 300 to 360 m3/hr total (3 or more pumps running), you're going to run out of metres head and start cavitating when you're at your stated tank level.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[davincigee]

Dear LittleInch,
Follow-up to your enquiries;
1. Usually during operation, we run all of them. They are all hooked to the VFDs hence the lead pump starts and then the others follow suit as lag. On a very busy day, we can have as many as all the 4 pumps running. Usually this problem is noticeable when 2 or more pumps are running at their full capacity (142m3/hr @ 62m head)
2. We don't have any differential gauge across the Y-strainer. But it is something we are seriously considering.
3. Oops, I apologise. The mesh size is 0.6mm.

The strainers are there because sometimes the gasoline from the tanks bear in them impurities. If we remove them we face the risk of wearing out our mechanical seals in no time.

 

[LittleInch]

That sounds about right then according to my rough calculation above. At anything past 300m3/hr, the calculations show that with a fairly modest 6m (about 0.4 bar) accross the filters, at 19m tank height you run out of NPSH and start cavitating.

Things you can think about to make it better - more or bigger pipes from the tanks - double up at the very least (2 x 12") or replace with a 20" or 24". 100 to 110m is quite a long way when only the atmospheric pressure and a few metres head is pushing it along at that flow rate.

Remove the filter mesh

Run less pumps

Place pumps in a pit 5 m deep

Place pumps much closer to the tanks (say within 25m)

Do nothing and replace your pumps on a regualr basis / not achieve your required flow rate

By the way- did you mean maximum velocity in the suction pipework or minimum in your post two above? If maximum that is only a guide and you need to do your own calulations for your particular circumstnaces. If minimum, what on earth are you talking about? There is no minimum for this type of pipework.



My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[davincigee]

Dear Littleinch,
Yeah very outrageous to request for minimum. (What the ****!). Anyway, thanks for the info. Was really helpful. By the way, can you please furnish me with the formula for calculating flowrate from the tank to the pump inlet. Not the Q=area x velocity. I don't even know the velocity of the fluid. Thanks

 

[LittleInch]

I just googled liquid flow rate calculation and input the conditions. engineering toolbox has some good ones to use for this sort of simple system.

As said, your key issue remains the filters. Putting these upstream of the pumps is a wholly bad idea when you have no real inlet pressure.

100+m from the tank is actually quite a long way.

Make your inlet pipework much bigger and your issues go away. Its just too small at the moment for your flow. On this sort of pipework aim for 1 to 1.5m/sec max.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way