PUMP MAGAZINE: Questions and Answers (101 - 110)

Question #101: Dr. Pump, -

I have been asked by my boss to analyze the pumps we have here at our water plant. We have three low lift pumps that our located in our "Intake Building". There are several areas to read pressures in this building. They then pump a distance of approximately 1000 feet (vertical lift is approximately 45 feet) to our treatment facility where the MGD's are read by a magmeter where a pressure gage is located. The piping then splits in two, passing through static mixers and then splits further where they discharge into four clarifiers  under head. These low lifts are 125 hp single bowl vertical pumps that are variable speed. Our high lift pumps are 400 hp three-bowl VFD vertical pumps whose MGD's are read by a venturi meter on plant property. If you could direct me to literature that would explain the methodology of field testing these pumps so I could derive present pump curve s, efficiency curve and BHP curves, I would be greatly appreciative.  

Chris Harrington

Answer: Dear Mr. Harrington:

Testing pumps in the field is rather different as compared to shop test at the manufacturer facility. We do a lot of work with water and waste treatment plants, and it is often a challenge to have enough flow meters to catch certain pumps, and often relying on a single meter with side branches and take-offs, forcing assumptions and approximations. There is little in terms of field testing, but the main idea is not to predict the efficiency exactly, but to establish certain baseline, and then monitor regularly, to see the trend. Ultimately, what matters is the change in pump, performance, that necessitates action: rebuild, adjustment, troubleshooting, etc.

Getting flow data is thus not easy, not to mention questionable instrumentation, age, and condition. For example, if a plant flow is registered as 100 MGD, with 4 pumps operating, can we assume it is 25MGD per pump? We often find pumps at 20+30+40+10 = 100 MGD, with low flow one beating itself to death, while a high flow is running out of its NPSHR at high end.

Regarding power, it is almost never the case: and, at best, you trust your amps, and volts, but then you must assume the power factor and motor efficiency. Within the range of such assumption, you can be off much further then the desired accuracy, indented in the first place to do such study.

We can help in two ways: first, attending our Pump School training, which is a great primer on pumps, with a lot of theory and hands-on. During class, you would also have an opportunity to discuss your specific problem, and will learn a lot: both from the instructor, as well as via interaction with others, from different plants and industries. Second,  we can come to your plant and do (or help do) a pump survey as you noted, regarding pumps performance, reliability, energy and efficiency. We do a lot of troubleshooting work similar to this at other plants, and I think you will find it helpful.

By the way, having pressure gages – in front (suction), and after (at discharge side) of pumps is critical in troubleshooting, including intermediate points.

Dr. Lev Nelik, P.E.

Pumping Machinery, LLC

Question #102: Hello,

I'm Johan Goossens, sales engineer at Sterling Fluid Systems Belgium.

I'm looking for the specifications for a baseplate according API 610.

What is so specific on these baseplates? Were can I find more info how these baseplates should be build?

Are there standard suppliers for these baseplates?

Many thanks in advance.

Johan

Sterling SIHI (Belgium)

Answer:

Johan, - API 610 specifies standard dimensions for the pump baseplate. The main intent is to make sure the bases are strong and robust, and resist deflections and damage the best way possible. 9^th Edition requires the equipment to be liftable with the base during transportation, without the damage to the baseplate. There used to be problems with baseplates, and still are, when they are not manufactured sufficiently strong, mainly due to cost cutting considerations. API now frowns on such baseplates, and wants to make sure these problems do not plaque API pump installations. Proper grouting methods are also very important, and API mentions the requirements as far as grout holes sizes and locations, as well as vent holes, rounding of bed corners to prevent damage to grout, etc.

As far as supplies, my recommendation to you is to inquire from your pump users purchasing people as to their experiences and contacts for the local fabrication shops that specialize in engineered fabrications.

If we have any additional feedback from our readers, we will be happy to let you know as well.

Regards,

Lev Nelik

Pump Magazine

Question #103: Hello,

I have a cantex lift station about 40 feet in the ground that needs both isolation valves replaced.  The problem is the force main is 10 inch and approximately 2 miles long.  I thought about inserting a plug in the force main where it dumps out at the sewage plant but some people tell me this will not keep the water from flowing back once the line is opened. Others told be to freeze the line above the valves and this will keep the water from coming back.  When I looked into this it was quite expensive ($1,000 per inch of pipe size).  I could pump the wet well down as far as it will go and then hold up on the check valve to allow the force main to drain back, but I don't have that much capacity in the wet well. Any and all suggestions would be welcomed on how to keep from flooding the dry well when replacing these valves.

Thanks,

Mark Lehnert

Defiance, OH

Answer: Mark, -

I have recently come across a similar challenge where a failed isolation valve need to be removed, and the plant was having a dilemma of how to “isolate the isolation valve”, - not an easy task. Freezing is a possibility, and they say it takes several days for a freeze to take hold, although the cost is reasonable. You want to take it safe, however; if the frozen line fails under pressure, how will you stop the fluid form coming back? I recommend you invite a construction company to visit your site to see the specifics. For such projects, I would work with reputable companies, that have done such projects before, and can give you several references you can call to verify. I will let you know if we get additional feedback from our readers on this.

Dr. Lev Nelik, P.E.

Pump Magazine

Question #104:

At the outset, let me express my sincere gratitude to Pump Magazine for a very helpful website on pumps. My question is about a 10-stage pump arranged with in line impellers.

1)What is the importance of balancing disc and balancing piston and how does it help in running of the pump?

2)How the running clearances between theses two is contributing to abnormal operation?

3)Is it required to have a pressure gauge in balancing line? In case of increased value of pressure what should be remedial action by maintenance crew?

I will be highly obliged to get these answers.

Regards,

L.N. PRASAD

Indian Oil Corporation

Panipat Refinery, Haryana

Answer:

Balancing device is critical for multistage pumps with in-line impellers. Each impeller develops an axial thrust due to difference in area between the hub and shroud sides, on which differential pressures act. These thrust components add up to a large number, and must be balanced by something. Bearings would be too huge for that, and thus most of the axial thrust is balanced by a disk positioned passed the last stage, with some small residual then taken by the bearing of more reasonable dimension. Thrust devices vary in design – disk, drums, or a combination (called stepped disk). A pressure differential across the disk is huge – a full pump differential, and the low pressure side is connected back to suction pressure. Sometimes there is a valve installed to this line, which is often a cause of big problem – if this valve is closed, or gets clogged, the balancing disk will not have any differential pressure, and thus no thrust balancing capability. All thrust will then go to small bearings (or sometimes none) which will get wiped out and pump destroyed.

Knowing pressure in balancing line is critical, and a pressure gage should be there, and better yet also a pressure transducer, to signal operators a potential problem.

Lev Nelik

Pump Magazine

Question #105:

If a customer wants to know the shear rate of a centrifugal pump stage to determine the potential for emulsion formation, how can the shear rate of the pump stage be determined? What factors should be observed?  Tip velocity, impeller diameter, RPM, viscosity, etc...?

Thanks for the help,

Sean Williams

Application Engineer

Answer:

Shear rate is a du/dn, as you may recall from school math. Even though it occurs mainly at the boundary layer, it is typically approximated in pumps as velocity change across the clearance gap. For example, for an open impeller, calculate the velocity of the tip, and divide by the clearance gap. You can then also approximately determine a shear stress, torque, and power, as well as efficiency loss, although it is not usually done – all they do is compare two pumps, or two designs, to compare the shear rates, which would obviously have dimensions of (ft/sec / ft) = 1/sec

There is another type of shear – that is a general; turbulence within the impeller blade cascade, - which mostly characterizes a “chopping” action, rather then product degradation impact, such as polymer solidification within the narrow clearances of the gear pump, or a screw pump, etc.

I can also recommend attending our Pump School, - these types of topics are discussed, and more. The schedule is on our web.

Dr. Lev Nelik

Pump Magazine

A follow-up with notes:

Can you correctly estimate the shear rate by the following...

(1) Calculate the tip velocity of the pump impeller. - yes

(2) Estimate the difference of liquid/impeller velocities. - Impeller velocity minus wall velocity (which is zero) – impeller velocity across the clearance gap

(3) Estimate the energy dissipation rate. ((2) / (impeller diameter/2)) - (2) divide by clearance. For example, your impeller velocity will be something like 80 ft/sec and impeller to wall clearance on the order of 0.012” = 0.012 / 12 = 0.001 ft. Then du/dn = 80/0.001 = 80,000 1/sec

(4) Estimated shear rate equals square root {(3) / (2)} - not sure why you use this

Thanks,

Sean

More follow-up…

OK... The bigger the casing the impeller sits in, the smaller the shear rate.  Theoretically, if the impeller was sitting in open space with no limitations, the shear rate would be zero.  However, if it was sitting in a twenty foot tank, there would be a shear rate, just very minimal.

Additional response notes:

Yes. There is a small correction on this, however, as you are obviously digging in deeper into the subject, a good thinking. Remember, I told you first that shear rate is du/dn – and a first approximation is deltaU/deltaN = V / T = velocity of the tip / clearance away from the wall. You are right, - in an infinite pool of liquid this term becomes zero. However, technically speaking, du/dn was only an approximation – good enough from the engineering point in most cases, but not good enough from the scientists view (the difference between engineers and scientists is that scientists need a 100% data to state the conclusion, and engineers usually have 40% data and 2 hours to make a decision). So, the du/dn differential term is technically a velocity gradient within the boundary layer, with no-slip condition assumed at the solid boundary, i.e. impeller blade being a first boundary (moving), with no stationary boundary present (it is far away). Depending on the conditions, this boundary layer can be laminar or turbulent, with Reynolds number characterizing the more specific velocity distribution within the layer. There is, thus, a du/dn there, right next to the blade, but calculations of that are now more involved, and pump manufacturers usually do not get into this nitch of discussion. In practice, there is very little impact of this shear, since the remainder of the pool of fluid around it so huge. However, in some instances, theoretically, it may matter. For example, if a sheared fluid changes characteristics from benign to toxic as it transforms from liquid to a solid-looking sheared residue (like glue), then even a very small percentage of that stuff could spoil the whole batch. Or, of a very thin layer of polymer is deposited by a pump over a film strip, then a movie projected on a screen, filmed observer such film strip, will have blinking dots, sort of a 1930s-looking quality movie, with occasional imperfections of the deposited layer even visible to the eye.

What type of an application do you actually have? – does it fall into a category where a 0.01% phase-change contamination matter?

My application is electrical submersible pump in crude oil with high water cut.  I'm trying to find the shear rate, in order to find the viscosity of an emulsion compared to viscosity of "clean" oil. I want to estimate an emulsion factor (emulsion viscosity divided by the normal mixture viscosity at a certain water cut). This would be helpful to use in our pump selection program. Below shows how I got my question in previous email.  Does this way of calculating shear rate in a pump sound correct to you?

Estimating the shear rate in the pump:

1. 1^st estimate – maximum value

Tip velocity of the pump impeller:

Effective difference of liquid/impeller velocities (estimated)

Estimated energy dissipation rate – maximum local value:

The maximum shear rate is

2. 2^nd estimate – average value

This estimate is performed for an ordinary centrifugal pump with the total head of 0.6 MPa = 6× 10⁵Pa.

The power output is estimated as follows:

The loss of power inside the pump (for the pump efficiency of about 75%):

The volume of media inside the casing:

Estimated energy dissipation rate – average value

The shear rate is

Thank you,

Sean

Sean, - your approach in calculations from the fluid boundary value theory is on the right track. A good reference book on this is classical work on boundary layer fluid mechanics and heat transfer by Schlichting. My feeling is you may still have to do some testing, or find a lab to do such testing for you. To predict a mixture viscosity from the viscosities of its components is not that difficult, and there are charts available (I can send you one, but it is basically just an interpolation), - but some fluids change their behavior drastically when mixed, and the resultant viscosity may be very different from either components, and may not even have Newtonian behavior; it may become dilatant, or thixotrophic, with very different shear characteristics.

Dr.Lev Nelik

Pump Magazine

Feedback from the readers:

I agree with your last comments, Lev – I would define each of the fluids/mixtures rheologically using a rotational viscometer and then apply the pump using HI viscous corrections. If the reader does not find the HI corrections close enough for his purposes, then further refinement of the corrections would probably have to be done empirically. Determination of shear rate in a pump seems to me to be a job for CFD – there are too many assumptions that could be way off. One thing that I don’t see discussed directly is the shear stress (other than as a portion of input power). A pump impeller sitting in a 20 foot tank is a mixer – low shear rate, but prodigious amounts of shear stress.

Dan Roll, Finnish Thompson Pumps

(formerly with Goulds Pumps, Pulp and Paper Engineering Group)

Question #106:

Hello Dr Pump,

I'm having some difficulty with calculating acceptable suction/discharge flange loads. We have in our section (21) Worthington 10HN 27 horizontal centrifugal pumps that we suspect are being overloaded by the pipe work. We have modeled the pipe work and supports in Auto Pipe to determined the forces and the moments. As the casings are manufactured in a AS 2074 (Australian Standard) austenitic stainless steel, where the original material was a cast iron, we can not go by the manufacturers specifications. But on comparing these loads to the API 610 standard (both table 2-1A and appendix F), all of our values fall well below the acceptable levels. Due to the fact that this design of pump has been in operation for 20 years, leads me to believe that the API 610 standard is too conservative for our situation. Would I be able to compare our pumps with pumps of similar material and configuration as a bench mark? Do acceptable loads exist for these pumps in a similar material?

Our pumps

Pump type: Worthington 10HN 27

Casing Material: AS2074 grade C7A-1, Austenitic stainless steel

Casing Material (OEM): Cast Iron

Mechanical Properties:  Tensile, 430 MPa

                  Yield, 230 MPa

                  Elongation, 22%

Your expertise would be greatly appreciated.

Chris Jensen

Reliability Engineer

QUEENSLAND ALUMINA LIMITED PARSONS POINT, Australia

Answer:

Chris, - I have asked a colleague of mine, who is active with the pump API work, to comment on your question. I am including his answer, and also a few of my own comments:

It is important to go back to basics and review the intent of the API-610, in light of the logic I would like to offer here:

a) API spec originates and grew from the needs of refineries, with tough pump applications, high temperatures, high pressures, and particularly stringent requirements on reliability and safety. Within the intended limits of the spec, refineries usually apply steel and alloy pump construction, and iron does not fit within the intent of the spec. API specifically mentions steel and alloy steel in 5.5.1 paragraph of Rev. 9. Iron is mentioned on a side note, and is mainly allowed (reluctantly) to the auxiliary applications, outside of main battery limits.

b) Other industries, as well as non-critical services within the refineries, do apply API spec, but the trend is essentially an “afterthought”, and a desire of choice for other industries, engineers, maintenance and reliability personnel working at those industries, who found API a good thing, and feel that compliance with API would provide them with a good, robust, and reliable pump.

c) Pump manufacturers did conduct nozzle load tests for API-designated pumps that they sell, and, over the years, improved such pumps’ features. For example, many single stage end suction pumps have much more robust flanges at the 8^th edition, as compared to the 7^th, or 5^th, - i.e. a result of continuous studies, tests, and refineries desire for more conservatism, which they perceive (and rightly so), as an added safety feature.

d) Any pump manufacturer, that desires to produce pumps with equal or similar conservatism, as are the pumps qualified for API services may do so. For example, if a manufacturer makes a pump from iron, but makes it such that nozzles deflections are small when API-tabulated loads are applied, - good for him, although still not an automatic acceptance to the API world. Such (hypothetical) iron pump most likely would have much thicker walls and shorter stubbier feet as compared to the steel API pumps, - but no matter – as long as the pumps remain strong, with little deflection; - you got a first shot at being considered, and (maybe) accepted. Of course, the other intents of API would have to be maintain, and not only loads. For example, a pump made from the very very thick cast iron material, may resist deflections well, but would still be susceptible to thermal cracking when thermally shocked – i.e. a no good situation, given the fact that many refinery applications indeed undergo such shocks. So, the cast iron would likely not cut the test, and, at least a ductile iron would be used. But even that would probably not be good enough, as other criteria of API would likely still rule it (iron) out, from a practical view.

Thus a comparison question you asked has a simple answer: as long as a pump manufacturer can prove that their pump complies with the API spec, i.e. can withstand nozzle loads specified, have small deflections and small casing and feet distortion, low susceptibility to cracking, etc. etc., - it would then be in compliance in spirit, although may still not be technically accepted (for example, having steel or alloy as a clause in API, would rule the “wood” out, even though a pump may have a wooden casing  - one mile wall thickness). However, you make a good point of the fact that your pumps, being not in strict API compliance (iron?), lasted for years, successfully. Even if the loads you have a low in comparison to API, - perhaps this is why your pumps did indeed lasted that long, and thus you should perhaps best stick with these loads, and do not tempt them.

Having 10 revisions over the years, has not come lightly – many tests, and manufacturers and refinery users’ thoughts and concerns went into its gradual and thorough development. To relax the conservatism, at this point, would be unnecessarily opening a door to lowering reliability, and stepping backwards into history of the trend.

Few in the refinery business would risk such relaxation, and thus, my feeling is, you would have to stay with the API requirements as they are, and insist on the pump supplier on compliance.

Best regards,

Dr. Lev Nelik, P.E.

Pump Magazine

A comment from the API industry expert:

Hi Lev,

Regarding the question from Australia, the way it is worded is confusing.

He says the nozzle loads they determine from 'Auto Pipe' are well below the values suggested in API 610, and yet he suspects the 610 standard is 'too conservative' for their situation.  If he feels 610 is 'too conservative', I would think the loads they have determined would be higher than those suggested by 610, wouldn't you?

As I recall, the Worthington HN pumps are SSDS pumps.  It is my experience that the API 610 suggested nozzle loads for all between bearings pumps are indeed conservative - meaning that the pumps can most likely carry significantly higher nozzle loads without either  a) causing excessive shaft deflection at the coupling, or  b) causing casing deformation that would result in internal clearance problems.  In fact, the nozzle loads are conservative in most instances by at least a factor of 2.  Remember, the premise upon which the API 610 nozzle load values are offered is that the pump will be designed to accommodate at least twice the values without distortion problems.  This is stated in clause 5.3.3 of API 610, 9th and 10th editions, and ISO 13709.

It would be helpful to know more precisely what problems are being encountered by the correspondent.  The fact that the casings are now manufactured from austenitic stainless steel instead of cast iron should not cause a stress problem, but it might cause distortion problems because of a possibly significant difference in 'yield strength' or proportional limit, resulting in permanent distortion and closing of internal running clearances.

The question of being able to compare the pumps in question with pumps of similar material and configuration as a bench mark is also confusing.  Is he asking for acceptable values of nozzle loads from other similar pumps that you might know of?  Or is he suggesting using known nozzle loads from other pumps that he is familiar with and applying them to the pumps in question?

If that is the case, and he truly has known nozzle loads on similar pumps and knows that these pumps have a successful track record of trouble-free operation, then I would think he has the makings of good bench mark data.

From what I read, I get the feeling he has some known nozzle loads that are in excess of API 610 recommendations and is experiencing some un-named pump problems as a result.  If these are indeed SSDS pumps, I would assume the operating temperatures are relatively low, so I would question where such high nozzle loads are coming from!

Hope this helps.

Question #107:

I have been looking for a good book on Plunger Pumps and Havent found one. Do you know where I could find a book on Plunger Pumps?

Alejandro Flores
SPM FLOW CONTROL, INC.
Ft. Worth.