Operational risks of operating centrifugal pumps in series 1

[Lan85]

I have been working on a hydraulic calculation for a client to check whether pumps provided by a vendor can do the intended job.
I need 34m head to transfer the water and each of the vendor pump (vendor provided two) have a head of 22m at the required flowrate.
I thought arranging pumps in series will be a solution, however our lead engineer has said that operating pumps in series can be problematic and not recommended.

What are some operational problems of operating centrifugal pumps in series? Anyone have practical experience on this?

Thanks

 

[Artisi]

Without any idea of the L E's concerns, it is a bit difficult to make much comment.
Operating pumps is series is a no-brainer, and with the low pressure you are talking about with these pumps I can't see what his concerns might be.
However, you haven't given any idea what the flow rates are - 10gpm or 10,000 gpm this of course could have some bearing on his concerns.

His concerns might be - how to start / stop, over pressurising the second pump casing, or gland / mech. seal area, give us a bit of feed back.

If it's a big problem / concern, change to a single pump for the duty.

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]

You need 34m and the vendor provides pumps only able to do 22? Looks like they can't do the job to me.

Series pumping is not in itself that problematic, but has it's own issues. Your issue here would appear to be that two pumps in series give you more than 34m and hence needs to be throttled to prevent over speed.

Red artisis post carefully, he has a lot of pump experience and come back with some more information.

My motto: Learn something new every day

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

 

[HPost]

I can't see a problem here, other than the potential slight compromise in pump efficiency.

Pump 1 will generate 22 metres of head above the inlet pressure, then pump 2 will generate another 22 metres of head. There will be some losses along the way, but with 44 metres available against a requirement of 34 metres, there will be 10 metres spare.

On starting, run pump 1 first, then pump 2.

As I say, there will be some efficiency losses, but this is effectively a de-constructed multi stage pump.

HPost CEng MIMeche

 

[Zonked]

I kinda have the same question as Lan85. I was told that if the mass-flow variations is too high a parallel system is preferred. (centrifugal pumps)

Also when it comes to redundancy. If one pump fails in a parallel system it should not be a great problem. In series, if one pump fails, the whole system will fail?

 

[LittleInch]

H post - the issue is where does that extra 10 m go? you either need to throttle it and hence throw energy and money away or the pumps will both continue to move along their pump curve until it meets the system resistance curve. This might be Ok or it might be at the end of their curve or further in which case the motor becomes overloaded, you pump a lot more fluid than your design point and you risk NPSH issues on the first pump.

Like a lot of these posts, you really need to design a pump system to do a duty and match it against the system curve to be relatively stable or work within a certain range. Doing this (using what you have and hoping it will work as you think it might), often leads to significant issues.

My motto: Learn something new every day

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

 

[HPost]

It only works if the pumps in series are the same specification in terms of flow and head. The system restriction needs to be the same on both pumps.

It doesn't work with a fluctuating flow and head requirement or different pumps performance curves.

 

[LittleInch]

What "only works if...."? The OP says they are the same (we assume when he says vendor supplied two), but my point remains - if you don't add a control valve ad then throw energy away, what happens to the 10m?

If they are in series, there is only one system restriction.

My motto: Learn something new every day

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

 

[Artisi]

For over head generation in series, you can always reduce the Impeller diameter of the second pump.

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]

Or you could just buy the right pump in the first place....

Loosing half it's current design head though is a bit much for an impellor size reduction when we don't know if it's close to max size or min size already.

You never know as the OP hasn't responded, but if the motor is big enough you might be able to put a larger impellor in the pump and run it in parallel as it was originally intended.

We prob need to assume that the operation is one on, one standby otherwise putting two in series will halve your flow rate if you get the head issue sorted.

My motto: Learn something new every day

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

 

[1gibson]

The extra 10m doesn't necessarily need to be throttled, the system resistance will take care of it, and you will end up running at a higher flow rate. Is this OK with the process? NPSH, motor power? Is the system curve flat (mostly static lift, flow could increase a lot) or steep (mostly friction, should handle itself without major flow increase)?

You can run dissimilar pumps in series (it's called a pipeline...) without any inherent hydraulic problems. You just need to make sure that the pumps are reasonably well matched in terms of flow rate, and then work out the operating logic for start/stop/bypass. If you run a low flow/low head pump in series after a high flow/high head pump, the smaller pump might act as an orifice, but the system will still "work." It would most likely damage one or both pumps, but that isn't a hydraulic problem.

It doesn't "not work" either, in the sense that parallel pumps would "not work" if they are too dissimilar. In parallel, one pump could be forced to shutoff or end of curve and you would not get the combined flow rate that might be expected. With series pumps, you can reliably predict how they will interact with the system. Add the heads together at a certain flowrate, when that head and flow match the system curve, there you go. The flow rate through both pumps always has to be the same (unless there is some piping in between that would allow fluid to be diverted.)

It will affect the thrust, and the seal chamber pressure, of the second pump.

 

[LittleInch]

Lan85 - Questions, questions. If you answer some of them you might get a bit further with your problem, but if you don't I think we've come to a full stop (I agree with all 1gibson says).

My motto: Learn something new every day

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

 

[Lan85]

Artisi: Flowrate is 60kL/hour (15850 gallon/hour). If you reduce the impeller diameter on 2nd pump, then it will act as an orifice as 1gibson pointed out and it can damage the pumps.

LittleInch: Client changed the pump location and purpose so the head required went up. Vendor initially supplied for a duty and standby unit for the required purpose at that time.

As 1gibson and Littleinch pointed out, extra 10m head does not need to be throttled but then the pump operating point will move towards the right of the curve (increased flowrate). If accurate flowrate is required, then throttling is also required. I guess, if the pumps are in series, discharge pressure of one pump will be the suction head for the other pump? I have attached the pump curve for reference. However, series system will be very risky as these are for a water treatment facility and there will be no feed water for the plant if one pump fails.

We have advised client to select new pumps for duty-standby arrangement and we are going ahead with it.

 

[Artisi]

The comment re the second pump acting as an orifice is not valid in this case, the comment was dissimilar pumps and that would raise another discussion if required. In your case if the pumps in series produce too much head you can reduce 1 or 2 of the impeller diameters to reduce flow withoutany detrimental effect on either.

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]

Lan 85. These are relatively small pumps, but looking at your curves and looking up the grundfos curves, you don't have much room, either to operate beyond your current duty point - prob up to about 90 m3/hr, but beyond the end of the curve. The pump and motor may well survive but you risk doing long term damage. The impellor is already at the low end of its range so there's not much joy there either to reduce impellor size.

Your decision to go for a new revised 1+1 system looks like a good choice to me.

My motto: Learn something new every day

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

 

[Artisi]

Seems to ge some conflict in data, suggest you advise flow rate and supply a system loss curve, from here we might make some sounds assumptions instead of guess work.

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.)

 

[1gibson]

The second pump acting as an orifice is a very extreme case. Extrapolate your pump curve until the line crosses 0 and head turns negative, that is the flow where the second pump would act as an orifice. You would basically need the first pump to be big enough to flow around 2x the runout flow of the second pump, major mismatch, should never happen. The point was, no matter what you do in series, you can predict how the fluid will behave.

 

[moltenmetal]

Wow, if you can't operate centrifugal pumps in series without generating problems, I guess the Grundfos side-channel/stacked impeller multi-stage centrifugal pumps that you find in water wells all over the world, must be a nightmare!

I've done this many times, with no problems. All you need is a decent size and capacity match, plus some thought to start-up as 1gibson mentioned. Is there an efficiency hit relative to a properly sized single pump? Well, that depends. Where on the curve are you operating? Multiple pumps in series are a preferred solution for stable operation at comparatively high head + low flow with a centrifugal pump rather than a PD pump. Operating a larger pump too close to its minimum flow may actually be a less efficient solution than two smaller pumps in series, each operating near its BEP.

Installing one pump is going to be cheaper than installing two smaller ones, unless the one larger pump is a specialty item and the two smaller ones are commodity items. Installing 1 two-stage pump will be cheaper than installing two stand-alone pumps, subject to the same caveat. So: what matters more to you: capital cost, including installation, or operating cost?

 

[Drexl]

One issue of operating pumps in series generally could be that say you have 2 parallel pumps, some piping and then 2 more parallel pumps. You are running all pumps and trip one pump. This can cause too high flow for the single pump. Problems could be too high power and cavitation.

 

[Artisi]

Drexl, if this was considered a concern, then the easy fix is to install an orifice plate down stream of the pump discharge to restrict flow to an acceptable level.

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.)