Parallel pumps operation 1

[geliberman]

I've recently integrated a new pump at a boiler feed water plant that is in addition to three existing pumps.
Three old pumps are identical(2 motor driven and one steam turbine driven pumps) with 73m shut off head.
The new pump (Steam turbine driven) is smaller and have 58m shut off head.
When we use 2 steam driven pumps,(new and old) there is no problem, but when the third pump starts it will cause the new pump to work without flow and the pump check valve will fluctuate,and after a while the casing temperature will increase. the cavitation also occurs in the new pump.
I know that we have made a mistake in pump selection, but how can we solve the problem.
How we can change the system curve to use 2 old pump together with new pump?
is that possible to use ARC valve or it is better to use a control valve?

 

[BigInch]

It does not change the pump curve. Just like any partially closed valve in the discharge piping, or a reducer, or a smaller diameter pipe segment, it changes the system curve, increasing the required head needed to move any given previous flowrate. It increases the steepness of the system curve.

Since the pump curve doesn't change, that also means that either a new operating point will be created, one at a lower flowrate then previous, due to the flow restriction of the orifice, or if you can keep the same flowrate, a higher head will now be required to move it. One way to keep the higher flowrate is to speed the pump up, which will also tend to increase discharge head.

Those two combinations, adding an orifice plate plus increasing speed, will tend to deliver higher head at any given flowrate. You will need to size the orifice exactly right, so that those two combinations will not increase the head too much.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[bimr]

The addition of an orifice somewhere in the discharge line of the pump does not actually change a pump's performance curve. What the orifice does is to change the system head curve by increasing the slope of the system curve. Then, the same pump curve intersects the system curve at a lower capacity. The only way you can change a pump curve is by changing the impeller diameter, changing the pump speed or increasing the viscosity of the pumped liquid.

No system curve has been presented in this thread, but it seems that since the pump curve is flat with a lower head then the existing pumps, that there is no way an orifice installation will allow this new pump to compete with the existing pumps. However, if the pump performance curve was steep, an orifice may work because then the curve would move the discharge head above that of the existing pumps.

The addition of an orifice plate is probably more successfully accomplished in a pump system where a single pump is operating. With a parallel pump system, moving the system curve leftward is self defeating because the intent is obtain more capacity.

http://www.engineeringtoolbox.com/pumps-parallel-serial-d_636.html

http://www.engineeringtoolbox.com/pump-system-curves-d_635.html

 

[SeanB]

I agree 100%. A while back someone asked me to take a look at a couple of pumps and determine if they were suitable for parallel pumping. These pumps had very flat curves. And for the required flowrate the pumps would have to operate in the flat range. My opinion was that the pumps were not a good fit - for parallel pumping you want to have a curve that continually rises to shut off. If the curve is flat then a very small change in system head can cause a large change in flow rate. So one pump could be operating fine, while the other is actually at its shut off head. Exactly what appears to be happening here.
But a similar statement was made saying can we add an orifice? I didn't understand how an orifice would help the situation. The orifice increases the system head pushing the pump further back on its curve and further into the flat portion - where we don't want to be. For the pumps to operate well we need to be further out on the curve - which means system head needs to be reduced.

 

[Pumpsonly]

It is correct you don't change the original pump performance by adding an orifice plate at the discharge. But it reduced the head (pressure of) available after the orifice and therefore also the pressure at the common discharge manifold. It has the same effect as throttling the discharge valve immediately after the subject pump It is stated clearly earlier you need to increase the pump speed or the impeller diameter.

Whether it modify the pump curve or the system curve depends on which side of the orifice you are measuring the pressure.

The posting at the below link has a good discussion on this subject and how is it done.

http://www.linkedin.com/groups/SHARING-EXPERIENCES-Modification-pump-curve-4014613.S.76105661

I fully agreed with that some of you will argue that it is inefficient way of operating a pump. But then again only if you have better alternative.

 

[BigInch]

"Whether it modify the pump curve or the system curve depends on which side of the orifice you are measuring the pressure."

Sorry, but no. An orifice always modifies the system curve. Just like a fixed-position valve. Only a change to the pump, physical change, or speed change, can modify the pump curve.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[JJPellin]

You are correct that the orifice would change the system curve, not the pump curve. I will explain the reason for my statement. There is a major API pump manufacturer that regularly incorporates an orifice plate at the discharge of their pump. They test the pump with the orifice plate and plot the results as the pump curve. They probably do this to meet a specification requirement that the pump curve be continuously rising to shut-off, or that it has a certain percent rise to shut-off. This pump manufacturer treats the orifice plate as if it is part of the pump even though it is located external at the pump discharge flange. Obviously, this is an unusual way to address the problem.

This pump manufacturer is correct in the fact, that the addition of the orifice plate (if it is treated as part of the pump) will create a pump curve (pump plus orifice) that is steeper and continuously rising to shut-off. That is the effect I was attempting to duplicate. As many of you already noted, this would also require an impeller diameter increase or speed increase to offset the head lost at rated flow.

Johnny Pellin

 

[BigInch]

It is unusual, but if it has a pump part number, OK, I guess it qualifies as changing the pump curve. Maybe. Ya. OK. It has a pump part number, right?

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[JJPellin]

Yes. It is listed on the Bill-of-Material and given a part number as if it is part of the pump. The point is accademic. Adding an orifice plate can change the interaction between the pump and the system. This can be used to make the new pump behave as if it is a better match to the existing pumps. It may not be an ideal solution, but it can work. I have done it.

Johnny Pellin

 

[BigInch]

Yes accademic it is, as long as you are not analyzing the hydraulics of the system. When analyzing the hydraulics, it is a thousand times simpler to consider the orifice as part of the piping system, rather than working out and entering a new pump curve. Any hyraulics program suitable for professional work is easily capable of accepting an orifice plate flow element, whereas correctly working out a new system curve will involve considerable, error-prone manual work.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[JJPellin]

Actually, I find it extremely simple to analyze the hydraulics by treating the orifice as part of the pump. I add the orifice to my spreadsheet program which generates a new pump curve (pump plus orifice). I overlay the curves for the old and new pumps and adjust the turbine speed and orifice size to get the best match over the required operating range. I add the pump curves together for the combinations of multiple pumps running making combined curves. We do this all the time with our crude booster pumps which tend to be constructed in groups (3, 4 or 5) running in parallel for different crude rate conditions. Easy as pie.

Johnny Pellin

 

[BigInch]

Exactly what I mean. All you need to do with a good program is to define the in/out connection and enter one number, a Cv. Not even a curve. No point for me to spend more than 1 minute with it.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[Pumpsonly]

Treating the orifice as part of the pump is much simpler and logical. In the event of a need to buy a replacement pump in the future, you just need to specify the original head and flow required of the pump.
If you treat the orifice as the system component with the increased head, some one who is not aware of the history will will end up buying a pump with too high head because the orifice is removed with the pump...and the problem recycle itself.

 

[BigInch]

This is getting silly.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[Artisi]

Agreed, even the OP has stopped responding.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)