Theoretical solution for the cross-flow heat

exchanger

A. Hofmann

Abstract
The theory of the cross-flow heat exchanger first was treated by Nuelt [1, 2] on the base of
heat balances for the two interacting fluids over an exchanging area element. This leads to a
partial differential equation. But the solution was an infinite row which could not be
expressed in a compact formula. In this paper it will be shown that a compact formula for the
infinite row is possible. All temperatures of the interacting fluids, for example the local fluid
temperatures and the mean outlet temperatures as well as the local temperature difference and
the mean temperature difference over the complete exchanger are now being available in
simple formulae which have the form of an infinite sum. The summation has to be stopped at
a finite value with a negligible deviation.
As all variables in the formulae are dimensionless, normalized diagrams are developed which
are generally valid and give a good overview over a wide range of exchanger conditions.

Crossflow Air HX
Introduction
Crossflow heat exchangers are commonly used in gas heating or cooling. A tube bundle carries
a heating or cooling fluid (either gas or liquid), normally perpendicular to a gas flow which
passes over the tubes and allows heat to be transferred between the fluids. Crossflow heat
exchangers can be of the mixed or unmixed type (see figure 1). The mixed type is the simpler
of these designs in which the gas is mixed and not separated into channels.

Figure 1: Crossflow heat exchanger types (Cengel, 2012)

Heat exchanger design

The exchanger design consists of several stainless steel tubes embedded in a boxed housing
(see figure 2). This unit is fitted into the stove pipe above the stove such that flue gas passes
vertically through the housing (driven by natural convection) and passes over the tubes. An
electrically powered fan sited at the rear of the tubes creates a high velocity air flow to
improve heat transfer characteristics (as opposed to those obtained by natural convection).
The performance of a heat exchanger of this type could be modified through varying tube
number, dimensions and spacing and fan volume capacity (i.e. how much air the fan can
move).

Parameter

Size [mm]

Tube length

300

Tube
diameter

30

SD

60

Sp

60

Table 1: Heat exchanger tube geometry

Figure 2: Basic crossflow heat exchanger design (Magic Heat Reclaimer, 2013)

Modelling crossflow heat exchanger

The method adopted for modelling the heat transfer performance of a crossflow heat
exchanger integrated into a domestic biomass stove flue pipe follows that of the LMTD
method.
- Logarithmic mean temperature difference (LMTD)
A theoretical explanation of this method can be found elsewhere and will not be explained
here (Holman, 2002). In general terms, the method quantifies the temperature difference
between two exchange fluids by the log mean temperature difference, T m(CF). This can be
expressed as a function of the inlet and outlet temperatures of the fluids as:

where Ta and Tb are defined as the difference between the hot fluid inlet temperature and
cold fluid inlet temperature, and hot fluid outlet temperature and cold fluid inlet temperature
respectively. The bracketed term (CF) denotes that this LMTD is for the case of a counterflow
heat exchanger. For other types of heat exchanger (i.e. crossflow) a correction factor F must
be applied.

- Correction factor, F
The correction factor is obtained from the calculation of two ratios P and R associated with
the inlet and outlet temperatures of the two fluids as shown in figure 3. The value of F is then
read from the graph. The value of Tm is then taken as:

Figure 3: Correction factor for a single-pass cross-flow heat exchanger with one fluid mixed
(Cengel, 2012)

- Convection correlations
In order to determine the heat transfer coefficients between the fluids and the heat exchanger
surface, convection correlations must be used. These correlations allow the calculation of the
dimensionless Nusselt number Nu, which can consequently be used to calculate the heat
transfer coefficient for the fluid. For the case of turbulent flow through a pipe as is the case
for the forced air flow:

The Reynolds number Re is defined:

where the terms are defined as fluid density , fluid velocity u, fluid dynamic viscosity , and
characteristic dimension l (pipe diameter). The Prandtl number Pr is a function of the fluid
properties and can be ascertained according to temperature. The convection heat transfer
coefficient is then calculated:

The temperature at which fluid properties are evaluated is taken to be the average of the inlet
and outlet temperatures.

For the case of flow over a bank of tubes (i.e. flue gas flow), the Grimson correlations are used
(Holman, 2002). For a flow normal to an in-line tube bank the maximum flow velocity occurs
through the minimum fontal area (Sn-d). Thus the maximum flow velocity is expressed

where u is the free stream velocity, d is the pipe diameter as previously and S n is as shown in
figure 4.

Figure 4: Heat exchanger tube geometry

The Reynolds number is then calculated using the equation previously described, with u
replaced by umax and the characteristic dimension l being the tube diameter d. For this analysis
it is normally recommended that properties should be evaluated at the film temperature T f
which is the average of the bulk fluid temperature (prior to the tube bank) and the tube
surface temperature. However, in order to simplify calculations, properties have been
evaluated at the average between the bulk fluid temperature the average air temperature
inside the tubes. This will give a lower heat transfer coefficient than would otherwise be
obtained, and thus the model can be considered a worst-case scenario in this respect. The
Nusselt number can then be calculated by the Grimson correlation:

The constants C and n are obtained according to Grimson as C=0.254 and n=0.632 (Grimson,
1937).
In the case of a tube bank with less than 10 rows a correction factor must be applied
according to table 2.

10

Ratio

0.68

0.75

0.83

0.89

0.92

0.95

0.97

0.98

0.99

1.00

Table 2: Ratio of convection coefficient h for N rows deep to that for 10 rows deep (Holman, 2002)

The convection heat transfer coefficient can again be calculated as before. Assuming that the
walls of the tubes are thin-walled and therefore conduction can be neglected, the total heat
transfer coefficient U is expressed:

Given specified inlet and outlet temperatures, the heat transfer can be calculated:

The mass flow rate of flue gas is dependent upon the stove, and the air flow rate upon the fan
power used. The flue gas specific heat capacity is assumed to be the same as air at the same
temperature. Then the total heat exchanger area A is related to the heat transfer by the
relation:

The total heat exchanger area A is also expressed as:

Here np is the number of pipes and d and l are as previously defined.

- Pressure drop
According to Basu (1999) the pressure drop across tube bundles can be expressed:

where K is known as the loss coefficient. For the case of a series of inline pipes this is
calculated by the relation:

where n is the number of rows and K0 is the loss coefficient for a single row of tubes. For the
spacing of tubes in this design, the loss coefficient is given by Perkov (1965 in Basu 1999) as:

Here s1 = Sp/d and f = (Sp-d)/(Sn-d), see figure 4.

There are additional relations for the pressure drop at the expansion on entrance to the heat
exchanger and exit. From Basu (1999) the loss coefficients for expansion and contraction are
found to be Kexpansion/contraction = 0.43. This is calculated according to the ratio of cross-sectional areas
of the flue pipe to the HX box.
The pressure drop for the expansion and contraction can then be expressed:

RESULTS
The performance of the heat exchanger was investigated for the case of a range of different
stoves. A basic study of the effects of heat exchanger design was also performed by modifying
fan power, and hence air flow rate through the tubes. In all cases the payback period is
calculated as shown in the economic analysis section.
A basic investigation was made into the optimum number of tubes and fan power based on
these tube dimensions and a 6 kW low efficiency stove.
- Tube number
Firstly, the optimum number of tubes was investigated. Tubes numbers were chosen as shown
in table 3 and the heat recovery and payback calculated. Note that each row of the heat
exchanger consists of three tubes, therefore an additional 3 tubes is equivalent to one
additional row.

Number of tubes

Heat transfer [W]

Payback period
[years]

542

1.13

12

633

1.06

15

703

1.04

18

761

1.04

21

810

1.10

24

852

1.11

Table 3: Effect of tube number on heat recovery and payback period

This shows that a heat exchanger with a tube number of either 15 or 18 gives the best
payback, though there are only small variations between all the systems tested. Although
increasing the number of tubes increases the overall heat exchange surface area and hence
the total heat transfer, it also reduces the velocity through the tubes, and increases material
costs. Thus increasing tube numbers beyond 18 does not offer any advantages in terms of
payback despite higher heat transfer.
- Motor power
The fan motor power was varied according to models available from ebm-papst (ebmpapst,
2013). Results are shown in table 4.

Motor consumption
[W]

Motor quoted flow

rate [m3/s]

Heat transfer [W]

Payback period
[years]

26

0.058

484

1.11

30

0.128

703

1.04

58

0.264

902

0.94

80

0.383

993

1.00

100

0.472

1039

1.00

Table 4: Effect of fan motor power on heat recovery and payback period

As can be seen, a higher air flow rate results in greater heat transfer due to the higher heat
transfer coefficient associated with a higher velocity. For example the heat transfer
coefficient between the air and tube wall using a fan motor with consumption 26W is 6.65
W/m2.K, and for a 58W fan motor it is 22.26 W/m2.K. However, due to the higher power
consumption and higher initial cost of a larger fan motor, increasing the motor above 58W does
not reduce the payback period and therefore it is concluded that 58W provides the optimum
air flow rate.
- Optimised system
Taking the results of these simple investigations into account, it was can be concluded that a
crossflow heat exchanger with 15 tubes and a 58W fan motor would provide the shortest
payback period. A system with these parameters was tested across a range of stove types and
the results obtained are shown in table 5.

Stove
power
[kW]

Stove
efficiency

Stove fuel

Energy
recovered
[W]

Efficiency
improvemen
t [%]

Payback
period
[years]

Low

Log

902

7.9

0.94

Medium

Log

629

7.2

1.95

High

Pellet

386

5.1

3.37

10

Low

Log

1136

0.73

10

Medium

Log

764

5.2

1.54

10

High

Pellet

483

3.8

2.50

Table 5: Crossflow heat exchanger performance

The results show that in terms of heat transfer, the integration of the cross flow heat
exchanger with a low efficiency 10kW log stove provides the highest overall heat recovery. This
can be explained to a large extent by the fact that the flue gas temperature in this stove is
considerably higher than both the medium and high efficiency models at 286 oC compared to
174oC and 132oC. However the greatest stove efficiency improvement was observed for the
case of a 6kW low efficiency log stove.

Other feasibility considerations

There are several other factors which may present barriers to the widespread adoption of the
style of heat exchanger detailed in this section.
- Noise
There is a significant issue with the noise levels created by a fan running during operation of a
stove. Generally the noise from a fan of the size suggested would be expected to produce a
noise level over 60 dB (Norton, 2003). This is confirmed according to the empirical relation:
Lw = 94 + 20 log( S ) - 10 log( Q )

where S is the rated motor power [W] and Q is the volume of air discharged [m3/s] discharged
(The Engineering Toolbox, 2013). This sound power level is referenced to the threshold of
human hearing ~10-12 W and is only an approximation.

Motor consumption
[W]

S [W]

Q [m3/s]

Lw [dB]

26

0.058

60.35

30

0.128

60.99

58

16

0.264

63.87

80

25

0.383

66.13

100

34

0.472

67.89

Table 6: Noise levels from different fan powers

The BS 8233 gives reasonable levels of noise in a living room as 40dBA (BSI, 1999) and the
BRE advises levels < 35 dBA (BRE, 1993).
Therefore the sound levels predicted to be produced by the fan could be considerably above
these guideline levels. It is possible that the noise would become a nuisance for occupants of
the dwelling and this factor is one of the main disadvantages of this particular heat exchanger
design. However, the limits suggested are not legally binding levels and so does not present
any obstacle for the installation of such a device if a homeowner so desired. Therefore it my
be the case that some people are willing to put up with extra noise due to the energy and cost
savings which are associated.
A possible measure which may be taken to reduce the noise level at specific times would be to
incorporate a dimmer switch which may be used to manually control the fan power such that
rotation and hence noise levels are reduced.
- Space
The retrofitting of the heat exchange unit into an existing stove will require a space of at least
the height of the unit (~0.27m) between the stove top surface and the flue exit point from the
room. In reality, it would be advisable to perhaps double this distance in order to reduce risk
of damage to the unit, and to allow ease of installation and maintenance.
- Maintenance

It is expected that the heat exchanger of the unit will require periodic cleaning in order to
remove creosote from the heat exchange tubes. This could be done by hand or by an additional
piece of sheet metal that can be moved over the tube surface using a small handle on the
front of the unit.

References
Basu, P., Kefa, C., Jestin, L. (1999) Boilers and Burners, Design and Theory, New York: Springer
BSI. (1999). " Sound insulation and noise reduction for buildings. Code of practice." BS 8233,
London, UK
Building Research Establishment (BRE) & Construction Industry Research and Information
Association (CIRIA) Sound control for homes (1993).
Cengel, Y. A., Turner, R. H., Cimbala, J. M. (2012) Fundamentals of Thermal-fluid Sciences, 4th
edition, London: McGraw Hill Higher Education
ebmpapst (2013) Q-motor multifunction motor technical specification sheet. [online] Available
at: http://docs-europe.electrocomponents.com/webdocs/052f/0900766b8052f3b9.pdf [Access
ed: 22 Apr 2013]
Grimson, E. D. (1937) Correlation and Utilisation of New Data on Flow Resistance and Heat
Transfer for Cross Flow of Gases over Tube Banks. Trans. ASME. 59 583-594
Holman, J. O. (2002) Heat Transfer, 9th edition, London: McGraw Hill Higher Education
Magic Heat Reclaimer, 2013. [Online] Available at: http://www.magicheatreclaimer.com/magic-heatreclaimers/ [Accessed 12 May 2013].
Northern Tool (2013) Magic HeatTM Reclaimer. [online] Available at:
http://www.northerntool.com/shop/tools/product_200331508_200331508 [Accessed: 22 Apr
2013]
Norton, M., Karczub, D. (2003) Fundamentals of Noise and Vibration Analysis for Engineers, 2nd
edition, Cambridge: University Press
The Engineering Toolbox (2013) Fans and Noise Power Generation. [online] Available
at: http://www.engineeringtoolbox.com/fan-noise-d_61.html [Accessed 22 Apr 2013]

Back to top
University of Strathclyde | Sustainable Engineering Group Projects
University of Strathclyde 2013
Web Template created with Artisteer.

The Crossflow Heat Exchanger

A car radiator and an air conditioner evaporator coil are examples of crossflow heat exchangers. In both cases heat
transfer is taking place between a liquid flowing inside a tube or tubes and air flowing past the tubes. With a car
radiator, the hot water in the tubes is being cooled by air flowing through the radiator between the tubes. With an
air conditioner evaporator coil, air flowing past the evaporator coils is cooled by the cold refrigerant flowing inside
the tube(s) of the coil. Crossflow heat exchangers are typically used for heat transfer between a gas and a liquid
as in these two examples.

Understanding Heat Exchangers- Cross-flow, Counter-flow (Rotary/Wheel) and

Cross-counter-flow Heat Exchangers

The core of heat recovery is the heat exchanger. There are various types of heat exchangers
availabe including cross-flow, counter-flow (includes rotary/wheel) and cross-counter-flow.

Through a heat exchanger, fresh-filtered air flows into the house continuously and an equal
volume of stale air simultaneously flows out of the home. These airflows are allowed to pass
by each other - separated only by a thin membrane. The longer the two streams flow past
each other, the higher the efficiency.

How a Cross-flow Heat Exchanger Works

The supply air does not reach quite as high a temperature as with the counter-flow heat
exchanger, as at two corners temperatures with high differences between them encounter each
other. The efficiency is therefore less, even when a very large exchange surface area is
provided. In an optimal scenario it reaches about 70% efficiency.
The large temperature difference at one part of the surface means this form of heat exchange
reaches its maximum even with small surface areas. The devices can be constructed to be
very compact. This is in contrast to the counter-flow heat exchanger, which is more effective
the longer it is.

How a Counter-flow Heat Exchanger Works

These can be distinguished in that the counter-flow zone comprises the largest part of the
device. At the beginning and end of the device there are very small zones with crossed air
streams, and here too the problem of joining streams must be resolved. The efficiency in the
case of very long dimensions is entirely dependent on the available surface area and in
practice reaches 95%.

How a Rotary/Wheel Heat Exchanger Works

A rotary/wheel heat exchanger is considered a counter-flow heat exchanger. The advantage of

the rotary heat exchanger is that the air distribution is more straight-forward than in some
other counter-flow heat exchangers. In the latter device there is the issue of the complex
guiding of air from one side into the counter-flow zone, and then on the other leading out
from the counter-flow zone. The counter-flow heat exchanger also involves relatively
complex membranes and complex problems of maintaining air tightness between the
membranes, as the two air flows are not to mix. However it's on exactly this point that
adrawback to the rotary/wheel exchanger arises. While the other devices presented above
took pains to ensure the air flows do not come in contact, the rotary exchanger design accepts
this will happen. The cell through which the return air flows will have outside air flowing
through it just a short time later. Therefore the airtightness is not good due to the moving
wheel and it's more susceptible to leakage between the fresh air and stale airstream.

How a Cross-Counter-flow Heat Exchanger Works

The thermally wasteful corners are omitted thanks to the counter-flow zone. The remaining
cross-flow zones do not play as crucial a role if the counter-flow zone has sufficient surface
area. With this geometry too, an efficiency of up to 95% can be achieved.

Figuratively speaking, the cross-flow heat exchanger is divided in the middle and pulled
apart.

The principle of airstreams passing each other (counter-flow) is applied to the space created.
The result is the cross-counter-flow heat exchanger.

This design has several benefits :

a relatively small size can be achieved;

the crossing of the air streams at the front and behind resolves the
problem of joining the multiple streams;

the counter-flow area in the middle gives high efficiency