Positive Displacement Pumps in Series 5

[PhilOos]

Good day,

I need some insight into positive displacement pump theory. I realise the differences between a normal centrifugal pump and a positive displacement pump for example, but I want to know how they perform in a system e.g. in parallel or in series. Nowhere can I find literature which explicitly states that positive displacement pumps does not perform or do perform the same as centrifugal pumps in a parallel or series arrangement.

Any help (proof or references) will be much appreciated.

Kind regards

Philip Oosthuizen
Company info:
SteinMuller Engineering Services

http://www.steinmuller.co.za/

 

[BigInch]

Not exactly the same, since the "curves" of each are quite different, but you can generally combine each of them in a similar manner. What you get stacking "curves" of a recips typically straight lines can look much different than the result of stacking the curves of a centrifical.

One should be carefull with the recips acceleration head requirements as well, which may cause some negative offset of the stacking to be accounted for. Centrifugals would require the same consideration, if suction losses to the second pump are significant.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[ione]

For PD pumps arranged in parallel, use the same rule as that for centrifugal pumps. PD pumps in parallel must have the same maximum pressure and the total capacity will be the sum of the capacities.
As far as I know PD pumps installed in series is not a common arrangement. I have seen many booster (centrifugal) pumps connected in series with piston pump. This configuration is often used in condensate recovery plants, where hot condensate is recovered and used to feed water-tube steam generators. In the case aforementioned the booster pump is used to increase the NPSHavailable and prevents cavitation.

 

[micalbrch]

PD pumps in a direct series is not common. "Direct" means that one pump feeds straight into the other without any buffer tank. That will not work with any piston, plunger, or diapragm pump due to the oscillating displacement. And I have doubts whether it works with peristaltic or eccentric screw pumps.

PD pumps in parellel work but the pulsation needs to be considered seriously with piston, plunger, diaphragm or peristaltic pumps. It is difficult to get the pumps synchronized. Therefore the pulsation dampener sizing must not only be done for each pump itself but also for all pumps together. If this is done, PD pumps in parallel will work fine.

 

[PhilOos]

Hi all,

Thank you for the response. I'm looking at a fuel oil system that has a low pressure pump house (to draw oil from the tanks, with a outlet pressure of 470kPa) and a high pressure pump house (supplies oil burners with oil at 38bar). Between the two pump houses are a pressure regulating station and a few manifolds. Both these pumping station is equipped with screw type PD pumps.

The current problem is that the HP pumps are not able to supply the oil at a pressure of 38 bar (due to a lower than expected oil viscosity), but supply the oil currently at a pressure of 34.5 bar. I've created a simulation model of the system in order to investigate the systems behaviour. I found that by increasing the set pressure point of the low pressure control station to 820kPa, the HP pumps are able to supply the oil at 38 bar.

Am I correct by saying:
The outlet pressure supplied by the PD pumps is a function of the back-pressure of the system and the slip flow rate inside the pump. By increasing the LP set pressure, I effectively reduce the pressure drop over the pump, which in turn reduces the slip mass flow rate inside the pump. This leads to a higher nett mass flow rate created by the pump, which will raise the back pressure (increase the pump outlet pressure)??

Kind Regards


Philip Oosthuizen
Company info:
SteinMuller Engineering Services

http://www.steinmuller.co.za/

 

[ione]

The behaviour of a PD pump is different from a centrifugal one. In a PD pump a decreased viscosity should produce a reduction in capacity and a decreased capacity should produce an increase in pressure. Am I wrong?

 

[micalbrch]

You are wrong ione. The higher the flow the higher is the pressure. A higher viscosity can lead to less flow (from a certain point on) as the filling efficiency becomes worse.

 

[TenPenny]

One way to think of it is that a PD pump doesn't 'create' pressure, it 'overcomes' it...ie, the discharge pressure of a PD pump will be whatever is required to move that volume through your piping.

If your PD pump isn't generating the pressure that you expect, the problem might not be the pump.

 

[hydtools]

Like TenPenny says, the positive displacement pumps deliver flow, they do not create pressure. They must be able to deliver flow against the system-created pressure, the resistance to flow.
For a given system, pressure will vary with delivered flow rate. If you are not seeing the pressure at your delivery point, you are not pumping enough flow.

Ted

 

[BigInch]

There is no point trying to think about a pump working without a system unless its on the test bench not connected to anything. We have to stop thinking about pumps producing pressure when they are not installed in a system. A pump running on a test bench produces all flows at atmospheric pressure, because its system curve on a test bench is only limited by speed; the system curve is 14.7 psia for all flowrates.

For any pump with a given flow, the pump discharge pressure will correspond to the system curve head required for that flowrate.

If you don't see pressure at the delivery point, your system flow is not what you think it is, or you don't know how to calculate your system curve, or you have no flow, or your pipeline is still line-packing.

A decreased viscosity will lower your system curve and thereby require less pump discharge pressure. Either a recip on pressure control, or a centrifugal, may "runout" producing more flow. The recip will go to max speed until some head is (if) finally produced, the centrifugal may runnout until head is near zero, or (if) head is finally produced, each by reaching (if) some amount of line packing.

The higher the flow, the more pressure is typically required for the system curve to make that flow, hence the pump discharge pressure is also higher.

A centrifugal pump has a pump curve, that is truely a curve, but does not produce any head, except the head where the centrifugal pump curve intersects with the system curve.

A reciprocating pump has a pump "curve" which is essentially a vertical line, and can produce any head, however it will not produce any head other than the head where the vertical line intersects the system curve.

Centrifugal pumps produce exactly the same head when installed in a system as a recip pump installed in the same system for any given flowrate.

So, tell me again, What's the difference between a centrifugal and a recip, other than the cylinders?

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[micalbrch]

Shouldn't we go back to PhilOos' question? Whether to use a centrifugal pump or a PD pump should be decided in dependence upon the application, not in dependence upon personal preferences (at least not in this forum). Each pump type has its right to exists. 90 % of all pump applications can be done (and should be done) with centrifugals. But the other 10 % are still a few thousand application where a PD pump is the better choice.

O.k. Phil has PD pumps. His question is why the increase of pressure of his low pressure PD pumps increase the output pressure of his high pressure PD pumps. I have to confess that I have problems to imagine how his system looks like but the high pressure PD pumps' output pressure is just the differential pressure between in- and outlet. And that's why the output pressure of the second stage pumps gets increased.

 

[MikeHalloran]

As screw pumps wear, the teeth get thinner, and the backlash between them increases, but they don't just stop pumping unless there's a mechanical failure. What does happen is that their flow capacity at a given pressure falls off, because of reverse flow in the space between the teeth.

I have seen systems where nobody noticed the reduced flow until half the tooth mass was gone. .. after decades of service.

Having to increase the interstage pressure in order to maintain the secondary discharge pressure is simply an indication that the second stage pump is getting a little loose, and it's time to start shopping for a rebuild. No hurry.



Mike Halloran
Pembroke Pines, FL, USA

 

[NickJ67]

It is true that it is unusual to see PD pumps in series, although using dynamic pumps to flood feed PD pumps can work very well.

It is possible to use PD pumps in series. The trick is to balance their flows well enough to maintain the pressure between the two within a sensible band.

The pumps must have the same displacement otherwise damaging pressures (high or low) will soon occur between them.

It helps alot if the pumps have a reasonably constant delivery - placing high pulsation devices like piston pumps or peristaltics in series is unlikely to go well without some kind of buffering device between them.

As mentioned already by others, PD pumps "overcome" pressure rather than creating it. If dead-ended they'll carrying on upping the pressure until something gives, be it the pipework, motor overload, seals or transmission. In normal use they'll all have some slip, but if they are any good, it shouldn't a big percentage. Some PD pump styles have more inherent slip than others, but in addition to that the percentage slip will depend on the differential pressure across the pump, it's operating speed, the product viscosity and the condition of the seals. In a direct series application some slip could actually help make the matching less critical by serving as a (rather crude) buffer.

In answer to the OPs question

"Am I correct by saying:
The outlet pressure supplied by the PD pumps is a function of the back-pressure of the system and the slip flow rate inside the pump. By increasing the LP set pressure, I effectively reduce the pressure drop over the pump, which in turn reduces the slip mass flow rate inside the pump. This leads to a higher nett mass flow rate created by the pump, which will raise the back pressure (increase the pump outlet pressure)??"

Almost; it's really the outlet volume that is the function of differential pressure and slip rate. As the differential pressure rises, the percentage slip increases until a stable system is acheived. The extreme end of that is a dead end situation that would require 100% slip (more likely something breaks first).

In the system described, stability is being achieved at 34.5 Bar because that is the point at which pump curve at system curve meet. IE the flow from the pump at 34.5 Bar matches the flow needed to give 34.5 Bar back pressure through the burner nozzles.

That it occurs at 34.5 Bar rather than the expected 38 Bar could be because the lower viscosity oil increases the percentage of slip in the pump causing it's output to fall. It could also be because the lower viscosity oil gives less backpressure through the nozzles even though the flow through them remains the same. Most likely it's some of each, but unless you can measure flow directly you won't know for sure.

Increasing the inlet pressure of the second pump, thus reducing the differential pressure across it will decrease it's percentage slip and increase it's output volume but may not create the expected pressure rise due to the effects of viscosity on the burner nozzles.

Regards

Nick

 

[BigInch]

Phil is correct.
And, Yes mical, if Phil raises the suction pressure at pump2, his flowrate and his pressure output to the system will tend to increase; as the differential head of pump 2 is added to its suction pressure. But nothing says the differential head of pump 2 will stay the same as it was before. It will change its differential head to provide only what it needs to send its new flowrate into the pipeline, provided it has the power available to do so.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[electricpete]

So would it be fair to say that putting pd pumps in series would be bad practice because it will be difficult to create sharing of the "load" (load means flow rate times dp). Sharing of the load will only occur if the two pumps are identical. Slight variations will cause one pump to deliver most of the dp. (correct?)

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[hydtools]

How did PhilOos increase the output pressure of the first pump/the input pressure to the second pump? He had to increase flow. Increased flow to the system increased the pressure required to push the flow through the system.

His unexpected lower vicosity reduced the system resistance to flow, hence the lower pressure he reports as a problem.

He has not said if he is getting enough flow at the lower pressure, lower viscosity. If he is getting sufficient flow at the lower pressure, what is really the problem?

Ted

 

[BigInch]

If one isn't power limited for the given flowrate, they will split the required head between them. If one is of less power rating then the other, its differential head is limited, so they may not be outputting the same head, hence wouldn't have the same load either, but you knew that because one was power limited. Otherwise, the pump group (together as a unit), output the power to the fluid and they just see the one required head to meet; that head required by the system for the given flowrate. Each pump has to be moving the same flowrate, and they have no mechanism to control the distribution of head between them (each of their "curves" is a vertical line, so you can't add them vertically) so, by symetry, the head contributed from each has to be equal. Since flow is equal too, then power is equal. Maybe it would help if you consider that the "curve" isn't precisely vertical; it leans to the left 1/2 degree from zenith. And consider that each pump is running at its power limit, so each has a definite and equal differential head output, now you can add the two "vertical" lines together, one on top of the other. They have to add to the head required by the system curve, because that sets the total power required to move the fluid down the pipeline. Each takes half.

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[hydtools]

So his simulation responded correctly. Increasing the input pressure to pump 2 by 3.5 bar increased the total delivered pressure by 3.5bar to the desired 38bar. The total delivered pressure is the sum of the pressures of the two pumps.

How will he, in practice, increase the delivery pressure of pump 1?

Ted

 

[BigInch]

I let the above "run free", which is unusual for a PD pump, because you're typically wanting to run at some set delivery rate and always controlling it to do so somehow, but there are PD pumps which are run either on/off at whatever the synchronous speed of the motor is, such as when filling the car up with gasoline.

He just turns up the speed of pump1.

Running free for PD pumps would mean adjusting the speed of each pump to run the flowrate we want at the moment. So if we had two filling station gasoline pumps in series, both with switchable speeds, we'd move both to speed2, otherwise the efficiency of pump2 would suffer if it remained at speed1 but was feed the speed2 flowrate.

If you didn't change the speed of pump2, its efficency would suffer thereby requiring more brake horsepower to be delivered from the motor per unit pumped, but its hydraulic head now would still track the system increase and be 1/2 of the new head needed by the system. So, motor2's load would change according to pump2's new efficiency, but if eff2 didn't change much (perhaps debateable), hydraulic power would still be roughly the same power required of pump1; hydraulic power still being distributed 1/2 to each.

You won't find 2 PDs in series here, but its a pretty good treatment of what a PD curve looks like and how it interacts with the system curve and how they can and cannot be controlled. You can probably run with the principles from there. Here you go,

http://www.driedger.ca/ce2_pdp/CE2_PDP.html

"We have a leadership style that is too directive and doesn't listen sufficiently well. The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.liv

 

[katmar]

Phil, I think it is wrong to conclude from the fact that the pressure at the burners is lower than it should be that you have a pumping problem. The posts above agree that a PD pump is a fixed volumetric flow device and that the head comes from the system back pressure when passing that flow. We do not know enough about your control system and the operating philosophy of your plant to really give definitive answers.

How do you control the pressure between the pumps? I guess with some sort of spill back pressure controller? What is the controlled parameter for the oil fed to the burners? Normally it would be a flow control, again with some sort of spill back device.

If you are controlling the flow rate to the burners, and at the required flow rate you are seeing a lower than expected pressure it could well be that the burner nozzles have the wrong characteristic for the low viscosity oil you are using vs what they were designed for. So, rather than trying to fix the pumps I suggest you look at the whole system and see what else could cause the pressure to be low.

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com