Pumps and Parallel Connections

From the description what is given, the presuures in the two lines b & C are different. The end use also may be different. If the discharge lines are connecetd then the pressure in line B&C will be same. If it does not affect the end application it may be OK. Otherwise not OK.

Although it is an interview question, the basic (from an engineering point is)

If you have 6 pumps with same inlet header and same outlet header, why to have 6 ? why not only one of large capacity.

The only possible explanation may be variable load condition thus you will be using the pumps (may be shutting off a few) based on the output load.

This way the overall system efficiency may be maintained in a long run (by not operating all the pumps at a inefficient flow but running a few pumps at a better charactersitcs curve point)

This alos will make the whole system availability better. Say a pump or 2 breakdown will not shut off the whole system.

The size of the pump inlet and outlet may matter , but a little, using proper reducer and proper design of the pipeline, these can be improved.

(As an Ex-Electrical Engg- I have a bad habit of equating the lines to Electrical Lines)

Say there are 6 Gens connected to grid - 3x52 MW and 2x72MW

They have diff OP Voltage (say 11KV for first and 22 KV for the other set)

The demands are met thriugh synchronisation,

Based on load conditions, some specific gens may be kept on and others off (of courses in Electrical we don't switch off easily since re-synch takes time but assume they are off)

Why we don't put a single 350MW ? answer is clear - A B/D means all 350MW gone, a low load of say 100MW will make the total system go unstable so we land up in combinations.

The outlet area we adjust thruugh different transformers ( For you it is reducers)

But this theory in pumps is a bit different.

-Under some specific applications when the pumps are feeding to different equipments, we use different pumps to feed different lines.(we have some of this types, especially in a few of hydrostatic lines)

Since the pump will generate pressure to overcome restrictions (as per the Ch Curve) which will not be easity duplicated by the header feed. (Unless we have properly and may be much costlier flow controllers or constant flow orifices) But these are another aspect (in some of our hydro stats).

To answer you in brief- Yes it is possible to connect them in parallel

The net final head at the outlet header will take care of the discharge velocity differences

But it all depends on the load characteristics to whom they are feeding - single load, or multiple load and their individual characteristics (the basic pump design and selection)

No problem for pumps (may be efficiency only) but problems for load is more expected.

But since it is an interview (and no questions were asked abou load) so I don't think we can discuss much on this, just a food for thought for you and others.

Hopefully I have not confused more

Thanks a lot

the reason that have 6 pumps why just only one big pump, they usually used 3 pumps at same time but in case of emergency they use all of them (or in case of failur).

as I told you that I am warried about the velocities if we connect 6 pumps with same discharge line.

for the end application there is no problem for the pressure outlet, and I suggest to them to change the outlet of the 2 pumps (which have total capacity of 144 m3/hr) by reducer in both outlet to the header just to keep the same velocit for the 4 pumps.

by the way, I have asked my supervisor for this and he sent me to see this arrangment in the plant and I found what I told you before and now I am trying to do some changes for the existing system.

The pump(s) when are running in parallel - has to ensure the load sharing

The velocity decrease I feel will not matter - since the centrifugal pump is affected by the increase in the back pressure and if you are going to use a reducer (in reverse way must be an increaser from 4" to 6") the resultant increase in back pressure may not be significant.

But you must check up the operating point of the system when you are going to run the pumps in parallel since the OC curves do not only matter about the capacity of pump but also on the condition of them, And improper load sharing may have effect on both the pumps.

So it is the whole system in the totallity that has to be taken into account, just replacing a pump or connecting the pumps in parallel will not do here.

Check on the load,- what is the demand (characteristics curves - flow rate vs head) and then you have to match it with the pump curves- individual and as group.

I can give reference of one of the papers on parallel operation check up.

http://www.lawrencepumps.com/newsletter/news_v02_i12_dec05.html

Usually at our works we rarely use pumps in parallel - in series yes - but completely parallel feeding to same outlet header - very rare.

By the way should you so much bother about velocity ? the load requires only the head and flow - The pump will develope the Q unless you build up the back pressure (usual problem with centrifugal pumps) and since you are not building it up - if the header is able to accommodate it then the reduction in V will not matter isn't it ?

the fundamental is to know pump's charasteristic curves. FLow vs Pressure & Kwh.

In partallel the head is important and velocities are immaterial. The head in the header pipe is constant for all the pumps in paralle.

The flow through these pumps shall be as per the charasteristic curve of the pumps at the header pressure. Just donot bother about the velocities which are a function of pipe diameter.

the actual flow shall be as per the added flow of all individual pumps at that pressure head. Of course keep NPSH( Net positive suction head) in mind.

So its desirable to put same charasteristic curve pumps in parallel but in any case, one can find the exactly what will be the flow through a given system curve.

I am surprised that none of the earlier answers have not touched the fundamenrtal aspect of pumps charasteristic curve & its behaviour.

we conduct detailed energy audits in organisations & many MNCs and are shocked to see absence of such basic awareness amongst engineers.

hope this clarifies the issue in question.

do not hesitate to contact on som_derashri@rediffmail.com

all the best

academy for conservation of energy

820 siddarth complex R C Dutt Road, baroda, Gujarat, Baroda 390007, tel:0265 2325024, 2325034

som derashri

uhhhhhhhhh - I think all of the replies talked about the "charasteristic curve of the pumps", just not in the exact phrase you have put.

And velocity is very relevant (essentially the same as flowrate for any given pipe diameter) to determine the frictional losses.

In addition -

This statement:

the actual flow shall be as per the added flow of all individual pumps at that pressure head.

is not accurate. You must find where the pipings system flow curve intersects the combined pumps curves. See the site from the first reply:

http://www.mcnallyinstitute.com/15-html/15-01.htm

Here is the way it works. Each pump has it's own characteristic curve that indicates what flow the pump will produce and at what head. If all 6 pumps are operating, being fed from the same suction pipe and discharging into the same discharge line, they each will still be bound by the characteristics of their own particular curve. The velocity in the discharge pipe will be whatever the pipe will allow at the flow passing through it. If, for instance, the pipe is a 6" schedule 40 PVC pipe and the flow being generated by the six pumps is, say 400 gallons per minute, then the velocity will be 4.30 feet per second - period.

Now, the head (pressure) required to produce the 400 GPM will be the product of the friction loss in the pipe plus the static head (vertical difference) of the piping system and the velocity head and possibly the viscosity and specific gravity if the fluid is not water. Each pump in the system will deliver a proportionate part of the total flow that will be determined by the curve again. However, each pump will be producing the same pressure as all the other pumps.

Let's assume in our demonstration that the flow from the four large pumps (as determined by looking at the curve at the point of the head being observed) is 75 GPM each. That means that those four pumps will be delivering 300 gpm of the available 400 GPM. The two smaller pumps would then be delivering 100 GPM between the two of them, and, if they are identical pumps, it would be 50 GPM each.

Further, let's assume the two smaller pumps are turned off. The piping system would then see a different set of circumstances. Depending on the characteristic curve of the remaining larger pumps, it is entirely possible the four larger pumps may still deliver 400 GPM since they would now be pumping further out to the right on their respective curves, and the result would be more flow. However, the pipe system will still only allow 400 GPM at the head first observed, so the four remaining pumps would now be delivering 100 GPM each but still at the same head.

What this demonstrates is that the economics of operating the six pumps together may be poor because the addition of the two smaller pumps may add little if any additional flow to the system, and they would simply run without benefit. Therefore, as has been stated before, it is important to analyze the system and the pump characteristics to determine the most efficient use of the six pumps.

The answer to the question of whether the six can be piped in the manner described, well of course they can. But will there be a benefit to operating the system in that manner? That remains to be determined.

Thank you All, I will try to reconsider the arrangment and comeback to you soon.

regards