Suction Energy Question

[patbaugh]

Here's the situation, I'm a non-mechanical project engineer on a job where we installed a surplus Goulds 3196 8x10-17 1780 rpm centrifugal pump in parallel with two split case horizontal pumps. (The other two pumps have been working fine before and after the installation of the third pump.) The 3196 made a tremendous racket when we first tried to start up (deadhead mode). Subsequant start ups were a little better but the noise still continued. The manufacturer was contacted and suggested we were attempting to run below minimum flows. Our team investigated further and determined that it wasn't a minimum flow issue. Everyone was also convinced it wasn't a NPSH issue either because our available was always greater than required.

Since I know very little about pumps, I started doing some internet research and came across the suction energy issue. Since the other engineers had never heard of this term, they dismissed the idea since no one had ever taken that into consideration before on other installations. Just for grins, I did some calculations and determined that our pump was in the very high energy range (SE = 280 x 10^6 with suction speed at 13,800) at a flow rate of a little above min. required. The corresponding NPSH Margin Ratio = 1.8. We did some flow tests and the pump worked best with our valve open around 50%. It made noise but it was almost tolerable. Opening or closing the valve just a bit increased the noise significantly. I did some more SE calculations and the only thing different about this flow was that the suction energy (273 x 10^6) and suction speed (13,470) dropped only at this setting, it was fairly constant at the other flow rates. The NPSH margin ratio was lower of course (1.4). Anyway, my question is could this suction energy issue really be the source of our problems? The other engineers feel that since it made noise while deadheading, it couldn't be a NPSH issue at all.

Thanks in advance,
Pat

 

[Artisi]

At this stage you need to forget about academic discussions regarding suction energy etc and get down to basics regarding the pump suitability in conjuction with the installation and compatabilty with the other 2 units.

How is the Goulds unit installed in relation to the HSC units?
What head is being generated by the HSC units in relation to what can be developed by the Goulds unit?
How do the performance curves of the 3 pump units compare with each other?
Is the inlet / discharge pipe work capable of the increased flows?
If possible can you post a sketch of the installation and the pump performance curves for the HSC and Goulds units - this will give us a lot of information and a chance to see what is going on.

 

[BigInch]

What other 2 pumps do you have running there?
How long did you run deadheading?

http://virtualpipeline.spaces.msn.com

 

[patbaugh]

Actually, only two pumps run at a time and during the test runs one of the HSC pumps (Goulds 3410) was taken offline. The third pump was added for future increased flows only, current flows are in the 3500 to 4000 gpm range. These pumps are in parallel. The pump only deadheaded a couple of minutes before the discharge valve (one for each pump) was opened. Here are the pump curves and a sketch.

Thanks,
Pat

 

[TenPenny]

Please confirm the impeller diameters.....if that 3410 has a 17" impeller, I'm not sure that your 3196 is ever going to pump anything.

 

[BigInch]

3410s disch at 310 ft and the 3196 disch at 230

Assuming suction pressures are roughly equal for all pumps, it could be a min flow issue. The 3196 discharging at Psuction + 230 ft is NOT going to want to flow into the other pump's discharge of Psuction + 310 ft.

The 3196 will tend be forced to back up on its curve by the 3410 pump's higher discharge pressure, but the 3196 can't ever reach the 3410's discharge head of Ps + 310, since its dead head discharge is Ps + 275.

The 3410 may have the ability to overrun out on the curve, depending on piping resistance it will tend to run down on its curve to the 3196's Ps + 275.

Since you are probably at high flowrates with two pumps, I'd say the piping resistance is also high and needs somewhere around the Ps + 310 to flow that rate and the 3196 has backed up on its curve and is at close to zero flow.

http://virtualpipeline.spaces.msn.com

 

[patbaugh]

TenPenny, yes the 3410 has a 17" impeller.

BigInch, the pump folks are out today. We will discuss your observations when they return.

Thanks everyone for your help.

Pat

 

[Artisi]

The 3410 and the 3196 are not compatable for operation on the same system. By superimposing the curves onto each other you will see that there is no point at which the pumps can ever operate together at a common head.

For the pumps to operate together you would have to throttle the discharge of the 3410 or reduce the impeller diam. to bring its peformance(in terms) of head to the same as the 3196

 

[BigInch]

Then patbaugh loses at least 35ft of head and he's got no flow through the 3196, so that's what I call back to square 1 with one pump.

http://virtualpipeline.spaces.msn.com

 

[KernOily]

Suggest you draw a system curve first. Then create a combination parallel-flow pump curve and superimpose that over the system curve. The two curves have to intersect at your desired operating point. This procedure will tell you right away whether your third pump can cut the mustard or whether you need a different pump.

In general it is not a good idea to operate hydraulically dissimilar pumps in parallel. It can be done, but it requires careful attention to design of the system, selection of the pumps, and setup of the controls.

 

[Artisi]

BigInch - Correct as usual, I was making the point (or attempting to) that the 3410 and the 3196 are incompatable in their current configuration on a common system.

 

[Artisi]

KernOily (Petroleum) I agree that drawing a system curve would be helpful and removes all guess work.

If the original poster can give the flow rate, discharge pressure and static head when the 2 x 3410 are operating - a system curve can be approximated.

Regarding parallel operation, you can only have a sensible combination parallel pump curve if the pumps have equal performance - which they don't, although it may be possible to run dissimilar pumps in parallel but at one duty point and only where the head of each pump is equal.

At any flow rate the 3410 is producing higher head than the 3196 except at 5000 gpm / 275 ft, well to the right of BEP for the 3410 and approx 2250gpm on the 3196 very close to minimum flow - but are we operating anywhere near that - as we don't have a system curve we are only guessing.

The above operating points for the 3196 is very close to minimum flow, as per the comments from the maunfacturer. Also, this operating point might correspond with the comments in the original posting re operating at 50% valve opening - but who knows.

A system curve will remove a lot of the guess-work.

Still think these units are incompatable for this operation.

 

[BigInch]

Artisi, no denying the credit is yours, just maybe a little better to have left it with the first paragraph. But I guess there is some hope that they can work at a little bit better than a 1 pump flowrate, if the control valve is mostly closed on the 1 pump flowrate and two CVs can be opened up substantially for the 2 pump flowrates to hold the required discharge head into the downstream piping lower than 200 ft. Looking at the original design and assuming all the curves matched previously, I just don't think a 3196/3410 will ever do very much more than can be done with one 3410 alone, but it might be possible in a pinch. Maybe you could get 7000-8000 out of the two, in lieu of 6000 gpm with one 3410, but head must be less than 200 or so feet to do it.

http://virtualpipeline.spaces.msn.com