6 replies to this topic

[kkala]

1. A centrifugal pump (35 m3/h at liquid head of 338 m) was recently specified, to inject water of 40-50 oC into the crude line to desalter. The pump was fed from an atmospheric drum, resulting in available NPSH=11 m (NPSHa); but in the future feed could be from reflux drum bottoms of higher vapor pressure (due to dissolved H2S). Future conditions were not quite clear, yet Client wanted to facilitate this (future) possibility as much as possible, so he requested that NPSHa should have a reasonably low value, at any case not higher than 2.4 m. (vague NPSHa of a similar pump - 21 m3/h). Type of pump was not to change, remaining an horizontal, single suction centrifugal pump.
2. So the question was “What is a reasonable minimum NPSHa value for the pump of 35 m3/h & 338 m”. Using suction specific speeds (US units) of 11000 (good impellers) or 7000 (usual impellers), I concluded that NPSHa=2.0 – 2.2 m can be a reasonable minimum, realized by either a good impeller at 2900 RPM rotational speed or a rather usual impeller at 1450 RPM. Attached “NPSH.xls” indicates it.
It is noted that NPSHa was considered as 0.6 m (2 ft) higher than NPSHr and alternative current has 50 Hz frequency in Europe.
3. Nevertheless Pump specialists (out of Process Dept) had following objections on the matter.
3a. Procedure of NPSHr based on suction specific speed is oversimplified, available (extensive) pump data base contains only one horizontal centrifugal pump suitable, with NPSHr substantially lower (so price would be too high).
3b. Rotational speed of 1450 RPM may not satisfy pump requirements, let supplier decide on RPM.
3c. In brief, a small decrease in NPSHr may affect pump cost substantially after a low limit, since suitable pumps get limited in the market; this limit of “reasonably minimum NPSHr” cannot be predicted.
4. As a consequence we issued a pump data sheet with two alternatives (only one to be chosen): (α) for NPSHa=7m (it is standard practice not to specify NPSHa>7m ) and (β) for NPSHa=2.4 m.
However I wonder what is the “reasonably minimum NPSHa for the pump”, or what way should be used for it.
Even though the issue of “reasonable minimum NPSHa” is a bit “cloudy”, there must be a practical answer. Similar cases are often faced in plant.
In 1990 I had assumed a suction specific speed of about 13500 to estimate NPSHa of a liquid propylene pump (hence the elevation of propylene sphere) and this was then accepted by all supervisors. It concerned basic engineering (sort of FEED) for another company.
Any advice or opinion to clarify above query would be appreciated .

Attached Files

•  NPSH.xls   18KB   261 downloads

[breizh]

HI Kostas ,

Have a look at this link , hope it can support your query, especially the suction specific speed .

http://www.gouldspum...m/cpf_0008.html
Regards

Breizh

[chemsac2]

You can refer to an article in Hydrocarbon Processing titled "NPSHA - how much is enough" by T. Henshaw, October 2004, page 75-76. It has given a modification to suction specific speed based NPSHR estimation and author claims methodology to be approved by pump guru Yedidah.

During design when we do not have any idea of pump NPSHR, suction specific speed based estimates or values from pump database seem to be the only option.

Regards,

Sachin

[Zauberberg]

If the client specifies 2.4m for the NPSHa in future, make sure to confirm from which reflux drum, at what height, and at which process conditions the water will be sucked into the pump. Why should you just accept the 2.4m figure as it is? It could be over-conservative, it could be inaccurate, it could be too pessimistic. People from Operator environments like the simplified approach very much: if they know that certain system work at certain conditions, they will usually request some sort of replication for all other cases, and that is - based on your description - the case here. But that doesn't necessarily need to be true.

I wouldn't be that much bothered that with the Suction Specific Speed, pump RPM, etc. What I would do if I were at your place, is to approach pump vendors by having the following items accurately specified by myself:

- Flow (minimum, normal, maximum)
- Fluid conditions
- NPSH available
- Discharge pressure

Upon contacting pump vendors and receiving proposals, you will be able to compare different (workable) solutions and see which one fits the best to your application.

[kkala]

1. Sincere thanks for the aid so far. Link by Breizh indicates suction specific speed is a reliable method for similar issues and pumps with values up to 9000 (English units) can be easily found in the market. I am also searching Hydrocarbon Processing of Oct 2004, as suggested by Sachin, which could be of significant help.
2. Way proposed by Zauberberg is straightforward and used in almost all cases. But we had to follow a somehow reversed road in the specific case due to uncertainty of future conditions. Only present situation is clear, where the pump takes water from an atmospheric drum. After future revamp (apparently not fixed in detail) there may be also feed from sour water stripper, or from the boot leg (to collect settled water) of crude tower overhead drum. Any assumed base for future NPSHa would be so hypothetical (with manhours not paid), that the concept of ‘reasonably minimum NPSH’ was a kind of solution, just to result in less modifications. This was Client’s wish, although it might be more practical to replace pump at that time. However such practices (approaching a new case roughly and letting fine modifications later) can be met in plant revamps.
Indeed, it is worth while looking critically into given information – written data of existing similar pump (normal 21.3 m3/h, max 25 m3/h at 232 m head, shutoff head 232 m ?, NPSHa=2.4 m ?, NPSHr=2.28 m) looks ambiguous. Yet suction specific speed is calculated at ca 6730 (reasonable) and under present conditions NPSHa~11m (no problem). Apparently this pump was not initially specified to take suction from an atmospheric drum and is not a good “example” for the new parallel pump of 35 m3/h. It seems to run far from best efficiency point, since operating and shutoff (diff) heads are so close. But no more data can be obtained.
3. A simplified expression of the original query could be: NPSHr=1.4 m (that is NPSHa=2.0 m) is possible for an horizontal single suction centrifugal pump (of 35 m3/h at 338 m head), or this is an exceptional (expensive) case? I would say it is possible, since this corresponds to a suction specific speed of 11474 for 2900 RPM, or 5740 for 1450 RPM. The latter looks quite feasible to me. However Pump specialists say that pump of requested flow, head and NPSH can be hardly found in the market, according to available data base. So I understand that actually this combination of flow- head- NPSH is unusual (but why? pump can have two stages and NPSHr is a matter of how smooth / rounded the impeller eye can be), or the data base is incomplete. Any further help on the issue would be appreciated.

Edited by kkala, 05 September 2010 - 12:20 PM.

[Zauberberg]

Kostas,

If you want to get yourself deeper into centrifugal pump calculations, I have attached a workbook published by Joe Evans and also available on his website, PumpEd 101, related to Suction Specific Speed and Suction Energy. You may wish to try it and compare the outputs with your personal work.

I'd suggest you also to check the following link and download the PSIM (Pump System Improvement and Modeling) tool - I have used it on several occasions and found it very useful: http://www.pumpfunda...oftware-aft.htm

They also have another interesting article on the subject which I have uploaded before: http://www.pumpfunda...peed_primer.pdf

Explore Joe's website http://www.pumped101.com/ and I'm sure you'll find a lot of quality stuff. Whenever there is a pumping issue in my plant, PumpEd 101 is one of the first spots I check for more information.

http://www.pumped101...mance_analyzers

Attached Files

•  NSS Calculator.xls   408KB   183 downloads

[kkala]

Sincere thanks to all members having replied to the thread. I wish pump specialists (of the company I offer services to) had been so willing to help!
Link suggested by Breizh gives a quick answer: Use suction specific speed of 9000, reduce rotational speed if necessary.
Links by Zauberberg offer a basket of useful elements, including hydraulic institute recommendation of 8500 as suction specific speed limit to avoid cavitation, while "good" impeller designs can increase the value up to 11000 (article of J Chaurette).
HP article suggested by Sachim (chemsac2)deals with impeller characteristics and available NPSH; as per Vlaming's data (in a simplified form excluding few cases), NPSHa should be NPSHr of a pump having about 20% higher flow and a suction specific speed of 8500 (?)-- data limited to cool water and stainless steel impellers.
At any case a suction specific speed of 8500 would result in NPSHr=0.83 m for the pump in reference (35 m3/h), if rotational speed is 1450 RPM, thus NPSHa=0.83+0.60=1.43 m. This indicates the the value of NPSHa=2.4 m written in the bid is conservative for 1450 RPM.
Mentioned suction specific speeds are in English units. Some references use NPSHa in the suction specific speed formula, some other NPSHr. I think NPSHr is the correct value to use.

Edited by kkala, 05 October 2010 - 07:12 AM.