Recip AND centrifugal pumps together 2

[Trond]

Hi all,

Connecting centrifugal and reciprocating pumps to the same discharge manifold - what would be the concerns?

Assuming same discharge pressures, check valves in place, pressure relief valves provided, etc. Operating pressure around 150 bar (water injection).

It was just suggested by our operations team. I have yet to receive data sheets of the respective pumps, but I have never seen this done before, and was wondering it there are any good reasons who not to do it.

 

[Bjegovic]

I have not seen this in practice before! What are the reasons for such concept? And what the system looks like?

My main concern would be that, due to different nature of the two pump types, centrifugal pump(s) may be forced into out-of-range operating conditions. (compare the pump curves and connect them in parallel, then add to this drawing all possible system curves...)

rated pressure is quite high for centrifugal pumps. what is the capacity?

 

[checman]

Trond: I would be very careful discharging both into the same header. I think the problems that you may encounter will depend on the type of multistage and performance curve of the centrifugal pump. My concern would be the effects of the discharge pulse of the reciprocating pump on the centrifugal. If the pressure pulse momentarily reduces flow through the centrifugal the possibility of damaging thrust load changes within the pump exists. Centrifugal pumps do not like sudden pressure changes quick opening or closing valves normally result in short bearing life if not periodic bearing failure.

Regards Mark

 

[Montemayor]

This procedure or application is done all the time.

Chemical additive pumps (or pH stabilizers, for example) are frequently applied to processes to add or control the composition of the fluids being circulated. This is a common requirement in many industries. However, the discharges of each of the two different types of pumps are not piped directly next to each other. This is done in order to take advantage of the normal turbulance and natural mixing action found in the discharge of a centrifugal device and take advantage of the fact that a check valve is normally installed on the direct discharge of the centrifugal pump, deterring any mixing from migrating upstream.

Normally, the reciprocating pump's discharge is connected sufficiently downstream of the centrifugal pumps such that the pulsations of the former are damped and do not affect the latter's performance. It is the centrifugal that is the most sensitive and a "pussy" compared to the more robust and consistant recip.

There would be reason for concern if the capacities of both types were equal; however this is seldom or never the case. Normally, the recip has a relatively insignificant capacity compared to the centrifugal and the the commingling of the two discharges is unnoticable. As in all process applications, there should be a concern for studying the potential of a hazard or problem in applying the two pumps together. However, on total and detailed review it will be found that there is no reason for safety concerns if the normal safety rules are followed.

As I stated it is the recip, inspite of it's weak flow capacity, that poses a potential hazard in it's positive displacement capability and robust nature. It can, if allowed, create a severe or dangerous over-pressure scenario. However, with an appropriate PSV on it's discharge leading back to the suction source, there should be no problem.

Needless to say, the application should be dealt with at least one Hazop - and an MOC (Management of Change) if this is a process change.

 

[25362]

As Montemayor says, precautions should be taken to dampen flow pulsations. Suction stabilizing and discharge dampening devices are also used to isolate, as much as possible, the sensitive centrifugal operating characteristics from the (displacement pump) flow pulsations.

The magnitude of the centrifugal's flow variations due to pulsations from the other pump is a function of the system head curve, the centrifugal characteristic curve, the degree of flow pulses, and the liquid compressibility.

Let's not forget that the "new" increased capacity for both pumps working in parallel could only be supplied, if the discharge pressure of the centrifugal pump is capable to satisfy the "new" system generated back-pressure at the new higher total flow.

 

[EdStainless]

The the centrif. is large and the flow from the PD is small then there probably isn't much concern.
You need to make sure that the centrif doesn't see the pulsation. The hydrol balance system wouldn't hold up very long.

= = = = = = = = = = = = = = = = = = = =
Corrosion never sleeps, but it can be managed.

http://www.trenttube.com/Trent/tech_form.htm

 

[rzrbk]

Are these pumps to be run at the same time, or as one pump and a spare?

Have seen a similar situation in injection water service, for continuous exchanger water wash, where the existing injection system used a PD pump. One new centrifugal pump was installed for reliability, but the PD pump was kept as the spare.

If this is the case, some of the above concerns would only come into play during pump switching.(if uninterrupted flow is required)

 

[hydrae]

I saw this in the design of a nuclear plant I worked at, primary loop charging (150 bar). The PDP was designed for normal operation and two centrifugals were for safe shutdown. The PDP never worked right, so the centrifugals were used for normal operation, there was talk of replacing the PDP with a centrifugal but other issues took precedence.

Hydrae

 

[Trond]

Dear all,

I just posted this before leaving work on Friday afternoon, and I am overwhelmed by the response - don't you people take weekends off?

Anyway, to clarify somewhat - the system in question is a water injection one, and the two existing WI systems (two recip & two centrifugal pumps in parallel) are discharging produced water to separate injection wells, but the objective is now to inject into a single well - with different pumps operating at the same time. According to drawings (which is all the info I have right now), the flow rate for the recip pumps is about half that of the centrifugal pumps (18,000 bpd for the latter).

The reciprocating pumps have indeed been fitted with pulsation dampeners. I feel that Marks comment on pulses still carry some merit, though. Also the issues of increased back pressures due to the higher flow are very valid.

My gut feeling at this stage tells me there could be problems, considering the different operating characteristics of recip & centrifugal pumps. Would appreciate any other comments, suggestions or sharing of past experience that you guys or girls may have.

Trond

 

[TD2K]

The discharge pulsation effect of the recips on the centrifugals is something I can't comment on and especially at these pressures but in my mind it's the biggest question out there.

Once that is answered, I don't believe operating the centrifugals and recips in parallel is an automatic no-go but you will need to analyze the operating envelope carefully as mentioned by checman. Any changes in your injection well's charateristics (choke valve being adjusted, downhole ability to take the water changing in any way, etc) is going to shift the system curve up or down changing the operating point for the centrifugal pumps. The recips won't care, they will just keep moving their design capacity till they hit their trip settings, PSV set point, etc. The centrifugals will take all of the load change moving up or down their curve. You will want to ensure you have the proper controls or spill back loops to ensure minimum flow through the centrifugals at all time. How flat the centrifugal pump curves will have a significant effect on the 'reactivity' of this system (to use that term) to changing injection pressure/flow requirements.

 

[pumpkind]

As already mentioned - the main concern is the effect of the pulsating discharge of the PDP on the centrifugal. In situations where the PDP flow is negligeable compared with the centrifugal - there is no problem - is like injecting a syringe into a tap flow.
As per your additional information, in your application the PDP flow is approximatively half of the centrifugal flow. This is significant.
In this situation the bearing life expectancy of the centrifugal pump will be affected drastically - the same goes for the mechanical seal life ( if the pump has one )
and motor bearings

 

[ccfowler]

Pulsation dampeners are very useful, but I would still pay very careful attention to the residual flow fluctuations. They are pulsation dampeners not pulsation eliminators. The relative capacity of the recip's to the centrifugals is certainly great enough to cause due concern.