PUMP MAGAZINE: Questions and Answers (111 - 120)


Question #111: James, - do you have any vibration guidelines for Progressive Cavity pumps?  Our pumps are used in the water treatment plant setting to transfer lime sludge.

Laith Hintz
Design Engineer
Advanced Engineering and Environmental Services, Inc.
Bismarck, ND

Although we have done considerable analysis on machinery at water treatment plants through the years, we do not have any specific vibration guidelines for Progressive Cavity pumps. We have evaluated a large array of pumps at both water treatment and wastewater management facilities.

I am not familiar with where you might find such guidelines, but I can refer you to someone who has considerable expertise with a wide cross section of pump designs. His name is Lev Nelik, who is President / Technical Director with Pumping Machinery, LLC based in Atlanta, GA. Lev been deeply involved with pump design and troubleshooting (including vibration analysis), and also he provides specific training on pump design and analysis. I am sure you can find assistance from Lev. I believe he can if anyone can.

James E. Berry, P.E.
Technical Associates of Charlotte, NC

            Dr. Nelik comments:

Laith, - I am not familiar with vibration guidelines specifically for progressing cavity pumps. There is a lot published and available on centrifugal pumps, including single stage, multistage, vertical, etc. From the rotary types, very little is published, both in open literature, as well as even via internal guidelines by the manufacturers. The names of some of the leading PC pump manufacturers are noted in your email, and, of course, there are others. Neither Hydraulic Institute, nor API, publish vibration guidelines for PC pumps, although they do cover centrifugal types very well.


In the absence of such data, the best advice I can provide is probably the same as implied by James, which they use extensively in their work as he noted. Such general guidelines are defined by various ISO spec, such as ISO 2372-1974, ISO 10816-3, ISO 2372-1974, and some others. Below is an example of one:


cid:image002.jpg@01C6FCD7.017EC050                 scan0035 



Keep in mind that two things effect vibrations in PC pumps, in the opposite ways: first, they are inheritably unbalanced, due to a so called “nutational” motion of the eccentric rotor in relation to a stator – such motion of the eccentric mass produces undesirable force; but secondly (fortunately) - they generally run at low speed, typically below 300-400 rpm. I have found that when a PC pump operates above approximately 400-500 rpm, vibration becomes a concern and readily “feelable”. It depends, of course, also on a pump size, length, number of stages, energy level, conditions of the support (which is very important for a PC pump, or for any other type for that matter), etc. - but 500 rpm seems to work well for me, as a rough guideline.


I must say that, generally, vibration-based troubleshooting is rarely done for PC pumps, although it is a vastly acceptable, known and important way of troubleshooting root causes of other pump types, especially centrifugal. Troubleshooting of PC pumps is typically more pragmatic and rudimental. One of the reasons is perhaps that there is relatively fewer PC pumps in the filed, as compared to a much wider population of centrifugal pumps. Vibration analysis of centrifugal pumps can help reveal and detect early issues with bearings headed to failure, unbalance, misalignment, blade pass, and numerous other issues. The issues with PC pumps tend to be less involved (less sophisticated perhaps a somewhat bolder statement), as far as application of methods of detection of internal faults, and thus vibration analysis did not enter the arena of PC world as wide as it did centrifugals. PC failures are more likely to be caused by dry running, stator chunk-out, overpressure due to bad (or sometimes absent!) relief valve, wear, and shear-off of the joint, etc. Rarely, the repeated cause of a failures, for PC pumps, are bearings, because other issues, as noted, usually show up first - while, again, it is a very common issue with centrifugals.


I hope this helps a little, and sheds some light on at least reasons why this subject is relatively obscure, although it still leaves your question open. It is an interesting subject, and I am forwarding it to the Editor of Pumps & Systems magazine Mike Riley, to consider publishing, as a discussion, and to ask our reader to comment, hopefully uncovering more additional sources, standards or published recommendations, as well as personal insights. In my own opinion, as a rule of thumb, given a lack of a specific guideline, I would use a 50% higher limit on PC pumps vibration acceptance level, as compared to centrifugal pumps. For example, if a centrifugal pump, based on its standard, results in, say, 0.2 in/sec velocity limit, I would use a 0.3 in/sec limit on a PC pump of similar horsepowers, and limit speeds to under 500 RPM. PC pumps would run rarely at speeds above 500 rpm, although they do occasionally, and I would then treat vibration limit on those as a special case. Even more importantly, and perhaps more useful, would be not the absolute level of vibrations, but trending. If a PC pump vibrates at 0.4 in/sec (a number generally considered too high for centrifugals), but pumps well and have been working fine for many years, taking periodic (overall vibration) trends may be all you need. When such trend begin to show signs of increase, it could be an indication of a problem beginning to develop. To avoid catastrophic failure, it might then be a good idea to schedule an overhaul or a repair. But if failures are frequent, you would likely find simpler explanation, even without a vibration analysis. For example, excessive wear may indicate incorrect (too tight interference between the rotor and stator. Too frequent “flaking” of the rotor may point out on improper coated rotor. Rubber chunk-out may be caused by dry running. These reasons are typically listed in troubleshooting guidelines by the manufacturers, and if these guidelines are followed, plus common sense, much trouble with PC pumps in field can be avoided.


Best regards,


Lev Nelik, Ph.D., P.E., APICS

President / Technical Director

Pumping Machinery, LLC



Note from the original question by Laith Hintz:

Thank you for the additional insight on PC pumps.  My original question on vibration guidelines for PC pumps stemmed from a general pump specification for a project that required vibration analysis for all pumps covered in that section (mainly centrifugal).  From your discussion, it sounds like requiring vibration analysis on PC pumps after installation may be useful but has not become common practice.  Thank you again for taking the time to answer my question. 




Additional input from Todd Brown, Moyno Progressing Cavity Pumps company:


Moyno does not have any published guidelines with respect to vibration values for our Progressive Cavity pumps.  Because of its inherent eccentric nature, the pump has natural movement, specifically as you near the suction housing and stator.  This situation coupled with the vast array of mounting arrangements, platforms and foundations, makes it very difficult to come up with a standard.  I know of no standard in the industry for Progressing Cavity pumps with respect to vibrations.  When asked, Moyno will perform vibration analysis on specific units on the bearing (pump, gearbox and motor) locations only.  I know of no instances where we have failed to meet this assuming the entire unit was rigidly mounted (i.e. not overhung).


Todd Brown

Moyno Pumps


Question #112:  Hello,

I am in Yellowknife, NT Canada and the temperature is dropping. How much flow do I need to keep a steel pipeline running at -400F (400C)? If I maintain a good flow (full pipe) will the line still freeze? I am currently running about 2500 – 3500 gpm.


Thanks, - if you can help me out or direct me to information about cold weather pumping.




John Carlsen

Canada Water, Inc.


Pump Magazine responds: we are forwarding John’s question to our readers, particularly to attention of those at pipelines and terminals. We would appreciate if you could take a moment offering your opinion and ideas, and perhaps share your methods of protecting pipelines from freezing in similar situations.


Let us know. Your input is appreciated in advance.


Lev Nelik

Editor, Pump Magazine On-Line


Blair Northen, Kinder Morgan Pipelines, Atlanta, Georgia comments:

We transport mainly petroleum products and don't normally have any water. I would suggest keeping the water flowing where possible and using insulation / burial of lines and heat strips on equipment that sees static conditions.


Otherwise - drain the water out!?


John Richards, Middlesex Water, New Jersey offers:

My climate here typically does not experience temperatures below 0 deg. F and not for extended periods.


I can not give you an actual flow rate that will work but for the most part if the water is moving it is above the freezing temperature.    Any insulation (soil, hay, foam, even snow in really cold temps) will help keep it above freezing.  Typically heat trace is not practical.  We have many instances where non insulated pipes are exposed directly to cold temps (particularly bridge crossings) and freezing has not been an issue for us.  A neighboring water utility has installed a small blow off on a pipe hanging from a bridge crossing a stream.  This “leak” occurs all winter and I am not aware of any freezing issues.  In some older cities in Europe, Berlin comes to mind, steel water mains are installed 12 feet above ground on supports without freezing.


In my water treatment plant, things are different.  The water flows by gravity and flows very slowly.  Where I have valves in open outdoor tanks, I need to keep the valves submerged under water all winter.  I learned the hard way what it takes to thaw out a 48 inch butterfly valve that is frozen closed with water behind it.  Keeping the valves submerged reduces the useable capacity but it is necessary to have any use of the tank.  For my indoor filter operations, the valves are wrapped with heat trace for the winter, it is inefficient and expensive but it works.  Once the tanks have frozen over, we need to keep the ice from pulling on the wires for the pumps; this is typically done with sledge hammers and manual labor.  I wanted to try a home made bubble system this winter but did not get around to it (maybe next year).  We also lower the water level under the ice because sagging ice is not nearly as destructive as heaving ice.


If you have any others questions or experiences you want to share, give me a call.  If nothing else we can complain about the water operators in Miami and San Diego for the next 3 months.  Good Luck!


Readers Feedback:

Thanks for getting me contacted with your network. Just for your info, we have been successfully running three Godwin pumps continuously as the inflow varies. The temperature was – (minus!) 42 with a 20 Km/hr breeze for a few days on site. It took some effort to combat freezing of the 1000M 12” steel pipeline. If anyone else ever asks about cold weather pumping, feel free to send them to me.  At  www.canadiandewatering.com  there is a lot of experience and expertise of fluid handling in extreme conditions.



John Carlsen

Canada Water


Question #112:  Good day Dr. Nelik,

Let me introduce briefly. I'm a sales guy working out of Italy for Peerless Pump and I would like to pose you a simple question due to the fact that many times I incurred in such cases.

As you may know we are fire fighting pumps manufacturers, and lately we are working on a project for fire pumps (diesel driven) that are required to be compliant to API 610. I'm not sure on the content of the norm but I was told that is applying to the process/chemical/petrolchemical pumps only.

Do you have any comment on why a fire pump should be API 610 compliant?

Looking forward to read from you.

Thanking you in advance for the time you'll be dedicating.

Best Regards,

Ing. Benvenuti Andrea
Peerless Pump Company
Torino - Italy

Dr. Nelik comments:
Mr. Andrea
, - as you know, fire pumps are typically split case or vertical turbine pump types, although other types are applied occasionally as well. They do not fall under a category of API-610, although are governed by a fire pumps spec, which is more stringent as compared to a similar pump not intended for fire duty. API-610 design is very tough and stringent, and applies, as names implies, to petroleum industry, such as refineries and other petroleum operations. Some other industries, however, have adopted API-610, to either complete, or partial, intent, recognizing the fact that pumps designed to API-610 specification are much more robust and reliable for tough applications. Example would be power generation industry, which invokes API-610, at least partially, for the boiler feed pumps – hot, high speed, demanding machine, with utmost priority to reliability.
API-610 covers a variety of technical issues, such as shaft deflections, nozzle loads, etc. etc., and thus not every pump manufacturer can comply.

But what sometimes happens is this. A typical refinery consists of two types of equipment – battery limits (where oil is actually being refined), and supporting equipment. A pump in the basement of the cafeteria, for example, supplying HVAC needs there, may never see situations as tough and critical as its brethren a mile away, in the battery limit area. But, a purchasing department may require a supplier to comply with the API spec, because the pumps are technically slated to the refinery. Fire pumps may come under similar considerations.

Thus, each application needs to be reviewed on its merit.

I am copying your note to Peerless folks in Indianapolis, who I know well. Pete Noll, for example, an old friend of mine, going back to years ago when we worked at Goulds. Pete has excellent background in pumps, and has worked and lived oversees, and should be able to provide you with further insight into this important topic. Andy Warrington is originally from Britain (although looks Italian), and thus has not been entirely spoiled by the American pumping preferences, and is indeed another excellent source for guidance and direction.

I hope this helps. Feel free if any questions. I might be in Italy this summer on pump troubleshooting assignment, and if in the area, will give you a call – I love Italian food.


Lev Nelik, Ph.D., P.E.

President / Technical Director

Pumping Machinery, LLC

Andrew Warrington, Vice President of Sales, Peerless Pump Company, comments:

Andrea - Well I am honored to be considered to "look Italian" by Dr Nelik. Of course Andrea knows that I do not dress like an Italian and therefore am always embarrassed by his countrymen's sense of sartorial elegance. I suspect that part of this is due possibly to an illicit liaison between one of my ancestors and some Roman soldier two thousand years ago or something like that?

We often meet this problem. In fact of course, API and NFPA rules are often even contradictory in many ways (e.g. materials where NFPA or at least UL or FM dictate a cast iron case and bronze impeller where most API specs would call for at least cast steel I would think). So it's a non sequitur to say I would like an FM/ UL approve API fire pump. Many try and we do include some of the API style requirements in our fire pumps but we (and I don't think anyone else) has ever made a fully compliant API UL/ FM listed fire pump.

When I was with SIHI for example we supplied many of their specialty pumps to refineries where the traditional API guys made the centrifugal pumps and for some reason they wanted a vacuum pump or a side channel pump or a fuel transfer pump. We would always fight the exception battle and supply something that was perfectly good for the application but did not fully comply with API (which by the way changes like the wind with all the new editions anyway). Usually we could win the battle but the oil companies were always reluctant as they were so used to getting API pumps.

Another example is when I was with SIHI again we sold ISO standard process pumps. A family friend was a very senior engineer with Bechtel. Again used to buying API pumps for the petro-business. He once told me how many API pumps they had bought. All Bechtel London's specs were written around API. Over half of the applications at least did not need API. He asked what they would save if they specified ISO5199/ ISO2850. I worked it out from the data he gave me - in fact they would save 2/3 of their spend on the pumps that did not need to be API. You know what they did? Absolutely nothing! They could not get off the API fixation.

Now in Peerless we meet it all the time and the solution is to ask why they need this or that feature. Our fire pumps built to our standard design are the best in the business at doing what they do - putting fires out. They are pretty much a time tested good design for that. They don't do too well supplying cracking reactors or pumping hot crude or finished petroleum fractions. Then again, that's someone else's business.

Anyway, I am off to put on my Armani suit and go out on the town.


Andrew Warrington
Vice President of Sales, Peerless Pump Company,
2005 Dr. ML King Jr St., Indianapolis IN  46202-7026
: 317 924 7225 Cell: 317 414 3758 Fax: 317 924 7333



It looks like Andrea got the help he was looking for, as he notes:
Thank you for your explanations and will certainly be happy to take you to one of our best restaurants!!

Best Regards,
Ing. Benvenuti Andrea


Question #113:  Dear Dr. Pump,


It is being told that the vibration levels specified in the 9th Edition ( V(filtered) = 0.67 X V(unfiltered) ) is by error and 11th edition is in the process of correcting the mistake. Is it so? Please elaborate.



Engineer, Export Services.

Kuwait Oil Company




Regarding the vibration levels being "wrong" or different in API 610, 9th edition than they will be in the next edition, here is the logic:


As published in the 9th edition:    vf < 0.67 vu,  for discrete frequencies


To be published in the 11th edition:  vf < 2.0 mm/s RMS or 0.08 in/s RMS


Obviously 2.0 mm/s and 0.08 in/s are 0.67 x the overall values of 3.0 mm/s and 0.12 in/s, so there is no change in the values between what is now published in the 9th edition and what WILL be published in the 11th edition.  However, there is a significant difference when one misinterprets what was meant by the simple vf < 0.67 vu. 


What has happened through some reported misinterpretations is this: the purchaser of equipment interpreted API 610 as meaning that the filtered vibration should be 0.67 x whatever actual overall vibration level was measured.  This meant that, with a very well manufactured pump that exhibited a very low overall vibration level, it could never pass the filtered values of 0.67 x the low measured overall level.  This was not the intention of API 610.  The intention is that the LIMIT for filtered vibration would be 0.67 x the limit for overall vibration.  The table in API 610 is a table of limits.  Because of the misinterpretation discovered, it was decided to revert to actual numbers for filtered vibration limits in the next edition.

This, by the way, is NOT the case for shaft displacement measurements and those figures have not been changed from the 9th edition.


By the way: the vibration limits published in API 610 are meant for performance testing in the vendor's shop, but they happen to work quite well as field vibration limits as well.


In addition to the above, I rarely rely on exact value as specified by spec, as in my experience the field situation differs considerable from theoretical. Filtered, unfiltered overall, rms, peak-to-peak, zero-to-peak, etc. can be very confusing to most folks at the plants, who do not use vibrations as the sole way of making a living. Normally, these days, reliability engineers are the same ones responsible for machinery reliability, mechanical, structural, vibrations, hydraulics, and other issues. Years ago, these topics used to be sliced around more people, but today, with plant’ personnel reduction, much fewer people carry more burden. Perhaps partly as a result of that, a more pragmatic (some exceptions are noted further below) field approach is to read vibrations by magnetic pick up probes (accelerometers) and convert the signal automatically by the instrument into velocity (RMS) values. While there should, theoretically, be differences on the allowables, depending on energy level, speed, etc., some general guidelines, in my view, are sufficient, and perhaps are as:


  • A. under 0.10 in/sec – fully acceptable
  • B. between 0.10 and 0.25 in/sec – normal
  • C. between 0.25 – 0.30 in/sec – cause of concern and more detailed analysis, but not a shut down
  • D. over 0.30 in/sec to 0.40 in/sec – to be considered as corrected during upcoming outage
  • E. over 0.40 in/sec – potential failure could be imminent


Or, for a more formal guidance you can refer to ISO specification:

CCF03182007_00002             CCF03182007_00003


These are overall. Individual harmonics (FFT) should be analyzed only when troubleshooting, such as over case (C), but below it I would not waste time. Only in such cases, a more involved study is invoked, and typically done by vibration specialists. At that time, preferences between velocity or acceleration, amplitude, time domain versus frequency domain, begin to emerge, etc., - but all within the realm of relatively rare, albeit important, group of situations. Perhaps 95% of field troubles do not invoke that level of detail, and, while remaining extremely interesting and fascinating, remains under the umbrella of somewhat academic mould.


Many critical machinery units have installed proximity probes, which read amplitudes, rather then velocity, picked up at the journals of the rotating shafts. These have more elaborate approach, taking into account electric run-out, etc., and fed into automatic plant system, which could trigger machinery shut-down in case of a problem, when vibration accedes a set value. Examples of such would be boiler feed pumps, and, true to your question, pumps at the refineries, for which, in fact, the API-610 spec was written originally for. But even for those, a debate over the exact value of the set points could questioned as important, as most practitioners are aware of what constitutes high, moderate, or low vibration.


As time changes, more elaborate schemes come out, but many of these are products of people continue painting the same painting over, never satisfied with what already exists. In vibrations, good data have been in existence for years, and charts, graphs and calculations are published by Vibration Institute, and other practical organizations. These reflect years of experience, and should be handled by vibration professionals. I doubt that a pump vibration is any more or less damaging to a pump in 1950 as it would be in 2007, and constant change of the acceptance criteria is, in my view, is a waste of time.


However, your question is a good one, and I thank you for that, and copying a President of a local chapter of Vibration Institute for in formation. John Visotsky will be speaking on the subject of vibration at the upcoming PumpTec-2007 Conference in Atlanta in September (24-25), and, I am sure, questions like yours would be of interest to discuss.


Lev Nelik,

President / Technical Director

Pumping Machinery, LLC


Question #114: 


My application is for a typical sanitary sewer lift station and force main design for small to mid size collection systems, say 6-inch to 16-inch force mains.


Quite often, our parameters require that the design pumps over a high point to the discharge manhole. We cannot construct gravity sewer from the high point many times due to many reasons.  In many cases, this high point governs the pump design at the desired flow rate, that is, there is enough elevation fall from the high point to the discharge manhole that the sewer will flow by gravity in the force main from this high point to the manhole.  Thus, my pump TDH design will limited to the length of pipe from my pump to the high point


The question is: what is the latest trend for the best (and safest-raw sewer) mechanism to use to break the vacuum at the high point to assure gravity flow and proper pump operation?  It seems that the industry standard is to use an air/vacuum valve, which I have seen used many places.  This seems to have worked, but I wonder about their application to be used over and over in a cyclic manner.  Are these valves more suited to be used in less frequent cases such as line filling and draining (not often) or for line breaks (rare) or for surge control?  Using them at the hydraulic discharge point of a force main makes the valve an important device that needs to work every time the pump operates.  Is there any one type of these valves that are better suited for daily use?  Is there a different way of breaking this vacuum now?


Russ Brink, P.E.

Engineering Management Incorporated

Lawrenceville, GA


      We have asked Chris Staud, a Senior Engineer with a City of Atlanta Wastewater Division, and a periodic contributor to Pump Magazine technical site. Below is Chris’ comments:




The City of Atlanta has a couple of relatively long force mains that pump raw and partially treated sewage for a couple of miles.  Several of these force mains go over hills and into dips and require air/vacuum release valves.  Almost all of the air release valves are located in pits and unfortunately some become “interesting” maintenance problems.


City personal retire and don't pass along where some of these pits are, or developers develop the land and change the entire look of what use to be a field, or your maintenance road disappears due to lack of use. Or, my personal favorite occurs when a developer buries your pit under several feet of fill. It is a challenge to find these manholes with their air-release valves especially if you don't know if you have the as-built drawings.  Opening the pits sometime create some interesting problems when you discover that a den of snakes have taken up residence in your air-vacuum release pit. Our people learn to move fast when confronted with a den of snakes.


I personally prefer high point stack vents if possible, as there is virtually no maintenance required. If however, you must put an air-release valve in, then Apco and GA Industry's have some impressive web site listing the virtues of their products.  As for an industry trend to go away from air release valves, I don't see one. It is unfortunately a high maintenance item that currently doesn't have a replacement. Gases still want to come out of solution when subjected to a vacuum.


Hope this helps.



Chris Staud, PE

City of Atlanta, Wastewater Division


Question #115: 


Dear Pump Magazine,

I have a question regarding the specific gravity limitation of a pumping fluid: Liquefied Petroleum Gas (LPG) - primarily mixture of propane and butane which is received from refinery FCC (Fluidized catalytic Cracker) unit. We have two-stage pumping:


1st stage:

Pump is a vertical centrifugal. Fluid is received in suction piping directly from LPG sphere. However, the impellers of the 3 stage pump are at 20 feet below from ground level in a canned structure to create the required NPSHA.


Suction Pressure in the piping: 50 PSIG (before entering the underground can)

Discharge pressure: 284 PSI

Rated Flow: 950 GPM


2nd  stage:

Pump is horizontal centrifugal pump. It receives same 950 gpm from discharge of the first stage(above) and discharges to underground cross country pipeline. We normally pump LPG in the specific gravity range of 0.535-0.545 at ambient 86 °F.

We recently got LPG from the refinery with specific gravity in the range of 0.345-0.485 for 5-6 hours. We do not have alarm/tripping of pumps right now on receiving low density of LPG. Hence, we felt the need to incorporate tripping of pumps on receiving low density since this is an issue having direct relation to the quality of product we are receiving from the supplier. Such a qualitative change may not be acceptable to the customer to whom we are supplying the LPG through cross country pipeline (1250 KM).


I would like to know if such a low density product is harmful for the pump operation, especially for the first stage pumping since we are having narrow NPSHA-NPSHR margin in first stage pumping. Also, if we receive low SG LPG, it indicates that the propane quantity may be higher than normal (and butane lower than normal) which results in higher vapor pressure since propane has higher vapor pressure as compared to butane.

The pump O&M manual indicates the SG range as 0.515-0.585 but does not write anything explicitly about possible problem in encountering low sp.gr. fluid. Please advice at the earliest.


Yours truly,

Somak Gandhi

GAIL Ltd., Jamnagar Gujarat, India




This is a multifaceted question. One involves low NPSH margin, which usually (not always) is not a problem in practice with hydrocarbons, because of the vapor-volume relationship. The second relates specifically to low specific gravity and possible effects on the pumps.  The API standards have always suggested larger running clearances for specific gravities below 0.7 because of loss of inherent lubricity and the consequent danger of wear part damage.  With good running material combinations, the lower specific gravity should not be a problem, especially in the situation described where the pumps were continuously running.  As a safeguard in these applications, it is beneficial to recommend superior running materials, such as graphite products (Graphalloy as an example) or other suitable nonmetals for stationary running parts (PEEK comes to mind as another option).  With this protection, specific gravity excursions are a non-issue.


I have also discussed your question with engineers I know at Colonial Pipeline company in US Georgia, from the viewpoint of the pump operators.  Pipelines move gasoline, oil and other products thru their pipelines, and utilize high energy pumps, including multistage pumps. Such pumps have long rotors, supported at the ends by journal oil lubricated bearings, and a center bushing (obviously product lubricated) to assist rotor support. Contact of the rotor to rings and bushings is always an issue, especially for low lubricity fluids, where fluid film in the clearances is weak and offers little support. Such ability to support the rotor in clearances is often characterized by a so-called Lomakin effect, i.e. ability of a bushing (or a wear ring) to develop sufficient fluid film to provide dampening and stiffening forces supporting the rotor from contacting stationary part. In such cases, occasional rotor to bushings contact takes place, a known problem and a challenge for the pipeliners. Self-lubricated materials, such as Grahpalloy, for example, provide significant benefit as a solution, to prevent catastrophic failures.


Lev Nelik

Pumping Machinery, LLC


Question #116: 

Does anyone know what C factor should be used for lay flat and rigid hose, or know where I can find that information? We are in the pump rental business and we need to calculate head loss in order to select the right pump for our customers. We have the option of using HDPE pipe, Quick Disconnect steel pipe (Bauer), Aluminum Victaulic and hose (rigid suction hose and layflat). The rigid hose has either nitrile or poly liner so the C factor is known. It’s the layflat hose that is unknown. There are a lot of commercially available TDH programs out there and I have researched many. I find them to be overly complicated for our needs and too expensive, so I wrote my own program. It works fine for what we need it for and the Hazen Williams formula is accurate enough for our purposes. Going Darcy-Weisback increases the complexity of the programming and I have neither the time, ability or need to go that route. I was hoping that someone would have some idea, any idea of a number that would work for layflat hose, 80, 100, 120, whatever. The key is that we don’t have to be dead-on accurate for our type of business. If you can suggest a number I would be more than happy to use it because I have had no luck to this point finding anything.


Pat Black

Engineering Manager







DFS FlowNet by ABZ had a good database on hoses and tubing but does not use C Factors. Some people found 0.000075 feet absolute roughness works good for calcs for new general purpose hose up to 4” in diameter. I would probably use 0.0001 feet or even higher to be safe for new stuff and higher if fouled by grit or slime.  I asked some of my colleagues also, and they did not find a value for a flat hose. Some other references do not state a roughness for hose. What did a hose manufacturer say? Unfortunately, some of them make good hoses, my often do not provide a roughness factor, but it is worth a call. I believe that hose is similar to HDPE pipe;  Nipak (a Driscopipe distributor in WV) recommended to use C=150 for HDPE. Vinyl hose would be similar to Polyethylene pipe and C factor may come close to reality on small pipes.  Seelye book lists the following on page 22-03:


For the Hazen-Williams Formula, for Fire Hose:


Extremely smooth:   C= 143

Robber Lined:       C= 125-140

Mill Hose:          C= 100-120

Unlined Linen Hose  C= 85-95


Please note that this info is taken from a book that was published in 1960 (originally 1945).  Elwyn Seelye was a Civil Engineer with over 35 years of experience when he edited this book. I do not know if PVC hoses were available at that time and believe that PVC would approach the C used by the HDPE manufacturers, C=150.


Are you evaluating an old installation, or are you considering a brand new design? If an old one, I can recommend some ways to assess the C-factor in a somewhat round-about, but pragmatic, way. You could easily check these values by installing two pressure gauges on a flat water-hose and just calculating the pressure loss this way.  You may need a “5 gallon bucket” to check the flow rates etc.




Lev Nelik, Ph.D., P.E., APICS

President / Technical Director

Pumping Machinery, LLC


Question #117: 

Dr. Pump,

Please advise the source of Asarcon-520 material.



Bill Bogdan

Crane Aerospace & Electronics


Our contributing affiliate, Luis Rizo comments:


Hi Bill,


Asarcon 520 was a lead impregnated bronze used for sleeve bearing in horizontal split pumps.  I have not seen it in use, since my days at Worthington Pumps in the late 70’s.  It was highly specified by the marine industry for on board ships, boiler feed pumps and fire pumps.  As you can imagine a lead impregnated bearings were capable of running for short times un-lubricated without the rotor disintegrating. The material was readily available in bar stock, however with the issues associated with lead contamination today, I do not believe it is as readily today.  I would call the local bearing or/and bar stock suppliers and see if a similar material is available under some other name. 


Years later, while I worked for Exxon as a Rotating equipment engineer we began using graphite impregnated carbon (Grapholloy(www.graphalloy.com/html/products.html?gclid=CK2M_Iuco5UCFQSsGgodwkGSkQ)  sleeve bearing for pumps and Thorlon for the slower running centrifuges.  We retrofitted some old packing pumps in hot oil services to mechanical seals.  These pumps, by design,  depended on the packing for shaft support.  In order to support the shaft and allow the mechanical seal to live, I installed steel encased Grapholloy encase in a steel sleeve with spiral grooves between the bearing house and the old packing box.  This eliminated the run out and allowed for the seals to run true and not leak.  The Graphalloy material can be used in plain stock or encased in a steel sleeve to add strength, depending on need.


In either case you must calculate the loads to assure that the bearing is appropriately designed and support the shaft loads and that it is not on a node in the flexural curve. 


I hope this helps,

Good Luck!

Luis F. Rizo, PE



For additional information on similar materials, also review www.pump-magazine.com/pump_magazine/q&a/faq1_20/faq1_20.htm (question #15)


Question #118: 

Dear sir,


I want to design the pumps handling LPG (liquefied petroleum gas).


The tank outlet from the top, not from the tank bottom.


My question is how to calculate the NPSHa for LPG pumps. I have a closed vessel with 12 bar pressure. Vapor and liquid at the vapor space are equalized. Liquid  level is above the pump suction centerline.


Is my calculation correct? Please go thru sketch and help me to calculate NPSHA.


Viswanathan Damodaran




Vis, - attached is your (modified) sketch. LPG gas is in the tank, pressurized to 12 bar(gage) pressure. The vapor pressure is 10.19 barg (I assume also given in gage units, not in absolute). If the LPG is at 37.8 deg.C, then the pressure in the tank would have been 10.19 barg, not 12 barg, i.e. the LPG is somewhat sub cooled, and the layer of gas, as I showed on the surface of the tank, does not exist.


npsha sketch

First, convert everything into same units, say meters. If there is pressure gage on top of the tank, it would show 12 barg as 265.5 meters, as shown by calculations, given the specific gravity of LPG as 0.50 (502.9/1000). At the pump inlet, pressure is somewhat greater – by the level of the full tank, less the 1 m rise from the floor, so it is 268 meters, assuming hydraulic losses in the pipe around 0.2 meters (you can calculate these for better accuracy).


The 10.19 barg (11.19 bar(absolute)) would equal to 228 meters, and thus the NPSHA = 268-228 = 40 meters. The pump you would select would need to have NPSHR of less then this NPSHA, - perhaps around 35 meters or so. You also may want to double check the units of pressure,  - are they in gage or absolute units? – or perhaps a mix (for example, I suspect your 10.19 bar is in absolute, not the gage units I used)? If not, convert to the same (consistent) units, and recalculate, following the sequence I outlined for you.


Note that in order for pressure in the tank to be 12 barg, temperature there would need to be less then 37.8 deg.C, otherwise, if you state it is in equilibrium with the gas, then the pressure would need to be 10.19 barg, not 12 barg. You need to check that, and redo the calculations as I showed. There are also examples of similar problems on our web site, under section Articles or section Q&A. Please also keep in mind that initially, when the pump first starts, it needs to pull the liquid up the pipe to prime the line, i.e. the pump needs to be either self-priming, or have some special ways for pushing the liquid in the tank thru the line, up, and then to the pump. This is sometimes done by pressurizing the tank with nitrogen blanket or similar means.


For more information, you are welcome to attend one of our Pump School sessions: either in the US, or in Israel. The schedule is at http://www.pumpingmachinery.com/pump_school/pump_school.htm



Lev Nelik, Ph.D., P.E., APICS

President / Technical Director

Pumping Machinery, LLC

Atlanta, GA

Tel. 770-310-0866

Fax. 770-350-9311

email DrPump@PumpingMachinery.com

web www.PumpingMachinery.com



Question #119: 

We got a tough problem, - can you help? - Suggestions on a good quality Degreaser? Our applications are wastewater lift stations, floating grease and such.

Jason Henderson

Russellville, KY


    Jane, this is somewhat outside my direct expertise, but I will forward it to some knowledgeable folks at the wastewater lift stations at DeKalb County,    with whom we work on pumps. They deal with a wide range of equipment, and perhaps can help you out with an advice.


Lev Nelik, Ph.D., P.E., APICS

President / Technical Director

Pumping Machinery, LLC


Mrs. Kisselbaugh,

FOG problems are always tough problems indeed. I have 65 lift stations and two treatment plants in my system and have been doing extensive testing of products for Odor Control as well as FOG removal and elimination. The link below is a natural solution that both eliminates grease and also reduces odor (most fat Oils and Grease cause odor). If you tell me what specific problem you are experiencing I may be able to suggest something better.




By the way I graduated from Woodford County high school in Versailles Kentucky J.



Merat Zarreii
F&T Division Manager
DeKalb County Department
, GA

Watershed Management

 Question #120: 

I know that cavitation starts when suction pressure drops below vapor pressure. Does a manner of suction pressure reduction matter?  For example, I can reduce pressure in front of a pump by pinching suction side valve, or – I can reduce pressure at the supply tank by pulling vacuum. What is a difference? As I see it, as long as the suction pressure drops to the NPSHR value, cavitation should start, no matter how I get this pressure. Until that point, as I also understand it, the flow should not change. Please help clarify.


Bob Carren

Chemical Plant



Bob – there some difference in a manner of low pressure creation at the pump inlet. However, there are other factors of interest to review to understand this issue in more depth. Take a look at one of the live hands-on exercises we conduct during our regular Pump School sessions: pump suction throttle versus vacuum - why different? - YouTubein this exercise, two tests are conducted: one dropping suction pressure in front of a pump by valve throttling, and another case by introducing vacuum at the supply tank. In this example, water is recirculated back to the supply tank (dash line version), instead of a more common situation where it is pumped from one tank to another:


PML.jpg  => would you (not) expect what is shown to the right – in both cases? (!)


Fig. 1  System illustration

http://www.pump-magazine.com/pump_magazine/q&a/faqq111-120/faqq111_120.htm(Question #120)



Let’s use this example as a Test Quiz (correct answer gets you a winning ticket to the next Pump School:



As suction valve Vs is throttled, suction gage Ps reads less pressure, and in fact gets below atmospheric. From what we know happens when suction pressure begins to drop, - a the flow and head remain constant for some time, until suction pressure drops significantly, when NPSHA reaches NPSHR, and then the total pump head begins to drop very quickly. However, as the video on the link shows, flow starts dropping immediately while the suction gage needle barely moves! – and yet, no cavitation is observed.


Alternatively, when we keep the suction valve open, but apply vacuum to the supply tank via vacuum control valve Vvac – the flow remains constant, as expected, until rather strong vacuum is reached, and a fully developed cavitation becomes obvious and strong, with flow (only then) dropping suddenly.


Can you explain why? – should, in both cases, the flow behave similarly – i.e. staying constant for a long time, and only starting to be affected at low values of NPSHA (low Ps).


P.S. This test is one of the standard exercises during our Pump School sessions – for schedule, visit the link above^


The best answers we received are published below, with a few minor comments following it:


Dr. Nelik:

I watched the YouTube video you posted for the pre-class problem, and I think I have the solution.


In the first experiment, the pumped flow rate decreased when you throttled the suction-side valve because closing the valve added dynamic head to the system. The result is that the operating point moved the left on the pump curve, and the flow rate decreased.


In the second experiment, the pump flow rate did not change when you pulled a vacuum on the tank headspace because the pump system is a closed system. The change of pressure in the tank affected the suction-side and the discharge-side equally, so no dynamic head or static head change occurred. The system curve did not move, and consequently the operating point did not move. However, cavitation became evident (air bubbles in the flow meter) because reducing the pressure in the tank decreased the NPSHa in the system to the point that the pump started to cavitate.


I’m looking forward to finding out if I’m correct when I get to your  Pump School training this week. See you then.


Jim Gagnon, P.E.
Senior Engineer

Cincinnati, Ohio



Dr. Nelik,

Here is my interpretation of the facts.


The system as it operates use the pump just to provide the dynamic head losses of the circuit (tank to pump and from pump back to the same tank - connections in the tank is at the same height so the static head is zero).


When the vacuum pump is put in service (CASE B), there is no difference in terms of system itself and the pump still delivers the same flow (no changes to the circuit - static head still zero and the circuit is not changed because all valves remained in the initial position). If vacuum is further reduced and gets low enough, then cavitation will start to take place (NPSHa approaching NPSHr) and at this point the performance will start to deteriorate and flow will drop.


CASE A is different because the system is being modified by closing the suction valve and it is not related to "classic" cavitation (at -10inHg, assuming that test was done with water at room temperature water still far away from its vapor pressure). Once valve starts being closed, system curve "travel" to the left on the pump curve reducing the flow rate and consequently increasing the head to compensate for the losses imposed by the partial closure of valve.


The further we throttle the valve, the lower will be the suction pressure and the system curve moves further to the left of the pump curve and at some point will eventually reach shutoff resulting in no flow. I could not observe in the video if there was any signs of cavitation for CASE A, but if it was present it was related to recirculation cavitation (not the "classic" one due to the explanation given above).


Generally NPSHr curves do not extend all the way to zero flow rate. At lower flow rates, NPSHr curve will start rising again. In this case (to the left of the pump curve, far from BEP) eddy currents begin to form at the eye of the impeller and initially no detrimental effect is observed but the eddy currents effectively reduce the flow area which leads to an increase in velocity of the liquid and consequently increase pressure drop. Thus the NPSHr increases. When the pressure drop is big enough (approaching liquid vapor pressure) the pump then will start showing signs of the "classic" cavitation due to the consequences of recirculation cavitation.


Respectfully, Rodrigo Cardoso


Bravo Jim and Rodrigo! – good work, excellent insight!


Normal flow control of the pump/system is done by closing or opening of the discharge side valve – almost never by its suction side. Closure of the valve increases the losses across the discharge valve, and its opening decreases the losses: new system curves are thus created which intersect the pump curve at a new operating points:




Fig. 2 Connection between the discharge performance (H-Q) and cavitation (NPSHR)


Now, once the discharge valves “moves” the pump  to a new operating point, its suction characteristic begins to change – requiring less NPSHR at a lower flow, and more at higher flow:


Fig. 3 Development of cavitation


Keep in mind that a Pump (differential!) Head is a difference between Discharge Head and Suction Head. If the same pressure reduction is applied to the supply tank and delivery tank (or as in our example – the same tank), the differential head (pressure) does not change: added vacuum cancels out on both sides. Both suction and discharge gage readings change – but by the same amount.


However, if only suction side is affected (suction valve closure), but the discharge side is forced to remain the same (the pump discharge side sees the same pressure due to the same tank level), the differential does change, - i.e. pump head (differential pressure) increases, and so, according to the H-Q curve, the pump hydraulically moves to lower flow, - and at lower NPSH required, i.e. farther away from cavitation.


There is a still another little tweak to that, and at our next Pump School we will discuss how NPSHR curve also changes at low flow (as compared to how it is shown at the graph here), if certain design features (which ones) of the pump are modified: http://www.pumpingmachinery.com/pump_school/pump_school.htm


Lev Nelik

Pump Magazine On-Line

January-April, 2014





Back to main Q&A List