CONVECTION HEAT TRANSFER

Convection heat transfer is the transfer of energy by the mass movement of groups of molecules.
It is restricted to liquids and gases, as mass molecular movement does not occur at an
appreciable speed in solids. It cannot be mathematically predicted as easily as can transfer by
conduction or radiation and so its study is largely based on experimental results rather than on
theory. The most satisfactory convection heat transfer formulae are relationships between
dimensionless groups of physical quantities. Furthermore, since the laws of molecular transport
govern both heat flow and viscosity, convection heat transfer and fluid friction are closely related
to each other.

Convection coefficients will be studied under two sections, firstly, natural convection in which
movements occur due to density differences on heating or cooling; and secondly, forced
convection, in which an external source of energy is applied to create movement. In many
practical cases, both mechanisms occur together.

Natural Convection

Heat transfer by natural convection occurs when a fluid is in contact with a surface hotter or
colder than itself. As the fluid is heated or cooled it changes its density. This difference in
density causes movement in the fluid that has been heated or cooled and causes the heat transfer
to continue.

There are many examples of natural convection in the food industry. Convection is significant
when hot surfaces, such as retorts which may be vertical or horizontal cylinders, are exposed
with or without insulation to colder ambient air. It occurs when food is placed inside a chiller or
freezer store in which circulation is not assisted by fans. Convection is important when material
is placed in ovens without fans and afterwards when the cooked material is removed to cool in
air.

It has been found that natural convection rates depend upon the physical constants of the fluid,
density , viscosity , thermal conductivity k, specific heat at constant pressure cp and coefficient
of thermal expansion  (beta) which for gases = l/T by Charles' Law. Other factors that also
affect convection-heat transfer are, some linear dimension of the system, diameter D or length L,
a temperature difference term, T, and the gravitational acceleration g since it is density
differences acted upon by gravity that create circulation. Heat transfer rates are expressed in
terms of a convection heat transfer coefficient (hc), which is part of the general surface
coefficient hs, in eqn. (5.5).

Experimentally, if has been shown that convection heat transfer can be described in terms of
these factors grouped in dimensionless numbers which are known by the names of eminent
workers in this field:
 Nusselt number (Nu) = (hcD/k)
Prandtl number (Pr) = (cp /k)
Grashof number (Gr) = (D32g  T /2)

and in some cases a length ratio (L /D).

If we assume that these ratios can be related by a simple power function we can then write the
most general equation for natural convection:

(Nu) = K(Pr)k(Gr)m(L/D)n (5.14)

Experimental work has evaluated K, k, m, n, under various conditions. For a discussion, see
McAdams (1954). Once K, k, m, n are known for a particular case, together with the appropriate
physical characteristics of the fluid, the Nusselt number can be calculated. From the Nusselt
number we can find hc and so determine the rate of convection heat transfer by applying eqn.
(5.5). In natural convection equations, the values of the physical constants of the fluid are taken
at the mean temperature between the surface and the bulk fluid. The Nusselt and Biot numbers
look similar: they differ in that for Nusselt, k and h both refer to the fluid, for Biot k is in the
solid and h is in the fluid.

Natural Convection Equations

These are related to a characteristic dimension of the body (food material for example) being
considered, and typically this is a length for rectangular bodies and a diameter for
spherical/cylindrical ones

(1) Natural convection about vertical cylinders and planes, such as vertical retorts and oven
walls

(Nu) = 0.53(Pr.Gr)0.25 for 104 < (Pr.Gr) < 109 (5.15)

(Nu) = 0.12(Pr.Gr)0.33 for 109 < (Pr.Gr) < 1012 (5.16)

For air, these equations can be approximated respectively by:

hc = 1.3(T/L)0.25 (5.17)

hc = 1.8(T)0.25 (5.18)

Equations (5.17) and (5.18) are dimensional equations and are in standard units [T in oC, L (or
D) in metres and hc in Jm-2s-1oC-1]. The characteristic dimension to be used in the calculation of
(Nu) and (Gr) in these equations is the height of the plane or cylinder.
(2) Natural convection about horizontal cylinders such as a steam pipe or sausages lying on a
rack
(Nu) = 0.54(Pr.Gr) 0.25 for laminar flow in range 103 <(Pr.Gr) < 109. (5.19)

Simplified equations can be employed in the case of air, which is so often encountered in contact
with hotter or colder foods giving again:

for 104 < (Pr.Gr) < 109 hc = 1.3(T/D)0.25 (5.20)

and for 109 < (Pr.Gr) < 1012 hc = 1.8(T)0.33 (5.21)

(3) Natural convection from horizontal planes, such as slabs of cake cooling

The corresponding cylinder equations may be used, employing the length of the plane instead of
the diameter of the cylinder whenever D occurs in (Nu) and (Gr). In the case of horizontal
planes, cooled when facing upwards, or heated when facing downwards, which appear to be
working against natural convection circulation, it has been found that half of the value of hc in
eqns. (5.19) - (5.21) corresponds reasonably well with the experimental results.

Note carefully that the simplified equations are dimensional. Temperatures must be in oC and
lengths in m and then hc will be in Jm-2s-1 oC-1. Values for , k and  are measured at the film
temperature, which is midway between the surface temperature and the temperature of the bulk
liquid.

EXAMPLE 5.7. Heat loss from a cooking vessel

Calculate the rate of convection heat loss to ambient air from the side walls of a cooking vessel
in the form of a vertical cylinder 0.9m in diameter and 1.2m high. The outside of the vessel
insulation, facing ambient air, is found to be at 49oC and the air temperature is 17oC.

First it is necessary to establish the value of (Pr.Gr).

From the properties of air, at the mean film temperature, (49 + 17)/2, that is 33oC,
 = 1.9 x 10-5 Nsm -2, cp = l.0kJkg-1oC-1, k = 0.025Jm-1s-1 oC-1,  = 1/308,  =1.l2kgm-3.
From the conditions of the problem, characteristic dimension = height = 1.2 m, T = 32oC.

(Pr.Gr) = (cp /k) (D32g  T /2)

= (L32g  T cp) / ( k)

= [(1.2)3 x (1.12)2 x 9.81 x 1 x 32 x 1.0 x 103 )/(308 x 1.9 x 10-5 x 0.025)

= 5 x 109

Therefore eqn. (5.18) is applicable.

and so hc = 1.8T 0.25 = 1.8(32)0.25

= 4.3 Jm-2s-1 oC-1
Total area of vessel wall A = DL =  x 0.9 x 1.2 = 3.4m2
T = 32oC.

Therefore heat loss rate = hc A(T1 - T2)

= 4.3 x 3.4 x 32
= 468 Js-1

Forced Convection

When a fluid is forced past a solid body and heat is transferred between the fluid and the body,
this is called forced convection heat transfer. Examples in the food industry are in the forced-
convection ovens for baking bread, in blast and fluidized freezing, in ice-cream hardening rooms,
in agitated retorts, in meat chillers. In all of these, foodstuffs of various geometrical shapes are
heated or cooled by a surrounding fluid, which is moved relative to them by external means.

The fluid is constantly being replaced, and the rates of heat transfer are, therefore, higher than for
natural convection. Also, as might be expected, the higher the velocity of the fluid the higher the
rate of heat transfer. In the case of low velocities, where rates of natural convection heat transfer
are comparable to those of forced convection heat transfer, the Grashof number is still
significant. But in general the influence of natural circulation, depending as it does on
coefficients of thermal expansion and on the gravitational acceleration, is replaced by
dependence on circulation velocities and the Reynolds number.

As with natural convection, the results are substantially based on experiment and are grouped to
deal with various commonly met situations such as fluids flowing in pipes, outside pipes, etc.

Forced Convection Equations

(1) Heating and cooling inside tubes, generally fluid foods being pumped through pipes

In cases of moderate temperature differences and where tubes are reasonably long, for laminar
flow it is found that:

(Nu) = 4 (5.22)

and where turbulence is developed for (Re) > 2100 and (Pr) > 0.5

(Nu) = 0.023(Re)0.8(Pr)0.4 (5.23)

For more viscous liquids, such as oils and syrups, the surface heat transfer will be affected,
depending upon whether the fluid is heating or being cooled. Under these cases, the viscosity
effect can be allowed, for (Re)> 10,000, by using the equation:
 (Nu) = 0.027( /s)0 14(Re)0.8 (Pr)0.33 (5.24)

In both cases, the fluid properties are those of the bulk fluid except for s which is the viscosity
of the fluid at the temperature of the tube surface.

Since (Pr) varies little for gases, either between gases or with temperature, it can be taken as 0.75
and eqn. (5.23) simplifies for gases to:

(Nu) = 0.02(Re)0.8. (5.25)

In this equation the viscosity ratio is assumed to have no effect and all quantities are evaluated at
the bulk gas temperature. For other factors constant, this becomes hc = k v0.8 as in equation 5.28.

(2) Heating or cooling over plane surfaces

Many instances of foods approximate to plane surfaces, such as cartons of meat or ice cream or
slabs of cheese. For a plane surface, the problem of characterizing the flow arises, as it is no
longer obvious what length to choose for the Reynolds number. It has been found, however, that
experimental data correlate quite well if the length of the plate measured in the direction of the
flow is taken for D in the Reynolds number and the recommended equation is:

(Nu) = 0.036 (Re)0.8(Pr)0.33 for (Re) > 2 x 104 (5.26)

For the flow of air over flat surfaces simplified equations are:

hc = 5.7 + 3.9v for v < 5ms -1 (5.27)

hc = 7.4v0.8 for 5 < v < 30ms-1 (5.28)

These again are dimensional equations and they apply only to smooth plates. Values for hc for
rough plates are slightly higher.

(3) Heating and cooling outside tubes

Typical examples in food processing are water chillers, chilling sausages, processing spaghetti.
Experimental data in this case have been correlated by the usual form of equation:

(Nu) = K (Re)n(Pr)m (5.29)

The powers n and m vary with the Reynolds number. Values for D in (Re) are again a difficulty
and the diameter of the tube, over which the flow occurs, is used. It should be noted that in this
case the same values of (Re) can not be used to denote streamline or turbulent conditions as for
fluids flowing inside pipes.

For gases and for liquids at high or moderate Reynolds numbers:

 (Nu) = 0.26(Re)0.6(Pr)0.3 (5.30)

whereas for liquids at low Reynolds numbers, 1 < (Re) < 200:

(Nu) = 0.86(Re)0.43(Pr)0.3 (5.31)

As in eqn. (5.23), (Pr) for gases is nearly constant so that simplified equations can be written.
Fluid properties in these forced convection equations are evaluated at the mean film temperature,
which is the arithmetic mean temperature between the temperature of the tube walls and the
temperature of the bulk fluid.

EXAMPLE 5.8. Heat transfer in water flowing over a sausage

Water is flowing at 0.3 ms1 across a 7.5cm diameter sausage at 74oC. If the bulk water
temperature is 24oC, estimate the heat-transfer coefficient.

Mean film temperature = (74 + 24)/2 = 49oC.

Properties of water at 49oC are taken as:

cp = 4.186kJkg-1oC-1, k = 0.64Jm-1s-1oC-1,  = 5.6x 10-4Nsm-2,  = l000kgm-3.

Therefore (Re) = (Dv/)

= (0.075 x 0.3 x 1000)/(5.6 x 10 -4)
= 4.02 x 104
(Re)0.6 = 580

(Pr) = (cp  /k)

= (4186 x 5.6 x 10-4)/0.64
= 3.66.
(Pr)0.3 = 1.48

(Nu) = (hcD/k)
= 0.26(Re)0.6(Pr)0.3

Therefore hc = k/D x [0.26 x (Re)0.6(Pr)0.3]

= (0.64 x 0.26 x 580 x 1.48)/0.075
= 1904 Jm-2s-1oC-1

EXAMPLE 5.9. Surface heat transfer to vegetable puree

Calculate the surface heat transfer coefficient to a vegetable puree, which is flowing at an
estimated 3m min-1over a flat plate 0.9m long by 0.6m wide. Steam is condensing on the other
side of the plate and maintaining the surface, which is in contact with the puree, at 104 oC.
Assume that the properties of the vegetable puree are, density l040kgm-3, specific heat 3980Jkg-1
o -1
C , viscosity 0.002 Nsm -2, thermal conductivity 0.52 Jm-1s-1oC-1.

v = 3m min-1 = 3/60 ms-1

(Re) = (Lv/)
 = (0.9 x 3 x 1040)/(2 x l0-3x 60)
= 2.34 x 10 4
Therefore eqn. (5.26) is applicable and so:
Nu = (hcL/k) = 0.036(Re)0.8(Pr)0.33

Pr = (cp  /k)
= (3980 x 2 x 10-3)/0.52
= 15.3
and so

(hcL/k) = 0.036(2.34 x 104)0.815.30.33

hc = (0.52 x 0.036) (3.13 x 103)(2.46)/0.9

= 160 Jm-2s-1oC-1

EXAMPLE 5.10. Heat loss from a cooking vessel

What would be the rate of heat loss from the cooking vessel of Example 5.7, if a draught caused
the air to move past the cooking vessel at a speed of 61m min-1

Assuming the vessel is equivalent to a flat plate then from eqn. (5.27)
v = 61/60 = 1.02 m s-1 that is v 5 m s-1
hc = 5.7 + 3.9v
= 5.7 + (3.9 x 61)/60
= 9.7 Jm-2s-1oC-1
So with A = 3.4m2, T = 32oC,
q = 9.7 x 3.4 x 32
= 1055 Js -1

OVERALL HEAT TRANSFER COEFFICIENTS

It is most convenient to use overall heat transfer coefficients in heat transfer calculations as these
combine all of the constituent factors into one, and are based on the overall temperature drop. An
overall coefficient, U, combining conduction and surface coefficients, has already been
introduced in eqn. (5.5). Radiation coefficients, subject to the limitations discussed in the section
on radiation, can be incorporated also in the overall coefficient. The radiation coefficients should
be combined with the convection coefficient to give a total surface coefficient, as they are in
series, and so:

hs = (hr + hc) (5.32)

The overall coefficient U for a composite system, consisting of surface film, composite wall,
surface film, in series, can then be calculated as in eqn. (5.5) from:

1/U = 1/(hr + hc)1 + x1/k1 + x2/k2 + ……+ 1/(hr + hc)2. (5.33)

EXAMPLE 5.11. Effect of air movement on heat transfer in a cold store
In Example 5.2, the overall conductance of the materials in a cold-store wall was calculated.
Now on the outside of such a wall a wind of 6.7 ms-1 is blowing, and on the inside a cooling unit
moves air over the wall surface at about 0.61 ms-1. The radiation coefficients can be taken as
6.25 and 1.7 Jm-2s-1oC-1on the outside and inside of the wall respectively. Calculate the overall
heat transfer coefficient for the wall.

On outside surface: v = 6.7ms-1.

From eqn. (5.28)
hc = 7.4v0.8 = 7.4(6.7)0.8
= 34 Jm-2s-1oC-1
and hr = 6.25 Jm-2s-1oC-1
Therefore hs1 = (34+6) = 40 Jm-2s-1oC-1

On inside surface: v = 0.61 ms-1.

From eqn. (5.27)
hc = 5.7 + 3.9v = 5.7 + (3.9 x 0.61)
= 8.1 Jm-2s-1oC-1
and hr = 1.7 Jm-2s-1oC-1
Therefore hs2 = (8.1 + 1.7) = 9.8 Jm-2s-1oC-1

Now from Example 5.2 the overall conductance of the wall,

U old = 0.38 Jm-2s-1 oC–1

and
1/Unew = 1/hs1 + 1/Uold + 1/hs2
= 1/40 + 1/0.38 + 1/9.8
= 2.76.
Therefore Unew = 0.36 Jm-2s-1 oC–1

In eqn. (5.33) often one or two terms are much more important than other terms because of their
numerical values. In such a case, the important terms, those signifying the low thermal
conductances, are said to be the controlling terms. Thus, in Example 5.11 the introduction of
values for the surface coefficients made only a small difference to the overall U value for the
insulated wall. The reverse situation might be the case for other walls that were better heat
conductors.

EXAMPLE 5.12. Comparison of heat transfer in brick and aluminium walls

Calculate the respective U values for a wall made from either (a) 10 cm of brick of thermal
conductivity 0.7 Jm-1s-1 oC–1, or (b) 1.3mm of aluminium sheet, conductivity 208 Jm-1s-1oC–1.
Surface heat-transfer coefficients are on the one side 9.8 and on the other 40 Jm-2s-1oC–1.

(a) For brick

k = 0.7 Jm-1s-1 oC–1
x/k = 0.1/0.7 = 0.14
Therefore 1/U = 1/40+0.14+1/9.8
= 0.27
U = 3.7 Jm-2s-1 oC-1

(b) For aluminium

k = 208 Jm-1s-1 oC–1
x/k = 0.0013/208
= 6.2 x 10-6
1/U = 1/40 + 6.2 x 10-6 + 1/9.8
= 0.13
U = 7.7 Jm-2s-1 oC–1

Comparing the calculations in Example 5.11 with those in Example 5.12, it can be seen that the
relative importance of the terms varies. In the first case, with the insulated wall, the thermal
conductivity of the insulation is so low that the neglect of the surface terms makes little
difference to the calculated U value. In the second case, with a wall whose conductance is of the
same order as the surface coefficients, all terms have to be considered to arrive at a reasonably
accurate U value. In the third case, with a wall of high conductivity, the wall conductance is
insignificant compared with the surface terms and it could be neglected without any appreciable
effect on U. The practical significance of this observation is that if the controlling terms are
known, then in any overall heat transfer situation other factors may often be neglected without
introducing significant error. On the other hand, if all terms are of the same magnitude, there are
no controlling terms and all factors have to be taken into account.

HEAT TRANSFER FROM CONDENSING VAPOURS

The rate of heat transfer obtained when a vapour is condensing to a liquid is very often
important. In particular, it occurs in the food industry in steam-heated vessels where the steam
condenses and gives up its heat; and in distillation and evaporation where the vapours produced
must be condensed. In condensation, the latent heat of vaporization is given up at constant
temperature, the boiling temperature of the liquid.

Two generalized equations have been obtained:

(1) For condensation on vertical tubes or plane surfaces

hv = 0.94[(k32g/) x (/LT)]0.25 (5.34)

where (lambda) is the latent heat of the condensing liquid in J kg-1, L is the height of the plate
or tube and the other symbols have their usual meanings.

(2) For condensation on a horizontal tube

hh = 0.72[(k32g/ ) x (/DT)]0.25 (5.35)

where D is the diameter of the tube.

These equations apply to condensation in which the condensed liquid forms a film on the
condenser surface. This is called film condensation: it is the most usual form and is assumed to
occur in the absence of evidence to the contrary.

However, in some cases the condensation occurs in drops that remain on the surface and then fall
off without spreading a condensate film over the whole surface. Since the condensate film itself
offers heat transfer resistance, film condensation heat transfer rates would be expected to be
lower than drop condensation heat-transfer rates and this has been found to be true. Surface heat
transfer rates for drop condensation may be as much as ten times as high as the rates for film
condensation.

The contamination of the condensing vapour by other vapours, which do not condense under the
condenser conditions, can have a profound effect on overall coefficients. Examples of a non-
condensing vapour are air in the vapours from an evaporator and in the jacket of a steam pan.
The adverse effect of non-condensable vapours on overall heat transfer coefficients is due to the
difference between the normal range of condensing heat transfer coefficients, 1200 - 12,000
Jm-2s-1oC –1, and the normal range of gas heat transfer coefficients with natural convection or low
velocities, of about 6 Jm-2s-1oC –1.

Uncertainties make calculation of condensation coefficients difficult, and for many purposes it is
near enough to assume the following coefficients:

for condensing steam 12,000 Jm-2s-1oC–1

for condensing ammonia 6,000 Jm-2s-1oC–1
for condensing organic liquids 1,200 Jm-2s-1oC–1

The heat transfer coefficient for steam with 3% air falls to about 3500 Jm-2s-1oC–1, and with 6%
air to about 1200 Jm-2s-1oC–1.

EXAMPLE 5.13. Condensing ammonia in a refrigeration plant

A steel tube of 1mm wall thickness is being used to condense ammonia, using cooling water
outside the pipe in a refrigeration plant. If the water-side heat transfer coefficient is estimated at
1750 Jm-2s-1oC–1 and the thermal conductivity of steel is 45 Jm-1s-1oC–1, calculate the overall
heat-transfer coefficient.

Assuming the ammonia condensing coefficient, 6000 Jm-2s-1oC–1

1/U = 1/h1 + x/k + 1/h2

= 1/1750 + 0.001/45 + 1/6000
= 7.6 x 10-4

U = 1300 Jm-2s-1oC–1.
 HEAT TRANSFER TO BOILING LIQUIDS

When the presence of a heated surface causes a liquid near it to boil, the intense agitation gives
rise to high local coefficients of heat transfer. A considerable amount of experimental work has
been carried out on this, but generalized correlations are still not very adequate. It has been found
that the apparent coefficient varies considerably with the temperature difference between the
heating surface and the liquid. For temperature differences greater than about 20oC, values of h
decrease, apparently because of blanketing of the heating surface by vapours. Over the range of
temperature differences from 1oC to 20oC, values of h for boiling water increase from 1200 to
about 60,000 Jm-2s-1oC–1. For boiling water under atmospheric pressure, the following equation
is approximately true:

h = 50(T)2.5 (5.36)

where T is the difference between the surface temperature and the temperature of the boiling
liquid and it lies between 2oC and 20oC.

In many applications the high boiling film coefficients are not of much consequence, as
resistance in the heat source controls the overall coefficients.