Forced Convective Heat Transfer

4.1 Principle of convection

Till now we have understood and solved the problem where heat transfer coefficient was known.
Now, we would emphasize on finding the convective heat transfer coefficient ​h​. Finding out the heat
transfer coefficient is not so easy in a given situation as it is a very complex term and depends in
many physical properties of the fluid like, thermal conductivity, viscosity, density etc. Moreover, the
flow field has a great impact on the convective heat transfer. The convective heat transfer requires a
good knowledge of fluid dynamics, and we expect that the reader has adequate knowledge of that.

We have seen that it is easy to form a differential heat balance equations. However, for practical
cases, it becomes tedious or impossible to solve the differential equations. Therefore, for practical
situations, the heat transfer coefficient is calculated using the empirical correlations, which are
developed over the years. Though these correlations have a good amount of error in most of the
situations but these are indispensible for the estimation of heat transfer coefficients.

The convective heat transfer may be

● forced convection: in which the motion in the fluid medium is generated by the application of
an external force, e.g. by a pump, blower, agitator etc.
● natural convection: in which the motion in the fluid is generated due to a result of density
difference caused by the temperature difference.

Moreover, in many applications, heat is transferred between fluid streams without any phase change
in the fluid. However, in some situations, the heat transfer is accompanied by phase change.

In this chapter we will consider the forced convection. Before we analyse the forced convection we
need to know its physical mechanism.

4.2 Forced convection mechanism: flow over a flat horizontal plate

Consider the flow of a fluid over an immersed flat horizontal plate of very large width, as shown in
fig. 4.1. The fig. shows that the flow is fully developed with no entrance losses. The bulk flow or free
stream velocity is represented by ​v​. Velocity of the fluid at the surface of the flat plate is zero at
no-slip condition. The velocity of the fluid in the y-direction from the plate increases from zero to free
stream velocity. 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. The thickness of this region is called the
boundary layer thickness and is generally denoted by ​δ.​ It can be seen in the fig.4.1 that the
boundary layer thickness increased with the distance ​x from the body edge. The boundary layer
thickness, δ, usually is defined as the distance from the surface to the point where the velocity is
within 1% of the free shear velocity. It should be noted that the y-axis of the fig. 4.1 is exaggerated
greatly to have a clear picture.
 Lecture 15

Fig.4.1. Boundary layer flow past a flat plate

The velocity profile merges smoothly and asymptotically in the free shear and the boundary layer
thickness is difficult to measure. However, the theoretical laminar boundary layer thickness can be
calculated using the eq.4.1.

where, ​Re​x is the local Reynold number at a distance ​x​. The derivative of ​δ can be found in any
stated boundary.

The turbulent boundary layer thickness can be calculated using eq.4.2.

Figure 4.2 shows the velocity boundary layer for the turbulent and laminar zone.
 Fig.4.2: Boundary layer flow past a flat surface (a) turbulent, and (b) laminar

In continuation to the above discussion and understanding, if the solid surface is maintained at a
temperature, ​Ts​ ​, which is different from the fluid temperature, ​T​f ,​ measured at a point far away from
the surface, a variation of the temperature of the fluid is observed, which is similar to the velocity
variation described. That is, the fluid temperature varies from ​T​s at
​ the wall to​T​f ​ far away from the
wall, with most of the variation occurring close to the surface.

If​T​s >
​ ​Tf​ ​, the fluid temperature approaches asymptotically and the temperature profile at a distance ​x
is shown in fig.4.3. However, a thermal boundary may be defined (similar to velocity boundary) as
the distance from the surface to the point where the temperature is within 1% of the free stream fluid
temperature (​T​f ​). Outside the thermal boundary layer the fluid is assumed to be a heat sink at a
uniform temperature of​T​f ​. The thermal boundary layer is generally not coincident with the velocity
boundary layer, although it is certainly dependant on it. That is, the velocity, boundary layer
thickness, the variation of velocity, whether the flow is laminar or turbulent etc are all the factors
which determine the temperature variation in the thermal boundary layer. The thermal boundary

layer and velocity boundary layer are related by the Prandtl number, is

called the momentum diffusivity and is called the thermal diffusivity; is less than unity, the
momentum boundary layer (or velocity boundary layer) remains within the thermal boundary layer. If
P​r >1, the boundary layers will be reversed as shown in the fig.4.4. The thermal boundary layer and
velocity boundary layer coincides at ​P​r​ =1.

Fig.4.4: The relation of two boundary layers at different Pr numbers

The above boundary layer theory will be helpful to understand the heat transfer in the process.
Through the boundary layers heat transfer is covered in a separate chapter, but the detailed
derivation and development of all the relationships having engineering importance for the prediction
of forced convection heat transfer coefficient is beyond the scope of the course. The reader may
consult any standard fluid mechanics and heat transfer books for detailed knowledge.

The purpose of this chapter is to present a collection of the most useful of the existing relations for
the most frequently encountered cases of forced convection. Some of these relations will be having
theoretical bases, and some will be empirical dimensionless correlations of experimental data. In
some situations, more than one relation will be given.

The discussion on heat transfer correlations consists of many dimensionless groups. Therefore,
before we discuss the importance of heat transfer coefficients, it is important to understand the
physical significance of these dimensionless groups, which are frequently used in forced convection
heat transfer. The table 4.1 shows some of the dimensionless numbers used in the forced
convection heat transfer.

Table-4.1: Some important dimensionless numbers used in forced heat transfer convection
4.3 Flow through a pipe or tube

4.3.1 Turbulent flow

A classical expression for calculating heat transfer in fully developed turbulent flow in smooth
tubes/pipes of diameter (d) and length (L) is given by Dittus and Boelter

(4.3)

where,

n = 0.4, for heating of the fluid

n = 0.3, for cooling of the fluid

The properties in this equation are evaluated at the average fluid bulk temperature. Therefore, the
temperature difference between bulk fluid and the wall should not be significantly high.

Application of eq. 4.3 lies in the following limits

Gnielinski suggested that better results for turbulent flow in smooth pipe may be obtained from the
following relations

When the temperature difference between bulk fluid and wall is very high, the viscosity of the fluid
and thus the fluid properties changes substantially. Therefore, the viscosity correction must be
accounted using Sieder – Tate equation given below
However, the fluid properties have to be evaluated at the mean bulk temperature of the fluid except
μ​w​ which should be evaluated at the wall temperature.

The earlier relations were applicable for fully developed flow when entrance length was negligible.
Nusselt recommended the following relation for the entrance region when the flow is not fully
developed.

(4.7)

c.

Applicability conditions, .

As different temperature terms will appear in the course therefore to understand these terms see the
following details.

Bulk temperature/mixing cup temperature​: Average temperature in a cross-section.

Average bulk temperature​: Arithmetic average temperature of inlet and outlet bulk temperatures.

Wall temperature​: Temperature of the wall.

Film temperature​: Arithmetic average temperature of the wall and free stream temperature.

Free stream temperature​: Temperature free from the effect of wall.

Log mean temperature difference​: It will be discussed in due course of time

Illustration 4.1

​
Pressurized air is to be heated by flowing into a pipe of 2.54 cm diameter. The air at 200o​ C and 2 atm
pressure enters in the pipe at 10 m/s. The temperature of the entire pipe is maintained at 220​o​C. Evaluate
the heat transfer coefficient for a unit length of a tube considering the constant heat flux conditions are
maintained at the pipe wall. What will be the bulk temperature of the air at the end of 3 m length of the
tube?

The following data for the entering air (at 200​o​C) has been given,

Solution 4.1

Reynolds number can be calculated from the above data,

The value of Reynolds number shows that the flow is in turbulent zone. Thus the Dittus-Boelter
equation (eq.4.3) should be used,

Thus ​h​ can be calculated for the known values of ​k,​ and ​d,​ which comes out to be

Energy balance is required to evaluate the increase in bulk temperature in a 3 m length of the tube,
Therefore the temperature of the air leaving the pipe will be at 210.81​o​C.

4.3.2 ​Laminar flow

Hausen presents the following empirical relations for fully developed laminar flow in tubes at
constant wall temperature.

The heat transfer coefficient calculated from eq. 4.8 is the average value over the entire length

(including entrance length) of tube .

Sieder and Tate suggested a simple relation for laminar heat transfer in tubes.

The condition for applicability of eq. 4.9:

length) of tube .

The empirical relations shown in eq. 4.2-4.9 are for smooth pipe. However, it case of rough pipes, it
is sometimes appropriate that the Reynolds analogy between fluid friction and heat transfer be used
to effect a solution under these conditions and can be expressed in terms of Stanton number.

In order to account the variation of the thermal properties of different fluids the following equations

may be used (i.e. Stanton number multiplied by ),

where, is the mean free velocity. The friction factor can be evaluated from Moody’s chart.

4.3.3 Flow through non-circular ducts

The same co-relations as discussed in section 4.4.1 can be used for the non-circular ducts.
However, the diameter of the tube has to be replaced by the hydraulic diameter or equivalent
diameter for the non-circular ducts. The hydraulic diameter is defined as

Where ​r​h​ is hydraulic radius.

4.3.4. Flow over a flat plate

Heat transfer in flow over a plate occurs through the boundary layer formed on the plane. Therefore
at any location the heat transfer coefficient will depend on the local Reynolds and Prandtl number.
For local heat transfer coefficient in laminar boundary layer flow, the following correlation can be
used to find the local Nusselt number. It depends upon the distance from the leading edge (​x​) of the
plate.

(4.13)

where, and are the local Nusselt and Reynold numbers, respectively.

An average value of the heat transfer coefficient over a distance ​l​ may be obtained by,
4.3.5 Flow across cylinders and spheres

4.3.5.1 Flow across a cylinder

The heat transfer coefficient can be found out by the correlations given by many researchers

Applicability of eq. 4.19: 10​2​ < ​Re​ < 10​7​, and ​Re Pr​ >0.2 .

However, the following equation (eq. 4.20) is more accurate for the condition where 20,000 < ​Re <
4,00,000 and ​Re Pr​ > 0.2.
4.3.5.2. Flow across a sphere

The above correlation is applicable to both gases and liquids.

Flow over bank of tubes is one of the very important phenomena in chemical process industries.
Heat exchanger, air conditioning for cooling and heating etc. involve a bank or bundle of tube over
which a fluid flows. The two most common geometric arrangements of a tube bank are shown in fig.
4.5.

Fig.4.5: Tube banks: (​a)​ aligned; (​b)​ staggered

In any of the arrangements, D is the diameter of tube, S​L is the longitudinal spacing, and S​T is the
transverse tube spacing.

The flow over a tube is quite different than the flow over bank of tubes. In case of bank of tube, the
flow is influenced by the effects such as the “shading” of one tube by another etc. Moreover, the heat
​ and ​T​∞​,
transfer for any particular tube thus not only determined by the incident fluid conditions, ​v∞
but also by ​D, SL​ ,​ and S​T and the tube positions in the bank. It is now clear that the heat transfer
coefficient for the first row of tubes is much like that for a single cylinder in cross flow. However, the
heat transfer coefficient for the tubes in the inner rows is generally larger because of the wake
generation by the previous tubes.

For the heat transfer correlations, in tube banks, the Reynolds number is defined by

where ​v​m​ is the maximum fluid velocity occurring at the minimum vacant area of the tube bank.

For the aligned tube arrangement,

Zukauskas has summarized his extensive for the heat transfer coefficients for fluid past a bank of
tubes,

(4.27)

The applicability of eq. 4.27: , and number of tubes are

The constants C ​ and ​m of co-relation 5.26 can be found out from any standard book on heat
transfer. It may be noted that the above relation is for the inner rows of bank, or for banks of many
rows.
4.4 Momentum and heat transfer analogies

Consider a fluid flows in a circular pipe in a laminar low. The wall of the pipe is maintained at T​w
temperature, which is higher than the flowing fluid temperature. The fluid being in relatively lower
temperature than the wall temperature will get heated as it flows through the pipe. Moreover, the
radial transport of the momentum in the pipe occurs as per the Newton’s law of viscosity. For a
circular pipe momentum transport and heat transport may be written in a similar way as shown in the
eq. 4.28,

Momentum flux = momentum diffusivity × gradient of concentration of momentum

4.28(a)

It may be noted that the fluid velocity is a function of radius of the pipe.

Heat flux= thermal diffusivity × gradient of concentration of heat energy

Now, the question comes, why are we discussion about the similarities? The answer is straight
forward that it is comparatively easy to experimentally/theoretically evaluate the momentum
transport under various conditions. However, the heat transport is not so easy to find out. Therefore,
we will learn different analogies to find the heat transport relations.

Equation 4.28 is for the laminar flow but if the flow is turbulent, eddies are generated. Eddy is a
lump/chunk of fluid elements that move together. Thus it may be assumed that the eddies are the
molecules of the fluid and are responsible for the transport of momentum and heat energy in the
turbulent flow. Therefore, in turbulent situation the momentum and heat transport is not only by the
molecular diffusion but also by the eddy diffusivities.

Thus, turbulent transport of momentum and turbulent transport of heat may be represented by eq.
4.29a and 4.29b, respectively.
The terms represent the eddy diffusivities for momentum and heat, respectively.

At the wall of the pipe, the momentum equation (eq. 4.29a) becomes,

Where ​f is the fanning friction factor (ratio of shear force to inertial force) and is the average
fluid velocity.

Equation eq.4.30 can be rearranged as,

Wall heat flux can be written as,

Where T​w is the wall temperature and the T​av is the average temperature of the fluid. Thus, the
dimensionless temperature gradient at the wall using heat transfer will be,

Where the heat transfer coefficient is represented by ​h and dimensionless temperature is

represented by .

Based on the above discussion many researchers have given their analogies. These analogies are
represented in the subsequent section.

4.4.1 Reynolds analogy

Reynolds has taken the following assumptions to find the analogy between heat and momentum
transport.

1. Gradients of the dimensionless parameters at the wall are equal.

2. The diffusivity terms are equal. That is

Thus if we use the above assumptions along with the eq.4.32 and 4.33,

Thus if we use the above assumptions along with the eq.4.32 and 4.33,

x
Equation 4.34 is known as Reynolds’s analogy.

The above relation may also be written in terms of the Darcy’s friction factor (fD) instead of fanning
friction factor (f​D​ = 4f)

Where Stanton number (St) is defined as,

The advantage of the analogy lies in that the ​h may not be available for certain geometries/situations
however, for which ​f value may be available as it is easier to perform momentum transport
experiments and then to calculate the ​f.​ Thus by using the eq.4.34 the ​h may be found out without
involving into the exhaustive and difficult heat transfer experiments.

4.4.2 The Chilton-Colburn analogy

The Reynolds analogy does not always give satisfactory results. Thus, Chilton and Colburn
experimentally modified the Reynolds’ analogy. The empirically modified Reynolds’ analogy is
known as Chilton-Colburn analogy and is given by eq.4.35,

It can be noted that for unit Prandtl number the Chilton-Colburn analogy becomes Reynolds analogy.

4.4.3 The Pradntl analogy

In the turbulent core the transport is mainly by eddies and near the wall, that is laminar sub-layer, the
transport is by molecular diffusion. Therefore, Prandtl modified the above two analogies using
universal velocity profile while driving the analogy (eq. 4.36).
4.4.4 The Van Karman analogy

Though Prandtl considered the laminar and turbulent laminar sublayers but did not consider the
buffer zone. Thus, Van Karman included the buffer zone into the Prandtl analogy to further improve
the analogy.