Bypass / Recirculation for Centrifugal Pumps

[gilus02]

Hi;

As I had a hot talk with my friend over the use of bypass in Cent. Pumps I want to examine whether I am right or not?
Based on my knowledge I know three cases in which we can use bypass for Cent. pumps:
1- When the minimum flow is prone to be less than the minimum flow which was determined by manufacturer.
2- When we have a large and expensive pump and we usually want to start it with bypass. (honestly I don't know why?)
3- When we want to use a Cent. pumps for flowrate less than 5 m3/h, we choose a bigger one and with bypass we can reach to the rated flow. (I know the loss of energy but in some case this solution is economical and very attractive for managers in order to get rid of an expensive PD pump)

But a friend of mine has said that when a pump operates out of the API-610 defined operating region, and is rejected because of this deviation (let 40% of BEP flowrate) we can use a technical trick and use a bypass to push the rated point to the right of curve in the case that we cannot change the pump type or size.
I disagreed with him because I have seen just three mentioned cases in which we had used bypass and not for this one, but unfortunately I couldn’t represent any technical or standard-based reasoning.

I appreciate you if make me winner!

 

[LittleInch]

Gillus02,

I think you will win. Going through your points,

1) Yes
2) I agree with you, I can't see why you would want to do this. Compressors yes, pumps no.
3) Not sure about this one as I don't see why you can't get pumps with lower capacity than this. Not common maybe, but certainly available and why do you want to get to the rated flow if the loss of energy through the bypass is more than the increase in efficiency plus the higher flow??

4) Your point is basically the same as point 3, you're using a pump really designed for a higher flow and just wasting energy even compared to the lower efficiency at the flow you want. Why not just get the right sized pump??

Putting in a bypass in the case of item 3 & 4 is just lazy engineering and will cost a lot more in wasted energy than any saving in the pump cost...

My motto: Learn something new every day

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

 

[mimentzos]

2)see the attachment

 

[LittleInch]

This is for axial flow pumps, whereas the op says centrifugal. I appreciate the info, but note the conflicting info, i.e. start with an open valve / empty pipeline, but still needs a throttling valve.

My motto: Learn something new every day

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

 

[mimentzos]

Ι think that all your comments i have already mentioned with red line.

 

[JJPellin]

I regularly require an automated bypass on large, high energy pumps. If the flow control valve fails closed or the automatic back flow chop valve closes the pump will burn up in a few seconds.

Johnny Pellin

 

[JJPellin]

In an oil refinery, the range of flows required is sometimes too wide for any commercially available pump. A bypass may be the only way to meet the process requirement and still achieve acceptable reliaibility.

Johnny Pellin

 

[JJPellin]

For large reactor charge pumps, they need to start up on bypass and introduce the charge very gradually to prevent a process upset that might trip the reactor and shut the unit down.

Johnny Pellin

 

[LittleInch]

Although lots of people here don't seem to like them, you can get a similar effect by use of a variable speed device, either mechanical or VFD....

I can understand why it is sometimes required, by my preference is to avoid bypasses wherever possible. We all have our preferred way of doing things.

My motto: Learn something new every day

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

 

[JJPellin]

Actually, in the examples I am describing, a VFD will not work. Most refinery application are pumping from a feed drum at constant pressure to a fractionator or reactor at constant pressure. VFD's do not control well for widely variable flow at nearly constant head.

Johnny Pellin

 

[LittleInch]

Fair enough - I still think there are more elegant ways of accomodating a variable flow though - the key is what is the lowest flow that a pump will actually accept on a continuous and /or short term basis and then stick a control valve on the outlet with some sort of low flow cut off or use multiple pumps (I know that it needs to be looked wrt the system curve).

Sometimes variable speed will work, sometimes it won't - the OP was only talking in the most general sense.

My motto: Learn something new every day

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