CHAPTER 3

HYDRAULIC PUMPS

3.1 Introduction

The term 'hydraulics' is defined as the science of the conveyance of

liquids through pipes. Most of the theory applicable to hydraulic pumps has

been derived using water as the fluid medium but this by no means

precludes the use of other liquids. Two types of pumps commonly used are

centrifugal and axial flow types, so named because of the general nature of

the fluid flow through the impeller. Both work on the principle that the

energy of the liquid is increased by imparting tangential acceleration to it as

it flows through the pump. This energy is supplied by the impeller, which in

turn is driven by an electric motor or some other drive. In order to impart

tangential acceleration to the liquid, rows of curved vanes or blades move

transversely through it and the liquid is pushed sideways as it moves over

the vanes as well as retaining its original forward component of velocity.

Figure 3.1 showed typical centrifugal and axial flow pump impellers, while

between these two extremes lie mixed flow pumps, which are a combination

of centrifugal and axial flow pumps, part of the liquid flow in the impeller

being axial and part radial.

28
 The centrifugal and axial flow pumps will be dealt with in turn in the

following sections. However, before considering the operation of each type

in detail, we will look at a general pumping system, which is common to

both types. This is shown in Fig. 3.2 where a pump (either axial or

centrifugal) pumps liquid from a low to a high reservoir.

At any point in the system, the pressure, elevation and velocity can be

expressed in terms of a total head measured from a datum line. For the lower

reservoir the total head at the free surface is HA and is equal to the elevation

of the free surface above the datum line since the velocity and static gauge

pressure at A are zero. The liquid enters the intake pipe causing the head loss

hin, with the result that the total head line drops to point B. As the fluid flows

from the intake to the inlet flange of the pump at elevation Z i, the total head

drops further to the point C due to pipe friction and other losses hfi. The fluid

enters the pump and energy is imparted to it, which raises the total head to

point D at the pump outlet. Flowing from the pump outlet to the upper

reservoir, friction and other losses account for a total head loss hfo down to

point E, where an exit loss hout occurs when the liquid enters the upper

reservoir, bringing the total head at the upper reservoir to point F at the free

surface.

29
 If the pump total inlet and outlet heads are measured at the inlet and

outlet flanges respectively, which is usually the case for a standard pump

test, then

(3.1)

This is the head that would be used for determining the type of pump

that should be selected, and the term “Manometric head” is often used. The

static head Hs is the vertical distance between the two levels in the reservoirs

and from Fig. 3.2 it can be seen that for the pipeline

30
Fig. 3.1 Types and shapes of turbomachines

31
 Fig. 3.2 Diagram of pumping system

where hfi and hfo are the friction head losses at inlet and outlet respectively,

hin and hout are the velocity head in the same order.

It is worth nothing here that, for the same size inlet and outlet diameters, V o

and Vi are the same, and in practice (Zo-Zi) is so small in comparison to (po-

pi)/g that is ignored. It is therefore not surprising to find that the static

pressure head across the pump is often used to describe the total head

developed by pump.

32
3.2 CENTRIFUGAL PUMPS

Figure 3.3 shows the three important parts of a centrifugal pump: (1)

the impeller, (2) the volute casing and (3) the diffuser ring.

The diffuser is optional and may or may not be present in a particular

design depending upon the size and cost of the pump. The impeller is a

rotating solid disc with curved blades standing out vertically from the face

of the disc. The tips of the blades are sometimes covered by another flat disc

to give shrouded blades; otherwise the blade tips are left open and the casing

of the pump itself forms the solid outer wall of the blade passages. The

advantage of the shrouded blade is that flow is prevented from leaking

across blade tips from one passage to another. As the impeller rotates, the

fluid that is drawn into the blade passages at the impeller inlet or eye is

accelerated as it is forced radially outwards. In this way, the static pressure

at the outer radius is much higher than at the eye inlet radius.

The fluid has a very high velocity at the outer radius of the impeller,

and, to recover this kinetic energy by changing it into pressure energy,

diffuser blades mounted on a diffuser ring may be used. The stationary

blade passages so formed have an increasing cross-sectional area as the

fluid moves through them, the kinetic energy of the fluid being reduced

33
while the pressure energy is further increased. Vaneless diffuser passages

may also be utilized.

Fig. 3.3 Centrifugal pump components

Finally, the fluid moves from the diffuser blades into the volute

casing, which collects it and conveys it to the pump outlet. Sometimes

only the volute casing exists without the diffuser; however, some

pressure recovery will take place in the volute casing alone.

In dealing with the theory of hydraulic pumps, a number of

assumptions will be made. At any point within the blade passages the

fluid velocity will in general have three components, one each in the

axial, radial and angular directions as indicated in Fig. 3.4.

The velocity may then be written as a function of the three components

V = f(r,,z)

However, we will assume that the following hold:

34
 Fig. 3.4 Cylindrical coordinates for a centrifugal pump

1. There are an infinite number of blades so closely spaced that .

That is, there is no flow in the blade passage in the tangential direction

and V = 0.

2. The impeller blades are infinitely thin, thus allowing the pressure

difference across them, which produce torque, to be replaced by

tangential forces that act on the fluid.

3. The velocity variation across the width or depth of the impeller is zero

and hence .

35
 4. The analysis will be confined to conditions at the impeller inlet and

outlet, and to the angular momentum change between these two

stations. No account is taken of the condition of the fluid between these

two stations.

5. It is assumed that at inlet the fluid is moving radially after entering the

eye of the pump.

Assumptions 1 and 2 mean that the velocity is a function of the radius only,

V = f(r), and now with these assumptions the velocity vectors at inlet and

outlet of the impeller can be drawn and the theoretical energy transfer

determined. Figure 3.5 shows a centrifugal pump impeller with the velocity

triangles drawn at inlet and outlet. The blades are curved between the inlet

radius r1 and outlet radius r2, a particle of fluid moving along the broken

curve shown. 1 is the angle subtended by the blade at inlet, measured from

the tangent to the inlet radius, while 2 is the blade angle measured from the

tangent at outlet. The fluid enters the blade passages with an absolute

velocity C, and at an angle 1 to the impeller inlet tangential velocity vector

U1 where Ul=r1,  being the angular velocity of the impeller. The resultant

relative velocity of flow into the blade passage is W1 at an angle to the

tangent at inlet. Similarly at outlet the relative velocity vector is W2 at angle

36
 from the tangent to the blade. By subtracting the impeller outlet

tangential velocity vector U2, the absolute velocity vector C2 is obtained, this

being set at angle a2 from the tangent to the blade. It is seen that the blade

angles at inlet and outlet do not equal the relative flow angles at inlet and

outlet. This is for a general case, and unless otherwise stated, it will be

assumed that the inlet and outlet blade angles are equal to their corresponding

relative flow angles.

From Euler's pump equation, the work done per second on the fluid per

unit weight of fluid flowing is

E = W/mg = (U.2Cx2-U1Cxl)/g (J/s per N/s or m) (3.2)

where Cx is the component of absolute velocity in the tangential direction. E is

often referred to as the Euler head and represents the ideal or theoretical head

developed by the impeller only. Now

Cxl = C1cos1 and Cx2 = C2 cos 2

Thus

E = (U2C2 cos 2-U1C1 cos 1)/g (3.3)

But by using the cosine rule, W2 = U2 + C2 – 2UCcos1 , then

37
and

substituting into Eq. (3.3) gives

(3.4)

The terms in Eq. (3.4) may now be examined in turn

represents the increase of the kinetic energy of the fluid

across the impeller, represents the energy used in setting the

fluid into circular motion about the impeller axis and is the

gain of static

head due to a reduction of the relative velocity within the impeller. The flow

rate is

38
where Cr is the radial component of the absolute velocity and is

perpendicular to the tangent at inlet and outlet while b is the width of the

blade (in the z direction). It is usually the case that C 1 = Cr1 and hence 1=

90°. In this case Cx1=0 where Cx1 is the component of the inlet absolute

velocity vector resolved into the tangential direction. Wx and Cx are often

respectively called the relative and absolute whirl components of velocity.

When this is referred to as the 'no-shock condition' at entry. In this

case the fluid moves tangentially onto the blade. When there is no

fluid slip at the exit.

39
 Fig. 3.5 Velocity triangles for centrifugal pump impeller

3.3 SLIP FACTOR

It was stated in the previous section that the angle at which the fluid

leaves the impeller, , may not be the same as the actual blade angle .

This is due to fluid slip and it occurs in both centrifugal pumps and

centrifugal compressors and manifests itself as a reduction in Cx2 m the Euler

40
pump equation. One explanation for slip is that of the relative eddy

hypothesis.

Figure 3.6 shows the pressure distribution built up in the impeller

passages due to the motion of the blades. On the leading side of a blade there

is a high-pressure region while on the trailing side of the blade there is a

low-pressure region the pressure changing across the blade passage. This

pressure distribution is similar to that about an aerofoil in a free stream and

is likewise associated with the existence of a circulation around the blade, so

that on the low-pressure side the fluid velocity is increased while on the

high-pressure side it is decreased, and a non-uniform velocity distribution

results at any radius. Indeed, the flow may separate from the suction surface

of the blade. The mean direction of the flow leaving the impeller is therefore

and not as is assumed in the zero-slip situation. Thus Cx2 is reduced to

and the difference is defined as the slip. Slip factor is defined as

)3.5(

Stodola proposed the existence of a relative eddy within the blade

passages as shown in Fig. 3.7. He proposed that if a frictionless fluid passes

through the blade passages it will, by definition, have no rotation; therefore at the

outlet of the passage rotation should be zero. Now the impeller has an angular

41
velocity  so that relative to the impeller the fluid must have an angular

velocity - in the blade passages to comply with the zero-rotation condition. If

the radius of a circle that may be inscribed between two successive blades at outlet

and at a tangent to the surfaces of both blades is e, then the slip is given by

Now the impeller circumference is 2r2 and therefore the distance between blades

is 2r2/Z if we have Z blades of negligible thickness. This may be

approximated to 2e/sin 2 and upon rearrangement

Fig. 3.6 Slip and velocity distribution in centrifugal pump impeller blades.

42
 Fig. 3.7 The relative eddy between impeller blades.

Now referring back to Fig. 3.6 for the no-slip condition

and substituting into Eq. (3.5) gives

43
 (3.6)

For purely radial blades, which are often found in a centrifugal

compressor, 2 will be 90° and the Stodola slip factor becomes

(3.7)

The Stodola slip factor equation gives best results when applied in the range

20° < 2 < 30°. Other slip factors are named after Buseman

(3.8)

where A and B are functions of 2, Z and r2/r1 and are best used in the range

30° <2<80°. The Stanitz slip factor given by

(3.9)

44
is best used in the range 80°<2<90°

When applying a slip factor, the Euler pump equation becomes

(3.10)

Typically, slip factors lie in the region of 0.9, while slip occurs even if the fluid is

ideal.

2.4 CENTRIFUGAL PUMP CHARACTERISTICS

Dimensionless groups were used to express the power, head and flow

relationships for a hydraulic machine. A well-designed pump should operate at

or near the design point and hence near its maximum efficiency, but the

engineer is often required to know how the pump will perform at off-design

conditions. For instance, the head against which the pump is operating may be

decreased, resulting in an increase in mass flow rate. However, before

examining this aspect of off-design performance, we will look at the losses

occurring in a pump and the differing efficiencies to which these losses give rise,

whether or not the pump is working at the design point. We will then examine

the effects of working at the off-design condition.

45
 2.4.1 Pump Losses

The shaft power, Ps or energy that is supplied to the pump by the prime

mover is not the same as the energy received by the liquid. Some energy is

dissipated as the liquid passes through the machine and the mechanism ol

this loss can be split up into the following divisions.

1. Mechanical friction power loss, Pm due to friction between the fixed and

rotating parts in the bearing and stuffing boxes.

2. Disc friction power loss, Pi due to friction between the rotating faces of the

impeller (or disc) and the liquid.

3. Leakage and recirculation power loss, Pl due to a loss of liquid from the

pump or recirculation of the liquid in the impeller.

4. Casing power loss, Pc

Impeller power loss is caused by an energy or head loss hi in the impeller

due to disc friction, flow separation and shock at impeller entry. This loss is

associated with the flow rate Qi through the impeller, and so the impeller

power loss is expressed as

Pi=gQihi (3.11)

However while the flow through the impeller is Qi this is not the flow through the

outlet or inlet flange of the machine. The pressure difference between impeller tip

46
and eye can cause a recirculation of a small volume of the fluid q, thus reducing

the flow rate at outlet to Q as shown in Fig. 3.8, and then

Q=Qi-q (3.12)

Fig. 3.8 Leakage and recirculation in a centrifugal pump

If Hi is the total head across the impeller, then a leakage power loss can be

defined as

Pl=gqHi (N m/s) (3.13)

Equation (3.12) shows that when the discharge valve of the pump is closed,

then the leakage flow rate attains its highest value.

In following from the impeller outlet to the pump outlet flange, a further

head loss hc takes place in the diffuser and collector, and since the flow rate

here is Q, then a casing power loss may be defined as

47
 Pc=gQhc (N m/s) (3.14)

Summing these losses gives

(3.15)

where the total head delivered by the pump is defined as in Fig. 3.2 and Eq.

(3.1).

A number of efficiencies are associated with these losses:

:The overall efficiency o is given by

(3.16)

Casing efficiency c is given by

c = Fluid power at casing outlet/fluid power at casing inlet

Fluid power at casing outlet/(fluid power developed by impeller =

-leakage loss)

=gQH/gQHi=H/Hi (3.17)

Impeller efficiency i is given by:

i = Fluid power at impeller exit/fluid power supplied to impeller

Fluid power at impeller exit/(fluid power developed by impeller =

+ impeller loss)

48
 )3.18(

:Volumetric efficiency v is given by

v = Flow rate through pump/flow rate through impeller

= (3.19)

:Mechanical efficiency m is given by

m=Fluid power supplied to the impeller/power input to the shaft

)3.20(

,Therefore

)3.21(

A hydraulic efficiency may be defined as

(3.22)

.The head H is also known as manometric head

Where the theoretical head (Hi+hi) is that obtained from Euler's equation and

Figure 3.9 shows how each of the power loss are subtracted from the initial

input power. The rectangle OABC represents the total power input to the

49
shaft while OADEFC is equivalent to the mechanical power loss. The

impeller loss gQihi is next removed and is represented by rectangle EFGI

next loss to be accounted for is the leakage loss pgHiq represented by

rectangle DJKI, and finally the casing loss pghcQ represented by rectangle

MLGK is removed. This leaves us with rectangle JBLM, which represents

the fluid power output or power developed by the pump pgQH.

Fig. 3.9 Losses in a centrifugal pump

3.4.2 The Characteristic Curve

50
 Euler's pump equation gives the theoretical head developed by the

pump, but if it is assumed that there is no whirl component of velocity at

entry then Cx1 = 0 and the actual theoretical head developed is

(3.23)

and if slip is accounted for, Eq. (3.23) becomes

(3.24)

Now

where A is the flow area at the periphery of the impeller and Cr is perpendicular to

it. Thus from Eq. (3.23) the energy per unit weight of flow becomes

and since U2, 2 and A are constants, then

E=K1-K2Q (3.25)

51
and this equation may be plotted as the straight line shown in Fig. 3.10.

If slip is taken into account, it is seen from Eq. (3.9) that as Cr2 increases

(and hence Q) then s decreases, thus reducing the value of E in Eq. (3.25) to

E=(K1-K2Q) s (3.25)

The loss due to slip can occur in both a real and an ideal fluid, but in a real

fluid account must also be taken of the shock losses at entry to the blades, and

the friction losses in the casing and impeller vanes, or indeed at any point

where the fluid is in contact with a solid surface of the pump. At the design

point the shock losses are zero since the fluid would move tangentially onto the

blade, but on either side of the design point the head loss due to shock increases

according to

(3.27)

where QD is the design flow rate. The friction losses are accounted for in the

form

hf= K4 Q2 (3.28)

52
 Equations (3.27) and (3.28) are plotted in Fig. 3.10 and the sum of them is

subtracted from the curve of Eq. (3.26) to give the final characteristic. This curve

is called the head-flow characteristic of the pump.

Fig. 3.10 Centrifugal pump characteristics

3.4.3 Effect of Flow Rate Variation

A pump is usually designed to run at a fixed speed with a design head and

flow rate and these conditions would normally occur at the maximum efficiency

point. However, it is not always the case in practice that the operating point lies

at the design point. This may be due to a pipeline being partially blocked, a valve

jammed partially closed or poor matching of the pump to the piping system.

Also in general a variable-speed motor is not available to correct for any

53
deviation from the design condition, so that in what follows it is assumed that

the speed of the pump remains constant.

Figure 3.11 shows the velocity diagrams that pertain for three possible flow

rates: normal design flow rate, increased flow rate and decreased flow rate. When

the flow rate changes, Cr2 changes, and since U2 is constant and the blade outlet

angle 2 is constant (assuming ), the magnitude of W2 and C2 must

change along with the angle 2. Since the effective energy transfer E depends on

Cx2, then E will change accordingly. Thus a reduction in Q gives an increase in Cx2,

while an increase in Q gives a reduction in Cx2. It follows that, should the head

against which the pump operates be momentarily increased, E and therefore Cx2

increase and Q decreases to give the new operating point at the increased head.

Similarly a reduction in the operating head gives an increase in Q.

Fig. 3.11 Effect of flow rate variation on outlet velocity

54
 At the inlet the effect of flow rate change is to cause eddies on the suction

surface of the blade for a reduced flow rate and on the pressure surface of the

blade for an increased flow rate. The design condition is the 'no-shock'

condition, which corresponds to the flow QD in Fig. 3.10. The corresponding

velocity diagrams can be seen in Fig. 3.12. In all cases it is assumed that Cxl is

zero.

Fig. 3.12 Effect of flow rate variation on inlet velocity

2.4.4 Effect of Blade Outlet Angle

The characteristic curve will also be affected by the blade angle at outlet,

the three types of blade settings being backward-facing, forward-facing and radial

blades. Figure 3.13 shows clearly the velocity triangles for each case with

Cxl=0.

Case (i). Backward-facing blades, 2 <90

55
 Fig. 3.13 Centrifugal pump outlet velocity triangles for varying blade outlet angle

Writing E as a head,

H = a- bQ (3.29)

and for pumps 2 typically lies between 15° and 90°.

Case (ii). Radial blades, 2 = 90°

H=a (3.30)

Case (iii). Forward-facing vanes, 2 > 90°

56
 H = a + bQ (3.31)

where 2 would be typically 140° for a multi-bladed centrifugal fan.

These equations are plotted in Fig. 3.14 as characteristics and they revert

to their more recognized curved shapes (for the reasons previously discussed)

as shown in Fig. 3.15.

For both radial and forward-facing blades the power is rising continuously

as the flow rate is increased. In the case of backward-facing vanes the maximum

efficiency occurs in the region of maximum power, and if, for some reason, Q

increases beyond QD, this results in a power decrease and therefore the motor

used to drive the pump may be safely rated at the maximum power. This is said

to be a self-limiting characteristic. In the case of the radial and forward-facing

vanes, if the pump motor is rated for maximum power, then it will be under-

utilized most of the time, and extra cost will have been incurred for the extra

rating, whereas if a smaller motor is employed rated at the design point, then if Q

increases above QD the motor will be overloaded and may fail. It therefore

becomes more difficult to decide on a choice of motor for these latter cases.

57
Fig. 3.14 Theoretical characteristics for varying outlet blade angle

Fig. 3.15 Actual characteristics for varying outlet blade angle

58
3.5 FLOW IN THE DISCHARGE CASING

The discharge casing is that part of the casing following the impeller

outlet. It has two functions: (i) to receive and guide the liquid discharged

from the impeller to the outlet ports of the pump, and (ii) to increase the

static head at the outlet of the pump by reducing the kinetic energy of the

liquid leaving the impeller. These two functions may be called collector and

diffuser functions. The former function may be used alone while the latter can

occur either before or after the collector function. In addition diffusion can

take place in a vaned or vaneless diffuser.

3.5.1 Volute or Scroll Collector

A simple volute or scroll collector is illustrated in Fig. 3.16 and

consists of a circular passage of increasing cross-sectional area. The

advantage of the simple volute is its low cost. The cross-sectional area

increases as the increment of discharge increases around the periphery of the

impeller and it is found that a constant average velocity around the volute

results in equal pressures around the pump casing, and hence no radial

59
thrust on the shaft. Any deviation in capacity (flow rate) from the design

condition will result in a radial thrust, which if allowed to persist could result

in shaft bending. Values of radial thrust are given by the empirical

relationship

(3.32)

Fig. 3.16 Simple volute or scroll collector

where P = radial force (N), H = head (m), D2 = peripheral diameter (m),

B2 = impeller width (m) and K = constant determined from the following

equation for a particular value of Q:

60
 (3.33)

A cross section of the volute casing is shown in Fig. 3.17. The circular

section is adopted to reduce the losses due to friction and impact when

the fluid hits the casing walls on exiting from the impeller. Of the

available kinetic energy at impeller outlet, 25-30 per cent may be

recovered in a simple volute.

Fig. 3.17 Section through volute casing

61
3.5.2 Vaneless Diffuser

Diffusion takes place in a parallel-sided passage and is governed

by the principle of conservation of angular momentum of the fluid. The

outlet tangential velocity is reduced as the radius increases, while the

radial component of absolute velocity is controlled by the radial cross-

sectional area of flow b. A vaneless diffuser passage is shown in Fig.

3.18.

With reference to Fig. 3.19 the size of the diffuser may be

determined as follows. The mass flow rate m at any radius r is given by

m = ACr = 2rbCr

where b is the width of the diffuser passage perpendicular to the

peripheral area of the impeller and is usually the same as the impeller

width. Letting the subscripted variables represent conditions at the

impeller outlet and the unsubscripted variables represent conditions at

any radius r in the vaneless diffuser, then from continuity

rbCr = r2b22Cr2

or

62
 (3.34)

Fig. 3.18 Vaneless diffuser passage

If frictionless flow is assumed, then by the conservation of angular

momentum

But Cx>>Cr (usually) and therefore the absolute velocity C is

approximately equal to Cx or

(3.35)

63
From Eq. (3.35), for C to be small, which is what we are trying to achieve,

then r must be large and therefore, for a large reduction in the outlet

kinetic energy, a diffuser with a large radius is required.

For an incompressible fluid, the inclination of the absolute velocity

vector to the radial line remains constant at all 6 since at the outlet from

the impeller (Fig. 3.19)

since rCr is constant from the constant mass flow rate requirement, and

Cxr is constant from the conservation of angular momentum requirement.

Thus the flow in the diffuser remains at a constant inclination a' to radial

lines, the flow path tracing out a logarithmic spiral, and if for an

incremental radius dr the fluid moves through angle d0, then from Fig.

3.19.

Integrating,

64
Putting ' = 78° and (r/r2) = 2, the change in angle of the diffuser is

almost 180°, giving rise to a long flow path, which may result in high

frictional losses, which in turn gives a low efficiency. So it is seen that

the length of the diffuser must be balanced by the pressure recovery that

is required and an optimum point is usually found based on either

economic or hydraulic friction loss considerations.

Fig. 3.19 Logarithmic spiral flow in vaneless space

65
3.5.3 Vaned Diffuser

The vaned diffuser shown in Fig. 3.20 is able to diffuse the outlet

kinetic energy at a much higher rate, in a shorter length and with a higher

efficiency than the vaneless diffuser. This is very advantageous where

the size of the pump is important.

A ring of diffuser vanes surrounds the impeller at the outlet, and

after leaving the impeller the fluid moves in a logarithmic spiral across a

short vaneless space before entering the diffuser vanes proper. Once the

fluid has entered the diffuser passage, the controlling variable on the rate

of diffusion is the divergence angle of the diffuser passage, which is of

the order of 8-10° and should ensure no boundary-layer separation along

the passage walls. The number of vanes on the diffuser ring is subject to

the following considerations:

1. The greater the vane number, the better is the diffusion but the greater

is the friction loss.

2. The cross section of the diffuser channel should be square to give a

66
 maximum hydraulic radius (cross-sectional area/channel

perimeter).

3. The number of diffuser vanes should have no common factor with

the number of impeller vanes. This is to obviate resonant or

sympathetic vibration.

Fig. 3.20 A vaned diffuser

67
 The collector and diffuser operate at their maximum efficiency at

the design point only. Any deviation from the design discharge will alter

the outlet velocity triangle and the subsequent flow in the casing. Figure

3.21 shows the contribution of each section of the pump to the total head

developed by the pump.

Fig. 3.21 Head rise across a centrifugal pump

3.6 CAVITATION IN PUMPS

68
 Cavitation is caused by local vaporization of the fluid, when the

local static pressure of a liquid falls below the vapor pressure of the liquid.

Small bubbles or cavities filled with vapor are formed, which suddenly

collapse on moving forward with the flow into regions of high pressure.

These bubbles collapse with tremendous force, giving rise to pressure as

high as 3500 atm. In a centrifugal pump, these low-pressure zones are

generally at the impeller inlet, where the fluid is locally accelerated over

the vane surfaces. In turbines, cavitation is most likely to occur at the

downstream outlet end of a blade on the low-pressure leading face. When

cavitation occurs, it causes the following undesirable effects:

1. Local pitting of the impeller and erosion of the metal surface.

2. Serious damage can occur from this prolonged cavitation erosion.

1. Vibration of machine and noise is also generated in the form of

sharp cracking sounds when cavitation takes place.

2. A drop in efficiency due to vapor formation, which reduces the

effective flow areas.

The avoidance of cavitation in conventionally designed machines

can be regarded as one of the essential tasks of both pump and turbine

69
designers. This cavitation imposes limitations on the rate of discharge and

speed of rotation of the pump.

A cavitation parameter is defined as

c = pump total inlet head above vapor pressure/head developed by the

pump or at the inlet flange

(3.36)

The numerator of Eq. (3.36) is a suction head and is called the net positive

suction head (NPSH) of the pump. It is a measure of the energy available

on the suction side of the pump, and H is the manometric head. The

cavitation parameter is a function of specific speed, efficiency of the

pump, and number of vanes. Figure 3.22 shows the relationship between

c and Ns. It may be necessary in the selection of pumps that the value of

c does not fall below the given value by the plots in Fig. 3.22 for any

condition of operation.

70
 .Figure 3.22 Cavitation limits for radial flow pumps

3.6.1 Net Positive Suction Head (NPSH)

Simply, NPSH is the minimum suction condition (pressure)

required to prevent pump cavitation. Conceptually, NPSH can be

imagined as the pressure drop between the pump inlet flange and the

point inside the pump where the fluid dynamic action, as it leaves the

71
impeller, causes a pressure rise. Sufficient NPSH allows for pumping

without liquid vaporizing in the pump first-stage impeller eye as the fluid

pressure drops due to pump suction losses.

A distinction is made between NPSH value required by the pump

(NPSHreq.) which corresponds to the suction pressure head of the pump

and the available NPSH (NPSHav.) which corresponds to the head of the

installation or plant. The available NPSHav is defined as follows:

Where pt,s is the total pressure at pump suction, pv is the vapor pressure of

the fluid.

For a pumping system with open reservoir:

where pat is the atmospheric pressure, pms is the manometric suction

pressure and in such case:

72
The required net positive suction head of a pump NPSH req. is a specific

value for the given pump and it depends on many design factors such as

shape of the impeller entrance, angles shape of blades, speed and

operation conditions of the pump.

The determination of the NPSH required is largely a matter of

experience involving NPSH tests.

The operation point of the a pump can only be taken as a point of

continuous operation if

In connection with NPSH value “ Thoma Cavitation Parameter” is

defined for the pump as:

3.6.2 Suction specific speed

The efficiency of the pump is a function of flow coefficient and suction

specific speed, which is defined as

Thus,

73
The cavitation parameter may also be determined by the following

equation

(3.37)

3.7 AXIAL FLOW PUMP

An axial flow pump consists of a propeller type of impeller

running in a casing with fine clearances between the blade tips and the

casing walls. In the absence of secondary flows, fluid particles do not

change radius as they move through the pump, however, a considerable

amount of swirl in the tangential direction will result unless means are

provided to eliminate the swirl on the outlet side. This is usually done by

fitting outlet guide vanes. The flow area is the same at inlet and outlet

and the maximum head for this type of pump is of the order of 20m. It

may be seen in Fig. 3.23 that the dimensionless specific speed of axial

74
flow pumps lies at the right-hand side of the pump spectrum, its

characteristics being one of low head but high capacity. The usual

number of blades lies between two and eight, with a hub diameter ratio

of 0.3-0.6. In many cases the blade pitch is fixed but most large

hydroelectric units have variable-pitch blades to allow for load

variations.

Fig 3.23 Characteristic curves for various pump designs.

Figure 3.24 shows an axial flow pump impeller. The section

through the blade at X-X is shown enlarged with the inlet and outlet

velocity triangles superimposed in Fig. 3.25. It will be noticed that the

blade has an aerofoil section and that the inlet relative velocity vector W1

75
does not impinge tangentially but rather the blade is inclined at an angle

of incidence I to the relative velocity vector W1. This is similar to the

angle of attack of an aerofoil in a free stream. It is assumed that there is

no shock at entry and that the fluid leaves the blade tangentially at exit.

Fig. 3.24 An axial flow pump

76
 Fig. 3.25 Velocity triangles

Changes in the condition of the fluid take place at a constant mean

radius; therefore

Assuming also a constant flow area from inlet to outlet

77
and noting that the flow area is the annulus formed between the hub and

the blade tips, then we may write

From the Euler’s pump Eq.

(3.38)

and for maximum energy transfer Cx1=0, i.e. 1 =90º and C1= Ca , the

absolute flow velocity at inlet being axial for maximum energy transfer.

Now

or

Hence substituting for Cx2 in Eq. 3.38 with Cx1=0, the maximum energy

transfer or head is

(3.39)

From constant energy transfer, Eq. 3.39 applies over the whole span

of the blade from hub to tip; that is it applies at any radius r between Rt

and Rh. For E to be constant over the whole blade length it is obvious

78
that, as U2 increases with radius, so an equal increase in UCa cot 2 must

take place and since Ca is constant then cot 2 must increase, and the

blade must therefore be twisted as radius changes.

Strictly speaking the work done per unit weight of flow through an

annulus of thickness dr should be considered and this then integrated

across the whole flow area from the hub to the tip.

or

But incremental mass flow rate dm is

and U=r. Therefore

(3.40)

Equation 3.40 can only be integrated if the relationship between 2 and r

is known.

For design purposes it is usual to select conditions for use in Eq.

3.39 at the mean radius (Rh+Rt)/2 along the blade. The whirl component

79
imparted to the fluid at exit from the blade is relatively small, resulting in

a low kinetic energy loss. An axial flow pump therefore tends to have a

higher hydraulic efficiency than the centrifugal pump.

3.8 PUMP AND SYSTEM MATCHING

It has been shown that a hydraulic pump has a design point at

which the overall efficiency of operation is a maximum. However, it may

happen that the pipe system in which the pump is being used is unsuited

to the pump and a different pump with a more suitable characteristic is

required. This section will examine how a pump and a pipe system may

be matched to each other, the effect of changing the pump speed and

diameter, and finally the effect of connecting pumps in series and

parallel.

Consider the pipe system in Fig. 3.2. On the suction side the losses

expressed in terms of standard loss coefficients are the sum of the minor

losses hin

80
and the friction loss

where f is the Darcy friction factor, li is the length of the inlet pipe and di

its diameter. Thus the total head loss is

On the delivery side the sum of the bend, friction and exit losses

that must beovercome is

Finally, the liquid must be moved from the lower reservoir to the

upper reservoir through the static head Hs; hence the total opposing head

of the pipe system that must be overcome in order to move the fluid from

the lower to upper reservoir is

H = H s + ho + hi (3.41)

Now from the continuity equation, the flow rate through the system is

proportional to the velocity. Thus the resistance to flow in the form of

81
friction losses, head losses, etc., is proportional to the square of the flow

rate and is usually written as

System resistance = KQ2 (3.42)

It is a measure of the head lost for any particular flow rate through the

system. If any parameter in the system is changed, such as adjusting a

valve opening, or inserting a new bend, etc., then K will change. The total

system head loss of Eq. (3.41) therefore becomes

H = HS + KQ2 (3.43)

and if this equation is plotted on the head-flow characteristic, the point at

which Eq. (3.43) intersects the pump characteristic is the operating point,

and this may or may not lie at the duty point, which usually corresponds

to the design point and maximum efficiency. The closeness of the

operating and duty points depends on how good an estimate of the

expected system losses has been made. In Fig. 3.26 the system curve is

superimposed on the H-Q characteristic.

82
 Fig. 3.26 System resistance and pump characteristics

It should be noted that if there is no static head rise of the liquid (e.g.

pumping in a horizontal pipeline between two reservoirs at the same

elevation) then Hs is zero and the system curve passes through the origin.

This has implications when speed and diameter changes take place.

Because of the flatness of rotodynamic pump characteristics, a poor

estimate of the system losses can seriously affect the flow rate and head;

whereas in positive displacement pumps, the H-Q curve is almost vertical

and, even if the head changes substantially, the flow rate stays almost

constant.

83
3.8.1 Effect of Speed Variation

Consider a pump of fixed diameter pumping liquid with zero static lift. If

the characteristic at one speed N1 is known, then it is possible to predict

the corresponding characteristic at speed N2 and also the corresponding

operating points. Figure 3.27 shows the characteristic at speed N1.

For points A, B and C the corresponding head and flows at a new

speed N2 are found thus. We have

and

since D is constant (3.44)

Similarly

and

(3.45)

Applying Eqs (3.44) and (3.45) to points A, B and C and letting the

corresponding points be A', B' and C',

84
 and or

 (3.46)

Fig. 3.27 Effect of speed variation on the operating point

and we see that all corresponding points lie on a parabola passing through

the origin. This means that for an operating point at A at speed Nl ,it is

only necessary to apply the similarity laws directly to find the

corresponding operating point at the new speed since it will lie on the

system curve itself. However, if there is static lift (Hs # 0) it will be

necessary to calculate and then plot the corresponding points A', B', C' at

the new speed, since the system curve will no longer pass through the

origin. The system curve is then drawn to find the new operating point at

85
its intersection with the N2 characteristic such that the corresponding

maximum efficiency at design point DP2 remains the same as at DP1 but

at a different head and flow rate as in Fig. 3.28.

Fig. 3.28 Effect of speed variation on the efficiency

3.8.2 Variation of Pump Diameter

A variation of pump diameter may be similarly examined through the

similarity laws. For a constant speed,

and

86
or

 (3.47)

This curve does not lie on the system characteristic and therefore part

of the new characteristic must be drawn through A', B' and C' at the new

diameter so that the new operating point may be found. The efficiency

curve moves across in a similar manner to before, the corresponding

efficiencies being equal.

3.8.3 Pumps in Series and Parallel

Should the head or flow rate of a single pump not be sufficient for an

application, pumps can be combined in series to obtain an increase in

head, or in parallel for an increase in flow rate. The combined pumps need not

be of the same design.

Figures 3.29 and 3.30 show the combined H-Q characteristic for the cases

of identical pumps connected in series and parallel. It will be observed that

the operating point changes in both cases. In Fig. 3.31 the combined

characteristics of two different pumps connected in parallel and series are

drawn.

87
 When axial flow pumps are connected in parallel, care must be taken to

ensure that the system characteristic does not cut the pump characteristic in

two places, otherwise instability may result. This arises due to the lateral

spreading of the pump characteristic illustrated in Figs 3.32 and 3.33.

88
 Fig. 3.29 Two similar pumps in series

Fig. 3.30 Two similar pumps in parallel

Fig. 3.31 Two different pumps combined in series and parallel

89
 Fig.3.32 Single axial flow pump

Fig.3.33 Axial flow pumps in parallel

90