Are there any software to help select seal-type, seal-MOC, API-Plan, etc.

• Group: Shaft Seals at LinkedIn
• 
  
• Discussion started by SL Abhyankar (SLA) - Are there any software to help select seal-type, seal-MOC, API-Plan for application-parameters such as shaft/sleeve size, rpm, box-pressure, liquid-characteristics, solids if any, temp., etc. ?

Response by Mr. DominicDempsey (DD) -
Most major seal manufacturers will have some form of bespoke selection software. However, the reliability of the software is extremely variable because there are so many factors that one must take into account in order to produce a "good" seal selection. A few basic rules exist but by far the most useful tool is the experience of the application engineer making the selection.

Further by SLA -
I am glad at your mention that "...A few basic rules exist...". I guess, it would be a good idea to make a reference list of those few basic rules at least. Can you please volunteer ? I hope very much that interaction at this group may make the list more and more comprehensive. May I very earnestly request you to please set the ball rolling !

Further response by DD -
... the first step is to look at the industry, i.e. API / CPI / W&I, in conjunction with the machine type, pump, vessel, compressor.... this will determine the initial seal type... my way of selection is to begin with the most basic selection for a particular industry (I am going to use JC seal types for an example since this is who pays my mortgage but other manufacturers seals could be used in this example if preferred)... so if this were an API 682 application my first thoughts would be toward a 1648-2 seal (Type A, Arrangement 1)... I would then look through the application data / customer specifications and try to find reasons not to select this design... sounds strange but it is a good starting point... there can be many reasons to omit the design, i.e. customer prefers bellows seals, CAT. 1 will suit the pump better, pressure is too high (Engineered Special perhaps), speed is too high (sprung stator design), temperature is too high (Type C design required), solids in the flow (might want to protect the springs)... all of which would preclude the Type A design... if I can find no reason not to select the default offering then that is what is selected... I hope this is making sense...

Materials is then the next factor... and again can preclude a design because some seals are not available in all materials...

In terms of seal face combination, PV factors are worth considering although most applications are well inside the default material combination limits and as such are only really applicable in high pressure/high speed combination applications...

Inverting the seal (sprung stator) is used on applications where the face periphery speed is >25m/sec...

Be careful with some specifications with regard to temperature capability of elastomers because there are a few that require a 25 C buffer...

In applications where the true vapour pressure is >5 Bar(a) be careful with the selection of a liquid tandem (dual unpressurised - Plan 52) because there have been issues in the past due to vapour locking of the system... see SHELL specifications...

Plan 53A can be problematic above 10 Bar(g) due to N2 soak into the barrier fluid... again, it's not an exact science but something one needs to consider... hence the reason that SHELL use 53B by default...

There are lots of rules of thumb... too many to list... and new ones come about because we've all been "bitten" as the saying goes...

I hope this gets the conversation going and I look forward to other points of view on this... we should be aiming to learn something new everyday...

SLA - Dominic ! That's fantastic ! And I appreciate the closing "..I hope this gets the conversation going..." I am feeling tempted to convert this into an algorithm. Dialogue-boxes should allow better word-processing features, I guess.

Here is an invite ... would anyone volunteer an algorithm ?

Re.:- VSD for chiller pumps

19 January Query from Ferry Marfang -

I want to ask what data must be completed before installing variable speed drives on chiller pump, because I was afraid would happen abnormal conditions on the chiller when installing VSDs on the pumps. As additional information that the data are available only electrical measurement data. Because the flow meter and pump characteristic curve is not available. another question is how to choose the right inverter.

This is the technical specification of chiller and pump:

No	Parameter	Chiller 1,3,4	Chiller 5,6
		30GBN400	30GTN420
1	Refrigeran	R22	R22
2	Chill water pump	90 KW	30 KW
3	Chiller capacity	400 Ton Ref	420 Ton Ref
4	Power input compressor	45 kW	45 kW
5	Chill water temperature	-	-
6	In let	13 oC	13 oC
7	Out let (setting)	5 oC	6 oC

20 January, SLA wrote -

Is it possible that 30 kw mentioned for Chillers 5,6 is 'per pump' and 90 kW mentioned for Chillers 1, 3, 4 is for 3 pumps together ? That will make all pump and chillers identical.

The difference in refrigeration capacities 400 and 420 seems marginal.

In fact there is no logic for pump(s) in 400 T refrigeration capacity to be 90 kW whereas the pump(s) in 420 T (higher) refrigeration capacity are 30 kW (1/3 of 90 kW).

The VSD is never common. It is always dedicated to each motor. Selection of VSD is primarily dictated by motor-rating and the range of speed-variation required in an application.

Also, VSD is useful for energy-saving and to avoid too many pumps in an installation. A combination of having number of pumps and having also VSD may not help much. It may rather complicate O&M by the increase in number of engineering devices in the system.

20 January, Ferry wrote -

the reason I installed VSD on chiller pumps is due to the percentage loading capacity of 90 kW pump is maximum 60%. with this condition discharge valve closed 50% this is because the pressure chilled water in Air Handling Unit (AHU) is very high.

Here I attach a detailed layout diagram of chilled water systems.

With a record for the booster pump outlet after the header on the image attachment does not use anymore.

20 January, SLA wrote -

Your query was about method of selecting VSD. My answer was that selection of VSD is primarily influenced by the motor rating. If you have already installed VSD with due diligence, your query is answered by yourself.

Now when you say that percentage loading capacity of 90 kW pump is maximum 60%, I wonder whether the system was over-designed and in turn the pump also became selected with excessive capacity.

Since the pumps cannot be replaced as easily, your logic of installing the VSD seems logical.

24 January, Ferry wrote -

my intention to install a VSD on the pump is to reduce energy consumption at the pump, the current state of the pump rotation speed constant while the chiller cooling load vary. I want to ask for input consideration what should I look for when installing the VSD so that the VSD to work as expected.

28 January, SLA -

Since you have number of pumps in parallel and now you also want to have VSD, I wonder whether you can incorporate VSD in the common control panel, maybe, with a properly programmed PLC. That should save you from having VSD dedicated to every motor.

With a PLC it should be possible to link operation of the VSD to correspond to the variation of load on the chiller. With today's advancements in technology, everything that is technically logical should be possible ! Check with a smart PLC guy !

I know that in hydro-pneumatic systems (called also as pressure-booster systems) they do have more than one pumps, cutting in and out of standby pumps and VSD and pressure-switch mode all integrated using PLC. What is pressure-switch mode in that system will be chiller load in your system. Yes, parametric control of operation of VSD should be possible. In fact that will give best energy-saving.

On 24 January Ansari wrote -

Most importantly the motor should be compatible to be fitted with a VSD. Speed reduction involves lot of heat generation and a suitable means of dissipation of the same should be present to avoid motor burnt out.

Mr. Abhyankar: I hope, I am right. You can comment on this a little more.

28 January, Response by SLA -
Dear Juned,

You are right. As with any other device, VSDs also have their pros and cons. I just did google search on "Problems with VFDs". And I got "1,430,000" results in 0.45 seconds".... !

-o-O-o-

Sizing Pumping system for LPG transfer

Sizing Pumping system for LPG transfer

On 25 January 2012 15:32, farid farid wrote:

Dear SLA
Good day. I need your assistant to check my steps to size a LPG (liquefied gas petroleum) pump as I know u have a very good experience in pump include for LPG service. The intent of the pump is to transfer the LPG from bullet mounded LPG tank to the other place. I’ll tell you the steps in order for me to size a pump and appreciate for you to correct me if got mistake.
1. First step is I need to determine the flow rate. Generally, the pump flowrate for LPG should be around 30-50m3/hr. but this flow rate is depends on the client. Let say if the client want to transfer the liquid 70m3/hr, I should size the pump based on that flow rate right?
2. Determine the static head. For the open system for me no problem in order to determine the static head. For the pressurized bullet tank, I am not so sure since the bullet tank is under pressure and for me of course it will affect the static head. Can you help me on this since I am a little bit confuse on this

3. Determine the friction head
The friction head depends on the flow rate, the pipe size and the pipe length and fittings along the pipe. If I not mistaken, for LPG systems, the pump should be as close as possible to bullet tank in order to minimize the friction head. It is because we deal with the LPG which is high vapor pressure (around 10 bar) and easily vaporize.
4. Calculate the total head
Sum of the static head and the friction head both in suction and discharge pipe
5. Select the pump
Select the pump based on the pump manufacturer’s catalogue information using the total head and flow required as from the calculation above

I hope you can correct me if got any mistakes and add some more suggestion if you have.

SLA’s response on 26th January, 2012 -

Dear Farid,

You have chalked out the sequence quite good !

The crucial point is the time required to unload the bullet-tanker lorry. If you unload a 70 m3 tanker at 35 m3/h, it will take 2 hours to unload. Is this too long a time ?

By the way, when you mention "...the client want to transfer the liquid 70m3/hr, ...", the client does not know the flow-rate as 70 m3/hr. He knows the volumetric capacity of the tanker, say 70 m3. He would also know the time that unloading operation should take. If 70 m3 have to be unloaded in 1 hour, then pump's flow-rate should be more than 70 m3/hr. Note, 70 m3/hr and 70 m3 are two different things. 70 m3/hr is flow-rate and 70 m3 is volume.

The pressure in the bullet and pressure at discharge of pump, both will have to be higher than the vapour pressure of LPG. If LPG is to be unloaded into a pressurized container, the differential pressure needs to be converted into equivalent metres of liquid column (mLC). These should be added into static head. Such addition becomes necessary if one wants to draw the system resistance curve.

It is always safer and necessary to ensure that the pump is as near to the bullet. Important thought is that there should be no chance for pressure of LPG to drop close enough to its vapor pressure.

I hope I have answered your points.

Rated and Normal pump flow

Rated and Normal pump flow

By Mr. Bharat Kadu, process Engineer at Jacobs Engineering, Mumbai.

For single stage centrifugal pump, if my rated flow is 135% higher than the normal flow; then what about my BEP and pump actual performance in the field? I am not having pump curves, we are just deciding over the flow design margin as our desing flows are 20% higher than normal, add on that we have taken more 10% margin of desing flow. is worth designing pump this way?
I am quite worried about pump performance and efficiencies!

Lionel S. • If the rated flowrate is (too) much higher than the BEP, then:
* reliability falls (due to increased wear and recirculation)
* pump efficiency decreases
* power consumption goes up (increasing the flowrate by 10% increases the friction losses by 21% in turbulent flow).
* NPSH Required increases.

So I suggest you not to over-specify the pump's flowrate and specify only the design flowrate (normal flowrate + design margin, usually between 10 and 25% to allow for calculation errors - to be checked with your Process Design Basis/Criteria document).

The manufactuer will then select the closest model pump that fits these conditions.

After, once you get the selected pump curve from the pump manufacturer, check that the rated flowrate (which is the actual capacity realized by the actual, supplied centrifugal pump when running at the design speed and TDH) falls betwen 80% and 110% of the BEP flowrate (caution: this range varies from pump to pump). In any case the BEP should be between the rated point and the normal operating point on the pump curve.

S. L. Abhyankar -->

@Bharat Kadu - I notice that you are Process Engineering at Jacobs Engineering, Mumbai. Taking clue from that I would like to point out that your phrase "Is it worth designing pump this way?" is wrong. I take it that you are designing a system and are wanting to work out the purchase specification for the pump to work in the system.

The terms RATED and NORMAL flow have specific significance for the system-designer. To speak by numbers, say, the NORMAL system-requirement is 100 m3/h. But you would like the system to work for an increased output of 135 m3/h. So with 35% margin over NORMAL flow, you may specify RATED flow for the pump to be 135 m3/h.

Coming to specifying the head, if the static head in the system is 20m. If piping is selected for Hf = 5m at NORMAL flow of 100 m3/h, at RATED flow of 135 m3/h Hf will increase in square proportion. It will become 5 * (1.35)^2 = 5 * 1.8225 = 9.1125. Then RATED head will have to be 29.1125 m, say 29.5 m

NORMAL system-requirement is 100 m3/h, 25 m. RATED duty will be 135 m3/h, 29.5 m. Power requirement will be higher by the ratio (135*29.5) / (100*25) = 1.593 !

If piping is selected to have Hf = 5 m at 135 m3/h, at NORMAL flow of 100 m3/h will be 5 / 1.8225 = 2.7435 m. Then also -

NORMAL system-requirement is 100 m3/h, 22.7435 m. RATED duty will be 135 m3/h, 25 m. Power requirement will be higher by the ratio (135*25) / (100*22.7435) = 1.484 !

Capital cost of piping when selected for Hf = 5 m at 135 m3/h will be higher. But if piping is selected for Hf = 5 m at 100 m3/h, the Operating cost to the user will be higher !

Pump-vendor will offer pump to get best possible efficiency at RATED duty in purchase specification. But it may not be the BEP of the pump. Effect on performance in actual operation will depend upon where the RATED and NORMAL duties will be with reference to the BEP of the pump.

How to find flow rate of pump if flowmeter not available

How to find flow rate in (m3/h) of pump if flowmeter not available at field but I have some data : Different Pressure suc. & disc. is 2.33 Bar , dia pipe@ suc 72 in & @ disc 60 in.

Started by Ferry Ruana Suteja, Indonesia

(1) Johnny Qing Sheng Ke, Canada • Fery, A draw down test will tell you the flow rate if there is not flowmeter available if you know the wetwell volume. Or if you can find the pump curve ( by giving a call to the supplier or manufacturer), you can also read from the pump curve too.

(2) Anis Hamdani Zuberi, Pakistan • I have a chart for estimating flow from horizontal and inclined pipes of 2" to 12" diameter. If you ant I can send you the same (let me have your email address). This method of flow measurement is very easy in the field without the use of flow meter.

(3) Saqib Yaseen, Pakistan • If u have V2 i-e- output velocity then you can find using equation

Q = A x V

Q is Discharge/Flow output
A is the cross-sectional area at discharge
V is output velocity

and in lay terms, u can place a tub of say 5 gallons and use stop watch to calculate the flow.

(4) Fernando Bagüés, Madrid, Spain • Can you install any pressure differntial transmitter? maybe iti can be useful this way. let me know.

(5) Kevin Bushnell, Canada • This will give you close estimate-First to determine the condition of the pump as related to the curve. Confirm the impeller Diameter installed, suction pressure. Have a gauge installed between the pump discharge and a discharge valve ( note height of gauge if this location is elevated from the pump discharge ). With pump running close the discharge valve, take/note the discharge pressure. Then within 10 seconds slowly open the discharge valve ( so you don't hammer the system ) or shut the pump off ( with valve still closed ) then open valve.
Take that pressure, minus the suction pressure, minus the elevation of the gauge to the pump discharge. With what you have left convert to Ft or meters and see where you are on the curve, so 0 flow at what TDH, If you meet the correct point for the trim dia then you can take operating pressure and see wghere your flow is. If you are below where the TDH should be you have pump wear, so no way to be sure what your flowrates are. if the TDH is higher ( but would not be by much ) then you're good

(6) Kevin Bushnell • do keep in mind you will end up with a flow +- 5-10%

(7) S. L. Abhyankar on 12 December 2011 in response to a personal email from Mr. Fery –

Mr. Fery also sent a schematic as shown in the figure, (bottommost of the 3 figures).

Since you have mentioned differential pressure across the pump to be 40 psi, (28.13 m), if you have the curve for the pump, you would be able to read total flow.

Since the flow is bifurcated into two, one to the condenser and the other to the cooling tower, these flows can be estimated to be in proportion of the CS areas of the branches

M/s. Secure Meters, Udaipur have a method of estimating flow of a pump by a method of thermal differential. I guess, their method also cannot find flow in branches.

If you would like non-intrusive method, I think an ultrasonic meter would do the job.

I wonder whether CFD can simulate the schematic and then estimate the component-flows also.

Considering that the line to the condenser is dia 30 only, it seems to be a small set-up.

(8) Another email from Mr. Fery on 12 December 2011 –

Now I am on analyzing pump performance at an power plant. Pump type is centrifugal pump double suction with vertical shaft.

The case is I should to measure pump performance base on existing instrument that installed in pumping system (Circulating Water), so it's not base on design analysis.

The question is :

1. What kinds of Key Performance Index for pump performance ?

2. How I should to identification what kinds of instrumentation who significantly can use to monitoring pump performance ? (e.g valve , Piping, Level Control, etc)

3. What kinds of parameter to determine pump performance ?

(9) My response on 15 December 2011 –

I wonder what is the scope of your "analyzing pump performance". Are you analyzing it to find potential for saving energy-consumption or for extending MTBF (Mean Time Between Failures) i.e. for improving reliability or for optimizing pump-performance for the system requirements.

"valve , Piping, Level Control" are not instruments, because they do not give any measurement or reading. Pressure gauges are instruments. In your schematic you did mention differential pressure across the pump as 40 psi (28.13 m)

In a power plant, they would certainly have pump's performance characteristics of a pump in the cooling water circuit. That should be you base reference. Pump's performance characteristics would also include curve for Power versus flow-rate. You can check up what was the power required from the curve and compare it with actual power being consumed. Readings of actual power being consumed would be available in the power plant. This comparison would give hints for potential for saving in energy-consumption.

If you want to analyze pump-performance in relation to performance of cooling tower and the condenser, then you may require to know how much of total flow delivered by the pump is getting bifurcated - part flow to condenser and major flow to cooling tower. In fact I am to wonder why part flow of the pump's discharge (before cooling) should go to condenser. But it is so seen in your schematic. One of your question mark refers to such flow recirculating to the condenser through a dia 30 branch from pump's outlet.

Basically it seems to be a condensate extraction pump drawing condensate from the condenser. I hence appreciate the installation being vertical so that the pump's suction has adequate NPSHa.

I am also left to wonder why pipe size from cooling tower changes from dia 54 to 72. What is meant by "inc" ?

Also, cooled water should go to boiler feed. Why does it return to condenser ? But condenser will also need some cooled water for condensation. Possibly the dia 30 branch should be from the cooled water, say from the dia 54 pipe below the cooling tower. In your schematic it appears from the pump's outlet.

Please clarify your schematic.

(10) Reply from Mr. Fery on

Here I attach clarified schematic as your request.

This is middle figure of the three figures.

I hope you can help me to solve this problem.

it's very help full if you attach with the calculation method & result .

(11) My reply on 9 January 2012 –

Your latest schematic shows the system to be much simpler. That is good. I also notice that the pump discharges into the cooling tower. Is that open end of pipe accessible ? If yes, you can do Google search on "measuring flow at open end of pipe". You will get info on interesting simple (though not accurate) method of measuring flow.

(12) One link w.r.t. above reply is –

http://smallhydro.com/200910/small-micro-hydro/measuring-the-flow-rate-q-from-an-openpipe/

How to Measure your Water Flow Rate Q from an Open Pipe:

There are times when it’s necessary to estimate the flow rate from a stream constrained to flow in a water jet flowing from an open pipe. This flow measurement method doesn’t require us to have precision fluid flow measurement instrumentation, other than a straight edge and plumb bob. Flow meters or weirs would likely be more accurate, but sometimes all you have is a filled pipe with a jet of water streaming out. If that’s the case, then try this method. See topmost of the three figures.

(13) On 16 January 2011 I have sent a further reply as follows -

On further thinking I realize that in your system a horizontal open end of pipe is not available. Also pipe size 60 in. is quite large. I think your best bet is an ultrasonic flow-meter.

I am wondering whether you can use details from the nameplate of the pump which should give flow-rate and head for which the pump must have been ordered by the original buyer. Your problem then would be to check whether the system has caused any difference to the original order specifications.

I am curious as to why there is the bypass line provided. I do not see any chance of the pump running at flow less than safe minimum flow. Only one valve on this line is usually required. 2.33 bar must be the pressure-setting for this valve on the bypass line to open automatically, if the pressure in the 60-in discharge line exceeds this pressure.

You need to have a pressure-gauge on the 60-in line. This will give you some idea of system's operating Head in m. Then by referring to pump's curve you can estimate the flow.

For estimating flow by using power measurement and pressure-gauge reading, one can do some indirect calculation -

(1) Power drawn by pump's motor (kW) = (rho) * (Q in liters per second) * (H in m) / 102 / (Pump efficiency) / (Motor efficiency)

(2) Assuming rho = 1,

(Q in liters per second) = (kW) * 102 * (Pump efficiency) * (Motor efficiency) / (H in m)

Pump efficiency of a cooling tower would be of the order of 0.75 and Motor efficiency would be of the order of 0.92. Using these values in the above equation will obtain an estimate of the flow-rate.

Hope this helps.

-o-O-o-

If NPSHa is lower than NPSHr

If NPSHa is lower than NPSHr, What Changes can be made on Pump only, without changing flow rate?

Posted by Amir Hamdan, UAE at Pumps & Systems Group of LinkedIn

Responses –

(1) Venkitaraman Govindarajan, Singapore • Hi, in my view it's like cutting the feet to suit / match the shoes. NPSH r is the requirement of the pump. We need to ensure that the system is modified to meet this requirement. Hence, please focus on the system rather than the pump.

(2) Carlos Troyo, Costa Rica • Not much you can do on the pump without changing flow. Use the propper pump. Or instal the pump in a lower location were more head is available, and revise pressure drop on the suction pipping (again system design).

(3) Randal Ferman, LA, USA • It depends upon the specific pumping conditions and particulars of the pump inlet and impeller eye geometry. If the NPSH Margin Ratio (R) of NPSHa / NPSHr is not much less than 1.0 then it’s likely that some effective modification of the pump inlet or the impeller eye can be done. Pump manufacturers routinely make such modifications during testing to meet performance guarantees.

(4) Asif kundi, Pakistan • NPSHa must be greater than NPSHr. Going for changing geomatery and adding modification its better to reselect the pump type. And only that pump in the manufacturer range must be selected which NPSHr is lower or suits your requirement.

(5) George Nicolaidis, Greece • if you can raise the tank you're pumping from it may do the trick, or as said further up if you can lower the pump. If you're not prepared to lose capacity not much you can do... Just a thought that occured to me just now! How about pressurising the tank you're pumping from! That may do it!

(6) Peter Berghs, Calgary, Canada • See if the pump manufacturer offers an inducer that will reduce NPSHr. What fluid are you pumping? It may be possible to suppress cavitation by playing with partial pressures and injecting air or another gas into the inlet (<1%). As always in these discussions we need more info to provide accurate answers.
Or...Put a low head, low NPSHr booster pump in front of the unit you already have...this depends on if a low enough NPSHr booster pump can be found that meets your process conditions.

(7) Amer Hamdan, (He started this) • Fluid is Caustic Soda ,other pump is for EDC,

(8) Peter Berghs • I had an amine booster pump that cavitated...adding a very small bleed of natural gas into the suction (needle valve cracked ! turn) just before the impeller significantly reduce the amount of cavitation damage. The effect of the gas could be heard. Check with your process staff to see if you can bleed in some nitrogen or other stable gas into your fluid....this will suppress the formation of pumped fluid vapor bubbles...I am not very familiar with caustic fluids so I can't say for sure.

(9) saurabh sitesh, Qatar • what can be done is to increase the NPSH availability..this can be done by using a vertical can type pump( VS6).. the length of the can is designed such that the difference in NPSHa and NPSHr is minimum 0.5m.

(10) Hans Fontijne, Netherlands • Its difficult to give you a answer thats the solution of your problem
normaly you have to check the situation

what is the medium / material you have to pump
temperature / viscosity?

high viscoity and low temperature can gives a low NPSHa

the length and diameter, how many elbow, valves, strainers are fitted in the suction pipe, al that gives more frictionloss.
Old pipe and others can also give more frictionloss

what give the suction gauge and pressure gauge,
the suction gauge can say the suction pipe is or get smaller (blocked). you have to clean the suction pipe inside

check first what is the real flow and what is the flow in the BEP (best efficiency point), is the fow more to the right
you need a bigger pump whit a NPSHr

if the flow more left of the BEP and the NPSHa is lower than NPSHr you have to redesign the diameter of the suction pipe (bigger diameter) or place the pump on a lower level, or use a other type of pump a vertical submersible.

Do not forget suction or persure cavitation can damaged your pump, think broken shaft, baerings, leaking seals.

(11) sunil gupta, Mumbai • if your NPSHr is lower than the NPSHa than first upon you have to analyse the route cause of the available NPSHa,it might be due to poor suction condition that are to be incorporate due to Various losses occur at suction pipe line.so due to which cavitation occurs and your required flow will be affect.Also it depends on type of media you are using for pumping.
And if your suction condition is poor than you have to design the new valve & valve seat assembly such that the NPHSr of your pump will be lower.so that it would not affect your flow.

(12) Simon Bradshaw, ITT, NY • Probably the best fix is to fit a low NPSHr impeller, specifically designed for the operating conditions.

If the NPSHr is not much below the NPSHa and you are not pumping an aggressive fluid then profiling and cutback of the impeller leading edges may yield sufficient improvement. However there are negatives to doing this.

If the inlet design is poor then it may be possible to rework it to reduce the losses and provide more uniform flow into the impeller eye. However most reputable manufacturers will have already optimized the inlet designs. CFD is your friend in evaluating if worthwhile inlet improvements are possible.

(13) Yasser Abdalla Grundfos Egypt • A bigger impeller that can be installed in the same casing will have lower NPSHr. This could be available if you consult your pump supplier.

(14) Anis Hamdani Zuberi, Pakistan • You can do the either of the following things:

1. Insert a smaller diameter Impeller in the pump (if the duty point is towards the right of best efficiency point with the existing impeller diameter)

2. The Pump foundation should be lowered (below the suction tank) to increase the NPSHa.

3. If hydraulics of the pump permit then install a lower speed pump (it will have lower NPSHR than the existing high speed pump)

4. Reduce the number of bends, increase the Suction Pipe Diameter to increase NPSHA

(14) John Yatcilla, Netschz, Philadelphia, USA • If you don't want to lose flow by slowing the speed of the pump down, then look into making the suction piping size larger to help with the NPSHa.

(15) Bo Dixon, Colorado, USA • Have you thought about adding a "booster/charge" pump into the equation?

(16) Todd Vencill, Wyoming, USA • Randal Ferman is correct, you can modify the A and B gap some of the eye of the impeller to improve NPSHr and EFF in some cases However, not much. Depends on the style of pump and impeller. I would really look at the system first. If an Axial split case Multi-stage API-610 you can do some of this for sure. Most manufacturers will do this as Randal states to meet hydraulic tests and increase EFF a point or two.

(17) Amer Hamdan (He started this) • For your information, the design or elevations of tank ,suction line, elevation of pump cannot be changed due to project requirements, even a booster pump is not acceptable, the fluids are EDC and Caustic Soda, the tanks are open to atmospheric pressure, Now I have two questions: 1- Is the Inducer is a common used solution without any defect on duty point or pumps components? 2- both Pumps are API 610 ,is the difference between NPSHr and NPSHa accepted to be .6 M? Kindly advise.

(18) Randal Ferman • Amer, I believe just about every pertinent option has been offered by knowledgable pump experts in the foregoing comments. An inducer is just one of a variety of possible solutions. Concerning NPSH margin, even 0.0 m could be acceptable if conditions are known with a high degree of certainty. In my opinion you are at the point where you need to have an experienced pump professional directly involved.

(19) Amer Hamdan • Thanks for your advise Mr. Randal, however OEM of these pump solutions were not acceptable after Technical evaluation of End user, As follows:
To use smaller capacity pumps with low NPSHr, as per the hydraulics of these pumps
: not in all cases the smaller pump has lower NPSHr ,it is related to design of casing and impeller, ( we are talking about ship loading pumps) and this will not be ok due to loading time required for each ship in port as it is time limited for loading ship, and the storage tanks in and out flow rate is calculated to keep certain minimum liquid level, so small capacity pump is not valid solution, and minimum liquid level cannot be changed

To add an orifice plate on discharge line of loading pump will lower the NPSHa but it will decrease flow rate by 20% and this is not ok, due to loading time limitations of each ship.

To lower Pump n: Speed is not valid as well due to loading time.

Finally OEM rejected to design a new pump with low NPSHr as the lead time of spares for these pumps in the future will not serve End user maintenance schedule.

Mr. Randal I am so happy to hear that 0 difference between NPSHr and NPSHa could be acceptable, can you explain this creteria please?

(20) George Nicolaidis • How about installing a second pump in parallel? This means that the NPSHav. will be the same. you will need to lower the size of your impeller therefore giving approx half the capacity for each pump. Half capacity should mean that you are going to have a lower NPSHr per pump so you should be ok.
If you don't want to play around with the impeller size of your current pump it may be possible to use an inverter to turn its' capacity down. In conjunction with a second pump (perhaps also inverter driven) you could pick and choose on your operating point according to pumped liquid / temperatures / head etc
Are you perhaps having problems because of low ambient temperatures?

(21) Amer Hamdan • No Problems with Ambient Temperature.

(22) Simon Bradshaw • Randal is correct in that if you know your NPSHa accurately, then very low margins can be tolerated (he also urged you to seek further review, as do I). However there are a number of important caveats to this:

1) NPSHa can change as a system wears and corrodes, so what is ok when new may not be ok later.

2) At 0m margin, you already have a 3% head impairment within the pump and extensive cavitation. It takes only a very small further reduction for the pump performance to "fall off a cliff" and go into complete head breakdown.

3) Depending on the liquid characteristics and the time you operate at 0m margin, the MTBR of your pump will be reduced by some amount, possibly a lot due to high vibration and cavitation damage.

For these reasons most end users will require at least a 15% (or 1m) NPSH margin for hydrocarbons. For pumps handling water, the necessary margin for reliable operation can vary from 15% to over 100% depending on the conditions of service.

(23) Bo Dixon • Amer, if a booster pump is not "acceptable" by your client, "not being able to pump the EDC and Caustic" will certainly be LESS acceptable. Running 2 pumps, in series, may be a solution. Your first pump needs to be a slower running pump, with a lower NPSHr than your current NPSHa. Then that first pump can feed a second pump that will complete your discharge requirements.
If your customer WILL NOT elevate the tank or possibly keep a higher level of fluid in the tank, both creating a higher NPSHa, then try try the 2 pump approach.

(24) Amer Hamdan • Well, thank you all for your advice. it was realy interesting discussion. OEM accepted to design an impeller with respect to pump casing to come over this issue without changing Flow rate.

(25) SL Abhyankar • People have used Suction Inducers. But a suction inducer is like a screw ahead of the centrifugal impeller. In other words it is like series operation of a PD pump upstream of a centrifugal pump. The inducer generates the head needed to meet the NPSHr of the centrifugal (and maybe some margin also). Since the inducer also develops head, say 3m, 6m, whatever, on the same Q, it demands additional input power proportionate to Q*dH

Before thinking of using an inducer one needs to check whether the motor has as much margin in power.

Also since inducer is as good as a PD pump upstream, and since PD pumps are fixed discharge designs, (in turn they become custom-designed device) the inducer may not help for off-design operations.

Furthermore, inducer as a PD pump will have limitations of operating speed.

Inducer is an option but not a straightforward option.

(26) SL Abhyankar • Another option is to install the pump down, providing static head which will provide margin over NPSHr.

Most installations of LPG pumps and pumps for condensate extraction are "constructed" with margin over NPSHr inbuilt in pump-design. These pumps are typically vertical pumps with pump-assembly fabricated into a can.

(27) Manish Hingrajia, Germany • Hello Amer, u can increase suction pipe dia to increase NPSH a, this will reduce suction losses, if possible lower pump elevation w.r.t suction vessel or increase vessel elevation. If possible from design point of view u can add inducer to the pump to reduce NPSHr. In reade suv. Vessel pressure.

(28) Johnny Qing Sheng Ke, Canada • Amer, 0.6 meter is the number I used in my water pumping design. However, NPSHa-NPSHr can be less than 0.6 meters in your case, since you are dealing with EDC and NaOH solutions, both have higher boiling points than water does, which is the testing media used by the manufacturer to measure the NPSHr. The nature of the NPSH is to avoid excessive air escaping from the liquid, which leads to the collapse of the tiny air bubble at the impeller surface (i.e cavitation). So you can relax the 0.6 m to 0.3 m or even lower, if you know the temperature of the liquid is no too high compare to the manufacturer's testing water temperature.

(29) Sina Sanjari, Vancouver, Canada • select vertical barrel type instead of horizontal if the differences is high,in less values delte the strainer at suction(if it is possible) and choose the bigger size of suction line.

check the curve and analyze it ,inducer can be used but check the curve firstly.

there many many case study in this case which you can easily find them in web.

(30) Chip Prybylski, Chicago • You might be able to add an inducer to the pump. This sometimes has the affect of lowering NPSHr by about 30-50% in some instances. Other than that, you will have to modify the system

(31) Amer Hamdan • Now it is solved , thank you all

(32) Mike Lombard, Denver, USA • Randall- From what I recall, the use of inducers will narrow the operating range of some pumps, i.e. mechanical issues or cavitation occuring if pump is throttled back or operated close to run-out? I've been out of the biz for a little while but I recall inducers sometimes causing problems in these areas.

(33) Sudhir Kulkarni • You may try de-congesting the impeller eye by cutting alternate vane at suction eye. This will reduce NPSH-R, to some extent. Of course, you will have to consultant pump designer for this step.

(34) SL Abhyankar • Amer, Congrats that the problem is solved ! Or is it that you had too many comments ?! :-(

But if the problem is really solved, can you please share, what the solution has been.

By the way, I am skeptical of suggestion of Mr. Randal that one can work even with zero margin. Mr. Simon Bradshaw was more correct in saying small margins can work. But NPSH-margins is an issue, which prompted Hydraulic Institute to have the standard HI/ANSI-9.6.1

I wonder whether your problem can get any guideline from this standard.

(35)

Randal Ferman • Mike Lombard, I agree with your comment about inducers narrowing the operating range. And, ‘yes’, the application of inducers must be limited to sizes and/or energy levels for which the higher levels of unsteady radial loads from flow recirculation and cavitation are mechanically tolerable.

SL, This discussion thread begins with negative NPSH margin, so perhaps a typical NPSH margin simply may not be available. The statement that 0.0 m of margin is possible only serves to indicate the boundary from which an acceptable solution can be determined. For the reasons that Simon mentions, 0.0 m is probably not acceptable.

(36) Mike Lombard • Thanks Randall. BTW, I recall from my Johnston Pump days, the use, on suitable applications, of a low NPSHr 1st stage impeller on multi-stage vertical turbines.

(37) SL Abhyankar • @ Randal - Right, "..thread begins with negative NPSH margin..". So, what is not available needs to be made available. And the options typically are - (1) Improve NPSHa by modifying the system, there again different options such as (a) lowering the pump below the suction-line, (b) increasing suction pipe size, (c) eliminating any unwarranted bends, etc. (2) adding an inducer (3) having the pump redesigned to less NPSHr (4) Replacing the pump with one of less NPSHr.

On Option (4), in a cooling water system in a steel mill, there were four pumps running in parallel. All pumps were rated for same Q-H. Three pumps were high specific speed (2900 rpm) design and one was less specific speed (1450 rpm) design. The 2900-rpm pumps had high NPSHr, 12.5m at given Q. The plant engineer used to replace impellers every third month ! The 1450-rpm running in the same system had no cavitation-damage problem. Its NPSHr only 7m at given Q.

I was skeptical and still am, about your mention of zero margin. I mentioned HI/ANSI 9.6.1 for "how much margin".

Parallel Operation

From : Mr. Rajesh Patel on 12 th January 2011

Please explain me why we cant achive double flow with two same capacity parallel connected pump.

Response :-

If one looks at combined characteristics of two identical pumps running in parallel, at a given head, the flow on the combined characteristics will be double of flow of single pump.

But in a given system (shown hypothetically to illustrate the point), the point of operation is the point of intersection of the system curve with the single pump curve or curve of two pumps in parallel or curve of 3 pumps in parallel.

As can be seen in the diagram, at 32m head flow from single pump is about 80 m3/hr. Flow from 2 pumps is about 160 m3/hr and from 3 pumps is about 240 m3/hr.

But if you see points of intersection with a given system curve the flow from single pump is about 115 m3/hr, from two pumps in parallel is about190 m3/hr (which is not double of 115 m3/hr) and from 3 pumps about 260 m3/hr which is neither 3 times of 115 m3/hr nor 1.5 times of 190 m3/hr.

So the difference happens in the way one looks at the diagram, whether against a given head or whether in a given system.