Centrifugal Pump - Reverse Flow and Idle Forward Flow

It seems that the one pump is used to supply a positive (or less negative) inlet pressure to the second pump. The question then seems to be - what will happen if the suction booster is not used.

With the suction booster off the atmospheric pressure will have to overcome the total static and friction losses from the surface of the liquid to the eye of the main pump plus the additional friction caused by the stopped booster. The main pump may cavitate and stop pumping.

The inclusion of the booster indicates that it was necessary.

Reverse flow can be prevented by a check valve and/or a bypass pipe and valve.

The situation is the opposite from your understanding. Let me try again.

* The booster pump is running (i.e., the condensate pump). The booster pump supplies NPSH to the feedwater pump during normal operations, and the feedwater pump further pressurizes the system in order to pump water into a boiler.

* Now, let's assume that the boiler is cold iron, i.e., there is no backpressure from the boiler. The feedpump is not running, but the flow valves are open.

* The booster pump will push water through the suction of the feedwater pump and out the discharge into the boiler.

* How much water will go through the deenergized Feedwater pump? The pump is free to turn, restricted obviously by friction and the interia of the connected motor.

The other part of the question is how much flow will there be in reverse through a centrifugal pump if there is no check valve system pressure is trying to shove water from the discharge back to the suction of the pump?

Adding to Hendrik's post, calculating the reverse flow is not usually carried out as it is not usually necessary. It could be measured during a test, though why one would want to send water backwards through a centrifugal pump is abstruse. Nearly all installations will have a check valve to prevent this happening, as correctly stated. In plant design one must always consider the ramifications of unwanted reverse flow, not just to the pump - to the upstream systems as well.

'Reverse Flow' is one of the important considerations applied during a HazOp Study.

All of this is difficult to calculate, but why not build the system correctly?

I don't know anything about boiler feedwater piping systems but I can tell you from a high energy pump perspective that pumps that have wear rings and center bushings need to have pressure drop across those parts to create a hydrodynamic bearing inside the pump unless the pump is built as a 'stiff rotor' (very rare). Most feedwater pumps are 'flexible rotor' design. So a slow rolling feedwater pump will have metal to metal contact and is slowing grinding itself into early failure. For reverse flow on the booster, it may be a simple overhung pump with an open impeller and have no wear rings but reverse flow can overspeed the electrical motor and cause the rotor to become damaged. Motors may only be 'braced' inside for 125% speed before they 'swell up'.

For cold iron where the system curve is flat consider a minimum flow bypass line. The pump still must pump in order to live. Look on the pump curve and you will see a minimum continuous stable flow curve. Operating left of that point will destroy the pump and slow rolling a multistage is a death spiral for sure unless the pump is a very stiff rotor designed to run dry.

Summary: Don't calculate a poor design system flow. Just fix it.

edykes,

I've operated, designed and maintained power plants for 30 years and never seen any formula to do what you ask. Any detailed calculation would depend on the specific construction of the feedwater pump, including number of stages, type of staging (series, counterflow, etc.) and diaphragm clearance. Because of the complexity of the flowpath through the feedwater pump, it would be difficult to accurately calculate the headloss without several headaches worth of integrating the volume of each stage and the inter-stage connectors. It would also be necessary to account for frictional losses in both the pump and motor bearings, as well as the abnormal axial position of the pump impeller since the high pressure discharge pressure is not balancing the thrust on the shaft. I suppose you could use turbine formulae, but many assumptions would be invalidated because impellers are not very similar to turbine blades.

If you're looking for a ballpark figure, you can simplify the process by only looking at the smallest cross-sectional area in the path through the feed pump. If it is significantly smaller than any other "orifice" in the path, you may be able to neglect the losses through the rest of the pump. If you can find that information, and the pump design warrants the assumption, the calculation becomes a pretty straightforward venturi-type fluid mechanics problem. You would need pump curves for the condensate pump to find the flow vs. discharge head for that particular unit. Also, remember to take account for the difference in height between the condensate pump discharge and the feedwater header connection into the boiler drum (rule of thumb: 1 foot of water = 0.44 psi of head).

A simpler solution may be to start the condensate pump and use your feed flow meter to measure the actual flow rate through the feed pump. If necessary, connect a low-range differential pressure cell to the flow instrument taps to get better accuracy.

We would experiment and find out except the plant is in design. Thus, we need to estimate. The thumb rule that we have used to date is 10 to 15% flow both reverse flow and forward flow, but no one can verify the thumb rule.

We understand the possible issues with bearings, etc. The situation is an abnormal condition, not normal operations.

We would prefer a full analytical approach, but just bounding the situation is OK, such as a calculation demonstrating that the flow would be 15% maximum.