Energy and Sustainability II 399

Mathematical models of air-cooled condensers

for thermoelectric units
S. Bracco1, O. Caligaris2 & A. Trucco1
1
Department of Machinery, Energy Systems and Transportation,
Genoa University, Italy
2
Department of Production Engineering,
Thermoenergetics and Mathematical Models, Genoa University, Italy

Abstract
The present paper deals with the use of air-cooled condensers in thermoelectric
units, such as steam power plants or combined cycle units. The paper describes
the main features of air-cooled condensers, showing the problems related to their
use, such as the presence of non-condensable gases inside the steam flow or the
condenser performances variability due to ambient conditions. The present study
refers to A-frame air-cooled condensers characterized by single-pass and
multiple-row arrangement. A phenomenological study has been done in order to
calculate the main physical parameters which describe the condenser operating
conditions. In particular, three mathematical models have been developed: while
both the first and the second model consider a condenser with tubes
characterized by the same finned surface, with or without the use of throttling
valves upstream of each tube, the third model examines the behaviour of a
condenser with different row effectiveness. The three mathematical models
permit one also to investigate the possible vapour back flow inside each row of
tubes, by calculating the axial coordinate value along each tube at which the
condensing steam mass flow rate becomes null; the third mathematical model
has been implemented in the Matlab/Simulink environment in order to couple the
simulator of the condenser with the whole plant simulation model.
Keywords: air-cooled condenser, non-condensable gases, simulation model.

1 Introduction
Nowadays air-cooled condensers are more and more installed in thermoelectric
units, taking the place of water cooled condensers, because their use allows to

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
doi:10.2495/ESU090351
400 Energy and Sustainability II

save a lot of cooling water, in accordance with strict environmental rules, and to
build power plants in sites far from rivers or the sea [1]. But, as reported by
Fabbri [24], Krger [5] and Larinoff et al [6], air-cooled condensers are also
characterized by several technical and practical problems: the heat exchange
process between the ambient air and the condensing steam needs large heat
exchange areas which have to be cleaned periodically; the condenser operating
pressure depends on the site conditions such as ambient temperature, humidity
and windiness; the large number of fans are characterized by high electricity
consumptions and noise. Furthermore one of the major problems of a single-pass
air-cooled condenser is the accumulation of non-condensable gases inside the
finned tubes, as analysed by Fabbri [2], Krger [5] and Berg and Berg [7]; this
undesired phenomenon determines the reduction of the effective heat exchange
area and possible condensate freezing in several finned tubes.
The aim of this paper is that of describing three different simplified
mathematical models which have been implemented in order to predict the
steady-state behaviour of an A-frame air-cooled condenser, considering different
design data and operating conditions. Both the first and the second model have
been implemented using the Fortran code and are used to study the condenser
from a phenomenological point view while the third simulator, tuned-up in the
Matlab/Simulink environment, has been created and validated taking into
account the design and operating data of a condenser installed in a 400 MW
combined cycle power plant still operative in the Italian electricity market.

2 The air-cooled condenser

The air-cooled condenser is a steam-air heat exchanger composed by several
modules such as that sketched by fig. 1.

Figure 1: The cross-section of a single module of the air-cooled condenser.

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
 Energy and Sustainability II 401

The exhaust steam, leaving the low pressure turbine, comes up to the
horizontal main duct and then enters the parallel-flow modules flowing down
inside banks of finned tubes where its condensation occurs owing to the heat
exchange with the ambient air moved by axial fans; the condensate accumulates
into two outlet headers, in order to be withdrawn, while the remaining
condensing steam goes to the dephlegmator characterized by the steam and
condensate in counterflow arrangement. The dephlegmator modules are also
equipped with steam jet air ejectors for the removal of non-condensable gases.
Each fan is usually moved by a two-speed motor and the finned tubes are
installed in staggered rows; tubes can be round, elliptical or flat and several types
of fins are on the market with the aim of optimising the technical performance of
condensers subjected to different operating conditions.

3 The mathematical models

The developed mathematical and simulation models are based on the NTU
effectiveness method which permits to determine the condenser operating
parameters beginning from the calculation of the rows effectiveness and the
condenser effectiveness, as suggested by Fabbri [2]:
T T Tair _ out Tair _ in
Erow _ i = i +1 i , Econdenser = (1)
Ts Ti Ts Tair _ in
where Ti is the air temperature upstream the ith row while Ts is the condensation
temperature. Furthermore the system thermal balance equation is:
( )
M air c p air Tair _ out Tair _ in = M s r xs (2)
where M air and M s denote the air and steam flow rates, c pair is the air average
specific heat at constant pressure, r the condensation latent heat and xs the steam
quality at the condenser inlet. In accordance with the model proposed by Fabbri
[2], the steam condensation rate per unit of tube length is assumed constant along
each tube:
M air c p air (Ti +1 Ti )
m si = (3)
r xs L
where L denotes the tubes length.
The first two models analyse the steady-state behaviour of a single module of
the condenser, coupled with the corresponding fan, and characterized by four
rows of finned tubes while the third one refers to a condenser composed of 21
modules (18 parallel-flow and 3 dephlegmator) with three tubes rows. The first
step of the analysis has been the design calculation of the condenser main
operating parameters: the steam mass flow rate entering each row, the pressure
drop inside the tubes, the air intermediate temperatures Ti, the rows effectiveness
and the axial coordinate value ai along each tube at which the condensing steam
mass flow rate becomes null and the pressure reaches the minimum value. All
models assume the pressure drop per unit of tube length proportional to the
square of the current steam flow rate and consider the same pressure drop

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
402 Energy and Sustainability II

between the inlet and the outlet of each finned tube, the main duct and the lower
headers being in common.

3.1 The first model

The main inputs of the first simulator are: the air inlet temperature, the exhaust
steam flow rate at the turbine outlet, the pressure in the main duct and the
geometrical data of the condenser. The model considers four tubes rows having
the same effectiveness. For this case, in absence of non-condensable gases, the
first rows, which are those in contact with the coldest air, have pressure
minimum values inside the tubes instead of at their outlet (ai < L); as a
consequence, there is vapour back flow from the other rows to the first ones. If
the presence of non-condensable gases is considered, these gases accumulate
along the first rows from the ai axial coordinate to the tubes outlet and determine
the reduction of the heat exchange effectiveness.
The following considerations refers to the simulation of a parallel-flow
module belonging to a condenser characterised by the next design data:
- Total thermal flux = 218.05 MWth;
- Total steam mass flow rate = 97.51 kg/s;
- Nominal condensation temperature = 41.53 C;
- Steam quality at the condenser inlet = 0.9308;
- Ambient temperature = 15 C, ambient pressure = 100.4 kPa;
- 20 modules each composed of 6 bundles;
- 4 rows per bundle;
- Condensing steam mass flow rate per bundle = 0.81 kg/s.

Figure 2: Condensing steam flow rates (E1=E2=E3=E4).

Assuming that each row has an effectiveness equal to 35%, the following air
intermediate temperatures have been calculated by means of eqn. (1): T2 = 24.2
C, T3 = 30.31C, T4 = 34.24C, T5 = 36.79C. The condensing steam mass flow
rate and pressure along each tube row are plotted by fig. 2 and fig. 3, considering
the presence of non-condensable gases inside the flow. It is possible to notice

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
 Energy and Sustainability II 403

that the steam condenses completely inside both row 1 and row 2, which operate
in better conditions in terms of air lower temperatures and higher steam mass
flow rates; the condensing steam flow rate becomes null at the end of the 3rd row
while the steam which enters the 4th row does not condensate completely inside
the tubes and goes to the dephlegmator unit. If non-condensable gases had not
been considered, the steam exiting the 4th row would have gone back into the
tubes of the first two rows. So it is clear that the vapour back-flow does not occur
inside the tubes where non-condensable gases accumulate.

Figure 3: The pressure along each tube (E1=E2=E3=E4).

Being the operating conditions depicted by fig. 2 and fig. 3 not optimal, in
terms of heat exchange effectiveness and steam flow rate distribution between
the tubes rows, it has been necessary to find better design data for the condenser
in order to optimise its technical performance; it would be better to have the
same flow rate at each tube inlet, the same air temperature increase through each
tubes row and the steam complete condensation at the end of tubes (ai L), in
order to avoid any vapour back flow and non-condensable gases accumulation.

3.2 The second model

Different solutions have been adopted in order to optimise the condenser

operating conditions. The first is relative to a condenser with tubes characterized
by the same finned surface and consists in installing a gauged throttle upstream
of each tube with the aim of completing the steam condensation not before the
tube end. The condensing steam mass flow rate and pressure along each tube row
are plotted by fig. 4 and fig. 5, considering gauged pressure drops upstream of
each tube and row effectiveness equal to 35%.
The results show that the steam flow rate at the tubes inlet is not the same for
all rows and it is higher for row 1 in contact with cold air; as a consequence, this
solution does not seem to be the most adequate. It appears clear that the best

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
404 Energy and Sustainability II

Figure 4: Condensing steam flow rates (E1=E2=E3=E4, gauged pressure

drops).

Figure 5: The pressure along each tube (E1=E2=E3=E4, gauged pressure

drops).

solution is that of designing condensers with tubes rows characterized by

different effectiveness values, as nowadays proposed by manufacturers.

3.3 The rows effectiveness optimization

From the considerations drawn in 3.1 and 3.2 it derives that a well-designed
multi-row air-cooled condenser needs differently finned tube rows; in particular,
it is necessary to increase the fin surface of the more external rows which are
those exposed to a hotter air flow. The optimization of the rows effectiveness for
the condenser described in 3.1 has determined the following best values: E1 =
20.7%, E2 = 26.2 %, E3 = 35.5 %, E4 = 55.0%. Considering these nominal
values, the condensing steam mass flow rate and pressure along each tube
become the same for all rows, as shown by fig. 6. Furthermore, there is not
vapour back flow and the air temperature increase is the same through each row.

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
 Energy and Sustainability II 405

Figure 6: The pressure and steam flow rate along each tube (E1E2E3E4).

3.4 The third model

The third simulation model, implemented in the Matlab/Simulink environment,

is called operative because it is going to be linked to the dynamic simulator of
the whole combined cycle power plant; it is used to calculate off-line the
condenser performance parameters in off-design operating conditions. In this
case the steam flow rate is an important datum while the pressure downstream
the turbine is unknown. In the following the main equations used to simulate a
single module of the condenser are reported.
The main inputs of the simulator are:
- the ambient temperature (T1) and pressure (p1);
- the exhaust steam mass flow rate and quality at the condenser inlet;
- the condenser design operating parameters: condensation temperature (Ts-nom
= 37.90C), ambient temperature and pressure (T1-nom = 15C, p1-nom =
101.6 kPa), exhaust steam mass flow rate and quality ( m s nom = 5.52 kg/s,
xs-nom = 0.9256), air volume flow rate and density ( Q air nom = 581 m3/s, air-nom
= 1.228 kg/m3), air intermediate temperatures (T2-nom = 20.74C, T3-nom =
26.48C , T4-nom = 32.22 C), rows effectiveness (E1-nom = 25.07%, E2-nom =
33.45%, E3-nom = 50.27%), condenser global effectiveness (Econdenser =
75.2%);
- the condenser geometrical data.
The heat balance equation for the ith row can be written as:
M air c p air (Ti +1 T1 ) = U i Si Tlm _ i i +1 (4)
where Ui is the global transmittance, Si the heat exchange area and Tlm the log-
mean temperature difference. Since the NTU (Number of Transfer Units)
corresponds to the ratio [8]:
U i Si
NTU i = , (5)
M air c p air

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
406 Energy and Sustainability II

after a little algebra one gets:

Erow _ i = 1 e NTU i . (6)
Considering constant the average values of Ui, Si and cp air, it derives that the
product ( M air NTU i ) must be kept constant and so the effectiveness values can
be calculated from the current air mass flow rate, which depends on the fan
rotational speed, as follows:
M air nom NTU i nom

M
Erow _ i = 1 e air
. (7)
The next step is the calculation of the condensation temperature Ts: it is
determined by eqn. (2), after having expressed the condensation latent heat as a
polynomial function of Ts and the air outlet temperature T4 as a function of E1,
E2, E3, T1 and Ts. The Ts calculated value permits to determine the condensation
pressure, by means of the water-steam properties, and the air intermediate
temperatures T2, T3 and T4, by means of eqn. (1). As a consequence the steam
condensation rates per unit of tube length can be determined by eqn. (3).
As proposed by Fabbri [2], it is necessary to distinguish the tubes
characterized by vapour back flow (bf) from those where the steam does not
condense completely (nbf). In the first case, the steam mass flow rate at the axial
coordinate x along the ith tube is calculated by:
( )bf
M si = m si (ai x ) (8)
where ai is the length of the tube segment in which the steam flows from the
main duct to the lower header. The steam flow rate crossing the outlet end of
each tube is equal to:
 = m (L a ) .
(9)
si si i
On the other hand the steam flow rate at the axial coordinate x along the tubes
without back flow is calculated by:
M si ( )nbf
= m si (L x ) +
 .
si (10)
The calculation of the pressure drop between the inlet and the outlet of each
tube has been done referring to the model reported by Fabbri [2]. In particular,
for the tubes with back flow the following equation has been adopted:
k m 2
(
(pi )bf = i si 2ai3 3ai2 L + 3ai L2 L3
3
(11) )
ki denoting a friction coefficient which depends on the tubes material.
On the other hand, the tubes without back flow are characterized by a pressure
drop equal to:
k m 2
(
(pi )nbf = i si L3 3ai L + 3ai2 L .
3
(12) )
Since all tubes rows have the inlet and outlet plena in common, the pressure
drops have to be the same:
(pi )bf = (pi )nbf i. (13)

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
 Energy and Sustainability II 407

In order to calculate pressure drops within the Matlab/Simulink model

efficiently, eqn. (12) and (13) have been implemented in non-dimensional form,
as done by Fabbri [2] by proper choice of reference variables, as a function of the
steam flow rates defined by eqn. (8) and (10). Assuming the steam total flow rate
at the exit of tubes without back flow equal to the one flowing back to the other
tubes, that is equivalent to not considering a dephlegmator module downstream,
the model calculates the pressure drop between the inlet and outlet plena. Once
this value has been determined, it is possible to solve eqn. (11) and (12) in order
to calculate the axial coordinate ai along each tube at which the condensing
steam mass flow rate becomes null.
As a consequence, the steam flow rate at each row inlet can be evaluated by
the following equation:
M si = m si ai . (14)
In order to consider the presence of a dephlegmator unit downstream of the
parallel-flow module, it is necessary to modify the tubes length as follows:
L* = 1L , < 1 (15)
where the coefficient is connected with the steam flow rate to the
dephlegmator:
M si _ dph (1 ) M si . (16)
Consequently, the steam total flow rate at the parallel-flow module inlet is
equal to:

M si =
i

m si ai =
i
M si _ dph + M si _ c
i

(17)
i
where:
M
i
si _ c = L* m
i
si (18)

is the steam flow rate which condenses inside the parallel-flow module.
The rows length increase is equivalent to impose the steam complete
condensation inside the system composed of one parallel-flow module and one
dephlegmator unit connected in series.
3.5 The condenser performance curves
The aim of the present analysis has been that of determining, by the third
simulator, the condenser performance curves which show, on the pressure
steam flow rate plane, how the condensation pressure ps is influenced by both the
air inlet temperature T1 and the fans rotational speed.
Figure 7 reports the performance curves considering all fans running at full
speed. It is clear that, for the same value of steam flow rate, the higher is T1 the
higher is the condenser pressure; on the other hand, considering the same
temperature T1, the lower is the steam flow rate the lower is the condenser
pressure. The curves in fig. 7 are not parallel lines but their slope increases with
the air inlet temperature.
Since this type of air-cooled condensers are equipped with fans running at
two different speed values, respectively 100% and 50% of the nominal rotational

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
408 Energy and Sustainability II

speed, it has been interesting to calculate the performance curves also for the
case of all fans at half speed, that is an operative condition to which the
condenser is subjected both during startup and cold days. The results are plotted
in fig. 8: the curves have the same shape as those depicted by fig. 7 but are
characterized by a higher slope. In fact in this case, considering the same values
of air inlet temperature T1 and steam flow rate, the air flow rate is lower and so
the air temperature T4 at the condenser outlet is higher; as a consequence, the
condensation temperature increases as well as the pressure.

Figure 7: The condenser performance curves with all fans at full speed.

Figure 8: The condenser performance curves with all fans at half speed.

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
 Energy and Sustainability II 409

The curves obtained by the third simulation model have been compared with
those supplied by Ansaldo Energia SpA, an Italian company which installs air-
cooled condensers in its power plants sites. The model validation has been done
with reference to different steady-state operating conditions and it is important to
remark that the calculated performance curves are in good agreement with
manufacturers ones.

4 Conclusions
The mathematical and simulations model described in this paper have been
developed to study a multi-row A-frame air-cooled condenser for combined
cycle power plants. The analysis has been focused on the simulation of the
condenser behaviour under different steady-state operating conditions with the
aim of evaluating the influence of the air inlet temperature on the system
performance. The problem of the non-condensable gases accumulation inside the
finned tubes rows has been analysed from a phenomenological point of view and
some technical solutions have been proposed in order to reduce the risk of their
formation, as well as the vapour back flow inside the tubes.
The condenser performance curves have been determined considering two
different fans operating conditions, full or half speed, and varying both the air
inlet temperature and the saturated exhaust steam flow rate. The calculated
curves validation phase has confirmed the model validity over a wide range of
operating conditions.
The next step of the study is going to be the development of a dynamic
simulation model to study the condenser behaviour under different transient
conditions; this model will be integrated with the combined cycle power plant
dynamic simulator and validated with experimental data coming from the plant.

References
[1] Nagel, P. & Wurtz, W., Dry cooling for power plants: an innovative
modularization concept. PowerGen Europe SPX Conference, Cologne,
2006.
[2] Fabbri, G., Analysis of vapor back flow in single-pass air-cooled
condensers. Int. J. of Heat and Mass Transfer, Vol. 40, No. 16, pp. 3969-
3979, 1997.
[3] Fabbri, G., Effect of disuniformities in vapour saturation pressure and
coolant velocity on vapour back flow phenomena in single-pass air-cooled
condensers. Int. J. of Heat and Mass Transfer, Vol. 43, pp. 147-159, 2000.
[4] Fabbri, G., Analysis of the noncondensable contaminant accumulation in
single-pass air-cooled condensers. Heat Transfer Engineering, Vol. 18, No.
2, pp. 50-60, 1997.
[5] Krger, D.G., Air-cooled Heat Exchangers and Cooling Towers, Penn Well
Corporation, 2004.
[6] Larinoff, M.W., Moles, W.E. & Reichhelm, R., Design and specification of
air-cooled steam condensers. Texas Chemical Engineering, 1978.

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)
410 Energy and Sustainability II

[7] Berg, W.F. & Berg, J.L., Flow patterns for isothermal condensation in one
pass air cooled heat exchanger. Heat Transfer Engineering, Vol. 1, No. 4,
pp. 21-31, 1980.
[8] Ricard, J., Equipement Thermique des Usines Gnratrices dEnergie
Electrique, Dunod, Paris, 1962.

WIT Transactions on Ecology and the Environment, Vol 121, 2009 WIT Press
www.witpress.com, ISSN 1743-3541 (on-line)