EMG 3207 Notes DeKUT

3 Flows in Turbines and Pumps

Fluid machinery (add or subtract energy from the fluid) can generally be classified into:
rotodynamic and positive displacement machines. Examples of machines that add energy to
the fluids are pumps, fans and compressors, while machines that extract energy from fluid
are turbines.
In rotodynamic machines, there is a free passage of fluid between the inlet and outlet of
the machine without any intermittent sealing taking place. All rotodynamic machines have
a rotating part called a runner, impeller or rotor.
In positive displacement machines, fluid is drawn or forced into a finite space bounded
by mechanical parts and is then sealed in it by some mechanical means. The fluid is then
forced out or allowed to flow out from the space and the cycle is repeated.
In this section we are going to looked at the fluid flow in rotodynamic machines and
the relationships between the rate of fluid flow and the difference in total head across the
impeller.

3.1 Types of Rotodynamic Machines

The direction of fluid flow in relation to the plane of impeller rotation distinguishes different
classes of rotodynamic machines. This include:

1. Axial flow machines - The flow is perpendicular to the impeller, i.e the flow is along its
axis of rotation; see Fig.12(a).

2. Centrifugal/radial - The fluid approaches the impeller axially but it turns at the ma-
chine inlet so that the flow through the impeller is in the plane of the impeller rotation;
see Fig.12(b).

3. Mixed flow machines - The flow is partly axial and partly radial.

3.2 One-Dimensional Theory

The real flow through an impeller is three-dimensional, i.e., the velocity of the fluid is a
function of three positional coordinates. The velocity distribution, therefore is very complex

66
EMG 3207 Notes DeKUT

Figure 12: Axial and centrifugal Impellers

and depend upon the number of blades, their shapes and thickness and the width of the
impeller.
The one dimensional theory simplify the problem by making the following assumptions:

1. The blades are infinitely thin and the pressure difference across them is replaced by
imaginary body forces acting on the fluid and producing torque.

2. The number of blades is infinitely large, so that the variation of velocity across blade
passages is reduced and tends to zero (axisymmetrical flow), i.e

∂v
= 0.
∂θ

3. There is no variation of velocity in the meridional plane, i.e across the width of the
impeller. Thus,
∂v
= 0.
∂z

The result of these assumptions is that the flow velocity is a function of radius alone, i.e
v = f (r).

3.2.1 Euler’s Turbine Equation

This is the general expression for the energy transfer between the impeller and the fluid based
on the one-dimensional theory.

67
EMG 3207 Notes DeKUT

Figure 13: Centrifugal impeller rotating at ω

Consider a centrifugal impeller rotating with constant angular velocity ω, as shown

below (Fig. 13);

At inlet, the fluid moving with an absolute velocity v1 enters the impeller through a
cylindrical surface of radius r1 and may make an angle α1 with the tangent at that radius.
At outlet, the fluid leaves the impeller through a cylindrical surface of radius r2 , absolute
velocity v2 and may make an angle α2 with the tangent at that radius.
From Newton’s second law applied to angular motion,

Torque = Rate of change of angular momentum = ṁvw2 r2 − ṁvw1 r1 = ṁ(vw2 r2 − vw1 r1 ),

where ṁvw1 r1 and ṁvw2 r2 are the angular momentum entering and leaving the impeller per
second, respectively.

68
EMG 3207 Notes DeKUT

The work done in unit time is given by:

Work done per second, Et = torque × angular velocity = ṁ(vw2 r2 − vw1 r1 )ω.

But u1 = ωr1 and u2 = ωr2 . Hence on substitution,

Work done per second, Et = ṁ(u2 vw2 − u1 vw1 ).

The unit is joules per second or watts.

The specific energy (energy per unit mass) is

Es = (u2 vw2 − u1 vw1 ),

and the Euler’s head E is given by

E = (1/g)(u2 vw2 − u1 vw1 ).

But vw1 = v1 cos α1 and vw2 = v2 cos α2 , so that

E = (1/g)(u2 v2 cos α2 − u1 v1 cos α1 ).

Using cosine rule,

2
vr1 = u21 + v12 − 2u1 v1 cos α1 ,
1
u1 v1 cos α1 = (u21 + v12 − vr1
2
).
2
Similarly,
1
u2 v2 cos α2 = (u22 + v22 − vr2
2
).
2
Substituting these expressions, then
1  2 
E=( (u2 + v22 − vr2
2
) − (u21 + v12 − vr1
2
) ,
2g
or
v22 − v12 u22 − u21 vr1
2 2
− vr2
E= + + ,
2g 2g 2g
| {z } | {z } | {z }
A B C
where A is the increase of kinetic energy of the fluid in the impeller, B is the energy used
in setting the fluid in circular motion about the impeller axis (forced vortex) and C is the
regain of static head due to a reduction of relative velocity in the fluid passing through the
impeller.

69
EMG 3207 Notes DeKUT

3.2.2 Application of Euler’s Equation on a Centrifugal Pump/Fan Impeller

Consider a centrifugal pump/fan impeller, as shown below;

Figure 14: Centrifugal pump or fan impeller

Since the flow at inlet and outlet is through cylindrical surfaces and the velocity compo-
nents normal to them are vf 1 and vf 2 , the continuity equation applied to inlet and outlet
is
m = ρ1 πr1 b1 vf 1 = ρ2 πr2 b2 vf 2 ,

and for incompressible flow,

πr1 b1 vf 1 = πr2 b2 vf 2 .

At inlet the following assumptions are made:

1. The absolute velocity is radial. Therefore, v1 = vf 1 and vw1 = 0.

2. The blade angle at inlet β1 is such that the blade meets the relative velocity tangentially.
This assumption is known as ’no shock’ condition.

At outlet the following assumptions are made:

70
EMG 3207 Notes DeKUT

1. The fluid leaves the impeller with a relative velocity tangential to the blade at outlet.
Thus, β2′ = β2 . Thus from the velocity triangle at outlet,

u2 − vw2
cot β2 = ,
vf 2

so that
vw2 = u2 − vf 2 cot β2 .

Therefore applying these assumption in the Euler’s equation

E = (u2 /g)(u2 − vf 2 cot β2 ). (163)

The total amount of energy transferred by the impeller is, thus

Et = ṁgE = ṁu2 (u2 − vf 2 cot β2 ). (164)

3.2.3 Application of Euler’s Equation on an Axial Flow Machine

Consider an axial flow machine, as shown below (Fig. 15);

Figure 15: Axial flow impeller and velocity triangles

71
EMG 3207 Notes DeKUT

The changes in velocity from inlet to outlet take place at the same radius and, hence

u1 = u2 = u = ωr.

Since the flow area is the same at inlet and outlet,

vf 1 = vf 2 = vf .

It is obtained from
m = ρvf π(R22 − R12 ).

The following assumptions are made with regard to velocity triangles:

1. There is no prewhirl at inlet and, hence, α1 = 900 , vw1 = 0, and v1 = vf .

2. The blade is set at an angle such that it meets the relative fluid velocity tangentially.

3. At the outlet, the relative velocity leaves the blade tangentially.

Thus from the velocity triangle at outlet,

vw2 = u − vf cot β2′ ,

which, on substitution into the Euler’s equation, gives

E = (u/g)(u − vf cot β2′ ). (165)

This equation applies to any particular radius r and is not necessarily constant over the
range from R1 to R2 . However, the blade can be twisted so that the increase in u with radius
is counterbalanced by an equal decrease in vf cot β2′ and, hence the equation is constant
throughout.
Exercise

3.3 Isolated Blade and Cascade Considerations

In these approaches, the number of blades is considered to be finite. For this method we
define:

72
EMG 3207 Notes DeKUT

1. Pitch - The distance between the adjacent blades s.

2. Blade solidity, σ - The ratio of the blade chord c to the pitch. It is the measure of the
closeness of blades.
c
σ=
s

If the blades are close to one another, the flow between them may be treated as conduits.
Whereas if the blades are very far apart, they must be treated as bodies in external flows.

3.3.1 Isolated Blade Considerations

In this case, the assumption is that the blades are very far apart, i.e s → ∞ and σ → 0.
In section 3.6.1, it was shown that lift is dependent on circulation. From Eqs (203) and
(204),
1
Γ = U∞ Acl . (166)
2
Consider the circulation around a single blade as shown below (Fig. 16);

Figure 16: Circulation around the blade

73
EMG 3207 Notes DeKUT

I I B I C I D I A
ΓABCD = vds = vds + vdl + vds + vdl. (167)
A B C D

But,
I C I A I B I D
vdl = − vdl, vds = −vw1 s1 , and vds = vw2 s2 .
B D A C

Therefore, the circulation around a blade, Γb = ΓABCD is given by

Γb = vw2 s2 − vw1 s1 . (168)

Now, consider circulation around two blades, as shown below (Fig. 17):

Figure 17: Circulation around two blades

I I B I C I D I E I F I E
ΓACDF = vds = vds + vds + vdl + vds + vds + vdl. (169)
A B C D E F

74
EMG 3207 Notes DeKUT

But,
I D I A I B I F I C I E
vdl = − vdl, vds + vds = Γb , and vds + vds = Γb ,
C F A E B D

so that,
Γ = ΓACDF = Γb + Γb = 2Γb . (170)

And for number of blade, say z;

Γ = zΓb = z(vw2 s2 − vw1 s1 ), (171)

but zs1 = 2πr1 and zs2 = 2πr2 , so that

Γ = 2π(vw2 r2 − vw1 r1 ). (172)

However, Euler’s Equation is

E = (ω/g)(vw2 r2 − vw1 r1 ). (173)

Comparing Eqs (172) and (173), we obtain

Γ Eg
= . (174)
2π ω

or
ωΓ ωzΓb
E= = . (175)
2πg 2πg
This equation may be used in conjunction with Kutta-Joukowsky’s Eq. (166).

3.3.2 Cascade Considerations

In this case, the assumption is that the blades are very close, i.e the solidity σ is significant.
The geometrically identical blades are arranged at the same distance from one another and
positioned in the same way with respect to the direction of flow to form a cascade.
A straight cascade (applied in an axial flow impeller) is where the blades are arranged
along a straight line. While circular cascade (applied in a centrifugal impeller) is where
the blades are arranged around the circumference.
Cascade can also be classified according to the way the operate as follows:

75
EMG 3207 Notes DeKUT

1. Impulse Cascade - when velocities upstream and downstream of a cascade are the
same in magnitude, so that the change in momentum is due to change in direction.
From Bernoulli’s equation there will be no pressure difference between upstream and
downstream side of the cascade.

2. Reaction Cascade - The pressure difference exists due to absolute velocities upstream
and downstream not being the same.

76
EMG 3207 Notes DeKUT

Straight Cascade
Consider a stationary straight cascade of height Z, as shown below (fig. 18):

Figure 18: A straight cascade

Let the upstream fluid velocity v1 making an angle α1 with the line of the cascade be
deflected so that the downstream velocity v2 makes an angle α2 . The deflection is

ε = α2 − α1 .

The fluid velocities v1 and v2 are resolved into components parallel and normal to the cascade,
vt and vn , respectively. Applying Bernoulli’s equation, the pressure difference is given by
1
p1 − p2 = ρ(v22 − v12 ). (176)
2
Applying continuity, the mass flow through the cascade is

ṁ = sZρ1 vn1 = sZρ2 vn2 .

It implies that for an incompressible flow (density is constant)

vn1 = vn2 = vn .

Thus, the change in velocity is entirely due to change of tangential velocity component, i.e

v22 = vt2
2
+ vn2 , and v12 = vt1
2
+ vn2 ,

77
EMG 3207 Notes DeKUT

⇒ v22 − v12 = vt2

2 2
− vt1 .

Therefore, Eq. (176), reduces to

1 2
p1 − p2 = ρ(vt2 2
− vt1 ) = ρvt (vt2 − vt1 ), (177)
2
where vt = (1/2)(vt1 + vt2 ) is the mean tangential velocity. The force acting on a single blade
of the cascade in the direction perpendicular to it is

Fn = sZ(p1 − p2 ) = sZρvt (vt2 − vt1 ) = ρZvt Γb .

The rate of change of momentum across the cascade is again due to a change in vt and,
therefore, gives rise to a force in the direction of cascade, Ft , i.e

Ft = ṁ(vt2 − vt1 ) = sZρvn (vt2 − vt1 ) = ρZvn Γb .

The resultant force on the blade is

q q
F = Ft2 + Fn2 = ρZΓb (vt2 + vn2 ),

and it act at an angle β given by

Fn ρZvt Γb vt
cot β = = = . (178)
Ft ρZvn Γb vn
But vt1 = vn cot α1 and vt2 = vn cot α2 . Therefore,
vn
vt = (cot α1 + cot α2 ) ,
2
which on substituting into Eq. (178), result in
1
cot β = (cot α1 + cot α2 ) . (179)
2
β is equal to β∞ , defined as the mean direction of flow and obtained by superposition of the
inlet and the outlet velocity triangles, as shown below:

The force F being perpendicular to the mean direction of flow is the lift on the blade.
Exercise

78
EMG 3207 Notes DeKUT

3.4 Energy Losses

There are two main source of losses in hydraulic machine:

1. The velocities in the blade passages and at the impeller outlet are not uniform owing
to the presence of blade and the real flow being three dimensional. This results in a
diminished velocity of whirl component and, hence, reduces Euler’s head.

2. There are losses of energy due to friction, separation and wakes associated with the
development of boundary layers.

4 Boundary Layer Theory

Boundary layer is a thin viscous region between the solid body and fluid. This arises from
”no slip condition” at the solid boundary, i.e, zero relative velocity between fluid and wall.
The fluid velocity ought to resume its normal value at a very short distance from the wall.
This result in a large velocity gradient in this region and hence friction plays a significant
role. For the vast region of flow field away from the body, the velocity gradients are relatively
small, and frictions plays virtually no role. This natural division of the flow into regions;
one where friction is much more important than the other, was recognized by the famous
German fluid dynamist Ludwig Prandtl in 1904.

4.1 Boundary Layer Properties

Consider the viscous flow over a flat plate as shown below; see Fig. 19:
The viscous effects are contained within a thin layer adjacent to the surface, the thickness
is exaggerated for clarity. In the diagram, vw = 0 is the flow velocity at the surface which is
zero (no slip condition), Tw is the wall temperature (the temperature of the fluid immediately
at the surface is equal to temperature of surface), δ is the velocity boundary-layer thickness,
and δT is the thermal boundary-layer thickness.
δ is defined as the distance above the wall where u = 0.99ue . Here ue is velocity at the outer
edge of the boundary layer. For the flow over a flat plate ue = v∞ . For a body of general
shape, ue is the velocity obtained from an inviscid flow solution evaluated at the body surface.

79
EMG 3207 Notes DeKUT

Figure 19: Development of the boundary layer along a flat plate, illustrating variations in
layer thickness and wall shear stress

δT is defined as the distance above the wall where T = 0.99Te . Here Te is temperature at the
outer edge of the thermal boundary layer. For the flow over a flat plate Te = T∞ . For a body
of general shape, Te is obtained from an inviscid flow solution evaluated at the body surface.
At any x station, the variation of u between y = 0 and y = δ, i.e, u = u(y) is defined as the
velocity profile within the boundary layer. This profile is different for different x stations.
Similarly, the flow temperature will change above the wall, ranging from T = Tw at y = 0 to
T = 0.99Te at y = δT . At any given x station, the variation of T between y = 0 and y = δT ,
i.e, T = T (y) is called temperature profile within the boundary layer.
The relative thickness of δ and δT depend on the Prandtl number, Pr. Various research have
shown, that if P r = 1, then δ = δT ; if P r > 1, then δT < δ; if P r < 1, then δT > δ.
For air at standard conditions, P r = 0.71, hence the thermal boundary layer is thicker than
the velocity boundary layer, as shown in the above diagram.
The consequent of velocity gradient at the wall is the generation of shear stress at the wall,
i.e !
∂u
τw = µ , (180)
∂y w
 
where ∂u
∂y w
is velocity gradient at y = 0 (at the wall).

80
EMG 3207 Notes DeKUT

Similarly, the temperature gradient at the wall generate heat transfer at the wall,
!
∂T
qw = −k , (181)
∂y w
 
where ∂T
∂y
is the temperature gradient at y = 0 (at wall).
w

Definitions

1. Displacement thickness, δ ∗
Consider point y1 above the boundary layer as shown below; see Fig .??:

The actual mass flow (per unit depth) across the vertical line connecting y = 0 and
y = y1 , mB is
Z y1
mB = ρudy (182)
0

If there were no boundary layer, so that the speed would be ue everywhere, then the
mass flow, mideal , would be
Z y1
mideal = ρe ue dy (183)
0

The difference between the two, will bring a ”missing mass flow”, i.e given as

Missing mass flow = ρe ue δ ∗ (184)

Missing mass flow = mideal − mB

Therefore,
Z y1
ρ e ue δ =
∗
(ρe ue − ρu)dy (185)
0

Z y1 !
ρu
⇒δ =∗
1− dy δ ≤ y1 → ∞
0 ρ e ue

2. Momentum thickness , θ
Consider the same diagram for analysis of displacement thickness.
The momentum flow across a small segment dy, MB is given by

MB = udm = ρu2 dy (186)

81
EMG 3207 Notes DeKUT

The same elemental mass, outside the boundary layer has a momentum, M ideal given
by
Mideal = ue dm = ρuue dy (187)

The total decrement in momentum flow across the vertical line from y = 0 to y = y1
result in a missing momentum given as
Missing momentum = ρe u2e θ, i.e
Z y1
ρe u2e θ = (ρuue − ρu2 )dy (188)
0
Z y1
ρu u
 
⇒θ= 1− dy δ ≤ y1 → ∞
0 ρe ue ue

3. Energy thickness, δ E
Assignment 1: show that
Z y1 2 !
ρu u

δ =
E
1− dy (189)
0 ρe ue ue

4.2 Laminar Boundary Layer

The flow in a boundary layer can be laminar or turbulent. As far as boundary layer is
concerned, the transition from laminar to turbulent occurs at Reynold numbers above 105
based on mean fluid velocity and distance measured from the entry to the duct. In this
section we shall consider properties of the laminar boundary layer.

4.2.1 Equation of Motion of Fluid in a Laminar Boundary Layer

For simplicity, we shall consider two-dimensional flow along a plane portion of the surface.
This plane is taken as xz-plane, with x-axis in the direction of flow. The velocity distribution
is independent of z, and velocity has no z-component. We also assume that the flow is steady.
The exact Navier-Stokes equations and continuity equations are then

1 ∂p
!
∂vx ∂vx ∂ 2 vx ∂ 2 vx
vx + vy =− +ν + , (190)
∂x ∂y ρ ∂x ∂x2 ∂y 2

1 ∂p
!
∂vy ∂vy ∂ 2 vy ∂ 2 vy
vx + vy =− +ν + , (191)
∂x ∂y ρ ∂y ∂x2 ∂y 2

82
EMG 3207 Notes DeKUT

∂vx ∂vy
+ = 0. (192)
∂x ∂y
Here vx and vy denotes the velocity components in x and y directions, respectively, p is the
pressure and ν is the kinematic viscosity.
Since the boundary layer is very thin, it is clear that the flow in it takes place mainly parallel
to the surface, i.e, the velocity component vy is small compared to vx .
The velocity varies rapidly along y-axis, an appreciable change in it occurring at distances of
the order of the thickness δ of the boundary layer. Along x-axis, the velocity varies slowly, an
appreciable change in it occurring only over at distances of the order of length l characteristic
to the problem (the dimension of the body). Hence y-derivatives of the velocity are large
in comparison to x-derivatives. Therefore, ∂ 2 vx /∂x2 may be neglected in comparison with
∂ 2 vx /∂y 2 . Also comparing Eqs. 190 and 191, it can be seen that derivative ∂p/∂y is small in
comparison with ∂p/∂x, i.e it can be assume that ∂p/∂y = 0. This implies that pressure is
a function of x only, i.e p(x) and instead of writing ∂p/∂x, it is written as a total derivative
dp(x)/dx. This derivative can be be expressed in terms of velocity of the main stream U (x).
Since we have potential flow outside the boundary layer, Bernoulli’s Equation, p + (1/2)ρU 2
= constant, hold, whence (1/ρ)dp/dx = −U dU/dx.
Thus we obtain the equations of motion in the laminar boundary layer in the form of Prandtl’s
Equation:
1 ∂p
!
∂vx ∂vx ∂ 2 vx dU
vx + vy −ν =− =U , (193)
∂x ∂y ∂y 2 ρ ∂x dx
∂vx ∂vy
+ = 0. (194)
∂x ∂y

4.2.2 Flow Over a Flat Plate (The Blasius solution)

Consider the incompressible, two-dimensional flow over a flat plate at 00 angle of attack as
shown below:

For such flow density is constant, µ is constant and dp/dx = 0 (because the inviscid flow

83
EMG 3207 Notes DeKUT

over a flat plate at 00 angle of attack yields a constant pressure over the surface). Energy
equation is not needed to calculate the velocity field for an incompressible flow.
Hence the boundary layer equations, reduces to
!
∂vx ∂vx ∂ 2 vx
vx + vy −ν =0 (195)
∂x ∂y ∂y 2
∂vx ∂vy
+ = 0. (196)
∂x ∂y
The independent variables (x,y) are transformed to (ξ, η), where
yq √
ξ = x, η= Rex , and ψ(x, y) = νxU f (η) (197)
x
and u = ∂ψ/∂y and v = −∂ψ/∂x (stream function definitions). A similarity solution can be
obtained as
2f ′′′ + f f ′′ = 0 (198)

The wall shear stress, τw is generally expressed in terms of nondimensional local skin friction
coefficient, cf as
τw
cf = (199)
(1/2)ρU 2
which can be shown to be
0.664
cf = q (200)
Rex
where Rex is the local Reynolds number. The total drag force on the top surface of entire
plate is the integrated contribution of τw (x) from x = 0 to x = c. Letting Cf denote the skin
friction drag coefficient, we obtain
1Z c
Cf = cf dx, (201)
c 0
which can be shown to be
1.328
Cf = q (202)
Rec
where Rec is the Reynolds number based on the total plate length c.
It can also be shown that for laminar flow, the boundary thickness δ is given by
δ
= 5.48Re−0.5 (203)
x
Example

84
4.3 Turbulent Boundary Layer
−1/5
Cf = 0.074Rel (204)

It can also be shown that for turbulent flow, the boundary thickness δ is given by

δ = 0.37xRe−1/5
x (205)

85