CONVECTION

With conduction and radiation covered, convection can now be discussed. But before proceeding
with this topic, a review on units is a necessity. Convection of heat (the film coefficient ℎ) is
concerned with dimensionless numbers; that is, group of parameters, which together have no
units when the units for the individual parameters are in accord with a consistent units. Remember
- a consistent unit is one based upon Newton’s F = ma/k law which mutually defines force and
mass when the proportionality constant is equal to unity.
While it is not the intent to belabor the topic of units, one must have a clear concept of the English
fps-system (both pound-mass and pound-force), the metric cgs-system, and the SI mks-system.
Unfortunately, tables, charts, and curves may be based on any of these systems and the values
taken therefrom must be resolved when used in solving heat transfer problems, especially when
convection is included. Because convection depends much on viscosity, dimensional analysis, and
the flow of fluids, these topics will be touched upon in the paragraph which follow.

Viscosity
The surface coefficient of heat transfer ℎ depends, among other factors, on the absolute viscosity
of the fluid. since a casual perusal of engineering literature will reveal five or more different units
used for absolute viscosity, since writers are sometimes not explicit concerning units, and since
consistency in units must be observed in heat transfer problems, it is desirable to know the
physical concept of viscosity.

v2
A = Area
F

dy

v1
Figure 1

Take an element of a flowing fluid (Fig. 1). If there is ant shearing of the fluid, there will be a relative
movement of one layer with respect to another layer of the fluid. Thus, the lower surface of this
element is moving with a velocity v1 and the upper surface is moving with a velocity v2. The change
in velocity is v2 – v1 = dv for a distance of dy between the planes. The velocity gradient is dv/dy,
and one of Newton’s law is that the shearing stress on the fluid, F/A, is proportional to the velocity
gradient, or
𝐹 𝑑𝑣
∝ 𝑑𝑦
𝐴
or in equation form,
𝐹 𝑑𝑣
=𝜇 [Newtonian Fluid]
𝐴 𝑑𝑦

where 𝜇 (mu), is the proportionality constant, is called the absolute viscosity or just viscosity.
Solving for 𝜇 and indicating its dimensions,
 𝐹𝑑𝑦 (𝑓𝑜𝑟𝑐𝑒)(𝐿)
𝜇 = 𝐴𝑑𝑣 → 𝐿2 (𝐿/𝜏)

where the unit of force has not been indicated. If the unit used for the force is based on the pound
force, the units of 𝜇 are
𝑃𝐿 𝑃𝜏 𝑙𝑏𝑓 −𝑠𝑒𝑐
𝜇′ → 𝐿2(𝐿/𝜏) = →
𝐿2 𝑓𝑡 2

If the unit of mass is taken as the pound P, the unit of force (F = ma) is obtained from PL/𝜏2; thus,
for 𝜇, the units become
(𝑃𝐿)(𝐿) 𝑃𝑚 𝑙𝑏
𝜇 → (𝜏2 )(𝐿2)(𝐿/𝜏) = 𝜏𝐿 → 𝑠𝑒𝑐−𝑓𝑡

The relation between these values of viscosity is 𝜇 = 𝑘𝜇′ or 𝜇′ = 𝜇/𝑘, where the time and length
units are the same in each case. These units of viscosity have no generally recognized names, but
the poisel is a name that has been suggested for the unit lb/ft-sec.
The cgs units of viscosity, the poise (dynes-sec/cm2) and the centipoise (0.01 poise), are also
frequently used, conversions may be made as follows:

(𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦, 𝑙𝑏𝑚 ⁄𝑠𝑒𝑐 − 𝑓𝑡)(1 490) = (𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦, 𝑐𝑒𝑛𝑡𝑖𝑝𝑜𝑠𝑒)

(𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦, 𝑙𝑏 − 𝑠𝑒𝑐/𝑓𝑡 2 )(47 800) = (𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦, 𝑐𝑒𝑛𝑡𝑖𝑝𝑜𝑖𝑠𝑒)

Kinematic Viscosity
Test values of viscosity are often presented in the literature as the kinematic viscosity ν (nu), which
is defined as the absolute viscosity divided by the density; 𝜈 = 𝜇/𝜌. The kinematic viscosity is the
same in either of the foregoing systems of units; hence
𝜇′ 𝑃𝜏/𝐿2 𝐿2 𝜇 𝑃/𝜏𝐿 𝐿2
𝜈 = 𝜌′ → 𝑃𝜏2/𝐿4 = 𝑎𝑛𝑑 𝜈 = 𝜌 → 𝑃/𝐿3 =
𝜏 𝜏
[POUND FORCE] [POUND MASS]

Dimensional Analysis and Dimensionless Numbers

The following discussion suggests the nature of many of the deduction regarding the flow of fluids
and the convection of heat. The method of attack is known as dimensional analysis. If one wishes
to study the factors which define the drop in pressure of a fluid in a pipe, a consideration of the
pressure drop per unit length of pipe Δp/x would be function of the following factors:
(1) the size or diameter D of the pipe, because the pressure drop will increase as the pipe becomes
smaller,
(2) the velocity of flow v, since the faster a fluid is pumped through a pipe, the greater must be
the pressure drop,
(3) the viscosity of the fluid 𝜇, because it was observed that, for example, viscous oils do not flow
as readily as viscous gasoline, and
(4) the density of the fluid 𝜌, because if there are any acceleration of the particles of the fluid, as
in turbulent flow, it is expected the energy will overcome these effects to follow from the done
as represented by the pressure drop.
The foregoing statement may be written in the form

∆𝑝
= 𝜙 (𝐷 𝑎 𝑣 𝑏 𝜇 𝑐 𝜌 𝑑 ) (1)
𝑥

where a, b, c, and d are exponents of unknown value, 𝜙 means function of, and x is the length of
the pipe. Equation (1) must have the same dimensions on each side if it is to be physically correct.
The next step is to substitute into equation (1) the dimensions of various quantities in place of the
quantities themselves. The units or dimensions must be consistent; either system yields the same
result. If the force unit is taken as the pound, density 𝜌 has the units 𝑃𝜏 2 ⁄𝐿4, and viscosity has the
units 𝑃𝜏⁄𝐿2. Since the units of pressure are 𝑃⁄𝐿2, the dimensional equation corresponding to
equation (1) is

𝑑
𝑃 𝐿 𝑏 𝑃𝜏 𝑐 𝑃𝜏2
= 𝐿𝑎 ( ) ( 2 ) ( )
𝐿3 𝜏 𝐿 𝐿4

If this equation balances dimensionally, the exponents of the P’s on the right side must reduce to
1, because the exponent of P on the left side is 1. Moreover, the exponent of L must be -3 on the
right side, since it is -3 on the left. The exponent of 𝜏 must be zero. Thus, using the exponents of
the P’s, the equation is

1=𝑐+𝑑 (2)

From the exponents of the L’s,

−3 = 𝑎 + 𝑏 − 2𝑐 − 4𝑑 (3)

and for the 𝜏’s,

0 = −𝑏 + 𝑐 + 2𝑑 (4)

In these three equations, there are four unknowns. Hence, in this instance, numerical values of a,
b, c, and d cannot be found. However, the number of unknown exponents may be reduce to one
by solving for a, b, and d in terms of c as follows.
From equation (2),

𝑑 =1−𝑐 (5)

from equations (4) and (5),

𝑏 = 2−𝑐 (6)

and from equations (3), (5), and (6),

𝑎 = −1 − 𝑐 (7)
Substituting these values of a, b, and d into equation (1),

Δ𝑝
= 𝜙(𝐷 −1−𝑐 𝑣 2−𝑐 𝜇𝑐 𝜌1−𝑐 )
𝑥

Δ𝑝 1 𝑣2 𝜌
= 𝜙 (𝐷𝐷𝑐 𝑥 𝑥 𝜇𝑐 𝑥 )
𝑥 𝑣𝑐 𝜌𝑐

Δ𝑝 𝑣 2𝜌 𝜇 𝑐
= 𝜙[ ( ) ]
𝑥 𝐷 𝐷𝑣𝜌

This is as far as dimensional analysis can be used. The for of the function 𝜙 must be found from
experimental data. Assume that the form of the function is

Δ𝑝 𝑣 2𝜌 𝜇 𝑐
=𝐶 (𝐷𝑣𝜌)
𝑥 𝐷

Δ𝑝 𝑣 2𝜌 𝜇 𝑐′
=𝐶 (𝐷𝑣𝜌) (8)
𝑥 𝐷

where C is a dimensionless constant. With two unknowns left, the exponent c’ (= -c) and the
constant C, it may appear at first that little has been accomplished. On the contrary, a very
significant grouping of variables had been discovered. The group 𝐷𝑣𝜌⁄𝜇, called Reynolds number
Re, should evidently be used as a group variable in analyzing experimental data. If this done and
the proper experimental data are available, the constants C and c’ may be determined, thus
establishing a general equation for the flow of fluids in accordance with the original assumptions.
The success therefore of dimensional analysis depends not only upon the experimental
determination of certain constants, but also upon the correctness of the assumptions in the very
beginning. One factor omitted in the foregoing analysis is the degree of roughness of the inside of
the pipe, since it is expected smooth pipes will offer less resistance to flow (that is, have a smaller
pressure drop) than rough pipes. Hence, equation (8) should be applied to data concerning pipes
that have a similar degree of roughness, that are geometrically similar.
Checking the dimension of the Reynolds number by substituting the dimensions of the various
quantities which make up this number,
𝐷𝑣𝜌 (𝐿)(𝐿 ⁄𝜏)(𝑃𝜏2 /𝐿4 )
𝑅𝑒 = →
𝜇 𝑃𝜏/𝐿2
where the units cancel and Re is therefore a dimensionless number. It follows that the group
(∆𝑝)(𝐷)/𝑥𝑣 2 𝜌 from equation (8) is also a dimensionless group, a fact the reader may check for
himself. Moreover, since the units of 𝐷 and 𝑥 are the same, ∆𝑝⁄(𝑣 2 𝜌) must also be a dimensioless
group. The Reynolds number is important in convected heat problems as well as in the flow of
fluids. It may be expressed in terms of kinematic viscosity (𝜈 = 𝜇 ⁄𝜌) or in terms of the so-called
mass velocity G (𝐺 = 𝜌𝜐 → 𝑙𝑏⁄𝑓𝑡 2 − 𝑠𝑒𝑐 ) for the pound-mass system.

𝐷𝑣 𝐷𝐺
𝑅𝑒 = =
𝜈 𝜇
So from the foregoing discussion, the dimensional analysis results in a logical grouping of the
involved variables into dimensionless groups, a result that is invaluable in analyzing phenomena
that have not been able to analyze by a direct mathematical approach, phenomena such as the
flow of fluids and heat.

Examples:

1467/p. 137 – Faires

Equation (b) topic 19.36, Text, is empirical for heat transfer by free convection from vertical plates,
that is, ℎ𝑖 = 0.27∆𝑡 0.25 where the units of ℎ𝑖 are Btu/hr-ft2-oF. If ℎ𝑖 is expressed on the SI basis in
W/m2-K, what should the constant be in place of 0.27 leaving ∆𝑡 in oF? If ∆𝑡 is also change to K?

Given: ℎ𝑖 = 0.27∆𝑡 0.25 (the empirical equation for free convection)

Required: convert the formula to SI leaving Δt in oF; also Δt in K

Solution:
If Δt will remain in oF:
𝐵𝑡𝑢 1 055 𝐽 (3.28)2 𝑓𝑡 2 1.8 𝐹 1 ℎ𝑟
ℎ𝑖 = 0.27 (𝐹 0.25 ) [ ][ ][ ][ ]
ℎ𝑟−𝑓𝑡 2 −𝐹1.25 𝐵𝑡𝑢 𝑚2 𝐾 3 600 𝑠

𝑊
ℎ𝑖 = 1.53 𝑚2−𝐾−𝐹0.25 (∆𝑡 0.25 )

If Δt in K:
𝐵𝑡𝑢 1 055 𝐽 (3.28)2 𝑓𝑡 2 (1.8)1.25 𝐹1.25 1 ℎ𝑟
ℎ𝑖 = 0.27 ℎ𝑟−𝑓𝑡 2−𝐹1.25 (𝐾 0.25 ) [ ][ ][ ] [3 600 𝑠]
𝐵𝑡𝑢 𝑚2 𝐾1.25

𝑊
ℎ𝑖 = 1.77 𝑚2−𝐾1.25 (∆𝑡 0.25 )

1468. The flow resistance R per unit area encountered by a fluid moving through a closed duct
depends principally upon the fluid properties of density 𝜌, absolute viscosity 𝜇, and velocity 𝑣;
also, the duct diameter 𝐷. In mathematical form, 𝑅 = 𝐾𝜌𝑎 𝜇𝑏 𝑣 𝑐 𝐷𝑑 , where 𝐾 is a dimensionless
proportionality constant and a, b, c, and d are constant exponents. Through dimensional analysis,
show that 𝑅 = 𝑓𝜌𝑣 2 where 𝑓 = 𝐶R-b, a dimensionless quantity called the coefficient of friction;
𝐶 is a constant and R the Reynolds number.

Given: If 𝑅 = 𝐾𝜌𝑎 𝜇𝑏 𝑣 𝑐 𝐷𝑑 ,

Required: Prove that 𝑅 = 𝑓𝜌𝑣 2 , where 𝑓 = 𝐶R-b, using dimensional analysis

𝑅 = 𝐾𝜌𝑎 𝜇𝑏 𝑣 𝑐 𝐷𝑑 (1)

If the pound is used as unit of force, then

 𝑎
𝑃 𝑃𝜏2 𝑃𝜏 𝑏 𝐿 𝑐
= ( 𝐿4 ) ( 𝐿2 ) ( 𝜏 ) (𝐷)𝑑
𝐿2

For the P’s:

1=𝑎+𝑏 (2)
For the L’s:
−2 = −4𝑎 − 2𝑏 + 𝑐 + 𝑑 (3)
For the 𝜏’s:
0 = 2𝑎 + 𝑏 − 𝑐 (4)

Expressing the exponents a, c, and d in terms of b, then

From equation (2),
𝑎 =1−𝑏 (2𝑎)
From equation (4),
0 = 2(1 − 𝑏) + 𝑏 − 𝑐

𝑐 =2−𝑏 (4𝑎)
From equation (3),
−2 = −4(1 − 𝑏) − 2𝑏 + (2 − 𝑏) + 𝑑

𝑑 = −𝑏 (3𝑎)

Substitute equations (2a), (3a), and (4a) into equation (1):

𝑅 = (𝜌1−𝑏 𝜇𝑏 𝑣 2−𝑏 𝐷 −𝑏 )

𝜌 𝑣2 1
𝑅 = (𝜌 𝑏 𝑥 𝜇𝑏 𝑥 𝑥 )
𝑣𝑏 𝐷𝑏

𝜇 𝑏
𝑅 = 𝜌𝑣 2 (𝐷𝜌𝑣)

𝐷𝜌𝑣 −𝑏
𝑅 = 𝜌𝑣 2 ( )
𝜇

𝐷𝑣𝜌
Since =R and 𝑓 = 𝐶R-b,
𝜇

∴ 𝑅 = 𝑓𝜌𝑣 2