PD Pumps in series 2

[farhadsh]

I have two PD pumps in series and need your input to see if there is any problem with operation:

A progressive cavity pump (upstream) will feed a Rotary triple screw pump (downstream). Pumps are feeding one pipeline with about 300 meter distance away from each other.
Pumps will be sized for same flow. Both will have VFD. Pressure relief valve will be installed between the pumps to protect system form over pressure and a pressure transmitter will be installed at suction side of downstream pump to control speed of upstream pump hence ensure enough flow is moving between the pumps.
Do you see any failure or concern here with safe operation?
Thanks,

 

[LittleInch]

I think your system is going to be difficult to get working. I would have the initial pump feeding about 10% more flow than the downstream pump. Flow control on the pressure relief line would be my suggestion. Any other way you will either feed too much our too little into the second pump and have constant hunting of the two vfds.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[farhadsh]

Thanks for your comment.
Are you suggesting using fixed speed pumps for this case?
10% extra flow ensures there is always enough flow feeding the second pump and with flow control valve located at pressure relief valve return line we can adjust amount of flow returning. Should I expect relief valve return line is always returning some flow back to the return line? Is it reliable design that relief valve return line is always running and part of the process design?

Our set up dictates to use VFD on both pumps since flow demand is variable through the year so we have to use VFD on both pumps. As you said VFD on both pumps might be always running to adjust flow and pressure and I am not sure this can cause any issue. Any comment on this?
Thanks,

 

[georgeverghese]

Whay have you got 2 pumps in series here - is the first one for NPSH boosting for the second pump ?

The control setup as you have described is not practical - we are dealing with incompressible liquid here, not gas.

Does the first pump have to be a positive displacement type ? It would be a lot easier to control if you have a centrifugal.

 

[farhadsh]

Thanks for comment
Due to high liquid viscosity only option is PD pumps. As you noticed first pump act as booster. I have no other choice than controlling PD pumps in series.

 

[DubMac]

Just a wayward, off the cuff comment here, but actually the slip in the PC pump (assuming it has some sort of elastomeric stator and not metallic), could act as a cushion in helping to feed the screw pump. Per Lil Inch's suggestion, you would need to oversize the volume of the PC pump to accommodate for the slip (which is the difference between the theoretical and actual volume of the PC pump). This would be wasted HP and is a function of your desire to impose this system upon the application; but it might be worth it to you.

Although the PC pump is a PD pump, by playing with the clearances it can be made to act somewhat like a centrifugal. You might want to purposely design in a larger clearance between rotor and stator to have the slip "match" the required pressure at entrance to the screw pump. A very soft stator may also provide some "adjustability" to the volume depending on downstream pressure. You have to make sure the seals on the 3screw pump can handle inlet pressure.

This is all very much conjecture and thinking outside the box. Without a doubt, these considerations should be left in the hands of the PC pump manufacturer; make sure it is a knowledgeable one.

You have a crazy application and there is probably a better way to do it; but I'd sure like to see what happens if you do go this route.

Please keep us posted on what you decide to go with.

 

[georgeverghese]

With a pd pump as NPSH booster, think this is getting difficult now.

If indeed there is no centrifugal option, then suggest a few other options:

a)See if a suction cooler will lower temperature and vapor pressure so that there will be more NPSH and first pd pump can then be eliminated

b)Put the second pump at a lower elevation to get more NPSH - then no need for first pump

c)Install a gas - liquid vessel in between the two pump and operate this on gas side pressure control to maintain pressure to second pump. Operate the first pump on level control from this vessel. Alternatively, float the gas pressure of this in between vessel (or pipe) on the gas header of some existing separator or other vessel - then you dont need the gas side pressure controller on this in between vessel. For this case, re adjust the diff press duty for these 2 pumps to suit.

In case (c), either fixed speed or VFD on the first pump would be acceptable ( Level control resetting speed control or level control operating a level recycle valve returning liquid back to source vessel).

 

[farhadsh]

Thanks for all comments but before proceeding to suggestions I would like to understand why my original set up is not working. I will put some numbers which make sense to explain my set up

PC PUMP:
Flow =60 m3/hr
Suction pressure (min/max)=0/100kPg
Discharge pressure (min /max)=600/900kPg
Relief valve set point discharge pressure=1200kPg
Motor sized based on relief valve set point

Friction loss between two pumps assume=400kpg

Triple screw pump:
Flow = 60 m3/hr
Suction pressure (min/max)=200/500kpg
Discharge pressure (min/max)=2000/2400kPg
Relief valve set point discharge pressure=2400kPg

I deliberately selected above process input which make sense for my system.
Now I add a pressure level transmitter at suction of Screw pump with low level set point of around 230kpg. As soon as pressure level at suction of screw pump goes down to 230kpg it increases PC pump speed let’s say 10%. This will ensure flow if enough between the pumps.
A high set point for pressure level transmitter such as 450kpg is selected. As soon as pressure at suction of screw pump increases to 450 it decreases PC pump speed let’s say 20%.
With above set up one can ensure safe operation between two series PD pump. Drawback as discussed by littleinch would be one VFD constantly running to adjust the speed. I am not sure how often VFD needs to adjust pump speed with this set up and if this can cause any damage or failure to electric motor or VFD itself. Please review my set up and advise your comment why this set up will fail.
Your input is appreciated
If I missed something pls ask me I will provide

 

[georgeverghese]

Okay, the excess pressure recycle back to source should work also. PCV on bleed line going back to source vessel.

So the first pump would be set with Qdesign at 66m3/hr or so. Pressure controller (with setpoint at 250kpag or so ) to sense suction pressure close to 2nd pump suction. Pressure bleed line tap off to be from close to 2nd pump suction. PCV to be sized to let down 6m3/hr or so at the desired dp across the PCV.

The speed control signal from source vessel going to 2nd pump VFD is to be biased upwards to provide the speed setpoint for the first pump. Speed setpoint bias to first pump VFD to be set to enable approx 6m3/hr more flow.

An alternative to the PCV on the bleed line would be a manual globe valve set in position to enable 6m3/hr bleed - that may be more stable than the PCV. The PIC - PCV arrangement allows the flexibility to startup the first pump from control room at no load also. With the manual valve, an operator has to be at the pump to fully open the manual valve to startup. The PIC - PCV arrangement, if found to be not very stable during normal operation, can be operated on manual at the PIC with PCV at fixed opening.

Okay?

 

[LittleInch]

The issue as I see it is that you have an incompressible fluid and two PD pumps. The feature of a PD pump is that it pumps a set volume for a given speed. The problem is one of dynamics, system response time and harmonics.

Using your system, the upstream pump, lets call it P1, is essentially trying to deliver a fixed pressure by means of the VFD controller. However as it increase speed, the pressure would rapidly increase for only a small change in speed as there is very little compressibility. The second pump, P2, may in reality accept slightly more flow even though it is on a fixed speed, but the margin is small. As the pressure increase, P1 slows down, pressure rapidly falls, it speeds up, pressure rises, slows down.... I think you get the picture. Essentially without introducing some sort of pressurised bladder arrangement as George suggested above, your system is far too rigid and could very easily suffer from hunting of P1 and also damage to p2 if the pressure fell below its NPSHR value. You need to allow some "slack" in the system. You might be able to get this to work after some time fiddling with the control parameters, but it will always be quite unstable and a real bitch to start.

The easiest way is simply to set P1 on a fixed speed and have a flow return control valve set on the pressure at P2 inlet. This way you can actually start the system as your max p1 pressures are higher than your max p2 inlet pressures (!). Thus on start up P1 is returning 100% of flow at a delivery pressure of 500KPag. As P2 starts and friction increases, the return line then reduces flow and controls on P2 inlet at 500 kPa. P2 speed controller can then control on flow with an override that if its inlet pressure falls below 200 it slows down a bit.

Does that make sense?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

The description of the control challenges with incompressible liquids from Little Inch has helped I hope

There is a possibility that with 2 PID type speed controllers, one for each pump, that there could be control interactions that could result in a speed mismatch - intermediate pressure would then drop.

Which is the reason for the suggestion that we use one speed controller for P2 and that P1 speed controller setpoint is always at a fixed bias to the speed of P2 - coupling interactions between these 2 are then eliminated. Another way to further reduce control swings on speed between P1 and P2 is as follows:

Set up only one SIC-2 for P2. The output signal from SIC-2 is also biased by a fixed amount to be the speed set to VFD-1. Bias setting to enable P1 flow to be 6m3/hr more than that at P2.

A few more suggestions :

a) It may not be advisable to size the bleed PCV to enable full flow at no load also - control valve may turn out to be too big and would be pinching down when at 6m3/hr during normal ops. Suggest we leave this at 6-10m3/hr and handle no load start through a separate remote loaded CV on the PCV bypass.

b) As you may well know, VFD motors can only turn down to about 50% of rated speed. Below that the windings heat up and the motor trips. For VFD motors fitted with aux. cooling fan, the motor can turn down to about 10%. So there is a choice to be made here. In either case, if you have to handle flow at less than the appropriate motor turndown, an additional level recycle loop will be required. This would be a low level LC on the source vessel that operates a recycle valve to return liquids from P2 back to source vessel. At the time when this LC gets activated into operation, both P1 and P2 VFD motors would be at min turndown speed, and the flow within the recycle loop would be that corresponding to these 2 pumps at min speed.

We cannot be certain that we have covered all significant details here - hopefully major pitfalls are averted now - a process controls engineer should assist to look closer at this scheme and see what else needs to be in place.

 

[farhadsh]

Thanks for your input. Few comments after reviewing all input

1) Slip: By checking PC pump curves I see slip is already considered in vendor’s performance data for instance 60 m3/hr pump @ x dp has 68m3/hr at zero dp. Are you still suggesting still first pump needs to be sized with additional flow to ensure 100% flow is fed into second pump?
2)PC pump though part of PD category but shows smoother curve compared to piston type. Not sure I should be worried a lot of pulsation concept
3) I might consider using bladder arrangement/stabilizer since there is limited option to connect relief return to supply and control it. It can be connected to first pump suction line.
Thx,

 

[EdStainless]

The drive on the LP pump needs to be slaved to the drive on the high pressure pump.
You can't rely on pressure sensing and the dynamics of tuning constants. They need to be linked, and in such a way that the first pump is always delivering 10-15% more flow than needed. Use your PCV to always spill back to source the excess fluid. It is like a canoe, if you are just drifting with the current you can't steer, you need some excess speed to work with.
You do need pressure sensing in between the pumps, if it goes too high or too low they both shut off, because that means that something is seriously wrong. If your PCV ever closes you are in trouble also.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[georgeverghese]

Ed,
With the SIC2 output signal bias to VFD-1, the speed on P1 will always be biased from that of P2 by the corresponding delta in flow

Re review points from Farhadsh,

1) We need this +6 to 10m3/hr at full flow and dp across P1 - not at zero dp. And pls check that this flow at design dp is at the highest flowing viscosity.

2) if there is a concern with pulsations on flow output from P1, then one or or more of the following may be required :
i) a suction pulsation damper at P2 suction - damper precharged with N2 at 70% of PIC setpoint.
ii) operate PIC - PCV on manual at PIC during normal operations to stop PCV from oscillating
iii) operate the PIC with a deadband range at the desired setpoint

3) Not clear what is stated here

What happens at cold start here with this viscous fluid ? Else the pump will trip out on overcurrent - Do you need to heat trace all the piping here to get it up to operating temp?

 

[farhadsh]

Is it okay to connect relief line to pump suction instead of supply?
Piping and pump will be heat traced for sub zero protection.

 

[LittleInch]

Yes it is ok to connect relief / return line to pump suction, but don't operate it at no forward flow for very long as your fluid will heat up fairly rapidly ( a few minutes).

If you're going to put in a bladder arrangement in make it as big as possible and put it between the two pumps to give your controllers a chance to operate properly - basically control P1 on P2 inlet pressure and start both of them more or less together (maybe P2 a few seconds after P1 start).

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

Presume heat tracing type and heating power delivery is adequate to bring piping contents back up to NORMAL operating temperature for startup ? If we are looking at heat tracing to get back up to about 90-110degC or so, self regulating electric tracing is the better choice. Else you may need to move up to constant wattage parallel resistance type.

What about internal casing heaters for the 2 pumps to warm up the pumps also? External steam tracing or electric tracing for the pump casings may result in hot spots and uneven heating. If this is a flammable liquid at operating conditions, some attention required to prevent overcooling recirc seal fluid on plan 53 shaft seals during cold weather for startup.

If you are in a coastal location, chloride corrosion under insulation can be avoided if you use TSA aluminum spray primer coat.

Looks like you need to ask the P1 vendor to run at approx 15-20% higher speed to get 66m3/hr at design case dp for the full range of operating viscosities. Yes, flow pulsation amplitude on PC type pumps should be a lot more manageable - however, an intermediate pulsation dampener will put to rest any concerns with this and make the job of the excess pressure bleed controller a lot easier.

 

[SNORGY]

When I first read Dubmac's suggestion, I wondered what he was thinking. When I read it again, I found it to be pretty smart - provided that folks treated it exactly as he set out for them to treat it. Not sure how long the upstream pump would last and operate reliably. Might consider a continuous spill back to suction through a restriction orifice so that the downstream pump isn't a "master" over the upstream pump so much as it is a "master" over the upstream spill-back rate. That way, the PC isn't the only thing in the system taking the hit.

 

[farhadsh]

Thanks to everyone.