Proceedings of

2000 International Joint Power Generation Conference

Miami Beach, Florida, July 23-26, 2000

IJPGC2000-15002

MONITORING OF AIR CONTENT IN A MIXTURE REMOVED FROM CONDENSERS IN APPLICATION TO

STEAM TURBINE DIAGNOSTICS

Marian Trela, Dariusz Butrymowicz, Jerzy Guch, Andrzej Gardzilewicz, Eugeniusz Ihnatowicz
Institute of Fluid-Flow Machinery, Polish Academy of Sciences,
Fiszera 14, PL 80-952 Gdask, Poland,
+48 058 341 12 71, mtr@imppan.imp.pg.gda.pl

Bogusaw Zieliski
B. Z. Elektronika
Reja 10 A, PL 81-874 Sopot, Poland

ABSTRACT
The paper concerns the problem of the influence of non condensing
gases (air) on the condenser performance and on the overall efficiency
of the power unit. An efficient method for the air inleakage
measurement is proposed. It allows the accurate measurement and
monitoring of air inleakage rate in the wide range from 0.5 to 300 kg/h
of dry air. The results of such measurement allow in the next step the
calculation of the` power loss due to the increase of the condenser
pressure. The monitoring system based on the above method has been
constructed and implemented in the power station for power units of
200 MW.

U overall heat transfer coefficient, W / m 2 o C

To terminal temperature difference
Subscripts:
1
initial
2
outlet
a
air
b barometric
c
condensate
f
fouling
m mixture
o
reference value
t
tube wall
v
vapour
vac vacuum
w water
log log mean difference

NOMENCLATURE
A area
cp specific heat, J / kg o C
B coefficient
C D orifice discharge coefficient
d
D
i
m

throat diameter
pipe diameter
specific enthalpy, J/kg
mass, ratio m
& a /m
& ao
m
& mass flow rate, kg/s or kg/h
N identification number of the orifice
p pressure, Pa
q heat flux, kW/m2, , specific heat consumption kJ/kWh
Q heat transfer rate, J/s

INTRODUCTION
The paper concerns the measurement of mass flow rate of air
removed from a condenser in a steam power plants. The excessive
amount of air in the condenser diminishes the heat transfer and leads
to the increase of the condensation pressure, thus reducing the
efficiency of the cycle. In the paper, a modern system of continuous
measurement and monitoring of the mass flow rate of air removed
from the condenser is described in detail. The system enables to
measure, precisely and in a wide range, the mass flow rate of
discharged air, with simultaneous data acquisition. Due to the
relatively wide measurement range, the system can be used in the
power plants including turbogenerator units of very low up to very
high outputs.

R thermal resistance, m 2 o C / W
T temperature
To reference temperature
t
V

co-ordinate

time
velocity

Copyright 2000 by ASME

The paper also presents the diagnostics of losses resulting from

the excessive air-inleakage rate in the condenser. The diagnostics is
performed in two ways, each way basing on the air monitoring
system. The first, simplified method analyses the increase of the
terminal temperature difference and the increase of the condensation
pressure, resulting from excessive air concentration. Using certain
technical parameters given by the manufacturer of TK 200 turbine
units, the unfavourable influence of excessive air concentration on the
performance loss is evaluated.
The second, more precise method is based on the method of
diagnosing thermal cycles and their components, used in the code
DIAGAR [8]. Then, the condenser is treated as a component of the
cycle, the operating parameters of which are known. The condenser
terminal temperature difference, condensation pressure, and
condensate subcooling, shown as a function of the air flow rate, is
applied to make characteristic operating curves of the condenser. As a
result of the computation, done for the entire cycle, the unit output and
specific heat consumption are evaluated, as well as their operational
losses, distributed over all cycle components including the condenser.
The above method has been practically verified in the electric power
station [12].

Taking into consideration the following relations:

- logarithmic mean temperature difference

Tlog =

(2)

- energy balance formulae

Q=m
& v (i v i c )
Q=m
& w c pw Tw

(3)

- Peclet equation

Q = UATlog

(4)

for m
& v = const one obtains

T w =

THE BASIC FORMULAS OF CONDENSER PERFORMANCE

The air enters the condenser along with the steam and through the
turbine leaks and other gaps in the vacuum part of the turbogenerator
unit. The excessive amount of air affects unfavourably the heat
transfer during vapour condensation, thus decreasing the overall heat
transfer coefficient, U. This leads to the increase of the condensation
temperature Tc, i.e. to the increase of the condensation pressure pc then
resulting decrease of the turbine performance. The excessive amount
of gases in the condenser promotes also the corrosion, resulting from
the dissolution of O2 and CO2 in the condensate. Therefore, the
concentration of air in the condenser has to be controlled and kept
below a certain limiting values, which depends on the mass flow rate
of vapour entering the condenser and on the condenser geometry.
The temperature distribution in the condenser is shown
schematically in Fig. 1.
The condensation temperature Tc depends on the inlet cooling water
temperature Tw1, the temperature rise of circulating water Tw, and the
terminal temperature difference, To,

Tc = Tw1 + Tw + To

Tw
Tw

ln1 +
To

m
&v
m
&w

iv ic

c pw

m
~ & v
m
&w

(5)

Taking into account equations (3) and (4) one arrives at the terminal
temperature difference

To =

Tw
exp(UB) 1

(6)

where: B = A (m
& w c pw ) .
It is seen that for A, m
& w , c pw = const , the temperature difference
To is the only function of the overall heat transfer coefficient, U

To = f (U )

(7)

THE EFFECT OF AIR CONCENTRATION ON OVERALL

HEAT TRANSFER COEFFICIENT
The above analysis and resulting relations (4) and (6) reveal that
the overall heat transfer coefficient is a dominant factor as far as the
heat ejection parameters are considered. Decreasing the heat transfer
coefficient U provides the increases of the values of
Tlog , To and pc . The total heat transfer resistance R is the sum of

(1)

all individual resistances and may be expressed as:

R=

1
= R w + R t + Rc + Rf + Ra
U

(8)

Figure 1. Relationship between the heat rejection

parameters

Copyright 2000 by ASME

coefficient, hc, by about 30% in this section, which leads to the

decrease of the local heat transfer coefficient, U, by 12% [2]. In
general, the problem of determining the influence of the noncondensable gases on the heat transfer during vapour condensation is
very complex. Butrymowicz & Trela [3] have developed a numerical
code for the evaluating characteristic pertinent quantities for a lowpressure regenerative feedwater heater being a part of a thermal cycle
of a 200 MW condensing turbogenerator unit. In such heat exchanger,
the heat transfer area was divided into a finite number of sections, for
which the energy and mass balance equations were formulated and
solved taking into account: the local influence of air concentration, the
vapour velocity, the bundle effect, and thermal resistance of fouling on
the heat transfer. The results of the computations are shown in
Figure 2 and 3. Figure 2 depicts the temperature distribution along the
cooling water flow path (co-ordinate z) for both: a clean exchanger
(without air and fouling effects) and the contaminated one (with air
and fouling effects). The next figure, Figure 3, shows, in the same
arrangement, the heat flux distribution along the exchangers for both
cases, i.e. for clean and for contaminated exchangers. These obtained
results revealed that the effects of air concentration and fouling are
dominant in respect to the temperature and heat flux distributions in
the regenerative heater during vapour condensation. The same
situation occurs in the condenser when the limiting values of air
concentration and fouling resistance are exceeded.
The excessive gas concentration in the condenser is also
destructive in respect to the corrosion process. According to Coit [4],
in American power stations in the late seventies the corrosion within
condenser and its components contributed to as much as 70 per cent of
overall internal corrosion in the entire power station. In order to avoid
these two unfavourable effects i.e. on the turbine performance and on
the corrosion, the gases concentration in the condenser should be
controlled and kept below a permissible limiting values. These
limiting values depend mainly on the mass flow rate of the vapour
entering the condenser and on the condenser design. The other
important factor is the venting system which has to operate properly.

a)

Tc

60
o

[ C]
50

Tw

40

30
20
0

10

20

30

40

50

60

70

80

90

100

50

60

70

80

90

100

z z o 100

b)

Tc

60

[ C]

Tw

50

40

30
20
0

10

20

30

40

z z o 100

Figure 2. Sample temperature distributions for a lowpressure regenerative feedwater heater (four-stage
horizontal unit) along the water flow path (zo is total length
of water flow path), according to [3].
&a m
& v = 0.31% , Rf =1.510-4 m2C/W,
a) m
&a m
& v = 0 , Rf = 0.
b) m
The thermal resistances involved in equation (8) can be calculated
from various semi-empirical heat transfer correlations which may be
found elsewhere, e.g. in [1]. In nominal operating conditions, when the
fouling contribution may be neglected, R f 0 , and the air

AIR-INLEAKAGE MEASUREMENT SYSTEM

The measurement system of gases removed from the condenser
by using a steam jet ejector consists generally of two parts: one
devices to measure the mass flow rate of vapour-air mixture leaving
the ejector and the other one to determine its physical parameters [5].
The condition of air-vapour mixture exiting the condenser can be
determined by accurately measured pressure and temperature at the
ejector vent connection. The amount of vapour to saturate noncondensibles can be calculated from the well known formula:

concentration in the vapour is negligibly small, R a 0 , then the

overall heat transfer coefficient U depends only on thermal resistances
R w and R c . The thermal resistance due to the presence of air in the
condenser depends not only on its concentration, but also on the
geometry of the condenser tube bundle and on the steam approach
velocity.

&
R a = f a , Vv , geometry
m
&v

mv
pv
= 0.622
ma
p pv

(9)

(10)

The mass flow rate of the mixture can be measured with the aid of one
of standard method, for instance by using an orifice.
Figure 4 presents a schematic diagram of a novel system for
continuous measurement, monitoring and recording of the mass flow

Usually, the air concentration in the vapour at the condenser inlet

equals 0.0001 or less and does not affect the process of heat transfer.
At the outlet (in the cooler) the concentrations is higher, and reaches
about 0.01. This corresponds to the decrease of the heat transfer

Copyright 2000 by ASME

80

a)
q

The temperature was measured with the aid of thermocouples.

Reference temperatures were measured by using resistor
thermometers.
The device identifying the orifice number N was mounted
beneath the movable disc. It was mechanically coupled with the disc.
The device plays a role of a potentiometer, which is a part of the
bridge connected to the amplifier system of 4 20 mA intensity

[kW/m ]
60

40

signal. Therefore the disc position gave an input signal for the device
identifying the number N.

20

0
0

b)
q

10

20

30

40

50

60

30

40

50

60

70

80
90
Z / ZO

100

200
2

[kW/m ]
150

100

50

10

20

70

80
90
Z / ZO

100

Figure 3. Heat flux distributions along the water path for

the regenerative feedwater heater (the same conditions
as in Fig. 2):
a) for air and fouling contaminated exchanger,
b) for clean exchanger.
rate and air contents in the air-vapour mixture leaving the steam jet
ejector. It was designed and implemented in an electric power station
by the present authors. The system provides accurate measure of air
mass flow rate within the range of m
& a = 0.5 300 kg / h . Fig. 5 shows
the view of that system during testing in the IF-FM PASc Laboratory.
The system consists of the following components:
8 measuring orifice plates, mounted on a movable disc,
probes for measuring: pressure difference p , temperature T,
orifice number N, barometric pressure pb, and reference
temperature To,
data acquisition and micro-processing system,
PC-type computer with a monitor.
The first three probes are located directly on the outlet pipe of the
ejector. The sensor amplifier has two- conductor output 4 20 mA .
The pressure tap is located at one pipe diameter upstream from the
orifice plate. The barometric pressure gauge is located in a box
containing the microprocessor-based measuring system. The probe
embraces a self-contained thermometer. Its role is to compensate the
temperature change of the probe. The thermal compensation is done in
by the software. The temperature coefficients of a sensor-amplifier
system were evaluated experimentally for each barometric pressure
probe and were stored in EPROM memory. In the course of the
measurement the pertinent corrections coefficient obtained during
calibration were used.

Figure 4. Schematic diagram of the monitoring apparatus.

Copyright 2000 by ASME

the period of 1 minute, are displayed on the computer monitor and

then recorded and stored on the computer hard disc. Such procedure is
used to calculate the averaged values of m
& a for one hour, day, etc.
Due to the very high accuracy of the probes used, the measurement
ranges for the orifices were 0.04 < p < 0.4 kPa for the orifices
marked by numbers N = 2 8 and 0.03 < p < 7 kPa for the orifice
number N = 1 . When these limits are exceeded, a message appears on
the monitor screen which urge the operator to make the necessary
change of the orifice into a new one marked by a given number N.
There is also a possibility for introducing a limiting value of the air
mass flow rate m
& a , beyond which the system alarms the operator.
In the version described above , the monitoring system can be
used for measurements and monitoring of the mass flow rate of air
within the range of m
& a = 0.5 300 kg / h for one jet ejector.
CONDENSER PERFORMANCE ANALYSIS WITH RESPECT
TO AIR CONTENT
The losses of the condenser performance may be caused by
several reasons. For the loss resulting from the excessive amount of air
in the condenser it can be evaluated very easily if one has an empirical
relation between the terminal temperature difference and the airinlekage.
Folwarczny & Magiera [6] obtained an empirical relation
between the change of the terminal temperature difference, (To) and
the vacuum loss rate (Fig. 6a) for the TK 200 turbine unit. This kind of
examination of the condenser tightness is used in some countries
together with the measurement of the mass flow rate of the removed
mixture. During this examination the air removal system has to be shut
down then the air inflow affects the increase of the condensation
pressure. On the other hand the results of the experimental
investigation performed by the authors of this paper show that for the
TK 200 turbine units the vacuum loss rate of 1 mm Hg/min
corresponds to the mass flow rate of ejected air by approximately 6
kg/h. It is obvious that the terminal temperature difference increase of
(To) corresponds to the condensation pressure increase by pc. It
may be shown, that within the pressure range pc = 5 12 kPa , an

Figure 5. The view of the monitoring system during testing

in IF-FM PASc laboratory

The analogue-to-digital converter has the resolution of 1/20000.

Converting is based on the double integration method. The converter is
constructed with the use the ICL7135 system, controlled by a
PIC16C73 microprocessor. The microprocessor converts output data
to the I2C-bus standard. The converting system is based on the 80C31
microprocessor.
All system components are integrated by the I2C bus (InterIC-Bus). The only exception is the RS-232 interface with optical
insulation, which is controlled directly by the microprocessor. Via this
interface the data are transmitted to the PC-type master computer
located in the turbine set control room.
In the box containing the microprocessor system, the temperature
and pressure displays are located as well.
In order to determine characteristic curves for eight measuring
orifices a series of laboratory tests were carried out in an aerodynamic
tunnel of the Institute of Fluid-Flow Machinery, PASc. These tests led
to the evaluation of the discharge coefficient C D for those orifices.

average pressure change pc due to the saturation temperature change

equals 0.423 kPa/K. This allows to derive the relation P = f (m ) ,
basing on the graph shown in Fig. 6a, which is shown in Fig. 6b
(where P = p c / p c o , m = m
& a /m
& ao ). The increase of the

During the tests the similarity rules were preserved. The test results
were presented in the form of the following formula
n

C D = k Re ,

condensation pressure by pc corresponds to the turbine power output

loss by N. This loss can be evaluated from the following formula [6]:

(11)

N = k c p c ,

where the coefficients k and n are the functions of the Reynolds

number and the ratio d/D. The Reynolds number was derived basing
on the internal exit pipe diameter D. The data was obtained for the

(12)

where the coefficient kc depends on the condenser geometry. For the

turbogenerator unit TK 200 it equals k c = 1.7 MW / kPa [6]. Figure

Reynolds number in the range: Re = (0.2 17 )103 .

With the help of the above mentioned data an numerical

algorithm was developed for the determination of the rate mass flow
rate of the air-vapour mixture leaving the steam jet ejector and its
fraction i.e. the amount of dry air in the mixture. The current values of
the mass flow rates of dry air, m
& a , and mixture, m
& m , averaged over

6c shows the final estimation of the power output losses resulting from
the excessive air-inleakage i.e. the function of the power loss versus
the air mass flow rate removed from the condenser [7].

Copyright 2000 by ASME

a)

Exact power losses N of the turbine unit due to the excessive air
inflow to the condenser can be evaluated basing on the method of
diagnosing thermal cycles and their components, used in the code
DIAGAR [11]. The code makes use of a verified numerical model of a
heat cycle, which is each time adjusted to the operation of an
individual power unit basing on its heat and flow measurements [9].
In the model, each component of the cycle is represented by its
geometrical data and characteristic operating curves. These curves can
be the following:
Experimentally derived coefficients, shown as functions of a
certain set of parameters, which may include the operating time,
Operating curves of particular apparatuses constituting the cycle,
shown as functions of selected thermal and flow quantities. In the
case of a condenser they can be, among other curves: terminal
temperature difference, condensation pressure and condensate
subcooling vs. condenser air content.
These characteristic data in a numerical model are used to evaluate
distributions of thermal and dynamic parameters in the cycle. This
information are applied to evaluate important quantities like:
Efficiency curves for all cycle (specific heat consumption, for
instance),
Power output for the entire cycle.
Then, basing on these data one can evaluate changes of characteristic
parameters describing the worsening of the power unit operation, and
distribute them over particular apparatuses within the cycle. For the
estimation of worsening of particular characteristic parameters it is
necessary to define the reference basis, with respect to which they are
estimated. In diagnostics, this basis bears the name of the reference
state and corresponds to the best possible geometrical characteristics
of particular unit components.
It is noteworthy that this reference state (power output and
specific heat consumption of the unit at reference conditions) can be
computed using the above described model. For the condenser
diagnostic analysis, the characteristics of its operation should be
known for an economically and technically justified minimum volume
of air at assumed conditions of load and unit environment. The actual
operating state of the unit (power output and specific heat
consumption) is determined on the basis of measurements, including
the air contents measurement in the condenser. The difference between
the actual and reference states is defined by quantities relevant from
the diagnostic point of view:
Power output loss, N,
Increase in specific heat consumption, q, resulting from the
measured air contents in the condenser.
Using this method in condenser diagnostics for a relatively long time
makes it possible to derive generalised time-relating changes of
operating curves, or other variables determining the operating
degradation of power apparatuses. It was also stressed in [8] that for
making the process of drawing conclusions more efficient - which is
essential for automatic expert systems, being the goal of the technical
diagnostics. Promising prospects are created by the use of neural
networks for diagnosing apparatuses, including condenser.

6
(To )

[K ] 5
4
3
2
1
0

10

12

14

16

0 1

18

Vvac

20

0.6

b)

0.5

0.4
0.3
0.2
0.1
9

m 10

2.5

c)

2.0

/N

[%] 1.5
1.0

0.5

01

m10

Figure 6. Influence of the air content on the condenser

performance:
a)The influence of the vacuum loss speed
V vac = p c / t [mm Hg/min] on the terminal
temperature difference [6],
b) Condensation pressure rise vs. air mass flow rate [7]
c) Decrease of turbine set output vs. air mass flow
rate [7].

The numerical results presented in Fig. 6c reveal a significant

influence of the excessive air amount in the condenser on the power
losses. Decreasing the air inleakage (by eliminating some leaks), for
instance from the value of m = 6 (often recorded in some power
stations) to m = 2.5 improves the output power by, approximately, 1%.

Copyright 2000 by ASME

code is now being implemented in Poland in power stations units of

200 MW output. It fulfils the requirements imposed for so called
scattered systems, which are considered the most promising for
controlling and diagnostics of complex technical units.
ACKNOWLEDGEMENTS

The authors are thankful to Dr. A. Krupa, Mr. J. Zembik and

Prof. A. Jaworek for their contributions to design and construction of
the monitoring system.
REFERENCES
1. Zhang C., Zhang Y., Sensivity analysis of heat transfer
coefficient correlations on the predictions of steam surface
condensers, Heat Transfer Engineering, vol.15, no.2, pp. 54-63,
1994.
2. Shklover G., Grigoriev V. G., Calculating the heat transfer
coefficient in steam condensers, Teploenergetica, vol.22, no.1
pp.67-71, 1975.
3. Butrymowicz D., Trela M., Heat exchange modelling in surface
heat exchangers during water vapour condensation (in Polish),
Internal IF-FM Report No. 458/98, Gdask 1996.
4. Coit R. L., A designers approach to surface condenser venting
and dearation, in Power Condenser Heat Transfer Technology
(ed. Marto P. J., Nunn R. H.) Mc Graw Hill Book Company,
pp.163-180, 1981.
5. ASME PTC.12.2-1996 Condenser Test Code (Draft) publication
ASME 1996.
6. Folwarczny C., Magiera E., The influence of air leaks in steam
turbine vacuum systems on the power station work, and the leak
location methodology (in Polish), Energetyka, nr 10, str. 3336,1985.
7. Trela M., Butrymowicz D,. Banasiewicz J., Monitoring system of
air removal from the condensers of turbogenerator units of 200
MW (in Polish), Proc. Conference of diagnostics in power
engineering stations, pp. 157-170, Solina 1998, Poland.
8. Guch J., Gardzilewicz A., The Analysis of Performance of the
Turbine Condenser with a Prognosis of Repair, Proc. of The
ASME International Joint Power Generation98, Vol. I pp. 235244, Baltimore (USA), August 1998.
9. Guch J., Verification of the Experimantal Coefficients of the
Turbomachinery Calculation Methods based on Measurements in
Power Stations, (in Polish) PhD. Thesis, Technical University of
Gdask, Gdask 1992.
10. Guch J., Krzyzanowski J., Searching of inefficient components
of complex power systems, ASME-TURBO-EXPO98, ASME
Paper No. 78-GT-98, Stockholm (Sweden), June 1998.
11. Gardzilewicz A., Guch J., Uzibo W., Bogulicz M.,
Jankowski T., DIAGAR numerical codes for diagnostics of
thermal cycles with propgnosis of apparatus repair, (in Polish),
Internal Report of Fluid-Flow Machinery Co. Ltd., Gdask, 1996.
12. Trela M., Butrymowicz D., Krupa A., Zembik J., Ihnatowicz E.,
Jaworek A., Monitoring system of air removal from the
condensers in the power station (in Polish), Internal Report of
Fluid-Flow Machinery Co. Ltd., No 8/98, Gdask, 1998.

Figure 7. Example of answer of the neural network for a

condenser at different power unit loads. The difference
between pattern and answer is indicated by the distance
between a white square and the related black dot. The
showed case concern detection of a number of plugged
tubes.

The neural networks can find two-fold application in the

condenser diagnosis:
As a tool for indicating power output loss or increase in specific
heat consumption, on the basis of measurements done in the
condenser, including the above discussed air content
measurement,
As a tool for detecting leakage in the condenser.
The first case is easier for carrying into effect and needs shorter
time for it. The neural network is trained basing on the air content
measurements and calculations of the cycle adjusted using the code
DIAGAR. This process needs a huge number of data for training [8].
Such a network will replace the above mentioned characteristic
operating curves. Its advantage is that it gives an immediate result,
without carrying out time consuming calculations. Then, these
networks reveal relatively high accuracy, as was shown for a turbine in
[10]. The disadvantage, however, especially in the preliminary stage of
use, may be a long time needed for collecting data for training. The
second application is much more difficult, as it needs the experience in
air content measurements.

CONCLUSIONS
The presented monitoring system of the air removed from the
condenser has been successfully implemented in the three power units
of 200 MW output in a power station in Poland. In the present version,
such system serves to the one power unit in which two steam jet
ejectors work simultaneously. There is no restriction, however, for
developing this system so as to operate with more steam jet ejectors.
The air monitoring system is of great applicability when the DIAGAR
diagnostic code is in use. The diagnostic methodology applied in this

Copyright 2000 by ASME