You are on page 1of 91

Technical paper index


SUBJECT Vol Paper Page
Affinity laws- what happens when you change the pump speed or
2 1
impeller diameter?

Affinity laws for positive displacement pumps 13 6

Alignment between the pump and driver 14 3

ANSI pump Standard 14 5

API Gland 15 9

API plans that I use 15 5

API standards, what's wrong with them? 12 5

Barrier or buffer fluid. The liquid we circulate between dual seals. 3 6 3

Bearing fit tolerances

Bearing seals 13 9

Bearings - analyzing the parts and wear tracks 5 3

Bearings - anticipated life and how they fail 5 3

Bearings - keeping the moisture out of them 4 12 3

Bearings - keeping the solids out of them 4 12 4 (1 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Bearing lubrication 14 01

Bearings - the main causes of overloading 4 12 1

Bearings - understanding what is meant by fatigue 4 12 1

Canned pumps 3 11 1

Carbon graphite seal faces - how they are manufactured 4 7

Carbon seal faces; which is best? Solid or pressed in a metal holder? 15 3

Cartridge seals - the major cause of overheating them 7 4

Cavitation 1 3

Cavitation - what causes the noise and damage? 9 10

Centrifugal pump formulas, rules and definitions 13 4

Centrifugal pumps - what is wrong with their design 3 10

Charts and graphs you will need

Chemical classification for easier sealing 2 12

Classifying seals by the operating conditions 4 3 1

Concentric dual seals 8 8 4

Condensate - why you must seal it 10 7 (2 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Controlling temperature in the pump stuffing box 4 6

Conversion tables 8 5

Converting head to pressure 1 5 2

Corrosion - problems with the 300 series of stainless steel 4 1

Critical speed - calculating the first critical speed 5 4

Dangerous fluids, how to seal them 13 5

Decision making 7 7

Density and specific gravity 16 2

Desirable features in any mechanical seal 4 3 2

Discharge recirculation - when to use it 3 6 1

Double seals - another name for dual seals 8 8

Double volute pump 14 6

Dual mechanical seals 8 8

Dynamically balancing the pump rotating components 9 1

Elastomers - selecting the correct O-ring for the application 4 9

Elastomers - temperature limits 4 9 2 (3 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Elastomers - where the special compounds do not work 10 6

Electric motor - the service factor 6 4 2

Electric motors - selecting the correct horsepower 6 4

Electric motors - the A.P.I. safety factors 6 4 3

Environmental controls 3 6

Environmental controls - controlling the temperature in the stuffing box 4 6

Environmental controls - an overview of the subject 3 2

Environmental controls - controlling pressure in the stuffing box 4 10

SUBJECT Vol Paper Page

Environmental controls - controlling the pressure and temperature in
3 2
the stuffing box

Environmental controls - how to use them 3 6 5

Face opening 10 4 1

Face to face dual sealing 8 8 4

Flow. How the resistance is affected by any change of flow in a pipe 16 6

Flow formulas 11 8

Flow through an orifice 13 12 (4 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Flushing 3 6 2

Flushing - usually the wrong choice of word 3 6

Flushing - when is it a good idea 3 6 2

Fretting 10 3

Fugitive emissions - the chemicals on the list 5 1

Glossry of pollutant terms 15 06

Glossary of seal and pump terms 8 8

Graphs and charts that you need

Grout for the pump base 7 5

Hard face - selecting the correct one 5 2

Head - calculating the total head in inch units 7 1

Head - calculating the total system head in meric units 14 10

Head - changes in the discharge head 1 1

Head - changes in the discharge head- Design problems 1 1 1

Head - changes in the discharge head- maintenance problems 1 1 2

Head - changes in the discharge head- operation problems 1 1 2 (5 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Heat - amount generated at the seal faces 1 4 1

Heat - amount generated by friction within the pump 1 4 3

Heat - how it affects the pump and seal 1 4

Heat - modifications that will lower the amount being generated in the
1 4 9

Heat - recommendations to lower the amount being generated in the

1 4 7

Heat - the affect on the liquid in the pump 1 4 4

Heat - the affect on the mechanical seal faces 1 4 5

High pressue and Mr. Bernoulli 16 8

Hot oil sealing 3 5

Hot water sealing 3 3

How wide is your vision? 5 11

Hydrodynamic gas seals 13 1

Hydrostatic sealing 12 2

Impellers, all about them 10 1

Impeller modification 12 6 (6 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Impeller, Open vs. Closed designs 14 2

Impeller shape vs. the pump curve 12 7

Installing pump piping 14 7

Installing pumps 14 8

Jacketing fluid 3 6 4

L3/D4 and premature seal failure 11 6

Magnetic drive pumps - their limits 3 11 1

Metal bellows seals 12 10

Mixer sealing 3 7

Monitoring the centrifugal pump 9 11

Net positive suction head (NPSH) 11 12

NPSH some more about it 15 10

NPSH required, allowable reductions 12 1

NPSHA - testing for it 16 4

NPSHR - testing for it 16 5

Non lubricant sealing - dry solids 4 8 2 (7 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Non lubricant sealing - gases 4 8 2

Non lubricant sealing - liquids 4 8 1

Non metallic seals 12 0

Non seal pumps 3 11

Orifice, flow through 13 12

O-rings - the special compounds 10 6

OSHA 1910 8 10

Oversized pumps 7 10

Oxygen problems in condensate 10 7 1

Ozone 8 4

Packing conversion - the advantages of cartridge seals. 9 4 3

Packing conversion - the advantages of mechanical seals 9 4 2

SUBJECT Vol Paper Page

Packing conversion - the advantages of packing 9 4 1

Packing conversion - why you should do it and the possible savings 2 10

Parallel piping for pumps 15 01 (8 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Piping Systems. What you should know 14 7

Piping systems. Increasing the diameter to save operating costs 16 7

Positive dispalcement pumps 12 3

Predictive maintenance for centrifugal pumps 6 11

Pressure - how to control it in the stuffing box 4 10

Pressure and vacuum sealing 5 7

Preventative maintenance - pumps 9 5

Preventative maintenance - pumps and seals 2 6

Preventative maintenance - seals 9 6

Priming the centrifugal pump 6 6

Pump curve - how to read one 2 3

Pump efficiency - figuring the heads 7 2

Pump efficiency - making the calculations 6 1

Pump efficiency - the affect of the oversize pump 7 10

Pump features. The ten most important. 15 02

Pump modifications to increase seal life 11 1 (9 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Pump Partnering 13 11

Pump rebuilding tips 15 08

Pump reliability 17 01

Pump standards - the problem with them 2 4 1

Pump standards - what is wrong with them? 3 10

Pump standards - what should be modified 2 4 1

Pump terms, using them 12 11

Pumps - additional information you need 11 4

Pumps - the best technology 9 3 2

Pumps installed in pits. How to correct some existing suction

14 12

Quenching - one of the environmental controls you need to know 3 6 5

Radial thrust calculations for centrifugal pumps 13 2

Repair of mechanical seals 5 8

Rotary pumps 12 3

Rubber bellows seal 11 11

Rules of thumb - pumps 2 7 (10 of 18) [12/22/2004 7:56:55 AM]

Technical paper index

Rules of thumb - seals 2 5

Rules of thumb -seals and pumps 8 6

Seal Application. Do you have one? 15 07

Seal cartridge - problems with overheating in some designs 7 4

Seal design - build components to compensate for operating conditions 4 11 4

Seal design - the questions you should ask 6 2

Seal face flatness- reading the optical flat and monochromatic light 6 3

Seal face hardness testing 12 8

Seal face lubrication 9 7

Seal face opening 10 3 1

Seal failure 2 2

Seal failure - analyzing the components. 4 11

Seal failure - chemical attack 2 2 4

Seal failure - damage to a seal component 2 2

Seal failure - preventing premature failure 10 4

Seal failure - some things you can do to prevent it 10 5 (11 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Seal failure - things that will open the lapped faces 2 2 2

Seal failure - things that will open the lapped faces 4 11 2

Seal features - the ten most important 10 11

Seal hydraulic balance - the conventional method 8 1

Seal hydraulic balance - two way balance for dual seals 8 2

Seal improvements in recent years 15 4

Seal installation 5 10

Seal installation errors 10 3 2

Seal life - how to get good seal life 2 8

SUBJECT Vol Paper Page

Seal life - why don't good seals wear out? 9 9

Seal metal parts - selecting the right ones 5 9

Seal misconceptions 11 3

Seal partnering 13 10

Seal repair - the correct procedure 5 8

Seal repair kits 4 3 4 (12 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Seal specifications 4 3

Seal value - how to get good value 7 11

Seal, preventing premature failures 10 5

Sealing hard vacuum 5 7 2

Sealing high pressure 5 7 1

Sealing products sensitive to a change in temperature or pressure 8 9

Sealing products sensitive to agitation 6 12

Seals - as supplied by the original equipment manufacturer- the

4 4

Seals - dual 8 8

Seals - preventing premature failure 11 5

Seals - the best technology 9 3 1

Seals - the most asked questions 10 2

Seals classification 2 11

Seals. Which one should you buy? 17 04

Self priming pumps 12 12

Shaft assembly - dynamic balance 9 1 (13 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Shaft deflection 6 5

Shaft deflection 1 6

Shaft deflection direction when operating of the BEP 15 11

Shaft deflection - operating off the BEST EFFICIENCY POINT

6 5

Shaft deflection - some calculations in inch and metric 10 8

Shaft deflection - three rules 8 12

Shaft displacement - common causes 4 11 3

Shaft displacement - the cheat sheet 9 2

Shaft displacement and original equipment seals (O.E.M.) 11 7

Shaft fretting and damage. The main causes 10 3

Shutoff head, how to estimate it for a centrifugal pump 13 7

Siphon, How it works 15 12

Slurry sealing 3 4

Specific gravity 7 12

Specific gravity - calculating for solids liquid mixture 7 12 2

Specific speed - how to calculate it 7 3 (14 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Specific speed vs. suction specific speed 9 12

Split mechanical seals 3 8

Stainless steel - exceptions to the rule 5 9

Stainless steel corrosion types 4 1

Standards 9 3

Standby pumps. Should they be run? 17 03

SUBJECT Vol Paper Page

Stationary vs. Rotary seals 11 9

Stuffing box 7 9

Suction recirculation 3 6 2

Suction specific speed 9 12

Suction throttling 16 11

System curve for positive displacement pumps 5 12 3

System curve - understanding it 5 12

System head - calculating both suction and discharge heads 7 1

Tandem style dual sealing 8 8 3

Technical term correlations 16 1 (15 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Temperature - causes of a rise in stuffing box heat 9 8

Training for pumps and seals 17 02

Troubleshooting positive dispalcement pumps 12 4

Troubleshooting pumps 1 2

Troubleshooting pumps - a new technique, the 5 whys 5 6

Troubleshooting pumps - analyzing the visible rub marks 5 5

Troubleshooting pumps - loosing suction 10 12

Troubleshooting pumps - maintenance practices 6 8

Troubleshooting pumps - not enough capacity 10 10

Troubleshooting pumps - not enough head 10 9

Troubleshooting pumps - operating practices 6 7

Troubleshooting pumps - selection 6 9

Troubleshooting pumps - using too many amps. 10 11

Troubleshooting seal leakage 4 2

Troubleshooting seals 9 9

Troubleshooting seals 8 11 (16 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Troubleshooting seals - a quick reference to common mistakes 8 6

Troubleshooting seals - a quick reference guide 4 11

Troubleshooting seals - an overview of the subject 3 1

Troubleshooting seals - problems with no apparent cause 4 5

Troubleshooting seals - selection practices that cause seal problems 6 9

Troubleshooting seal parts 3 9

SUBJECT Vol Paper Page

Vacuum pumps 14 9

Variable speed drives 13 8

Venting horizontal pumps 16 9

Venting vertical pumps 16 10

Venturi, flow through 16 3

Vibration in a centrifugal pump - causes and cures 2 9

Viscosity 7 8

Viscosity corrections for Centrifugal Pumps 14 4

Water horsepower - how to make the calculation 16 12 (17 of 18) [12/22/2004 7:56:56 AM]

Technical paper index

Link to the Mc Nally home page (18 of 18) [12/22/2004 7:56:56 AM]

Shaft deflection

SUBJECT : Solving a major cause of shaft deflection in volute type pumps 6-5

To understand the following paragraphs, you must understand three rules about fluids:

1. As the velocity of a liquid increases the pressure (measured 90 degrees to the flow) decreases and
as the velocity decreases the pressure will increase. This is the same principle we use when we
place a venturi in a water hose, so that we can spray chemicals on the lawn.
2. Pressure, working against an area, will cause a force. (Pressure x Area = Force)
3. For non turbulent liquid flow to occur the velocity of the liquid times the area it is passing through
must remain a constant

The following illustration describes a volute pump. It is called a volute pump because the impeller is
mounted off center. The impeller vane clearance is closest at the cut water and increases as you move
towards the discharge.

For this pump to operate properly the pumped liquid must move at a constant velocity around the
impeller, even though the volute area is increasing. Since the impeller area (at the outside diameter) is a
constant, the pressure generated by the constant velocity of the liquid will not cause any radial forces on
the impeller (rule #1). We control this liquid velocity by the design and speed of the pump.

Three possible conditions can be present:

Condition #1- The liquid is fed between the impeller vanes in just the right proportions, and there is just
the right amount of resistance, or head at the discharge of the pump to keep the liquid moving at a
constant velocity around the impeller causing a constant pressure at the impeller outside diameter (rule
#1). We call this "operating at the best efficiency point" (B.E.P.) and there is no unbalanced radial force
acting on the impeller, thrusting it in a radial direction

Now we will investigate two other common operating conditions (1 of 3) [12/22/2004 7:57:40 AM]

Shaft deflection

Condition #2 - The pump is operating to the right hand (high capacity) side of the pump curve with little
or no resistance or head at the discharge side of the pump.

As the liquid travels 180 degrees from the cutwater location it increase in velocity due to the lack of
resistance at the pump discharge. As the velocity of the liquid increases the pressure will decrease at
approximately 240 degrees from the cut water, causing a radial force (rule #2) to be generated 60 degrees
from the cut water (in the direction of shaft rotation).

Condition #3 - The discharge valve is shut. No flow is entering or leaving the pump casing.

For steady flow to occur the velocity of the trapped liquid times the area of the volute casing must remain
a constant (rule #3). Since the area immediately following the cutwater is very small, the liquid must
increase in velocity&emdash; causing the pressure to decrease, with a resultant force being generated at
240 degrees from the cut water. You will note that this is exactly 180 degrees from the previous force.

The exact points at which the forces will be generated is determined by the Specific Speed (shape) of the
impeller. Francis vane impellers (the most popular shape) deflect at approximately 60 and 240 degrees
measured from the cutwater, in the direction of shaft rotation. Radial vane impellers deflect at close to 90
and 270 degrees. Axial flow impellers deflect close to 180 and zero degrees from the cut water.

Any time a centrifugal pump operates away from its best efficiency point a radial force is generated that
will attempt to bend the shaft. This can cause a rotating component, such as a wear ring or mechanical
seal to contact a stationary component causing damage to either or both of them.

You can recognize the problem when you inspect the damage at the point of contact. There will be a
mark all around the rotary unit and a mark at either 60 or 240 degrees on the stationary component.

The excessive deflection can cause a lot of other problems including:

Opening up the mechanical seal faces as the rotating portion of the seal contacts a stationary
Overloading of the bearings, especially the radial bearing.
Damage to the impeller and volute.
Excessive wear ring wear and loss of pump efficiency as the gap increases. This is a major
concern with "vertical&endash; in line" designs.
Excessive shaft fretting (wear) at the bearing seal locations.
Damage to the bearing seals
Packing sleeve wear.
Excessive packing leakage.
Overheating of the packing.
Damage to the stuffing box throat bushing. (2 of 3) [12/22/2004 7:57:40 AM]

Shaft deflection

Damage to an A.P.I. gland disaster bushing.

The breaking of a stationary seal face.

Here are some things you can do to help reduce the deflection:

Shorten the shaft.

Go to a larger diameter shaft. You can do this by either replacing the present power end with a
larger diameter shaft or in some cases you can replace the sleeved shaft with a solid version.
Remove the packing and substitute a sleeve bearing in its place. The seal can be relocated
between the face of the stuffing box and the bearing case. Any time you get the seal closer to the
bearings you are better off.
Install a recirculation line between the pump discharge and a low pressure point in the system.
This will work for throttled applications if you are prepared to lose some of the pump's efficiency.
Go to a double volute pump design. The slight loss in efficiency is worth it.
If the main head is "system head" a variable speed motor would make sense.
Tell the operator to operate the pump at its best efficiency point. (Good luck with that one!)
You will notice that I did not recommend up grading to a different shaft material. Unfortunately
all of the common shaft materials have approximately the same modulus of elasticity, so they will
all have the same bending problem.

Link to Mc Nally home page (3 of 3) [12/22/2004 7:57:40 AM]


SUBJECT: Bending of the pump shaft 1-6

When a centrifugal volute type pump is operating at its best efficiency point (B.E.P.) the bending forces
are evenly distributed around the impeller.

If the pump discharge is throttled from this B.E.P. then the fluid velocity is changed and you will
experience an increase in pressure at approximately 240 degrees from the cutwater in the direction of
shaft rotation.

It also follows that if the pump capacity increases because of a lack of sufficient head then this change in
flow will cause an increase in pressure in the opposite direction, or at approximately 60 degrees from the

The following illustration shows these forces. (1 of 6) [12/22/2004 7:57:48 AM]

PUMP SHAFT DEFLECTION (2 of 6) [12/22/2004 7:57:48 AM]


The formula for the calculation of a multi-diameter shaft looks like this: (3 of 6) [12/22/2004 7:57:48 AM]


Y = Shaft deflection at the impeller center line measured in inches

F = Hydraulic Radial imbalance, pounds ( "P" in the previous calculation)
M & N = Distances from the impeller centerline to the steps on the shaft, inches
L = Distance from impeller centerline to centerline of the inboard bearing, inches
X = Span between bearing centerlines, inches
IL, IM, IN, IX = Moments of inertia of the various diameters, inches 4
E = Modulus of elasticity of the shaft material (psi.)

How much the shaft bends depends upon the length of the shaft and its diameter. The strength of the
shaft has nothing to do with this. The strength only determines if the shaft will break.

The following formula is a simplified version of this longer formula. It is the formula we use for a single
stage centrifugal pump with a solid, round shaft and an over hung impeller.

You will note that shaft strength does not enter into the formula. The important number is the Modulus
Of Elasticity (E), and as you will learn just about every shaft we use in the pumping business has the
same modulus.

F = The force we calculated in the previous formula plus the weight of the impeller.
L = The length of the shaft from the center of the inboard bearing to the center of the impeller.
3 = A factor used for an end suction centrifugal pump. A double-ended design would use a
different number
E = Modulus of elasticity; 28 to 30 x 106 psi. ( 0,196 to 0,201 X 106 N/mm2) for most metals with
the exception of Titanium
I = Moment of inertia for a solid, round shaft =

Substuting the Moment of Inertia in to the formula gives us:

Since the F is the same in both pumps, along with 3, and 64; and since the Modulus of Elasticity is just
about the same for all shaft materials, we can cancel out those terms and we are left with Y = L3 /D4 This (4 of 6) [12/22/2004 7:57:48 AM]


ratio then becomes a logical method of comparing two competing pumps that have different shaft sizes.
This ratio is often called the stiffness ratio, slenderness ratio, or the flexibility factor.

If we assume that the length of the shaft from the center of the inboard bearing to the center of the
impeller is a fixed amount (8 inches or 200 mm.) then we can easily see the affect of varying the shaft
diameter in the stuffing box area. The following table shows the relationship:

If you keep this ratio below 60 (2 in the metric system) you will not have too much trouble with shaft
bending. If, however, you do not have a low L3/D4 you will have problems with the shaft packing,
mechanical seals and the pump bearings. Keep in mind that we are measuring the shaft diameter. If there
is a sleeve on the shaft, do not measure the diameter of the sleeve.

Pump packing has a very poor memory and is not able to follow the bending or deflection of a badly
designed shaft. Some packings can be readjusted for the changing leak rate, but in almost every case
additional heat will be generated requiring even more flush water to remove the extra heat. If the packing
is not re adjusted with the changing of the pump discharge head then excessive leakage will follow, and
along with it&emdash; all of the problems associated with too much leakage.

A bending shaft can be deadly to a mechanical seal because it increases the opportunity for the rotating
part of the seal to contact a stationary portion of the pump, causing the lapped seal faces to open and let
solids penetrate. If we can keep the lapped seal faces together the seal will not leak and solids cannot
penetrate between them. The more shaft movement we have, the more likely the faces are to open.

Bearings are affected by the shaft movement in the same way they are affected by pump/ motor
misalignment. There will be an increase in the bearing loading and a corresponding increase in the
lubricating oil temperature.

A fourth problem with shaft movement is often overlooked. The pump has several critical tolerances and
shaft movement changes them. The most obvious are wear ring clearance, impeller clearance, bearing fit,
and seal face loading. These changes can cause additional heat generation, loss of capacity and loss of
efficiency. (5 of 6) [12/22/2004 7:57:48 AM]


In summary then, shaft deflection is certainly undesirable. If the L3/D4 is too high you will not be able to
operate in slight cavitation or very far from the Best Efficiency Point and in the real world that is not
very practical.

Be careful of most small pumps, they often operate at above 3000 rpm and their L3/D4 ratio is atrocious.

Link to the Mc Nally home page (6 of 6) [12/22/2004 7:57:48 AM]

centrifugal pumps, shaft deflection

Some additional information about shaft deflection 15-11

Whenever you troubleshoot a centrifugal pump, there are only two things visible to the observer:

Evidence of parts rubbing against something.

Evidence of damage to a pump component.

This knowledge makes the troubleshooting task a bit easier. About the only other thing we ever see, is
that the product has attached its self to the impeller or volute. Until this buildup throws the rotating
assembly out of dynamic balance that can cause rubbing and damage problems, or interferes with the
hydraulic flow in the pump, product attachment is not of much consequence.

Shaft deflection occurs as a result of pipe strain, misalignment, lack of dynamic balance, thermal growth
and the one we want to talk about now, operating off of the pump's BEP

In an other paper we discussed this type of shaft deflection in detail. It can cause rubbing and damage,
especially to mechanical seal and precision bearings. To be effective troubleshooters we must be able to
confirm the problem and then come up with a sensible fix to prevent it from happening again.

In that other paper I mentioned that Francis Vane impellers deflect from the shaft centerline towards
approximately 60 and 240 degrees depending upon the percentage of flow through the pump. This
direction is always measured from the cutwater, and in the direction of shaft rotation. In this paper I want
to get a little more specific.

Please take a look at the following chart. It will show you the direction of deflection, as a function of
capacity, for four different specific speed impellers.

Impeller specific speed Capacity, as a % of Deflection, from the shaft

number the BEP centerline towards:
2370 0% 35 degrees
2370 55% 40 degrees
2370 75% 60 degrees
2370 80% 85 degrees
2370 110% 140 degrees
2370 130% 250 degrees
1735 0% 65 degrees
1735 60% 89 degrees
1735 140% 300 degrees (1 of 3) [12/22/2004 7:57:53 AM]

centrifugal pumps, shaft deflection

785 0% 110 degrees

785 50% 140 degrees
785 135% 250 degrees
530 0% 135 degrees
530 90% 145 degrees
530 78% 180 degrees
530 40% 265 degrees
530 170% 320 degrees

If you are experiencing this problem, you will observe a continuous rub mark all around some part of the
rotating assembly that is running very close to a stationary piece, and a partial rub mark on the stationary
part. The bottom of the stuffing box is a good example of this type of rubbing. You will frequently
observe that the bottom of the stuffing box has become egg shaped as a result of contact with the rotating
shaft, although a dial indicator verifies that the shaft is straight. Remember the shaft is deflecting or
bending, it is not bent! You will see these same rub patterns on the wear rings used with closed impeller

Whenever you are pumping close to the pump's best efficiency point, the radial thrust is insignificant. It
never really hits zero, but it gets close.

There are a lot of reasons why a pump operates off its BEP and some of them are difficult to correct:

This is a standby pump that starts and stops a lot. At each start and shut down, the pump is
operating on either side of its BEP.
The process is always changing, due to fluid demand. The operator frequently opens and closes a
control valve to satisfy the changing needs of production.
The pump is being used as an accumulator to keep a head on a system. The pump is started each
time the level in the head tank falls to a predetermined causing the discharge head to gradually
increase as the level in the tank rises
We should be using a positive displacement pump in this application but PD pumps do not have
enough capacity for our needs. The head is constant in this application because we are pumping
into a pressurized container. The capacity is varying with demand.

You have a couple of choices if you want to lessen the shaft deflection problems. Pick the one that makes
the most sense in your application:

Go to a double volute design.

Purchase a pump with a lower L3/D4 ratio shaft.
Install a bypass line. When the control valve begins to throttle the discharge, the bypass will open (2 of 3) [12/22/2004 7:57:53 AM]

centrifugal pumps, shaft deflection

and recirculate the unwanted capacity to a storage tank. You should not bypass to the pump
suction because it can heat the incoming fluid.
If the head is mainly system or friction head, you could use a variable speed driver.
Use multiple pumps in parallel, and run them as needed to meet the changing capacity needs.

Link to Mc Nally home page (3 of 3) [12/22/2004 7:57:53 AM]

Shaft deflection calculations

SUBJECT: Shaft deflection and the pump best efficiency point. 10-8

We all know that L3/D4 is a convenient method of talking about shaft deflection and this number has
proven to be an accurate method of predicting premature seal and bearing failure. In an earlier addition of
this Technical Series I gave you the formula we use to calculate the force on the end of the shaft of a
single stage centrifugal pump with an overhung impeller. This is the most popular pump being used in
the process industry today. Here again is the formula we use to calculate the hydraulic force on the end of
the pump shaft:

P = The resultant force in pounds

K = The radial thrust factor. This number comes from a chart that relates to specific speed.
H = Total head at Q gpm. measured in feet.
D2 = Outside diameter of the impeller measured in inches.
B2 = Width of the impeller in inches.
Sg. = The specific gravity of the fluid
2.31= The conversion from feet of head to pounds/ square inch
Kq = A capacity factor equal to:

Q = The capacity in gpm at which the radial thrust is to be calculated.
Qn = The capacity in gpm at the BEP of the pump

As I have in past papers I will be working the numbers in both the imperial and metric systems. First we
will work the numbers in the imperial system and at the end of this paper we will make the same
calculations in the metric system.

I will use a direct conversion to metric to show you that the conversion works. In reality we would not be
using these exact numbers, but it is important to develop confidence in your ability to work in either
system. Because I am working with a direct conversion I will continue to use 1750 rpm or the numbers
will come out differently. I am well aware that your calculations will probably be at 1450 or 2900 rpm.

We are now going to use this formula to make an actual calculation of the shaft deflection on a typical
ANSI standard pump at shut off. This is a typical starting method for centrifugal pumps of this type. The
following information would have been read off the pump curve that came with the pump and a radial
thrust factor chart (K) that is shown in the Technical Series.

P = The resultant force (in pounds)

K = 0.37. from the chart (1 of 5) [12/22/2004 7:58:11 AM]

Shaft deflection calculations

H = 184 feet.
D2 = 13 inches.
B2 = 1 inch.
Q = 0 gpm at shut off.
Qn = 300 gpm
Speed = 1750 rpm.
specific gravity 1.0

Putting these numbers into the formula we get:

= 383 pounds

If we add the weight of the impeller estimated to be ten pounds, the total force on the end of the shaft
becomes 393 pounds.

Now that we have the total force, we will use this information to calculate how much the overhung shaft
will bend. To make the calculation we will use the following bending formula:

Substituting this term into the bending formula we get:

Y = The amount of shaft bending in inches.

F = The total force on the shaft.
L = The length of the shaft from the center of the radial bearing to the center of the impeller.
E = The modulus of elasticity. The numbers for common shaft materials will vary from 28 to 30
million psi. (28 - 30 X106)
D = The diameter of the solid shaft under the sleeve, if there is sleeve on the shaft.
I = The moment of inertia for a solid round shaft

If we simplify the formula we would get: (2 of 5) [12/22/2004 7:58:11 AM]

Shaft deflection calculations

Thirteen thousands of an inch bending is enough bending to cause problems with the impeller, wear
rings, mechanical seals and bearings.

The impeller could hit the pump volute or the back plate.
The stationary and rotating wear ring components could come into contact.
The shaft could it the end of the stuffing box.
The rotating part of the mechanical seal could hit the inside of the stuffing box. The rotating shaft
could contact the inside diameter of the stationary seal face.
The bearings could become overloaded.
It could cause excessive movement of both stationary and rotating seal designs.
Shaft fretting will be accelerated.

Here is the metric force formula:

Here are the numbers converted to metric:

P = The resultant force in kilograms

K = 0.37. from the chart
H = 56.08 meters.
D2 = 33.02 centimeters.
B2 = 2.54 centimeters
Q = 0 m3/hr. at shut off.
Qn = 68 m3/hr.
Speed = 1750 rpm.
specific gravity 1.0 (3 of 5) [12/22/2004 7:58:11 AM]

Shaft deflection calculations

Putting these numbers into the formula we get:

If we add the weight of the impeller estimated to be 4.54 kg., the total force on the end of the shaft
becomes 182.12 kg..

Now that we have the total force, we will use this information to calculate how much the overhung shaft
will bend. To make the calculation we will use the following bending formula:

Y = The amount of shaft bending in centimeters

F = The total force on the shaft in kilograms.
L = The length of the shaft from the center of the radial bearing to the center of the impeller in
E = The modulus of elasticity. The numbers for common shaft materials will vary from 1.96 to
2.1 million kilograms per square centimeter)
D = The diameter of the solid shaft under the sleeve, if there is sleeve on the shaft in centimeters.
I = The moment of inertia for a solid round shaft

Now lets put in the actual numbers and see how much the shaft will bend with 182.12 kilograms force on
the end of it:

F = 182.12 kilograms

L = 22.86 centimeters

D = 3.75 centimeters. (4 of 5) [12/22/2004 7:58:11 AM]

Shaft deflection calculations

Link to the Mc Nally home page (5 of 5) [12/22/2004 7:58:11 AM]

SUBJECT: The concepts you need to understand centrifugal pumps 8-12

In my seminars I talk about the three magic formulas you need to know if you want to understand how
centrifugal pumps function.

Here they are:

1. As the velocity of a liquid increases, the pressure, ninety degrees to the flow, will
decrease, and as the velocity of a liquid decreases the pressure, ninety degrees to the flow
will increase.

2. Pressure acting on an area creates a force.

3. Velocity times area must remain a constant if liquid is to flow.

Let's will look at each of these formulas in detail:

Formula number one explains how airplanes fly. It all started when the Wright brothers discovered the
correct wing shape for an aircraft.

Look at the following diagram. You will note that the air is flowing under the wing at some velocity. The
air going over the top of the wing has a longer path to travel, so its velocity must increase if it is to join
with the air coming underneath the wing.

The air underneath the wing is at atmospheric pressure, but since the velocity is greater on the top of the
wing the pressure falls to some value below atmospheric pressure. This causes the atmospheric pressure
to push on the bottom of the wing lifting it, the airplane, and all the people inside up into the air. It will
continue to do so as long as the wing is moving forward and the configuration of the wing does not
change. Gravity offsets this lifting force and the aircraft flies between these two forces.

This same principle explains how an automobile carburetor works, why the shower curtain comes into (1 of 5) [12/22/2004 7:58:21 AM]

the bath tub when you take a shower, and how a sailboat can sail faster than the wind.

Formula number two explains why the wing lifted into the air:

Pressure x Area = Force

Pressure is measure in pounds per square inch (kilograms per square centimeter)

Area is measured in square inches (square centimeters)

The units for force then become pounds (kilograms) because the square inches (square centimeters)
cancel out.

lbs / in2 x in2 = pounds

It is important for us to know the forces being generated because force over distance, in a given time
period, is a measure of work, energy expended, or heat, depending upon which units we use.

Formula number three explains the action of a venturi. As the area inside a venturi decreases the velocity
of the fluid increases. This causes the pressure to decrease (formula #1) allowing atmospheric pressure to
push a fluid into the venturi. Look at the following diagram

We use the venturi principle to add chemicals to a lawn, remove air from a condenser, add chemical to a
boiler etc. It is the same principle we use to get fuel to the carburetor of your automobile.

Now we will look at the cross section of a centrifugal pump and these three formulas will explain why (2 of 5) [12/22/2004 7:58:21 AM]

mechanical seals have so much trouble with shaft deflection.

This picture describes a volute pump because the impeller is not in the center of the casing. You will note
that there is less clearance between the impeller and the cut water than there is between the impeller and
the rest of the casing. You will also note that this area is increasing as you move from the cutwater,
around the casing, to the discharge nozzle. Circular pumps have an equal area around the impeller. They
are used to pump larger quantities of liquid, without having to create a head. The volute design is the
most popular design because it will produce a head.

When we removed the packing from a centrifugal pump we lost a big part of the shaft support system. It
therefore becomes very important that we keep the forces equal around the impeller to prevent shaft
displacement. If the force increases on one side of the impeller it will deflect the attached shaft and
interfere with the performance of the mechanical seal and pump bearings .

Since the impeller is symmetrical in shape (the area is the same all around the impeller) It is important
that we do not let the pressure vary around the impeller or the resultant forces will not be equal. (Formula

To keep the pressure equal around the impeller, you have to keep the velocity of the liquid constant
around the impeller. (Formula #1).

Take another look at the cross section of the volute pump and you will note that the area (volume)
surrounding the impeller is increasing as you move, in the direction of shaft rotation, from the cut water
to the discharge nozzle. Formula #3 states that the velocity of the liquid times the area must remain a
constant, so that means that the velocity of the liquid is decreasing as the area is increasing.

If the velocity of the liquid decreases, the pressure increases (Formula #1.) Pressure times area creates a
force (Formula #2) and this force displaces the impeller and shaft in a direction towards 60 from the cut
water. (3 of 5) [12/22/2004 7:58:21 AM]

In other words there is a constant force displacing the shaft that will impact on the performance of the
mechanical seal.

If you design the impeller perfectly, and manufacture it just as it was designed, It is possible for the
rotating impeller to continually add just the right amount of liquid to this volute area and prevent the
velocity of the liquid from changing. When that occurs we say that the pump is operating at its best
efficiency point. (B.E.P.) and there is no shaft deflection.

Centrifugal pumps seldom run at their best efficiency point (BEP). Let's look at what happens when we
go off the BEP:

If there is not enough head in the system (you are operating on the right hand side of the curve)
the velocity of the liquid will increase as it approaches the discharge nozzle, causing the pressure
to decrease and a force will be generated that will displace the impeller at 60 towards 240 from
the cutwater.
Likewise if there is too much head on the system (you are on the left side of the pump curve),
some of the fluid will recirculate back through the cutwater causing the velocity of the liquid to
increase, making the pressure fall and a force will be generated pushing the impeller towards 60
from the cutwater.

Look at the diagram again and note those deflections:

In other papers on this web site I talk about methods of stabilizing the shaft for these "off design"
operations. But the fact remains that shaft deflection continues to be a major source of mechanical seal
problems, and will continue to be until the pump manufacturer accepts the responsibility of building a
sensible pump. (4 of 5) [12/22/2004 7:58:21 AM]

Link to the Mc Nally home page (5 of 5) [12/22/2004 7:58:21 AM]

Double volute pump

The double volute pump 14-6

In all of my classes and writings I discuss the advantages of using a double volute centrifugal pump to
eliminate radial shaft deflection caused by operating off the best efficiency point (BEP) of a single stage
centrifugal pump.

In this paper I will answer those questions I get the most frequently asked about double volute designs

The single volute pump impeller will deflect either 60 or 240

from the cut water depending upon which side of the pump's best
efficiency point (BEP) you are operating.

These numbers can change if you are using a low or high

specific speed impeller, but they are good numbers for the high
percentage of Francis vane impellers (SS 1500 to 4000) we find
in industry. You can read about specific speed (SS) in my Paper
7-3 .

The double volute design is actually two single volute designs

combined together.

Although this drawing does not show it clearly, the total throat
area of the two volutes is the same as the single volute design.

Double volute pumps were created to eliminate most of the

radial thrust caused by operating off the pump's best efficiency
point (BEP).

In its simplest form the double volute design tricks the impeller into thinking that it is located in a
circular casing. A circular casing does not generate any significant radial forces.

Let's take a look at a few of the specifics:

Testing has shown that the double volute does not entirely eliminate the radial forces, but they are
reduced greatly. Although the volute is symmetrical around its centerline, the two passages
directing the liquid to the discharge nozzle are not. This means that the radial forces do not
exactly cancel and a slight radial force does exist. (1 of 2) [12/22/2004 7:59:12 AM]

Double volute pump

Testing has further shown that a double volute pump will be 1% to 2% less efficient at its best
efficiency point (BEP), but 2% or more efficient on either side of the best efficiency point (BEP).
This means that the double volute will have an overall higher efficiency than its single volute
cousin. Unfortunately many pumps are purchased with the efficiency given at the best efficiency
point not the actual operating point used to make the purchasing decision.
Double volute pumps should never be specified for low flow (less than 400 gpm or 90 m3/hr.)
operations especially if there are solids in the product. The narrow passages behind the dividing
rib can easily clog with solids. They should, however, always be specified for larger volume
The rib can cause some production problems with the castings especially in small sizes.
Testing has shown the minimal radial thrust was experienced when the dividing rib did not extend
all the way to the volute discharge flange.
On large pumps there has been some problems with the rib cracking at the center when the pump
is subjected to high hydro test pressures. The reason for this is not really understood. Some
manufacturers ignore this because in operation they know that the pressure will be the same on
both sides of the rib. Other manufacturers leave a gap of 2-3 millimeters in the center to prevent
the cracking.
Triple volute casings have been tried, but haven't proved to be effective enough to justify their
high manufacturing cost.
If you have an occasion to repair the double volute cutwaters (and you can with some of the
newer metal repair compounds), be sure the cutwaters are located physically 180 degrees apart.
Many large double ended pumps have atrocious L3/D4 shaft numbers and are therefore supplied
with a double volute as a standard.

Why do we see so many end suction and smaller double ended pumps being supplied without this double
volute? The answer is easy. The lower efficiency at the pump's BEP (best efficiency point) has just about
eliminated the double volute as a design that will be quoted in this era of high efficiency. As a consumer
you should be looking for three features from your purchased items: performance, reliability and
efficiency in that order.

Unfortunately most purchasing decisions specify efficiency first, assuming that reliability and
performance are inherent in the product. Unfortunately they are not!

Link to Mc Nally home page (2 of 2) [12/22/2004 7:59:12 AM]

pump curve

SUBJECT : How to read a pump curve 2-3

Please look at the above illustration. You will note that I have plotted the head of the pump against its
capacity. The head of a pump is read in feet or meters. The capacity units will be either gallons per
minute, liters per minute, or cubic meters per hour.

According to the above illustration this pump will pump a 40 capacity to about a 110 head, or a 70
capacity to approximately a 85 head (you can substitute either metric or imperial units as you see fit)

The maximum head of this pump is 115 units. This is called the maximum shutoff head of the pump.
Also note that the best efficiency point (BEP) of this impeller is between 80% and 85% of the shutoff
head. This 80% to 85% is typical of centrifugal pumps, but if you want to know the exact best efficiency
point you must refer to the manufacturers pump curve.

Ideally a pump would run at its best efficiency point all of the time, but we seldom hit ideal conditions.
As you move away from the BEP the shaft will deflect and the pump will experience some vibration.
You will have to check with your pump manufacturer to see how far you can safely deviate from the
BEP (a maximum of 10% either side is typical)

Now look at the following illustration: (1 of 4) [12/22/2004 7:59:21 AM]

pump curve

Note that I have added some additional curves to the original illustration. These curves show what
happens when you change the diameter of the impeller.

Impeller diameter is measured in either inches or millimeters. If we wanted to pump at the best efficiency
point with a 11.5 impeller we would have to pump a capacity of 50 to a 75 head.

The bottom half of the illustration shows the power consumption at various capacities and impeller
diameters. I have labeled the power consumption horsepower, but in the metric system it would be called

Each of the lines represents an impeller diameter. The top line would be for the 13 impeller the second
for the 12.5 etc. If we were pumping a capacity of 70 with a 13 impeller it would take about 35
horsepower. A capacity of 60 with the 12 impeller would take about 20 horsepower. (2 of 4) [12/22/2004 7:59:21 AM]

pump curve

Most pump curves would show you the percent of efficiency at the best efficiency point . The number
varies with impeller design and numbers from 60% to 80% are normal.

When you will look at an actual pump curve you should have no trouble reading the various heads and
corresponding capacities for the different size impellers. You will note however, that the curve will
usually show an additional piece of information and that is NPSHR which stands for net positive suction
head required to prevent the pump from cavitating.

Depending upon the pump curve you might find a 10 foot (3.0 meter) NPSH required head at a capacity
of 480 Gallons per minute (110 cubic meters per hour) if you were using a 13 inch (330 mm.) diameter

You should keep in mind that the manufacture assumed you were pumping 20 C ( 68 F ) fresh water
and the N.P.S.H. Required was tested using this assumption. If you are pumping water at a different
temperature or if you are pumping a different fluid, you are going to have to add the vapor pressure of
that product to the N.P.S.H. Required. The rule is that Net Positive Suction Head Available minus the
Vapor Pressure of the product you are pumping (converted to head) must be equal to or greater than Net
Positive Suction Head Required by the manufacturer.

Suppose we wanted to pump some liquid Butane at 32 degrees Fahrenheit (0 degrees Centigrade) with
this pump. If we look at the curve for Butane on a vapor pressure chart similar to the one shown in the
charts and graphs section of this web site you will note that Butane at 32F needs at least 15 psi (1,0 Bar)
to stay in a liquid state. To convert this pressure to head we use the standard formula :

In other words Butane at this temperature would not vaporize as long as I had the above absolute heads
available at the suction side of the pump. (3 of 4) [12/22/2004 7:59:21 AM]

pump curve

Link to the Mc Nally home page (4 of 4) [12/22/2004 7:59:21 AM]


SUBJECT : All about specific speed 7-3

Specific speed is a term used to describe the geometry (shape) of a pump impeller. People responsible for
the selection of the proper pump, for their application, can use this Specific Speed information to :

Select the shape of the pump curve.

Determine the efficiency of the pump.
Anticipate motor overloading problems.
Predict N.P.S.H. requirements.
Select the lowest cost pump for their application.

Specific speed is defined as "the speed of an ideal pump geometrically similar to the actual pump, which
when running at this speed will raise a unit of volume, in a unit of time through a unit of head".

The performance of a centrifugal pump is expressed in terms of pump speed, total head, and required
flow. This information is available from the pump manufacturer's published curves. Specific speed is
calculated from the following formula, using data from these curves at the pump's best efficiency point

N = The speed of the pump in revolutions per minute (rpm.)

Q = The flow rate in liters per minute ( for either single or double suction impellers)

H = The total dynamic head in meters

Please refer to the following chart: (1 of 5) [12/22/2004 7:59:28 AM]


Pumps are traditionally divided into three types: radial flow, mixed flow, and axial flow. When you look
at the above chart you can see there is a gradual change from the radial flow impeller, which develops
pressure principally by the action of centrifugal force, to the axial flow impeller, which develops most of
its head by the propelling or lifting action of the vanes on the liquid.

In the specific speed range of approximately 1000 to 6000 double suction impeller are used as frequently
as the single suction impellers.

If you substitute other units for flow and head the numerical value of Ns will vary. The speed is always
given in revolutions per minute (rpm.). Here is how to alter the Specific Speed number (Ns) if you use
other units for capacity and head :

United States ....Q = G.P.M. and H = feet. Divide the Ns by 1.63

British ............Q = Imp.G.P.M. and H = feet. Divide the Ns by 1.9
Metric ............Q = M3/hour and H = meters. Divide the Ns by 1.5

As an example we will make a calculation of Ns in both metric and U.S. units :

Q= 110 L/sec. or 396 M3/ hour or 1744 G.P.M.

H = 95 meters or 312 feet
Speed = 1450 rpm. (2 of 5) [12/22/2004 7:59:28 AM]


If the above results were describing an actual application, we would notice that it was a low specific
speed, radial flow pump, meaning It would be a large pump with a low efficiency. Going to 2900 rpm. or
higher would increase the Ns to 1000 or more, meaning a smaller pump with a much higher efficiency,
but this higher rpm. would have other possible consequences :

The higher efficiency would allow you to use a less powerful driver that would reduce your
operating costs.
A smaller pump makes associated hardware cheaper. For instance, a smaller diameter shaft means
a lower cost mechanical seal and lower cost bearings.
Cavitation could become a problem as the increase in speed means an increase in the N.P.S.H.
If you are pumping an abrasive fluid, abrasive wear and erosion will increase with increasing
Many single mechanical seals have problems passing fugitive emission standards at the higher
pump speeds.
High heat is a major cause of bearing failure. The higher pump speeds contribute to the problem.

The following diagram illustrates the relationship between specific speed and pump efficiency. In
general, the efficiency increases as Ns increases. (3 of 5) [12/22/2004 7:59:28 AM]


Specific speed also relates to the shape of the individual pump curve as it describes head, capacity, power
consumption and efficiency.

In the above diagram you will note that :

The steepness of the head/ capacity curve increases as specific speed increases.
At low specific speed, power consumption is lowest at shut off and rises as flow increases. This
means that the motor could be over loaded at the higher flow rates unless this was considered at
the time of purchase.
At medium specific speed the power curve peaks at approximately the best efficiency point. This
is a non overloading feature meaning that the pump can work safely over most of the fluid range
with a motor speed to meet the B.E.P. requirement.
High specific speed pumps have a falling power curve with maximum power occurring at
minimum flow. These pumps should never be started with the discharge valve shut. If throttling is
required a motor of greater power will be necessary.

Keep in mind that efficiency and power consumption were calculated at the best efficiency point
(B.E.P.). In practice most pumps operate in a throttled condition because the pump was oversized at the
time it was purchased. Lower specific speed pumps may have lower efficiency at the B.E.P., but at the
same time will have lower power consumption at reduced flow than many of the higher specific speed

The result is that it might prove to be more economical to select a lower specific speed design if the
pump had to operate over a broad range of capacity. (4 of 5) [12/22/2004 7:59:28 AM]


Link to the Mc Nally home page (5 of 5) [12/22/2004 7:59:28 AM]

L3/d4 and premature seal failure

SUBJECT: The relationship between the pump L3/D4 and premature seal failure. 11-6

Some pump and mechanical seal sales people talk about L3/D4 . How important is the number when it
comes to selecting a pump? Well that is what this paper is all about, but keep in mind that any discussion
of L3/D4 is limited to single stage, end suction centrifugal pumps.

L = the length of the shaft measured from the center of the impeller to the center of the radial or
inboard bearing. This measurement must be in inches or millimeters.
D = the outside diameter of the solid shaft measured beneath the sleeve, if one has been installed
on the shaft. The measurement is in inches or millimeters.

The frame #1 pump is commonly supplied with a six inch impeller and turns at 3500 rpm (150 mm at
3000 rpm) The pump is used in applications that require a high head (pressure) and modest capacity.

If we compare the L3/D4 numbers of some shafts that are used in this very popular and competitively
priced U.S. frame I pump, we would find the following:


Duriron Mark II Group I solid shaft

Duriron Mark II Group I with a sleeve


Goulds 3196 ST with a solid shaft


Goulds 3196 ST with a sleeve


Worthington CNN frame 1


If you looked at the European and Asian versions of this same design you would find that their L3/D4
numbers are in the range of 3 to 5. At my schools I teach that the number should be less than 60 (2 in the
metric system). Does this mean that these models are not acceptable as good quality process pumps?

Of course not. It means that these pumps are designed for different purposes, in the same way a Porche
sports car is designed differently than a Mazda Miata. They are both two door sports cars, but they sell
for radically different prices.

If you want the feel of a sports car as you drive around town, the Miata is a good choice, but if you
intend to drive down the German autobahn at 200 kilometers per hour, the more expensive Porche would
probably be a more sensible selection. (1 of 3) [12/22/2004 7:59:43 AM]

L3/d4 and premature seal failure

Pumps are like that. If you are going to run a pump twenty four hours a day, 365 day a year, and not open
and close system valves, these lower cost pumps would be a logical choice. All you are required to do is
size the pump correctly and then the shaft displacement, at the best efficiency point (B.E.P.), would be

If you are going to do any of the following a pump with a shaft L3/D4 number less than 60 (2 in metric)
would make a lot more sense.

Start and start the pump a lot. Batch operations as an example.

If tank levels are going to change. Loading and unloading pumps often experience this.
Operate the pump with a variable speed motor and the application is not a circulating system or a
system where the head is predominately system head.
Run the pump throttled because it was purchased too large for the application anticipating the
need for a larger pump in the future.
Let inexperienced people size the pump and add in safety factors because they do not trust their
skill in pump selection.
Operate at different points on the pump curve by opening and closing valves in the system. Some
applications require the isolation of parts of the system as a normal routine. The valves are
opened, closed or throttled to satisfy the local demand.
Start the pump with the discharge valve throttled or shut to save power.
Experience occasional cavitation problems.
Fill a tank from the bottom instead of the top. This is a common occurrence if the pump is putting
a head on the system.

The conventional automobile water pump is attached to a vibrating engine. The shaft is pulley driven and
the service is intermittent. At best, a very difficult application for the mechanical seal we find on all of
these applications.

What kind of a L3/D4 number do we find on the shaft of this pump? Less than fifteen is typical in the
imperial system.

Check with your pump supplier to learn the L3/D4 number of the pump you are about to purchase. Often
you can get the correct L3/D4 by specifying the pump with a solid shaft rather than with a sleeve, but in
other cases you may have to go to a more expensive heavy duty model.

Link to the Mc Nally home page (2 of 3) [12/22/2004 7:59:43 AM]

L3/d4 and premature seal failure (3 of 3) [12/22/2004 7:59:43 AM]

Variable speed drive

Subject : Using a variable speed motor to control flow in a centrifugal pump.13-8

If you operate too far off the pump's BEP(best efficiency point) the shaft will deflect radially and that
could lead to both seal and bearing problems. So what should you do if you have to vary the capacity of a
centrifugal pump? The classic discharge control valve is not a logical choice because if you alter the
capacity of a centrifugal pump the head alters also, and in most cases this will guarantee you will be
operating off the pump's best efficiency point (BEP). It turns out there are several possible solutions to
preventing the problem of shaft deflection while running off the pump's best efficiency point.

Switch to a double volute pump design.

Install a pump discharge bypass line back to the storage tank such as the type commonly used on
boiler feed pumps.
Convert to a low L3/D4 pump design that will operate in a wide window.
Install a support bushing in the end of the stuffing box, move the seal closer to the bearings and
accept a small amount of shaft deflection.
And the big question, "how about a variable speed drive"?

You have several options when selecting a variable speed drive:

A variable speed electric motor

A gasoline or diesel engine.
A variable pulley arrangement.
A changeable gear box.
Electrical switch gear
A hydraulic coupling.

So what is the correct answer ? Is the variable speed drive a sensible choice? The only correct answer is
"sometimes"! Let's take a look at what alters when you change the speed of a centrifugal pump. In the
following drawing the "H" axis is the head (feet or meters) and the "Q" axis is the capacity (gpm or

Changing the speed of a centrifugal pump has just about the same affect
as changing the diameter of the pump impeller. The "Affinity Laws"
allow you to predict the results of this change.

The area within the curved lines (ABCD) is called the operating window
of the pump. Notice that the sloping best efficiency point line intersects
the capacity leg (Q) at an angle. This slope causes a problem with many
pumping applications. (1 of 5) [12/22/2004 7:59:51 AM]

Variable speed drive

For the variable speed drive to be a sensible solution to your pumping application the system curve
would have to fall on, or close to this best efficiency point line or you will experience radial loads that
will translate to shaft deflection. Most pump companies want you to operate within 5% to 10% of the
BEP(best efficiency point). Heavy duty pumps that have a low L3/D4 (shaft diameter to shaft length
ratio) have a much larger operating window.

The above diagram shows that the head is going to have to increase at a predetermined rate as the
capacity increases. In Technical Paper 7-01 you learned that there are three kinds of head that will have
an affect on the pump's capacity:

Static head. The distance from the discharge of the pump center line to highest liquid level minus
any distance caused by the siphon affect.
Pressure head. The head caused by pumping into a pressurized vessel.
System or friction head. The loss of head caused by friction in the piping, valves and fittings.

In paper 5-12 you also learned that a system curve is constructed by the end user of the pump and
describes the head/ capacity relationship over the desired operating range of the pump that is going to be
specified. The pump manufacturer places his pump curve on top of this system curve and the point where
they intersect is where the pump is going to operate.

Lets take a look at a system curve for a typical boiler feed pump or any pump that will be discharging
into a constant pressure vessel or tank:

The boiler is running at a constant pressure, but the steam demand is

changing. The boiler feed water capacity must vary with the steam
demand, but the pressure or head must remain constant.

The system curve is a straight horizontal line because the dominant head
is the pressure head. The amount of piping and elevation is minimal. (2 of 5) [12/22/2004 7:59:51 AM]

Variable speed drive

Laying the best efficiency point (BEP) sloping line from a varying
speed drive on top of the system curve (EF) would show that we are at
the best efficiency point only at one point.

Allowing the tolerances of the operating window (ABCD) you can see
that we are operating efficiently over only a portion of the desired
system curve. A similar application would be pumping a varying
capacity to a very high tower or elevation where the static head is the
dominant head.

A hot or cold water circulating system describes a different type of

system curve. The dominant head in this example is the friction head
and that varies by almost the square of the capacity.

In other words, two times the capacity gives you four times the head, or
three times the capacity would give you nine times the head. If you plot
this on a piece of chart paper you would get an "exponential curve" as
shown on the left.

If you lay the best efficiency point line on this "exponential curve E-F"
you would get a pretty good match and just about all of the system
curve falls within the operating window (ABCD), so this becomes the
ideal variable speed application.

In other words you use a variable speed drive any time the system head
is dominated by friction in the piping, fittings and valves.

You will find this last curve in many common pumping applications:

Circulating hot or chill water systems.

Loading liquid cargo or fuel to a ship from a distant tank farm.
A piping system with many outlets and a varying demand for product.
A fire hose. (3 of 5) [12/22/2004 7:59:51 AM]

Variable speed drive

Many systems are a combination of all three types of heads. You are
going to have to decide which head is the dominant one.

One of the most common methods of varying pump shaft speed is to use a Variable Frequency Drive
(VFD). These drives take advantage of the fact that torque, speed and horsepower of an AC electric
motor are related to the frequency and voltage of the electrical power supply. Here is the relationship:

hz = frequency or number of cycles per minute. In the U.S we run 60 cycles per minute (hertz),
most of the rest of the world uses 50 hertz.
Torque capability = F(volts/hz)

Horse Power Capability =f(Torque x speed)

VFDs convert incoming alternating current (AC) to direct current (DC) and then invert the DC power
into variable frequencies and voltage AC power. Most VFDs produce a constant voltage/frequency (hz)

A low L3/D4 shaft is still your best protection against damage caused by operating off the pump's best
efficiency point. Any pump that experiences frequent starts and stops has this problem.

If the dominate head in the system is pipe friction losses a variable speed device can have some

They can deliver a broad range of head / capacity figures so your estimate of flow needs does not
have to be exact.
You can eliminate the need for a throttling valve. Valves can leak and they require maintenance.
Often an inefficient bypass line can be eliminated.
Throttling a pump discharge produces unwanted heat in a pump that can be eliminated by
changing the pump speed instead. This heat can cause the pumping fluid to:
Vaporize or flash


Change viscosity

Coke or build a film on sliding seal parts. (4 of 5) [12/22/2004 7:59:51 AM]

Variable speed drive

Become more corrosive.

The heat can also change critical dimensions and could cause lapped seal faces to go "out
of flat".

The bad news is that pumps with variable speed drives have several potential problems:

The fluid viscosity can change with speed if it is a non Newtonian fluid. As an example, the
viscosity of dilatants increases with agitation requiring additional power.
The shaft can hit a critical speed on its way to the ideal operating speed.
You can dial in too much capacity that can, in turn, burn out the electric motor.
Operating off the BEP can cause shaft deflection.
Explosion proof motors must be approved to operate over the entire operating range. At the lower
rpms the cooling fan is often not rotating fast enough.
Variable speed demands may affect the electrical power distribution system by reducing electrical
The mechanical seal has to be designed to operate over the entire speed range. At higher speeds
the design has to be of the stationary type with the spring face load reduced.
At higher shaft speeds the NPSH requirement is increased to prevent cavitation problems. You
may have to install an inducer on the pump's impeller
Higher speed almost always dictates increased maintenance costs because of increased wear and
They cannot be used if the pump or equipment feeds multiple users because more than one flow
cannot be controlled by a single control unit.
The pump or mechanical equipment must be able to operate at reduced speeds. A liquid ring
vacuum pump could have trouble at lower shaft speeds because many of these designs will not
produce a vacuum below 80% of their rated speed.
Remember that a variable speed drive is another piece of equipment installed in the system that
will experience its own set of problems and require its own maintenance.

Link to Mc Nally home page (5 of 5) [12/22/2004 7:59:51 AM]

Pump radial thrust calculations

Subject: Pump shaft radial thrust alternative calculations (in Imperial dimensions) 13-2.

In past papers I have given you the formulas for calculating the radial thrust force on an end suction
centrifugal pump shaft. It sometimes becomes confusing when you read other publications and find
alternative formulas. Which ones are the best, and are they all correct?

The fact is they are all close enough for the work we do in the field, so here are a couple of more to add
to your collection. Use those formulas that you find the most convenient and less complicated.

If you want to approximate the shaft radial thrust at shut off conditions (the pump is running and the
discharge valve is shut or almost shut), use the following formula:

If you do not have shut off conditions, the following formula is a good approximation


Rso= Radial thrust (in pounds) at shut off.

R = Radial thrust (in pounds) at operating conditions.

Kso= Thrust factor at shut off (see figure #1 below )

K = Thrust factor at operating conditions (see above formula).

Hso= Total head at shut off (in feet).

H = Total head at operating conditions (in feet).

sg.= Specific gravity of the liquid.

D2= Impeller diameter (in inches). (1 of 3) [12/22/2004 8:01:59 AM]

Pump radial thrust calculations

B2= Impeller width at the discharge including shrouds (in inches).

Q = Capacity at operating conditions (in gpm.).

Qn = Capacity at the best efficiency point (in gpm.).

x = An exponent varying between 0.7 and 3.3 established by testing. You probably do not have actual
test data so you can safely assume a linear path between 0.7 at a specific speed of 500 and 3.3 at a
specific speed of 3500. Check my Technical paper Volume 9 Number 12 if you are not familiar with the
term specific speed.

PLEASE NOTE: D2 and D2 are not the same thing. D2 means to multiply D by its self (52 = 25). D2
means that particular D because sometimes (not in these examples) there is more than one D (shaft
diameter) in the equation.

FIGURE #1 (2 of 3) [12/22/2004 8:01:59 AM]

Pump radial thrust calculations

Link to Mc Nally home page (3 of 3) [12/22/2004 8:01:59 AM]

Calculating the system head

SUBJECT : Calculating the total system head in USCS units 7-1:

USCS stands for "United States Customary System Units" as opposed to the SI (Le Syst`eme
International d`Units) or metric units that have been adopted by the International standards Organization
(ISO). In a future paper I will present another paper using the metric units, but for the moment it is not
convenient to present it in both systems.

It turn out that "head" is a very convenient term in the pumping business. Capacity is measured in gallons
per minute, and each gallon of liquid has weight, so we can easily calculate the pounds per minute being
pumped. Head or height is measure in feet, so if we multiply these two together we get foot- pounds per
minute which converts directly to work at the rate of 33,000 foot pounds per minute equals one

Pressure is not as convenient a term because the amount of pressure that the pump will deliver depends
upon the weight (specific gravity) of the liquid being pumped and the specific gravity changes with
temperature, fluid, and fluid concentration.

If you will refer to FIG 1, you should get a clear picture of what is meant by static head. Note that we
always measure from the center line of the pump to the highest liquid level

To calculate head accurately we must calculate the total head on both the suction and discharge sides of
the pump. In addition to the static head we will learn that there is a head caused by resistance in the
piping, fittings and valves called friction head, and a head caused by any pressure that might be acting on
the liquid in the tanks including atmospheric pressure, called " surface pressure head". (1 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

Once we know these heads it gets simple, we will then subtract the suction head from the discharge head
and the amount remaining will be the amount of head that the pump must be able to generate at the rated
flow. Here is how it looks in a formula:

System head = total discharge head - total suction head

H = hd - hs

The total discharge head is made from three separate heads:

hd = hsd + hpd + hfd

hd = total discharge head

hsd = discharge static head
hpd = discharge surface pressure head
hfd = discharge friction head

The total suction head also consists of three separate heads

hs = hss + hps - hfs

hs = total suction head

hss = suction static head
hps = suction surface pressure head
hfs = suction friction head

As we make these calculations, you must sure that all calculations are made in either "feet of liquid
gauge" or "feet of liquid absolute". In case you have forgotten "absolute means that you have added
atmospheric pressure (head) to the gauge reading.

Now we will make some actual calculations:

Figure #2 demonstrates that the discharge head is still measured to the liquid level, but you will note that
it is below the maximum height of the piping.

Although the pump must deliver enough head to get up to this maximum piping height, it will not have to
continue to deliver this head when the pump is running because of the "siphon effect". There is of course
a maximum siphon effect. It is derived from: 14.7 psi (atmospheric pressure) x 2.31 feet / psi = 33.4 feet
maximum siphon effect. (2 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

We will begin with the total suction head calculation

1. The suction head is negative because the liquid level in the suction tank is below the centerline of the

hss = - 6 feet

2. The suction tank is open, so the suction surface pressure equals atmospheric pressure :

hps = 0 feet gauge

3. You will not have to calculate the suction friction head, I will tell you it is:

hfs = 4 feet at rated flow

4. The total suction head is a gauge value because atmosphere was given as 0,

hs = hss + hps - hfs = -6 +0 -4 = -10 feet of liquid gauge at rated flow

The total discharge head calculation

1. The static discharge head is:

hsd = 125 feet (3 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

2. The discharge tank is also open to atmospheric pressure, thus:

hpd = 0 feet, gauge

3. I will give you the discharge friction head as:

hfd = 25 feet at rated flow

4. The total discharge head is:

hd = hsd + hpd + hfd = 125 + 0 + 25 = 150 feet of liquid gauge at rated flow

The total system head calculation:

H = hd - hs = 150 - (-10)= 160 feet of liquid at rated flow

Note: did you notice that when we subtracted a minus number (-10) from a positive number (150) we
ended up with a positive 160 because whenever you subtract minus numbers it is the same as adding
them? If you have trouble with this concept you can learn more about it from a mathematics book.

Our next example involves a few more calculations, but you should be able to handle them. In this
example we are going to learn how to handle a vacuum application. Pipe friction numbers are taken from
the Hydraulic Institute Engineering Data Book. You can get a copy of this publication from your library
if you want to see the actual charts. I have some of this information in the chart section of this web site. (4 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head


1. Transferring 1000 gpm. weak acid from the vacuum receiver to the storage tank

2. Specific Gravity - 0.98

3. Viscosity -equal to water

4. Piping - All 6" Schedule 40 steel pipe

5. Discharge piping rises 40 feet vertically above the pump centerline and then runs 400 feet horizontally.
There is one 90 flanged elbow in this line

6. Suction piping has a square edge inlet, four feet of pipe, one gate valve, and one 90 flanged elbow all
of which are 6" in diameter.

7. The minimum level in the vacuum receiver is 5 feet above the pump centerline.

8. The pressure on top of the liquid in the vacuum receiver is 20 inches of mercury, vacuum. (5 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

To calculate suction surface pressure use one of the following formulas:

inches of mercury X 1.133specific gravity = feet of liquid

pounds per square inch X 2.31specific gravity = feet of liquid
Millimeters of mercury X 122.4 x specific gravity = feet of liquid

Now that you have all of the necessary information we will begin by dividing the system into two
different sections, using the pump as the dividing line.

Total suction head calculation

1. The suction side of the system shows a minimum static head of 5 feet above suction centerline.
Therefore, the static suction head is:

hss = 5 feet

2. Using the first conversion formula, the suction surface pressure is:

hps = -20 Hg X 1.133/ 0.98 = -23.12 feet gauge

3. The suction friction head, fs, equals the sum of all the friction losses in the suction line. Friction loss in
6" pipe at 1000 gpm from table 15 of the Hydraulic Institute Engineering Data Book, is 6.17 feet per 100
feet of pipe.

in 4 feet of pipe friction loss = 4/100 x 6.17 = 0.3 feet

Friction loss coefficients (K factors) for the inlet, elbow and valve can be added together and multiplied
by the velocity head:


6" Square edge inlet 0.50 32 (a)

6" 90 flanged elbow 0.29 32 (a)

6" Gate valve 0.11 32 (b)

Total coefficient, K = 0.90 (6 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

Total friction loss on the suction side is:

hfs = 0.3 + 1.7 = 2.0 feet at 1000 gpm.

4. The total suction head then becomes:

hs = hss + hps - hfs = 5 + (-23.12) - 2.0 = -20.12 feet, gauge at 1000 gpm.

Total discharge head calculation

1. Static discharge head = hsd = 40 feet

2. Discharge surface pressure = hpd = 0 feet gauge

3. Discharge friction head = hfd = sum of the following losses :

Friction loss in 6" pipe at 1000 gpm. from table 15, is 6.17 feet per hundred feet of pipe.

In 440 feet of pipe the friction loss = 440/100 x 6.17 = 27.2 feet

Friction loss in 6" elbow:

from table 32 (a), K = 0,29

from table 15, V2/2g = 1.92 at 1000 gpm.

Friction loss = K V2/2g = 0.29 x 1.92 = 0.6 feet

The friction loss in the sudden enlargement at the end of the discharge line is called the exit loss. In
systems of this type where the area of the discharge tank is very large in comparison to the area of the
discharge pipe, the loss equals V2/2g, as shown in table 32 (b).

Friction loss at exit = V2/2g = 1.9 feet

The discharge friction head is the sum of the above losses, that is:

hfd = 27.2 + 0.6 + 1.9 = 29.7 feet at 1000 gpm.

4. The total discharge head then becomes: (7 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

hd = hsd + hpd + hfd = 40 + 0 + 29.7 = 69.7 feet, gauge at 1000 gpm.

c. Total system head calculation:

H = hd - hs = 69.7 - (-20.2) = 89.9 feet at 1000 gpm.

Our next example will be the same as the one we just finished except. that there is an additional 10 feet
of pipe and another 90 flanged elbow in the vertical leg. The total suction head will be the same as in the
previous example. Take a look at figure # 4

Nothing has changed on the suction side of the pump so the total suction head will remain the same:

hs = -20.12 feet, gauge at 100 gpm.

Total discharge head calculation

1. The static discharge head "hsd" will change from 40 feet to 30 feet, since the highest liquid surface in
the discharge is now only 30 feet above the pump centerline.(This value is based on the assumption that
the vertical leg in the discharge tank is full of liquid and that as this liquid falls it will tend to pull the
liquid up and over the loop in the pipe line. This arrangement is called a siphon leg).

2. The discharge surface pressure is unchanged: (8 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

hpd = 0 feet

3. The friction loss in the discharge pipe will be increased by the additional 10 feet of pipe and the
additional elbow.

In 10 feet of pipe the friction loss = 10/100 x 6.17 = 0.6 feet

The friction loss in the additional elbow = 0.6 feet

The friction head will then increase as follows:

hfd = 29.7 + 0.6 + 0.6 = 30.9 feet at 1000 gpm.

The total discharge head becomes:

hd = hsd + hpd + hfd

= 30 + 0 + 30.9

= 60.9 feet, gauge at 1000 gpm.

5. Total system head calculation

H = hd - hs = 60.9 - (-20.12) = 81 feet at 1000 gpm.

For our last example we will look at gauges. Take a look at FIG 5: (9 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head


Capacity - 300 gpm.

Specific gravity - 1.3
Viscosity - Similar to water
Piping - 3 inch suction, 2 inch discharge
Atmospheric pressure - 14.7 psi.

Divide the heads into two sections again:

The discharge gauge head corrected to the centerline of the pump, in feet of liquid absolute is found by
adding the atmospheric pressure to the gauge reading to get absolute pressure, and then converting to
absolute head:

hdg = (130 + 14.7) x 2.31 / (1.3 Specific Gravity + 4 ) = 261.1 feet, absolute

Note the 4 foot head correction to the pump centerline.

The discharge velocity head at 300 gpm. is found in table 9 of the Hydraulic Institute Engineering Data

hvd = 12.8 feet at 300 gpm.

The suction gauge reading is in absolute terms so it needs only to be converted to feet of liquid, absolute.

hgs = 40 x 2.3 / 11.3 +2 = 73.08 feet absolute (10 of 11) [12/22/2004 8:02:06 AM]

Calculating the system head

Note the 2 foot head correction to the pump centerline.

The suction velocity head at 300 gpm. is found in table 11 of the Pipe Friction Manual:

hvs = 2.6 feet at 300 gpm.

The total head developed by the pump is:

H = (hgd + hvd ) - ( hgs + hvs ) = (261.1 + 12.8) - (73.08 + 2.6)= 198.22 feet absolute at
300 gpm.

Link to the Mc Nally home page (11 of 11) [12/22/2004 8:02:06 AM]

Pump system curve

SUBJECT: Understanding the system curve 5-12

Every pump manufacturer would like to recommend the perfect pump for your application. To do this he
would like you to provide him with an accurate system curve that would describe the capacity and head
needed for your various operating conditions. Once he has your system curve, he can plot his pump
curves on top of the system curve and hopefully select something that will come close to your needs.
Without this system curve, neither one of you has much of a chance of coming up with the right pump.

To create a system curve we plot the desired capacities against the required head over the total
anticipated operating range of the pump. The head will be measured in feet or meters and the capacity
will be measured in gallons per minute or cubic meters per hour.

Some of the confusion begins when we realize that there are three different kinds of head:

STATIC HEAD This is the vertical distance measured from the center line of the pump to the height of
the piping discharge inside the tank. Look at figure "A" and note that the piping discharge is below the
maximum elevation of the piping system. We do not use the maximum elevation in our calculations
because the siphoning action will carry the fluid over this point once the piping is full of liquid. This is
the same action that lets you siphon gasoline out of an automobile to a storage can.

The pump will have to develop enough head to fill the pipe and then the siphoning action will take over.
The pump operating point should move back towards the best efficiency point (B.E.P.) if the pump was
selected correctly.

FIGURE "A" (1 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

DYNAMIC OR SYSTEM HEAD As the liquid flows through the piping and fittings, it is subject to the
friction caused by the piping inside finish, restricted passages in the fittings and hardware that has been
installed in the system. The resulting "pressure drop" is described as a "loss of head" in the system, and
can be calculated from graphs and charts provided by the pump and piping manufacturers. These charts
are not included with this paper, you can find them in the Hydraulic Institute Manuals. This "head" loss
is related to the condition of the system and makes the calculations difficult when you realize that older
systems may have "product build up" on the piping walls, filters, strainers, valves, elbows, heat
exchangers, etc., making the published numbers some what inaccurate.

A general "rule of thumb" says that the friction loss in clean piping will vary approximately with 90% of
the square of the change in flow in the piping, and 100% of the square with the change of flow in the
fittings and accessories. You calculate the change in flow by dividing the new flow by the old flow and
then square the number. As an example: (2 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

In the original application system, loss was a combination of the loss through the piping and the loss
through the fittings for a total of 100 feet at 200 gallons per minute. When we increased the flow to 300
gallons per minute our system head changed to a total of 208.13 feet. This change would have to be
added to the static and pressure heads to calculate the total head required for the new pump.

Please note that the pump is pumping the difference between the suction head and the discharge head, so
if you fail to consider that the suction head will be either added to or subtracted from the discharge head,
you will make an error in your calculations. The suction head will be negative if you are lifting liquid
from below ground or if you are pumping from a vacuum. It will be positive if you are pumping from a
tank located above ground. If the suction head is pressurized, this pressure must be converted to head and
subtracted from the total head required by the pump.

A centrifugal pump will create a head/capacity curve that will generally resemble one of the curves
described in figure "B" The shape of the curve is determined by the Specific Speed number of the

Centrifugal pumps always pump somewhere on their curve, but should be selected to pump as close to (3 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

the best efficiency point (B.E.P.) as possible. The B.E.P. will fall some where between 80% and 85% of
the shut off head (maximum head).

The manufacturer generated these curves at a specific R.P.M.. Unless you are using synchronous motors
(you probably are using induction motors on your pumps) you will have to adjust the curves to match
your actual pump speed. Put a tachometer on the running motor and record the rpm. difference between
your pump and the speed shown on the pump manufacturer's published curve. You can use the pump
affinity laws to approximate the change.

POSITIVE DISPLACEMENT PUMPS have a different shaped curve. They look something like Figure

In this system, the head remains a constant as the capacity varies. This is a typical application for:

A boiler feed pump that is supplying a constant pressure boiler with a varying steam demand. This
is a very common application in many process systems or aboard a ship that is frequently
changing speeds (answering bells).
Filling a tank from the top and varying the amount of liquid being pumped, is the normal routine (4 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

in most process plants. The curve will look like this if the majority of the head is either static or
pressure head.

The second system is the ideal one, Figure "E" describes it:

In this system the entire head is system head so it will vary with the capacity. Look for this type of curve
in the following applications:

A circulating hot or cold water heating/ cooling system.

Pumping to a non pressurized tank, a long distance from the source with little to no elevation
involved. Filling tank cars is a typical application.

System curve "G" is a common one. It is a combination of static, pressure and system heads. (5 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

Once the pump manufacturer has a clear idea as to the shape of your system curve, and the head and
capacity numbers needed he can then select the proper centrifugal pump. The shape of his curve will be
pretty much determined by the specific speed number of the impeller.

In addition to specific speed he can select impeller diameter, impeller width, pump rpm., and he also has
the option of series or parallel operation along with the possibility of using a multi-stage pump to satisfy
your needs.

The sad fact is that most pumps are selected poorly because of the desire to offer the customer the lowest
possible price. A robust pump, with a low L3/D4, is still your best protection against seal and bearing
premature failure when the pump is operating off of its best efficiency point. Keep the following in mind
as you select your pump:

A centrifugal pump will pump where the pump curve intersects the system curve. This may bear
no relationship to the best efficiency point (B.E.P.), or your desire for the pump to perform a
specific task.
The further off the B.E.P. you go, the more robust the pump you will need. This is especially true (6 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

if you have replaced the packing with a mechanical seal and no longer have the packing to act as a
support bearing when the shaft deflects. Shaft deflection is always a major problem at start up.
When you connect pumps in parallel, you add the capacities together. The capacity of a pump is
determined by the impeller width and r.p.m.. The head of a centrifugal pump is determined by the
impeller diameter and rpm. If the heads are different, the stronger pump will throttle the weaker
one, so the impeller diameters and rpm's must be the same if you connect pumps in parallel.
Check the rpm's on these pumps if you are experiencing any difficulties.
If you connect the pumps in series, the heads will add together, so the capacities must be the same
or one of them will probably cavitate. You could also have a problem operating too far to the right
of the best efficiency point with a possible motor "burn out".
When you vary the speed of a centrifugal pump, the best efficiency point comes down at an angle.
The affect is almost the same as changing the diameter of the impeller. This means that the
variable speed motor will work best on a system curve that is exponential (Figure "F").
Unfortunately most process and boiler feed pump system curves are not exponential.
Pump curves are based on a speed of 1750, 3500, 1450, or 2900 r.p.m.. Electric induction motors
seldom run at these speeds because of "slip". You can estimate that a 2% to a 5% slip is normal in
these pumps with the "slip" directly related to the price of the motor.
You should also keep in mind that if the motor is running at its best efficiency point that does not
mean that the pump is running at its B.E.P..

Since you will be using pumps that were supplied at the lowest cost, you can do the following to resist
some of the shaft displacement:

Use a solid shaft. Sleeves often raise the L3/D4 number to over 60 (2 in the metric system), and
this is too high a number for reliable seal performance.
Try to keep the mechanical seal as close to the bearings as possible. It is the mechanical seal that
is the most sensitive to shaft deflection and vibration.
Once the seal has been moved closer to the bearings, you can install a sleeve bearing in the
packing space to support the shaft when the pump is operated off of its B.E.P. This is especially
important at start up, or any time a pump discharge valve is operated.
Stop the cavitation if you are experiencing any.
Balance the rotating assembly.
Check that the shaft is not bent or the rotating assembly is not out of dynamic balance.
Use a "C" or "D" frame adapter to solve pump- motor alignment difficulties.
A center line design wet end can be used if pipe strain, due to temperature expansion, is causing
an alignment problem.

Do not trust the system prints to make your calculations. The actual system always differs from that
shown on the print, because people tap into the lines, using the pumped fluid for a variety of purposes
and after having done so forget to change or "mark up" the original system print. You are going to have
to "walk down" the system and note the pipe length, the number of fittings, etc., to make an accurate
system head calculation. Do not be surprised to find that the discharge of your pump is hooked up to the
discharge of another pump further down the line. In other words, the pumps are connected in parallel and (7 of 8) [12/22/2004 8:02:13 AM]

Pump system curve

no body knows it. Pressure recorders (not gauges) installed at the pump suction and discharge is another
technique you can use to get a better picture of the system or dynamic head. They will show you how the
head is varying with changes in flow.

Pump selection is simple but not easy. Do not depend upon the knowledge of the local pump salesman to
select the correct pump for you. In many cases he is prepared to sell his pump at cost&emdash;to get the
spare parts business. If you are purchasing pumps at too big a discount&emdash;something is wrong,
there is no free lunch. Keep in mind that if several people are involved in the selection process each of
them will commonly add a safety factor to the calculated pump size. These factors added together can
cause you to purchase a pump that is very much over sized.

Link to the Mc Nally home page (8 of 8) [12/22/2004 8:02:13 AM]

Specific speed

SUBJECT: The difference between specific speed and suction specific speed 9-12

The best way to describe the shape of an impeller is to use its specific speed number. This is a
dimensionless number that was generated by the formula :

Ns = Specific speed

N = Pump shaft speed

Q = Capacity in GPM.

H = Total head in feet

The following chart gives you a graphic picture of the impeller shape represented by this number:

The major use of the specific speed number is to help you specify pumps that are more efficient.

The maximum pump efficiency is obtained in the specific speed range of 2000 to 3000.
Pumps for high head low capacity occupy the range 500 to 1000. While low head high capacity
pumps may have a specific speed of 15,000 or larger.
For a given head and capacity the good news is that the pump having the highest specific speed,
that will meet the requirements, probably will be the smallest size and the least expensive. The
bad news it that it will run at the highest speed where abrasive wear and cavitation damage
become a problem. (1 of 4) [12/22/2004 8:02:30 AM]

Specific speed

Efficiencies start dropping drastically at specific speeds below 1000. Also smaller capacities
exhibit lower efficiencies than higher capacities at all specific speeds.
In propeller and other high specific speed impellers (axial flow) it is not practical to use a volute
casing. Instead, the impeller is enclosed in a pipe like casing.
The lower the specific speed number, the higher the power loss you get with wear ring clearance.

The clearance between the impeller and the tongue of the volute has a bearing on efficiency, pressure
pulsations and cavitation. For high efficiency you would want a small clearance, but this produces larger
pressure pulsations and the increased flow in this area can reduce the fluid pressure enough to cause
flashing of the product and a type of cavitation known as " The vane passing syndrome".

For impellers up to fourteen inches in diameter (355 mm) this clearance should be a minimum of four
percent of the impeller diameter. If you are using greater than fourteen inch diameter impellers the
clearance should be at least six percent of the impeller diameter. Also remember that as this clearance
increases the impeller experiences some slippage. That is the major reason that we do not like to remove
more than ten percent of the impeller diameter when trimming is called for.

If you work in both metric and imperial units as I do, the subject of specific speed becomes very
confusing because both systems use the same specific speed numbers to describe the impeller shape.
They do this even though they use a different set of units to arrive at the same number.

In the metric system the capacity is calculated in liters/ minute and the head in meters. Knowledgeable
people in this area feel that if the calculations are done in imperial or other metric units the final number
should be reduced by the following amount:

U.S. Gallons/ minute and feet divide the result by 1.63

U.K. Imperial gallons and feet divide the result by 1.93

M3/hour and meters divide the result by 1.50

SUCTION SPECIFIC SPEED is another number that we use in pump selection. The formula looks the
same as the regular specific speed formula, but in this formula we use the NPSH required number rather
than the total head produced by the pump.

Ns = Specific speed

N = Pump shaft speed (2 of 4) [12/22/2004 8:02:30 AM]

Specific speed

Q = Capacity in GPM.

NPSH = Net positive suction head required to prevent cavitation. Remember that this number is for sixty
eight degree F. (20C) fresh water. You are going to have to add the vapor pressure of you product to this
number to get the real number that you will be using.

We use this number to predict cavitation problems with your impeller selection.

The flow angle of the inlet vanes and the number of vanes affect this number.
A desired value would be below 8500 with impellers having a flow angle of about seventeen
degrees and five to seven vanes. The higher the flow angle number, the faster the liquid will travel
and the lower suction head (pressure) we will get.
Boiler feed and condensate pumps often require suction specific speed numbers as high as 12,000
to 18,000 because of the temperature and pressure of the water. To get to these values the impeller
inlet flow angle is reduced to a low as ten degrees and the number of vanes reduced to as little as
four . Fewer and thinner vanes help to reduce the blockage in the impeller inlet. A disadvantage to
these low flow angles is that the pump will probably run very rough at below fifty percent of
Water applications can run at these higher numbers because the amount of fluid expansion is very
low for hot water. Mixed hydro-carbons have this same advantage because unlike a single
product, the flashing of the mixed hydro-carbons does not take place all at the same time.
The higher the suction specific speed number the narrower the stable window of operation.
Inducers have been used successfully with suction specific speed numbers of approximately 24,
Should the available NPSH be so low that a suction specific speed number of more than 18,000 is
required, then a separate axial flow impeller (an inducer) can be used ahead of the centrifugal
impeller to prevent cavitation. Its flow angle is some where between five and ten degrees with
typically two vanes and no more than four. In other instances a booster pump can be installed
between the pump and the source.
In their desire to quote a low NPSH required some manufacturers will cut away the impeller inlet
vanes to reduce fluid drag and thereby lower the NPSH required. If this has been done with your
application, you must insure that the impeller to volute clearance is adjusted correctly with open
impeller designs and the wear ring clearance meets the manufacturers specifications with closed
impeller designs, or you will experience internal recirculation problems and cavitation at the
impeller outlet vane tips. Keep the suction specific speed number below 8500 and this problem
should never comes up.

A pump's suction specific speed (SSS) number is a constant. You can re-arrange the formula to calculate
a new NPSHR: (3 of 4) [12/22/2004 8:02:30 AM]

Specific speed

In the metric system we calculate the capacity in liters/sec and the NPSH in meters. You should try to
keep the final SSS number below 5200. Above 7800 you are going to have trouble with internal
recirculation and cavitation.

Link to the Mc Nally home page (4 of 4) [12/22/2004 8:02:30 AM]

Seal troubleshooting

SUBJECT : A quick reference guide for mechanical seal failure 4-11

Of all the seal related activities, analyzing mechanical seal failure continues to be the single greatest
problem for both the consumer and the seal company representative. I have addressed this problem in
several of my other technical papers. If you will take a little bit of time to familiarize yourself with the
following outline you should feel a lot more comfortable the next time you are called upon to do some
seal troubleshooting.

I should mention here in the beginning that as you look over the failed seal components keep in mind that
a rebuilt seal may have some marks that occurred during a previous failure, making them especially
difficult to analyze, but regardless of the design mechanical seals fail for only two reasons:

Damage to one of the components

The seal faces open prematurely.

We will start with damage. This damage is almost always visible. Look for :

Corrosion - The elastomer swells or the other seal parts become "sponge like" or pitted.

The product you are sealing is attacking one of the seal components.
The attack is coming from the cleaner or solvent used to clean the lines between batches or at the
end of a "run".
The attack is coming from lubricants put on the elastomers or seal faces. Petroleum grease on
Ethylene Propylene O-rings will cause them to "swell up".
Galvanic corrosion - Happens with dissimilar materials in physical contact and connected by an
electrolyte. As an example: stainless steel can attack the nickel binder in a tungsten carbide face.
Oxidizers and Halogens attack all forms of carbon including black O-rings.


Physical damage.

Wear or rubbing of a flexible component.

Thermal shock of some seal face materials. Especially those that are hard coated or plated.
Thermal expansion of the shaft or sleeve can break a stationary seal face or interfere with the free
movement of a dynamic elastomer.
The rotating seal hits something because of shaft deflection.
Temperature extremes (both high and cryogenic) will destroy elastomers and some seal face
Erosion from solids in the product you are pumping.
Fretting caused by the dynamic elastomer removing the passivated layer from the corrosion
resistant shaft or sleeve. (1 of 6) [12/22/2004 8:48:32 AM]

Seal troubleshooting

Fluid abrasion that can weaken materials and destroy critical tolerances.
A discharge recirculation line circulates high velocity liquid with entrained solids that can break a
metal bellows and injure lapped seal faces, as well as interfere with the free movement of the seal.
The elastomer or rubber part can swell and breaks the face.
Problems at installation. This includes mishandling, setting at the wrong compression, putting the
wrong lubricant on the elastomer etc.
Fatigue of the springs caused by misalignment.

The seal faces opening prematurely is the second cause

Scoring or wear of the hard face is the most common symptom of this failure. The scoring occurs
because the solids imbed into the softer carbon face after they open. The seal faces must stay in contact,
but there are all kinds of conditions that are trying to force or pull them open.

Physical causes

Axial shaft movement (end play or thrust). This is normal at start up.
Radial shaft movement (run out or misalignment)
Operating off of the pump's best efficiency point.
Hysteresis caused by a viscous (thick) product.
Centrifugal force tries to separate the faces in a rotating seal application.
Hydrodynamic forces generated between the lapped faces.
Pressure distortion caused during pressure peaks such as water hammer and cavitation.
Thermal distortion that can cause the seal face to separate from its holder or "go out of flat".
A failure to provide equal and opposite clamping across the stationary seal face will cause
A hardened sleeve can cause the seal set screws to slip.
A wrong initial setting of the face load.
Springs can clog if they are located in the product.
Loose set screws. If the sleeve is too soft they can vibrate out.
Shaft tolerance and finish is out of specifications.
The rotating shaft or seal hits something.
The discharge recirculation line can force open the faces.
Outside springs painted by maintenance people.
A cartridge seal installation method can compress one set of faces and open the other.
Fretting hang up.
Cartridge mounted stationary seals move excessively unless they have some type of "built in" self
aligning feature.

Product problems . With a loss of an environmental control the fluid can: (2 of 6) [12/22/2004 8:48:32 AM]

Seal troubleshooting

Vaporize between the lapped faces forcing them open and causing a "chipping" of the carbon
outside diameter as well as leaving solids between the lapped faces.
Become viscous preventing the faces from following normal "run out".
Solidify between the lapped faces or around the faces.
Crystallize between the faces or around the dynamic portions of the seal.
Build a film on the sliding components or between the faces causing them to separate.
Be a slurry and/ or abrasive
Operate in a vacuum causing the ingestion of air between the faces of some unbalanced seal
Swell up the dynamic elastomer, locking up the seal .
Cause slipstick between the faces if the sealed fluid is a non, or poor lubricant

The common causes of shaft displacement.

Operating off the pump's best efficiency point (B.E.P.).

Misalignment between the pump and its driver.
The rotating assembly is out of balance.
A bent shaft.
A non concentric sleeve or seal.
Slip stick
Passing through, or operating at a critical speed.
Water hammer in the lines.
The stuffing box is not square to the shaft, causing misalignment problems.
Pipe strain.
An impeller adjustment is made to compensate for normal impeller wear.
Thermal growth of the shaft in both a radial and axial direction.
Bad bearings or a poor bearing fit.
Two direction axial thrust at start up is normal.
The motor is finding its magnetic center.
Cavitation - there are five separate types of damage that can be observed.
The sleeve moved when the impeller was tightened.
The unit is pulley driven causing excessive side thrust
The impeller is positioned too far from the bearings. This is a severe problem in mixer or agitator

How to preventing product problems that cause premature seal failure.

Control the environment in the stuffing box. (3 of 6) [12/22/2004 8:48:32 AM]

Seal troubleshooting

Control the temperature in the seal area

Use the correct spring or bellows compression.
Use only hydraullically balanced seals.
Select a low friction face combination.
Avoid "dead ending" the stuffing box.
Jacket the stuffing box
Quench behind the seal with the correct temperture steam or fluid
Use a gland jacket
Utilize two seals with a barrier fluid between them
Use heat tape around the stuffing box
Use a heat pipe to remove heat from the stuffing box.
Vent the stuffing box, especially in a vertical application
Flush in a cool compatible liquid.
Control the pressure in the seal area
Be sure to use only hydraulically balanced seals.

Discharge recirculation will raise the pressure if you put a restrictive bushing into the

bottom of the stuffing box.

Suction recirculation will lower the pressure in the stuffing box.

Use two seals and let the barrier fluid control the pressure between the seals.

Cross connect the stuffing boxes to equalize the stuffing box pressures in a multi stage

Stage the stuffing box pressure with tandem seals.

Impeller pump out vanes will lower stuffing box pressure.

Give the seal more radial space

Bore out the existing stuffing box if it is possible.

Make or buy a new back plate with the large stuffing box cast into it.

Make or buy a large bore stuffing box and attach it to the back plate after you have

machined the old one off.

Flush the product away if you are unable to control it.
Suction recirculation will bring fluid into the stuffing box from behind the impeller, where

it is usually cleaner. This works on most closed impeller pump applications and those open
impeller pump applications where the impeller adjusts to the volute rather than the back
Flush with a clean liquid from an outside source.

A pressurized barrier fluid between two seals can keep solids from penetrating between the

faces if the faces should open. This application will also work if the solid particles are less
than one micron in diameter (Kaoline is such a product).

Build the seal to compensate for operating extremes.

Slurry features that can be part of seal design.

Springs out of the fluid (4 of 6) [12/22/2004 8:48:32 AM]

Seal troubleshooting

Teflon coating the metal parts so particles will not stick to sliding components.
The elastomer moves to a clean surface as the face wears.
Keep the sealing fluid on the outside diameter of the seal to take advantage of centrifugal force
that will throw solids away from the lapped faces.
Rotate the fluid with the seal to prevent erosion of the seal components. A simple vane
arrangement can accomplish this.
Use two hard faces if you find it impossible to keep the lapped seal faces together.
Use a pumping ring to keep solids away from the faces.
Mount the seal closer to the bearings to diminish the affect of shaft deflection.

Design for higher temperature capability

Eliminate elastomers when ever possible.

If you cannot eliminate elastomers, the O-ring location becomes important. Try to move the
elastomer away from the faces.
Hydraulically balanced seals generate less heat.
Select low friction faces.
Fool proof, correct installation dimensions are necessary. A cartridge design is your best choice.
Keep a good product circulation around the components.
A good lapping technique will keep the faces flat at high and cryogenic temperatures.
Pumping rings will keep fluid circulating between two seals. If you are using balanced seals a
simple convection tank is usually more than adequate. An air operated diaphragm pump can be
used in the line to increase the circulation. Try to avoid the use of petroleum based fluids as the
barrier or buffer fluid between the seals. Petroleum based fluids have a very low specific heat that
will increase the temperature between the seals,
Gland features such as quenching, recirculation, venting and flushing help.
Choose well designed faces that will resist thermal distortion. The closer you get to a "square
block" design, the better off you are going to be.
Do not insulate the faces with an elastomer.

Design for pressure resistance

Limit the number of diameters in any single seal component

Laminated bellows will allow you to keep a low spring rate while maintaining pressure capability,
if you are using a welded metal bellows design.
Finite element analysis of the seal components will prevent pressure and temperature distortion.
Use more mass to resist hoop stresses.
Higher modulus materials will resist bending and deformation.
Use a tandem seal design for pressure break down between two seals.

Design for corrosion resistance (5 of 6) [12/22/2004 8:48:32 AM]

Seal troubleshooting

Choose good materials, clearly identified by type and grade.

Eliminate elastomers when possible. Elastomers are the most corrosion sensitive part of the seal.
Design non stressed parts when ever possible
Try not to weld any of the metal components. If it is necessary, monitor the temperature to
prevent inter granular corrosion
Control the temperature. Corrosion increases with temperature.
Use non metallic materials for non metallic equipment.
Watch out for galvanic corrosion when using dissimilar materials.
Do not use stainless steel springs. Stick with Hastelloy "C" if the metal parts of the seal are
manufactured from iron, steel, stainless steel, or bronze. If the seal is manufactured from a
different metal, use springs manufactured from that material.
Do not depend upon flushing to provide corrosion resistance. Use the correct materials, keeping in
mind that solvents and steam are sometimes used to flush the lines. Any materials that you select
must be compatible with these flushing or cleaning fluids also.

If you need cryogenic capability

Go to a welded metal bellows configuration to eliminate all elastomers.

You will need a special carbon/ graphite face that has an organic material impregnated to assist in
the release of the graphite.
Avoid plated or coated hard faces. Differential expansion will cause them to crack.
Always lap the faces at a cryogenic temperature.
Do not coat the faces with grease or oil. It will freeze at cryogenic temperatures.

Link to Mc Nally home page (6 of 6) [12/22/2004 8:48:32 AM]