Chapter 4

FORCED CONVECTION

The concept and definitions of heat transfer coefficients, individual and overall, have been
introduced in Chapter 3. We explained with examples how simple problems involving
convective transport of heat can be solved if the heat transfer coefficient is known. But the
question remains as to how we can obtain the heat transfer coefficient in a given situation. The
heat transfer coefficient is not as simple a quantity as a fundamental property of a body or a
system. It depends upon many factors-the physical properties of the fluid (like thermal
conductivity, density, viscosity), the velocity field, the geometry of the system and its
characteristic dimensions, etc. Fortunately, a large number of empirical correlations have been
developed over the years by heat transfer researchers. These correlations cover a great majority
of situations encountered in industrial practice, and are used for the calculation of heat transfer
coefficients. No doubt, the use of any correlation involves some error-that may be even 25-
30% in some cases. However, in the absence of more reliable data or information in specific
cases, these correlations are indispensable for the estimation of heat transfer coefficients.

In this chapter we will first qualitatively analyze convective heat transfer in a few simple
systems in order to provide further understanding of this important mode of heat transfer.
Later, we will discuss the more important heat transfer correlations and illustrate their uses.
Convective heat transfer may be of two types forced convection and free or natural
convection. Both are caused by motion in the medium. If the motion or velocity in the medium
is generated by the application of an external force (e.g. by a pump, a blower, an agitator, etc.),
heat transfer is said to occur by forced convection. But if the motion in the medium occurs as a
result of density difference (which may be caused by a temperature difference), the concerned
mode of heat transfer is called free convection. In this chapter we will confine our attention to
forced convection. Free convection will be discussed in Chapter 5.

4.1. FORCED CONVECTION IN SYSTEMS OF SIMPLE GEOMETRIES

Here we identify a few simple flow situations for understanding the physical mechanism of
forced convection heat transfer, without going into the mathematical analysis. Although the
simplest way to visualize convective heat transfer to or from a surface is to assume the existence
of a stagnant fluid film at the wall, no such film does exist in reality. It is true, however, that in
the case of heat transfer from a surface, most of the resistance to heat transfer is offered by a
narrow zone of fluid near the surface, called the boundary layer, but the fluid in this zone is
never stagnant. The concept of boundary layer is of vital importance in understanding the flow
and heat transfer characteristics near a wall.

4.1.1. Flow over a Flat Plate

Let us consider the flow of a fluid over an immersed wide flat plate [Fig. 4.1(a)] at zero angle of
incidence (which means that the plate is oriented along the direction of flow of the bulk fluid).

Prepared by M. Sc. Eng KHOUN Rithymean 31 Forced convection

Fig. 4.1 (a) Boundary layer over a flat plate (at zero angle of incidence). (LBL: laminar boundary layer, TZ:
transition zone, TBL: turbulent boundary layer.)

Fluid velocity at the surface of the plate is zero (because of no-slip condition), and gradually
increases with distance from the plate. At a sufficiently large distance from the plate, the fluid
velocity becomes equal to the ‘free stream velocity’, V. The region above the plate surface
within which this change of velocity from zero to the free stream value occurs is called the
boundary layer, or more specifically, the velocity boundary layer (also called the momentum
boundary layer or the hydrodynamic boundary layer). The thickness of this region is called the
boundary layer thickness and is denoted by . The boundary layer thickness increases with the
distance x from the leading edge of the plate, i.e  =  (x).

A theoretical analysis of boundary layer flow indicates that the transverse distance from the plate
where the free stream velocity is attained is very large. As a result, the thickness of the boundary
layer is rather arbitrarily defined as the distance from the plate at which a certain percentage,
usually 99%, of the free stream velocity is attained.

The characteristics of boundary layer flow depend upon the Reynolds number defined as Rex =
Vx/, where x is the distance from the point O or the leading edge. Because x is a variable, Rex
is called the ‘local Reynolds number’. The fluid motion in the boundary layer remains laminar
up to a point on the plate where 105 < Rex < 5  105, which is the critical Reynolds number (it
depends upon the roughness of the plate) for boundary layer flow. When Rex increases beyond
this value, transition to turbulent boundary layer begins. At Rex = 106, the boundary layer
becomes fully turbulent. However, these values should not be taken too strictly because they
depend upon several factors including the surface roughness of the plate. The laminar,
transition as well as the turbulent portions of the boundary layer are shown in Fig. 4.1(a). The
velocity profile in the laminar region is also shown.

The turbulent boundary layer is visualized, on the basis of reported experimental

measurements, as consisting of three zones-a laminar sublayer adjacent to the wall, a buffer
zone and the turbulent zone. Change in the velocity from zero to V is virtually confined within
the first two zones.

4.1.2. Thermal Boundary Layer

Heat transfer from a hot plate to a flowing fluid occurs principally by convection. Similar to
the velocity boundary layer, a thermal boundary layer is formed in the liquid that includes
laminar, transition and turbulent zones in the sense of heat transport depending upon the
variation of the Reynolds number along the plate. The thickness of the thermal boundary layer
is, however, different from that of the hydrodynamic boundary layer, and considerably depends
upon the thermal properties of the fluid. Just at the surface of the plate, the temperature of the
fluid will be the same as the temperature of the plate (continuity of temperature). Again, the
Prepared by M. Sc. Eng KHOUN Rithymean 32 Forced convection
fluid temperature will be very close to the bulk temperature at the edge of the thermal
boundary layer (i.e. at the edge of the thermal boundary layer, the difference from the bulk
temperature remains within 1% of the total temperature drop across the boundary layer). Thus,
it is not through the so-called stagnant film, but through the boundary layer that convective
transfer of heat from a surface occurs.

Velocity distributions in a two-dimensional boundary layer [i.e. how the x- and the y-
components of the velocity, u and v, depend upon the position (x, y)] govern the heat transfer
coefficient in the boundary layer. A correlation that can be used to calculate the heat transfer
coefficient (which is a function of x) will be described later. The thermal boundary layer over a
flat plate is shown in Fig. 4.1(b). It may be noted that when the dimensionless group called the
Prandtl number, Pr = / ( = / is called the momentum diusivity, and  = k/ cp is called the
thermal diffusivity) is less than unity, the momentum boundary layer remains within the
thermal boundary layer. If Pr > 1, the reverse becomes true.

Fig. 4.1(b) Laminar velocity and temperature boundary layers over a heated plate.

4.1.3. Flow across a Cylinder

Transverse flow of a fluid across a single cylinder or a bank of cylinders is an important flow
situation in industrial heat transfer. For example, a hot process stream flowing through a pipe
may be cooled by blowing air across the pipe. Here too, a boundary layer forms over the cylinder
as shown in Fig. 4.2. The point A is called the stagnation point because the fluid velocity is
zero at this point and the pressure is maximum (the physical reason being conversion of velocity
head to pressure head). The distance of a point on the cylinder from the stagnation point is given
by the linear measure x or the angular measure . As one moves away from the stagnation point
along the periphery, the boundary layer decelerates and its thickness increases. The velocity
profile within the boundary layer becomes gradually more and more flattener. Farther away from
the stagnation point, the retarded layer is unable to overcome the static pressure and the
boundary layer ‘separates’ from the cylinder surface. Downstream of this point of separation,
reverse flow sets in near the surface but forward flow continues in the outer part of the boundary
layer. On the rear side of the cylinder there are wakes, and the flow becomes turbulent.

Prepared by M. Sc. Eng KHOUN Rithymean 33 Forced convection

 Fig. 4.2 Velocity profile for cross-flow over a cylinder.

For practical heat transfer calculations, an ‘average heat transfer coefficient’ rather than the local
coefficient is used. We will now discuss the methods and the theoretical basis of development of
correlations for heat transfer coefficients in forced convection.

4.2. DIMENSIONLESS GROUPS IN HEAT TRANSFER

Before we discuss the important heat transfer correlations available in the literature, it is
pertinent to make a list and also to mention the physical significances of the dimensionless
groups which are frequently used in forced convection heat transfer (Table 4.2).

Prepared by M. Sc. Eng KHOUN Rithymean 34 Forced convection

The above table needs some more explanation. In these expressions (except the last one, Gz), l
denotes the characteristic length. A characteristic length can be defined based on the geometry of
the system. For example, in the case of flow, heat or mass transfer in a pipe, the pipe diameter is
taken as the characteristic length. For flow past a flat plate, the distance from the leading edge is
the characteristic length. So, in the case of pipe flow we use d as the characteristic length in
place of l. In the expression for Graetz number, L is the pipe length over which heat transfer
occurs.

4.3. CORRELATIONS FOR THE HEAT TRANSFER COEFFICIENT-INTERNAL

FLOWS

Systems of various shapes and different flow conditions are encountered in practical heat transfer
calculations. The common shapes or geometries are cylindrical, spherical or flat (or planar). The
most common flow problem involves flow through a circular pipe; the other important ones are
flow through a non-circular duct (e.g. a rectangular channel), through a packed bed (e.g. of
catalysts or of some inert material forming a regenerative bed), in an agitated vessel with
internals, through the jacket over a reaction kettle or vessel, across a bank of tubes, etc. Again,
the flow of a fluid in a system may be internal (e.g. flow through a pipe) or external (e.g. flow
across the outer surface of a pipe). Here, we describe the more important correlations along with
the ranges of applicability for the calculation of heat transfer coefficients. Some of these were
developed even more than half a century ago, but are still widely used in engineering
calculations. The important correlations for the case of internal flow are discussed below.

4.3.1. Laminar Flow through a Circular Pipe

Laminar flow (Re < 2100) may occur in industrial equipment for heating or cooling of a
considerably viscous liquid or of a liquid or solution sensitive to shear stress. Because it is
possible to analytically solve the problem of heat transfer to or from a fluid in laminar flow
through a pipe, a large volume of theoretical research work has been published in this area.
Details are available in Shah and London (1978). For fully developed flow and a constant
temperature of the pipe wall, the following correlation of Hausen (1943), although outdated, is
recommended. This correlation incorporates a term called the ‘thermal entry length’. (It is the
distance from the point of entry to a pipe or channel over which the ‘boundary layer
approximations’ are applicable; this will be discussed later.)

(4.1)
where
L is the pipe length
d is the pipe diameter
h is the average heat transfer coefficient over the pipe length L including the entry length.

A simpler correlation was suggested by Sieder and Tate (1936).

(4.2)

Prepared by M. Sc. Eng KHOUN Rithymean 35 Forced convection

The above correlation, Eq. (4.2), is applicable when
(a) 0.48 < Pr < 16700,
(b) the viscosity ratio is within the range, 0.0044 < /w < 9.75, where  is the viscosity
of the fluid at the bulk temperature, and w is that at the wall temperature, Tw
(c) (Pe d/L) > 108.

The fluid properties have to be known or evaluated at the mean bulk temperature of the fluid.

4.3.2. Turbulent Flow through a Circular Pipe

For ‘fully developed’ turbulent flow through a pipe, the following correlation suggested by
Dittus and Boelter (1930) is used widely.

(4.3)
where
n = 0.4 for heating (Tw > T)
n = 0.3 for cooling (Tw < T).

The conditions for applicability of this correlation are: (a) 0.7  Pr  160; (b) d/L < 0.1; and (c)
Re  10 000.

The above correlation, Eq. (4.3), is applicable for moderate values of the temperature difference
betwen the wall and the bulk, Tw - T. The fluid properties are evaluated at the arithmetic mean of
the bulk temperatures (i.e. the average of the inlet and the outlet temperatures of the fluid). The
maximum error in the predicted value of the heat transfer coefficient does not exceed 25%.

When the temperature difference between the wall and the bulk is substantial, its effect on the
fluid properties, particularly on viscosity, needs to be taken into account. For such cases, the
Sieder-Tate equation (1936) is recommended.

(4.4)

Here, /w is the viscosity correction factor that has to be used when the wall viscosity at the
wall temperature, w, is substantially different from that, , at the bulk temperature. The
conditions of applicability are: (a) 0.7  Pr  16700; (b) Re  10 000; and (c) d/L  0.1. The bulk
fluid properties have to be evaluated at the arithmetic mean bulk temperature.

Whitaker (1972) suggested the following equation which is based on the experimental data of a
large number of researchers.

(4.5)

In the case of heat transfer to or from a liquid metal, flowing through a pipe, at a constant wall
temperature, the correlation of Seban and Shimazaki (1951) may be used.

Nu = 5.0 + 0.025 (Pe)0.8 (4.6)

Prepared by M. Sc. Eng KHOUN Rithymean 36 Forced convection

Equation (4.6) is applicable when Pe > 100 and L/d > 60. All the properties of the liquid metal
should be evaluated at the mean bulk temperature.

4.3.3. Flow through a Non-circular Duct

Ducts of non-circular cross-sections, rectangular or square, are often used in the industrial and
other applications. Flow through the annulus of a double-pipe heat exchanger, flow through the
shell side of a shell-and-tube heat exchanger and flow of hot exhaust gases through rectangular
or square ducts are a few common examples. The foregoing equations for estimation of the
Nusselt number are generally applicable for heat transfer calculations for flow through non-
circular ducts, but the equivalent diameter of the duct is to be used in the calculation of the
Reynolds number. The equivalent diameter de is four times the hydraulic radius, rH. The
hydraulic radius is defined as

and
de = 4rH (4.7a)

For example, in the case of flow through a rectangular duct of sides l1 and l2 the equivalent
diameter is

It should be noted that the calculation of the wetted perimeter of a duct in the case of heat
transfer calculation may be different from that used in the case of determination of pressure drop
(Kem, 1950; Ludwig, 1983). To illustrate this point, we consider the flow through the annulus
between two concentric tubes in a double-pipe heat exchanger (the construction of this kind of
heat exchanger will be described later). Let the outside diameter of the inner pipe be d1, and the
inside diameter of the outer pipe be d2. If our objective is to calculate the pressure drop for flow
through this annulus, the wetted perimeter is (d1 + d2), because both the walls forming the
annulus contribute to the frictional pressure drop. But if we consider heat transfer from a hot
fluid flowing through the inner pipe to a cold fluid flowing through the annulus, wetting of the
outer wall of the inner pipe only becomes relevant to the heat transfer coefficient in the annulus.
So, here the wetted perimeter is simply d1. The equivalent diameters of the annulus in the
respective cases are:

For pressure drop calculation,

(4.7b)
For heat transfer calculation,

(4.7c)

4.4. CORRELATIONS FOR THE HEAT TRANSFER COEFFICIENT-EXTERNAL

FLOWS

Convective heat transfer in external flows is encountered in many practical situations. Typical
cases are flow over a flat surface, along or across a single pipe or a pipe bundle, over a sphere or
a bed of spheres (e.g. convective heat transport from a wall, heating of air by a bundle of steam

Prepared by M. Sc. Eng KHOUN Rithymean 37 Forced convection

pipes, heat transfer in a packed bed of catalyst or that in a fluidized bed, etc.). Several important
correlations that cover more common cases of external flows are discussed below.

4.4.1. Flow over a Flat Plate

Heat transfer in flow over a flat plate occurs through the boundary layer formed on the plate. A
physical description of the boundary layer including the laminar, transition and turbulent zones
has been given in Section 4.1. The following correlations are suggested for heat transfer
calculations in the boundary layer flow.
For heat transfer in laminar boundary layer flow, the following correlation for the local Nusselt
number can be obtained from an approximate solution of the boundary equations of momentum
and energy. It depends upon the distance from the leading edge of the plate.

(4.8a)
where
Nux is the local Nusselt number = xhx/k
Rex is the local Reynolds number = xV/

Here x is the distance from the leading edge of the plate, and hx is the local heat transfer
coefficient. An average value of the heat transfer coefficient over a distance L may be obtained
as

(4.8b)

For heat transfer in turbulent boundary layer flow, a simple correlation in common use is

(4.9)

4.4.2. Flow across a Cylinder

Quite a few correlations for the average Nusselt number are available for the case of convective
heat transfer in flow across a cylinder. For cross flow of a liquid over a cylinder, the heat transfer
coefficient can be calculated by using:

(4.10)

(4.11)

(4.12)

where Prw is the Prandtl number at the wall temperature. These correlations are applicable to
both gases and liquids.

(c) The correlation of Churchill and Burnstein (1977)

Prepared by M. Sc. Eng KHOUN Rithymean 38 Forced convection

 (4.13)

The above correlation is based on the data for flow of air, water and liquid sodium. In the
intermediate range of Reynolds number, however, the following correlation predicts the heat
transfer coefficient more accurately.

(4.14)

(4.15)

4.4.3. Flow past a Sphere

The following important correlations are available for the case of heat transfer in flow over a
single sphere:

(a) For flow of liquids past a sphere, Kramers (1946) proposed the following correlation:

(4.16)

(b) Whitaker (1972) suggested the following correlation:

(4.17)

where  is the viscosity of the fluid at the bulk temperature, and w is that at the wall
temperature. This correlation is applicable to both gases and liquids, and the error in the
prediction remains within 30%. The number ‘2’ arises out of the contribution of conduction
only when there is no motion in the medium, i.e. when Re = 0. The Reynolds number is based on
the diameter of the sphere.

4.5. HEAT TRANSFER WITH A VARIABLE DRIVING FORCE-COCURRENT AND

COUNTERCURRENT OPERATIONS

The calculation of the rate of heat transfer becomes very simple if the overall heat transfer
coefficient, the area of heat transfer and the temperature driving force all remain constant. But in
many cases, one or more of these quantities may vary. Examples of heat transport through a
variable area (in the case of radial transport through a cylinder or a sphere) or with a variable
heat transfer coefficient (in the case of boundary layer heat transfer) have been discussed.
However, variation of the temperature driving force with position in a heat transfer equipment or
device is more common. In order to explain this, we will consider a very simple type of heat
transfer device-the double-pipe heat exchanger.

Prepared by M. Sc. Eng KHOUN Rithymean 39 Forced convection

The schematic of a double-pipe heat exchanger is shown in Fig. 4.5 (cocurrent) and Fig. 4.6
(countercurrent). The device consists of two concentric pipes welded at the ends and provided
with flow nozzles so that a hot and a cold fluid can be brought in thermal contact in order to
effect transfer of heat. One of the fluids flows through the pipe and the other through the
annulus. If both the streams, the hot and the cold, flow in the same direction, the flow
arrangement is called parallel or cocurrent. If the hot and the cold streams flow in the opposite
directions, the arrangement is called countercurrent or counterflow. The countercurrent flow
arrangement is more common because it allows exchange of more amount of heat between the
streams than that by the cocurrent configuration.

Fig. 4.5 (a) Double-pipe heat exchanger in parallel or cocurrent flow and (b) temperature distribution.

Fig. 4.6 (a) Double-pipe heat exchanger in countercurrent flow and (b) temperature distribution.

Let us consider the schematic of a cocurrent flow system shown in Fig. 4.5. The following
notations will be used:

m' is the flow rate of the fluid stream, kg/s

cp is the specific heat, J/kg °C

Prepared by M. Sc. Eng KHOUN Rithymean 40 Forced convection

 T1, T2 are the temperatures of the fluid stream at the ends 1 and 2 of the device,
respectively
T = Th – Tc = the driving force at any section.

The subscript ‘h’ refer to the hot fluid and ‘c’ to the cold fluid, and Th and Tc are the ‘local
temperatures’ of the respective streams. We arbitrarily assume that the hot fluid flows through
the inner pipe and transfers heat through its wall to the cold fluid flowing through the annulus as
shown in the figure. The temperatures of both the streams vary with the position along the
device.

Now consider a ‘thin’ section of the device having a heat transfer area dA (i.e. the area of the
wall of the inner pipe in this thin section is dA). The local temperatures of the fluids are Th and
Tc for the hot and the cold fluid, respectively. If U is the overall heat transfer coefficient
(assumed constant), the rate of heat transfer, dQ, at steady state through the small area dA may
be written as

dQ = U dA (Th – Tc) = U dA T; T = (Th – Tc) (4.24)

Also, a heat balance over the thin section gives

dQ = - mh' cph dT h = mc' cpc dT c (4.25)

Here, dT h and dT c are the changes in the temperatures of the hot and the cold streams over the
thin section because of exchange of heat. The hot stream cools down as it flows through the pipe
and hence dT h is negative. So a negative sign is incorporated in the term in the middle of Eq.
(4.25) to make it consistent with respect to sign.

Next we put Th = T + Tc in Eq. (4.25) and eliminate Tc. Thus

Therefore,

(4.26)

Substituting for dQ from Eq. (4.26) in Eq. (4.24) and arranging, we get

(4.27)
We now integrate Eq. (4.27) from end 1 to end 2 of the system

At end 1, A = 0 and T = Th1 – Tc1 = T1 (say)

At end 2, A = A and T = Th2 – Tc2 = T2 (say).

Therefore,

Prepared by M. Sc. Eng KHOUN Rithymean 41 Forced convection

 (4.28)

Now we write the heat balance equations for both the streams, the hot stream and the cold
stream, over the heat exchange device. If Q is the total rate of heat transfer, we have

(4.29)
or

Therefore,

(4.30)

If we define Tm as an appropriate mean driving force over the entire length of the device, we
have
Q = U A Tm (4.31)

From Eqs. (4.30) and (4.31), we get

(4.32)
Comparing Eqs. (4.28) and (4.32), we get

(4.33)

Equation (4.33) is one of the most important equations in heat transfer calculation. It shows
that when the temperature driving force varies from one end of a heat exchange device to the
other, the log mean temperature difference, Tm, given above is the applicable mean driving
force.

Next we will consider the case of heat exchange between a hot and a cold fluid in
countercurrent flow as shown in Fig. 4.6. The notations used are very much the same as those
used in the case of cocurrent flow.

The rate of heat transfer at steady state through the elementary area dA is

dQ = U dA (Th – Tc) = U dA T (4.34)

Also, if dTh and dTc are the temperature changes of the streams over this elementary area, we
have
dQ = - mh’ cph dTh = mc’ cph dTc (4.35)

Prepared by M. Sc. Eng KHOUN Rithymean 42 Forced convection

The heat balance equations for the countercurrent flow situation are similar to those for the
cocurrent flow. It is easy to verify that the final form of the mean driving force remains the same
as Eq. (4.36) and the rate of heat transfer is then given by Eq. (4.31). The log mean driving force
is abbreviated as LMTD. This is left as an exercise for the students.

Prepared by M. Sc. Eng KHOUN Rithymean 43 Forced convection

 PROBLEMS

1. Nitrogen gas is heated at a rate of 2000 kg/h before passing it through the trays in a tray
drier in which an organic product is being dried. Medium pressure steam available from a
waste heat boiler at a pressure of 5.7 bar (saturation temperature = 160°C) condenses
within the tubes of a ‘finned tube’ heater. The gas flows outside the tubes. The heat duty is
38,700 kcal/h. The specific heat of nitrogen can be taken as 0.239 kcal/kg °C. The heat
exchanger has an area of 10 m2 and the overall heat transfer coefficient is estimated at 70
kcal/h m2 °C. The fin efficiency is 63%. Calculate the inlet and the outlet temperatures of
the gas.

2. Hot water is flowing through a 3.5 cm (1-1/4") schedule 40 steel pipe at a velocity of 1.8
m/s. The inlet temperature is 110°C, and the length of the pipe is 15 m. A 2 cm thick layer
of insulation (kc = 0.12 W/m °C) covers the pipe. The outside film coefficient is 10 W/m2
°C, and the ambient temperature is 20°C. Calculate the drop in the temperature of the water
over this section of the pipe.

3. A 41 mm i.d. (1-1/2" nominal bore) schedule 40 pipe carries water flowing at a rate of 1
kg/s. Water enters the pipe at 28°C and is heated by a stream of hot flue gas in cross flow
over the pipe. The gas velocity is 10 m/s. The arrangement essentially aims at recovering
a part of the waste heat of the gas stream which has a bulk temperature of 250°C. The
pressure is essentially atmospheric. The length of the pipe is 20 m. At what temperature
does the water leave the pipe? The properties of the flue gas are about the same as those
of the air.

4. Lubricating oil used in the gearbox of a 14,000 rpm high speed blower is being recycled
continuously through a double-pipe counterflow heat exchanger for cooling. The oil is to
be cooled from 70°C to 40°C at the rate of 1000 kg/h using water entering at 28°C. The
water temperature at the exit should not exceed 42°C. The specific heat of oil is 2.05 kJ/kg
°C and that of water is 4.17 kJ/kg °C. Calculate the required rate of flow of water. If the
heat exchange area is 3.0 m2, calculate the overall heat transfer coefficient.

Prepared by M. Sc. Eng KHOUN Rithymean 44 Forced convection