Parallel Operation of Pump

Consult the respective pump curves for their performance characteristics at varying flows and pressures.

The best solution will depend on the water requirement, system curve etc.

If you need the high flow most of the time it would be better to select the optimal efficiency at that duty point

For the lower flow the pressure will be lower and the duty point will move to the right and most probably at a slightly lower efficiency.

If the low flow is the norm your design strategy should be the other way round.

Why not have two different centrifugal pumps each designed to do the duty most effective.

One can even have different pumps operating in parallel. the duty point is only determined by the back pressure at the pump and not the pump on the other side.

Tornado and Henrik are right again. The best way to verify what is going on is to measure the current of the motors (assuming electric drive).

The basics is if one pump can provide presure and flow on its own use it and just swap the pumps often enough to prevent problems with the "spare" pump. The reason why I am saying this is if you use 2 pumps to do a job achievable by one you will be paying for 2 lots of fixed losses (I squared R losses, Iron losses, Windage losses, Bearing losses, and drag in the pump).

If you are closing the discharge valve as your post seems to suggest when 2 pumps are running, then you are simply wasting money.

Duty and standby / backup (choose your own words) has been the standard in pump control for so long and it has proven pretty good. The modern version of Duty / Standby includes a V/S drive on both so no valve throttling occurs.

Way back when we had no better ways, discharge throttling was used to balance multiple pumps, (chiller for instance) this was regarded as OK back then as the current drawn drops somewhat as flow reduces in centrifugals. The problem is it is not a linear relationship.

Greater flow, yes, greater head no

To respond I need some more information:

1. What is the rated capacity and discharge head (pressure) of the pump?

2. How much flow rate you really require and at what pressure?

3. What type of pump you have, centrifugal or positive displacement?

Assuming inlet conditions are not compromised by running two pumps, they can't be positive displacement. Volume would double, consumed power would double, and head would not be affected.

I regularly specify positive displacement pumps to meter multiple ingredients into in line mixers. When variable speed drives are added the proportions can be varied with a high degree of accuracy and flow is directly related to frequency for the middle 60-80% (dependent on size) of the rated pump curves.

Reply of your question is 1-Rated capacity of single pump is 60cu.m/hr with discharge pressure of 40Kg/sq.cm 2-Flow rate sometimes increase to 67 to 70 cu.m/hr so we have to operate other pump in parallel 3-It is six stage centrifugal pump. I think its sufficient? Moreover, these pumps are equipped with minimum flow devices with by-pass feature.

Why the discharge valve is reducing to 60% when in parallel??

If you have a pressure sensor and you are controlling to maintain a fixed head at discharge, then I can understand the 100% to 60% valve operation.

Now: If the 100% opening is keeping the Head as required, then no need to run both pumps since the flow from One pump is enough to maintain the pressure. If not, then yes the running of both pumps is required.

the adverse effect is only on the power consumption being higher than if you had a single pump capable of providing the required flow at the required head. (Already mentioned by ...).

if the flow requirement is variable, then you also have the problem of modulating the discharge valve to below the 50% if both pumps are running! This means one of the pumps is superfluous at that flow... more power loss %tage...

Here is another option that may work for you. At work, our primary air compressor is a duel compressor design. Two 5 HP compressors mounted on a single tank. There is an alternator that alternates between each compressor. This keeps wear equal on both compressors. But if the pressure in the air system drops too low, the second air compressor will kick in and both compressors run at the same time to keep up with occasional high volume use.

This arrangement of controls might work out for you also. Low power when only low water flow. And higher flow only when needed. Just some food for thought. Good luck.

Thank you for replying with positive and informative answer Holzfeller

Something I wanted to comment on that seems to have bneen overlooked in your question. You stated "adverse affects of operating two similar pumps in parallel regardless of the fact that parallel operation gives higher flow and greater discharge head". However, pumps in a parallel configuration do not give any higher discharge head pressure, just increased flow. To increase discharge pressure you must place the pumps in a series configuration (stacked cans).

With your valves opened to 60% in parallel, you are going to be losing more head to the valves, as you must operate with in a velocity range (typically maximum in the manifold around 5 fps). What I do not understand is why design a discharge manifold to operate with valves open to only 60% of full. Operating in a throttling configuration all the time is problematic for the valves (increases O&M). Typically, in a mainfolded system you want to operate as near BEP as possible at the peak discharge rate required for that pump. So ideally parallel pumps will give a cumulative system curve equalling the sum of the discharge flow rates at any give pressure (thus for 2 equal pumps the system pressure looks the same as the pump curve but at twice the flow rate). (If the pumps have dissimilar discharge pressure capcities, you are only going to operate to the maximum pressure of the lowest capacity pump). However, throttling increases the head the pump must push against, reducing pressure by the loss across that valve and the flow in the system from each pump (wasted energy). It is usually better to stage the pumps to operate in a sequence, such that as one pumps capacity is exceeded you bring the next online. Rather than trying to throttle the flow, I would suggest using some variable frequency drive pumps if a continuous range of discharge is necessary.

The only reason you have pumps in parellel is to have one on standby, one on duty... I have to agree with Holzfella, I too cannot see the logic or reason for two pump running together.. so let not start trying to redesign the wheel and adding VSD's into the equation as they, on thier own give you a whole new set of problems to deal with..

for your info: EVEN with a VSD, you still have to control the flow rate by either a flow control valve or orrifice plate.

You do not need flow control valves on variable frequency drive (VFD) pump systems, unless you use a pump bypass system with a modulated control valve to control flow, which substantially reduces system efficiencies (I almost never see a modern large pump station still designed this way as VFDs are in common usage, except by a few older engineers). All they need are inline check valves, transient control valves and some form of shut off valve for pump isolation. Also if you have a large flow demand regime you will utilize multiple pumps as a VFD only effectively operates over a range of 33% to 100% of pump capacity, and is recommended to operate over a range of 50% to 100% of capacity. Thus if you need a range of say 15 cfs to 360 cfs, you must employ a multiple pump system in parallel (each staged at about 50% of the capacity of the next larger pump in the system). Also, in many scenarios (particularly in large pumping systems) a few smaller pumps can be more cost effective than a very large pump and motor (particularly in life-cycle cost analysis).

sorry can't agree.. please come to the oil fields with me and we will remove the "chokes" and see how long the pumps will last. All a VSD does is change the operating range, it DOES NOT keep the pump at its BEP and in it operating range.. if you think it does then please let me know and I'll give you a Master Class in pumps.

I see that you have also posted on another topic.. in which I posted pump curves, in particular a curve of one pump at different frequencies. Please look at that curve, it only gives you the operating range at the different frequency, and not the BEP. Every pump must be kept at it BEP as much as possible to prevent wear in either up thrust or down thrust. That is and an only be controlled by back pressure or a flow control valve.. commonly known in the ESP industry as "Chokes" heres the link if you can't find the last topic we discussed http://cr4.globalspec.com/thread/66571?frmtrk=cr4sd#comment701705

Actually already have a Master's degree in the subject. You have already indicated your own response in your answer "Every pump must be kept at it BEP as much as possible to prevent wear". Obviously from your own statement youi don't have to oeprate at a BEP, it just helps reduce wear from transients (actually it is called a best efficiency point primarily because it produces the best application of pumping energy relative to the energy input into the system, and there are multiple BEPs for a pump depending on speed). As you implied it doesn't have to be set at a BEP, and most manufacturers supply pumps that aren't exactly set on the BEP for operation, but rather as close as possible to the BEP for the most common operating capacity or sometimes at maximum pump capacity. Ideally on a VFD system you can set a operating curve that ranges across one of the BEPs such that the most common pumping rates are as close as possible (weighted average of pumping rates).

Attempting to maintain maximum eficiency is why a VFD should not fall below 50% of pumping capacity for a pump (VFDs are already slightly less efficient than CSDs anyways). However, a consideration of more concern than wear is the best efficiency point, so you get the highest efficiency in pumping for the power input (electricity or fuel cost relative to pumping). Wear from transients can be controlled and reduced by other means than just operating at one point. Besides there are multiple BEPs for a pump based on the specific speed you set the pump at. There is a BEP on each different speed curve. If you choose correctly, you can employ a VFD across a broad range of flow rates without any throttling control valves inline. This is actually commonly dones in large water system. As a matter of fact there is a whole Text on the subject of pump station design that discusses the use of VFDs, that is possible the most coommon reference employed. It is entitled "Pump Station Design" by Jones, Tchobanoglous, Sanks, and Bosserman. As indicated, however, the applications of VFDs in this respect is a relatively new occurence within the last 20 years, so some older designers (or those employing old design processes) do no utilize them. Obviously, larger municipalities and water conveyance systems do, but some industries and fields are resistant to change.

Actually there is no particular reason to employ VFDs, if you are just going to control flow through throttling or a fixed orifice diameter. Plus the throttline back pressures can cause more damage as the pumping pressure must be increased to account for the head lost to throttling and the system design pressures (the pump must supply pressure to overcome against the throttling head and the required downstream head). Plus then you need a bypass with modulated control to make any substantial flow control changes in the downstream conveyance

The only instances where pump discharge head would not increase are:

1. If the system curve is totally flat (ie zero friction losses).
2. If the duty point on the pump curve is already at maximum head.

Putting pumps in parallel will normally always move the duty point to the left on the pump curve, ie less flow/more head.

I have to disagree with your last statement.

In fire service it is parallel= more volume, less pressure.

Series = more pressure, less volume.

Pump speed is then adjusted to maintain desires pressure needed to obtain desired flow, and or reach.

It's been bothering me a bit too, this 'one or the other' demarcation argument.

What I'd suggest is two pumps paralleled into a line, is more of both volume and pressure, due increased velocity, so drag, so back pressure to overcome.

But I'm really not sure what the OP is asking.

Is the two 60% throttled to give 120% of flow? If so why not a have a pump sized to that and VFD it back to 'usual' flow?

2 pumps paralleled in a line instead of one large one would not necessarily cause the flow rate to be any different. He did not indicate that he wanted to increase the flow velocity in a pipeline of the same size. You would manifold the pumps together. The conveyance systems head losses will increase or decrease based on how you size the mainfolding, intake pipes and downstream pipelines (darcy-weisbach equations).

additive pump curves for a system curve are actually very simple 2 pumps in parallel causes the system curve to increase additively by the flow rates of each individual pumps ratings at specific pressures. In other words, much like parallel in batteries increase amperage (not the voltage), parallel in pumps increases flow rate for a specific pressure. Series increases pressure (sort of like voltage) but maintains flow rate. This is why when you buy increase pressure pump systems they are built as stacked cans, which is a series of pumps stacked up. This is actually very basic engineering, that every civil engineer had to take in water resources in college (at least in ABET certified or canadian equivalent programs).

Now in this case he wants to be able to operate 2 pumps instead of one that is throttled to 60%, with one pump offline when operating at 60%. Thus it sounds like he wants to have one pump moving the same water flow rate as the large pump throttled to 60%. The throttling burns a lot of head, so the 2 pump system would have to be designed to account for the system head needed down stream, but it increases the head delivered by the pump as it moves down the pump curve Q.

"Thus it sounds like he wants to have one pump moving the same water flow rate as the large pump throttled to 60%."

It seems to me you have missed the point of the OPs question entirely. He mentions two similar pumps, and compare your statement above with that of the OP below:

"But considering the fact that discharge valve opening reduces to 60% when pumps are running in parallel against 100% opening when one pump is running."

And the rest is so mind-bogglingly poorly phrased, as if it were specifically designed to confuse. Do you have anything to add to what has already been said?

Yes I read that backward in his statement. I assumed he was throttling the one pump versus using two parallel pumps that could be operated to meet demands (a common practice). I read it the way an engineer would design a pump station. What he said "But considering the fact that discharge valve opening reduces to 60% when pumps are running in parallel against 100% opening when one pump is running. " make very little sense, why operate 2 pumps and then throttle to 60% than operating 1 pump. The throttling is just a waste of energy and increased O&M cost that build up in the life-cycle cost for the facility. If you only need flow from one pump to meet the demands at that time, you only use flow from that pump, you don't throttle the system, (unless of course the pumps are CSDs and you have a highly rapidly variable demand and a highly demand sensitive system, which is extremely rare).

As far as being mind boggling well it is possible, as a quickly drafted response can be poorly drafted for the audience or person you are responded to. Something I frequently have to bear in mind is the audience's capacity and background knowledge on which they can draw. Sometimes i skip discussions on related subject matter that seems rudimentary, and should elaborate to connect points for some. This is problematic when communicating to others outside this field of engineering.

I will often phrase a comment as a 'dumb question' rather than a statement.

If possible: it's couched to a 'neutral poster'; the reason for this is to encourage revisiting the base data by one, or both, 'other parties', without 'lending support' to either.

I.e. you were supposed to go back and read the OP and rethink the 'absolute stances' being taken. But leaping in, and patronizing, is an option.

As my wife will attest, I am probably more often wrong than right, but I'm pretty certain about this one. If someone can prove me wrong I'd be happy to learn from it, but otherwise I must stand by this.

Rather than go into long explanations, I suggest you look at this , which doesn't do a bad job at explaining system curves and pump curves.

Remember that a centrifugal pump can operate at any point along its' curve, from closed-valve (not recommended, but that's another issue) to End of Curve, and the point at which it operates is determined entirely by the 'system' that the pump is presented with. If there is even the slightest element of head loss (friction loss) in the total head, then increasing the flowrate through the system (by adding a pump in parallel) will increase friction losses, which will increase total head, which will move the duty point on the pump curve to the left (ie more head/less flow). The only exceptions are as I mentioned before, either the duty point is already at the far left of the curve, or there are no friction losses so increased flow does not affect head.

"In fire service it is parallel= more volume, less pressure." I have the utmost admiration for the fire services all over, but I fail to see how even they can defy the laws of physics, and I'd be interested to know how pumps in parallel can produce less pressure.

I may be setting myself up for a hose down, but the fire service maxim "more flow less pressure" would work if it refers not to pumps, but to hoses taken off the same pump outlet manifold.

No I don't think your are 'setting yourself up" at all.

Holzfeller is speaking only of pump performance at the outlet.

Bob is speaking of the reality of water delivery in 'hose and branch terms' - line losses, nozzle restriction, mass transfer rates.

Bob is right in 'whole system outcomes'

They would agree, if 'on the same page'.

"If there is even the slightest element of head loss (friction loss) in the total head, then increasing the flowrate through the system (by adding a pump in parallel) will increase friction losses, which will increase total head, which will move the duty point on the pump curve to the left (ie more head/less flow). "

This is correct if the system is not properly sized for the added parallel pump. However, most main conveyance systems represent only a small portion of the overall system losses, and most of these tend to occur in the valving and other appurtenances on the manifolding system (even over pipelines of many miles). So when you add the second pump in and then the subsequent manifolding, you size the manifolding to the apporpriate losses and offset for any added loss in the main conveyance system. A little less loss in the properly sized manifolding accounts for any additional sklight losses that might occur from the increase in velocity in the main conveyance. The proper design of the manifolding of parallel systems is the key to the issue of system head loss. You could actually design the manifolding such that the system head loss decreases by increasing flow and adding in a parallel pump.

A good reference i use, that happens to be on my desk at the moment, to understanding pumps is "Pump Handbook" by Karassik, Messina Cooper and Heald. If that doesn't help I can go look behind me on my shelf and find some others.

No, this statement is not conditional on correct or incorrect sizing of the system. Even a system correctly sized for the flowrate of two pumps in parallel will show an increase in head at the higher flowrate if friction head forms a part of total head.

"...offset for any added loss...accounts for any additional sklight losses ...You could actually design the manifolding such that the system head loss decreases by increasing flow and adding in a parallel pump."

I don't see how the pump manifold design can influence system losses downstream (unless of course it constricts the flow, and reduces head losses by reducing flow, which defeats the purpose). If you know of a manifold design that specifically results in a decrease in system head loss with a higher flowrate, I'm sure a lot of people would be interested to know about it.

Please correct me if I am wrong -

However I understood maximum head to be the height of a vertical column of water that the pump could support - ie the point at which flow drops to zero. - ie no friction except in the pump itself.

This is not affected by the number of pumps in parallel since they can all support only the same pressure and are all subjected to the same pressure.

At any height less than this the flow rate will be approximately doubled because each pump can supply the same flow rate at the specified height below the maximum head.

However I understood maximum head to be the height of a vertical column of water that the pump could support - ie the point at which flow drops to zero. - ie no friction except in the pump itself.

Yes, this is true, which is why I indicated earlier (#14) that this is one of the two conditions where pump head will not increase with pumps in parallel. For a given impeller diameter and pump speed, a centrifugal pump cannot exceed its' maximum head. However, particularly with the relatively flattish curves you see today, this point often (or usually) occurs before the flow has dropped to zero.

This is not affected by the number of pumps in parallel since they can all support only the same pressure and are all subjected to the same pressure.

True.

At any height less than this the flow rate will be approximately doubled because each pump can supply the same flow rate at the specified height below the maximum head.

Not at "any height". It depends entirely on the nature of the system curve. A flat system curve means system head will not increase with increased flowrate, so the overall flowrate will be approximately doubled. But a rising system curve will mean increased system head at the higher flowrate. The overall flowrate will be approximately the sum of the individual pump flowrates at the new head.

We should consider the fact that pumps are not working isolately they are part of system. Whenever we run both pumps in parallel due to increase in plant load, the both pumps deliver the combine flow rate but discharge valve level controller (LC) closes automatically to maintain the required level of suction vessel. So when discharge valve closes partially head automatically rises.

Holzfeller has given the most knowledgeable opinions thus far. The rest are a mixed bag of sometimes partly right, other times largely wrong, and still other times incomplete; with much confusion thrown in.

It is only the following statement that I disagree with;

Putting pumps in parallel will normally always move the duty point to the left on the pump curve, ie less flow/more head.

This is counter to what I have earned a living doing for the last 30 years.

I don't agree with you. Holzfeller is right in his statement. I think you are comparing with the duty point of single pump operation. It is to be compared with parallel two pump operation. If we keep discharge pressure same as with one pump, then flow rate will practically double. But in this way, duty point will be out from the system curve. Assuming system curve as same, duty point will slide towards left (lesser flow rate than double) and at the same time it will slide upwards too, delivering higher discharge pressure.

UsmanS gave some interesting informations at post 18 (not replied to my questions at post 5):

1. They run second pump also to meet increased plant load.

2. Discharge control valve is not for controlling discharge flow or pressure, but it is used for controlling the level of suction vessel (may be by taking input from level transmitter). Level of suction vessel to be controlled to prevent running out of fluid.

I can only say that it is wrong method of control. I strongly suggest to control the inlet valve of the suction vessel, keep discharge fully open and install VFD for at least one pump to run the system efficiently.

In reply to the question of original post i.e.What is the effect on performance and efficiency of pumps after parallel operation, I can say that efficiency will be much lower, as pumps have to operate at part load & higher head (which is not necessary as you are throttling the discharge valve). Performance of pumps also will decline as life of parts like bearings and seals will be lower for this type of operation.

I will have to dig one of my pump theory books out and attach to post. Afraid not till Monday. BUT, you can all hate me all weekend long if you would like. Sorry.

We won't hate you for being thorough, bob c, and checking things out properly, so take your time.

In fact, we are more likely to hate Tornado for showing us up with his efficiency and economy in the words per GA dept. One day I'll be as good as him.

The series-parallel operation is explained reasonably well on pages 1-5, and 1-6.

http://www.haleproducts.com/_Downloads/hale/product_manuals/Muscle%20Pump%20Operating%20and%20Maintenence%20Manual.pdf

This is how two stage fire pumps are designed to be used. In large flow scenarios, fire engines are used the same way. That is, if there is a long lay of hose to a fire, There will be one fire engine at the start of water flow, and a second fire engine at the other end of the long hose. Each fire engine will use some of its power to pump the water. From a hydrant, or lake to the first engine. That first engine will usually build up half of the pressure that the second fire engine needs to deliver the required water pressure. That way each truck is doing half the work. Likewise, if high flow is needed, two or more fire engines will pump water into a single high flow water nozzle, called a monitor.

In fire service, it is always series for pressure, and parallel for volume.

That's true, but it is irrelevant to the controversy in this thread. In particular, it doesn't say anything about pressure decreasing from one pump alone versus two in parallel on the same manifold.

I never made any comment concerning one pump versus two in parallel.

My objection was to the line below;

Putting pumps in parallel will normally always move the duty point to the left on the pump curve, ie less flow/more head.

Precisely: Hozfeller is talking "per pump" is "less" and as Bob knows it's not 'twice'.

It's just a 'same page issue'.

Good job - GA

To a certain extent yes, but not entirely. I should have made it clear that I was looking at the effects on individual pump curves, and my statement should have read:

Putting pumps in parallel will normally always move the duty point to the left on each pump curve, ie less flow/more head.

My apologies if this caused confusion. However, we are not on the same page as long as some people claim that the pressure will stay the same, or will decrease, and this is the point I was trying to clarify.

Depends on who is in 'we'

Perhaps I should have phrased that better:

"I am not on the same page as those who claim that the pressure will stay the same, or will decrease."

How's that?

Very well put, Tornado.

UsmanS replied in post 36 as follows

"Reply of your question is 1-Rated capacity of single pump is 60cu.m/hr with discharge pressure of 40Kg/sq.cm 2-Flow rate sometimes increase to 67 to 70 cu.m/hr so we have to operate other pump in parallel 3-It is six stage centrifugal pump. I think its sufficient? Moreover, these pumps are equipped with minimum flow devices with by-pass feature."

This requirement is very common - Starting a second pump is very normal

The flow will increase

the head will increase.

and the flow control discharge valve will close slightly.

The original question seems to be about inverse affect - Nothing expected.

Any effect if the pump is used individually again - NO.

The effect on the efficiency - could be lower but rather check the pump curve at 35 m3/h vs 60m3/h