Professional Documents
Culture Documents
1
Q
B
W
1
W
2
1 Q
R
=Q
B
4
where Q
B
is the heat transfer administrated to
the _ m stream in the boiler, before state 1. For
simplicity, we recognize that in modern power
plants the ratio Q
R
/Q
B
is of order 1/10, and this
allows us to approximate Equation (4) as
ZQ
B
W
1
W
2
1
W
1
Q
B
_ _
5
where Q
B
is xed because state 1 is xed (P
H
, T
H
).
Equation (5) can be written further as
Z
Q
B
_ mc
P
T
H
Z
1
1
P
i
P
H
_ _
R=c
P
_ _ _
Z
2
1
P
L
P
i
_ _
R=c
P
_ __
1
_ mc
P
T
H
Q
B
Z
1
1
P
i
P
H
_ _
R=c
P
_ _ _ _
6
The expression shown above can be maximized
with respect to P
i
, by solving @Z/@P
i
50, and in the
limit _ mc
P
T
H
=Q
B
oo1 (which is consistent with
Q
R
ooQ
B
), the optimal IP pressure is
P
i
P
H
P
L
1=2
Z
2
Z
1
_ _
c
P
=2R
_
1
_ mc
P
T
H
Q
B
_
1
2
1
Z
2
Z
1
_ _
Z
2
Z
1
_ _
1=2
P
L
P
H
_ _
R=2c
P
__
7
Figure 4. Train of two turbines, high pressure and low pressure.
Y. S. KIM, S. LORENTE AND A. BEJAN 992
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er
Simpler forms in the limit represented by
Equation (7) are available for Z
2
=Z
1
1 (note:
Z
1,2
o1),
P
i
P
H
P
L
1=2
1
_ mc
P
T
H
Q
B
1
P
L
P
H
_ _
R=2c
P
_ _ _ _
8
which in the limit _ mc
P
T
H
=Q
B
0 reduces to
P
i
P
H
P
L
1=2
9
The analysis that follows is based on the simpler
version, Equation (9), in order to make the
demonstration entirely analytically. In reality,
the limit _ mc
P
T
H
=Q
B
0 is not reached. For
example, the boiler of a 500 MW power plant
receives _ m 473 kg s
1
and Q
B
510
6
kW, while
T
H
5813 K [13]. In this case, the dimensionless
group _ mc
P
T
H
=Q
B
is equal to 1.26. However, its
effect on Equation (8) is weak because the quantity
shown between fg is essentially constant and of
order 1.
The isentropic efciencies increase mono-
tonically with the sizes of two turbines (M
1
, M
2
),
cf. Figure 3. The Z(M) curves must be concave
because they both approach Z 51 in the limit
M-N. It is reasonable to assume that near Zp1
the Z(M) data for turbine designs are curve
tted by
Z
1
1 a
1
e
b
1
M
1
10
Z
2
1 a
2
e
b
2
M
2
11
where (a, b)
1,2
are four empirical constants that
depend on the pressure level of the turbine. The
total power produced by the two turbines is
W Z
1
W
1;rev
Z
2
W
2;rev
, where the functions
Z
1
M
1
and Z
2
M
2
are known. The total mass of
the ensemble is xed,
M M
1
M
2
12
There is one degree-of-freedom in the making
of Figure 4, namely, the dividing of M into M
1
and M
2
. The optimal way to divide M is
determined using the method of Lagrange
multipliers. We form the aggregate function
F W lM 13
for which W and M are the expressions (1, 2)
and (12). We solve the system @F=@M
1
0 and
@F=@M
2
0, eliminate the multiplier l, and
obtain
b
1
M
1
b
2
M
2
ln
a
1
b
1
W
1;rev
a
2
b
2
W
2;rev
_ _
14
Equations (14) and (12) pinpoint the mass
allocation fractions M
1
/M and M
2
/M. The rst
and most important conclusion is that there must
be a balance between M
1
and M
2
.
For example, if we use the HP and IP data
of Figure 3 for Z
1
of Equation (10), we obtain
approximately a
1
0:2 and b
1
4:6 10
5
kg
1
.
If we use the LP data of Figure 3 in conjunction
with Equation (11), we estimate that a
2
0:13
and b
2
2:1 10
5
kg
1
. If, in addition,
Equation (9) holds, then W
1,rev
5W
2,rev
and this
means that on the right hand side of Equation (14)
we have a
1
b
1
W
1,rev
/(a
2
b
2
W
2,rev
)34, which is
considerably greater than 1. In conclusion,
Equation (14) states that b
1
M
1
is greater than
b
2
M
2
, namely, b
1
M
1
b
2
M
2
13.5.
Combining Equations (12) and (14) yields the
optimal mass distribution equations for HP and
LP turbines
M
1
0:044 M 7:3 10
4
kg 15
M
2
0:956 M 7:3 10
4
kg 16
M
1
equals M
2
when M51.6 10
5
kg. In
conclusion, there is an optimal way to distribute
mass along the train of turbines. According to
Equations (15) and (16), the total mass is allocated
in a balanced way; with more mass at the HP end
when total mass M is small, and with more mass at
the LP end when the total mass is large. What we
showed here for a group of two turbines also holds
for groups of three or more.
The penalty associated with not using a train of
turbines can be calculated with reference to
Figure 4. We form the ratio W(M
2
50)/W
max
,
where W
max
is the W maximum that corresponds
to the constructal design, Equation (9). The ratio is
WM
2
0
W
max
1
2
1
P
L
P
H
_ _
R=2c
P
_ _
o1 17
This ratio is always smaller than 1. In conclusion,
the penalty is more signicant when the overall
pressure ratio is greater.
DISTRIBUTION OF SIZE IN STEAM TURBINE POWER PLANTS 993
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er
3. MULTIPLE TURBINES
Consider the limit where the number of turbines is
extremely large (i 51,2,y,N) and their individual
sizes are small, Figure 5. Each turbine and
subsequent reheater can be modeled as an
isothermal expander receiving the heat transfer
rate Q
i
and the steam _ mT
H
; P
i
DP
i
, and
delivering the power W
i
and the steam _ mT
H
; P
i
.
With reference to the elemental system (i) shown in
Figure 5, the rst law requires
W
i
Q
i
_ mh
i1
h
i
18
Because the steam is modeled as an ideal gas,
the enthalpy of the inow is the same as
the enthalpy of the outow, and the rst law
reduces to
W
i
Q
i
19
If the isothermal expander (i) operates
reversibly, then
Q
i;rev
_ mT
H
s
out
s
in
i
_ mT
H
R
DP
i
P
i
20
and W
i;rev
Q
i;rev
. The actual system operates
irreversibly with the efciency
Z
i
W
i
W
i;rev
o1 21
and its power output is
W
i
_ mT
H
RZ
i
DP
i
P
i
22
where DP
i
P
i1
P
i
. Of interest is the total
power delivery
W
T
_ mT
H
R
N
i1
Z
i
P
i1
P
i
P
i
23
which in combination with the boiler heat
input (Q
B
, xed) and the total heat input to
the isothermal train (Q5W), yields the efciency
ratio
Z
W
Q
B
W
24
In order to maximize Z we must maximize total
power W
T
given by Equation (23). Attractive are
larges value of (DP/P)
i
and Z
i
, in each stage of
isothermal expansion. The data of Figure 3
suggest that, in general, Z
i
depends on both M
i
and P
i
, such that
@Z
i
@M
i
40 and
@Z
i
@P
i
40 25
The masses of the N turbines are constrained
by the total mass of the turbine train, which
Figure 5. Train of N turbines.
Y. S. KIM, S. LORENTE AND A. BEJAN 994
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er
is xed
M
T
N
i1
M
i
26
Another constraint is the DP
i
s must all up to the
overall pressure difference, DP
T
5P
H
P
L
:
DP
T
N
i1
P
i1
P
i
27
In summary, we must maximize the sum (23)
subject to the constraints (26) and (27). According
to Lagranges method of undetermined
coefcients, this problem is equivalent to seeking
the extremum of the function
F
N
i1
Z
i
P
i1
P
i
1
_ _
lM
i
mP
i1
P
i
_ _
28
where Z
i
5Z
i
(M
i
, P
i
), and l and m are two
Lagrange multipliers. The function F is a linear
combination of the sums (23), (26) and (27). It
depends on 2N12 variables, namely, l, m,
M
1
,y,M
N
and P
1
,y,P
N
. Its extremum is found
by solving the system of 2N equations
@F
@M
i
@Z
i
@M
i
P
i1
P
i
1
_ _
l 0
i 1; 2; . . . ; N 29
@F
@P
i
@Z
i
@P
i
P
i1
P
i
1
_ _
Z
i
P
i1
P
2
i
m 0
i 1; 2; . . . ; N 30
This system establishes the N masses (M
i
) and the
N IPs (P
i
) as functions of the undetermined
coefcients l and m. In principle, these
coefcients can be determined by substituting the
solutions for M
i
(l, m) and P
i
(l, m) into the
constraints (26) and (27).
Here, we make analytical progress on a simpler
path by linearizing the function Z
i
(M
i
) in the
vicinity of the design range in which all the
turbines are expected to operate (namely,
near Z
t
1 in Figure 3). We write that for all
N turbines the efciencymass relationship is
unique,
Z
i
a bM
i
cP
i
31
where (a, b, c) are constants. The rst result of the
linearization is that the N Equations (29) reduce to
P
i1
P
i
constant 32
In view of Equation (27), we conclude that the
total pressure interval DP
T
is divided into
N pressure intervals such that
P
i1
P
i
P
H
P
L
_ _
1=N
33
or
DP
i
P
i
P
H
P
L
_ _
1=N
1 34
This is in agreement with what we found earlier in
Equation (9).
The distribution of M
T
among the N turbines
follows from Equation (30), in which we substitute
Equations (31) and (32). The result is
Z
i
P
i
P
L
P
H
_ _
1=N
m c
P
H
P
L
_ _
1=N
1
_ _ _ _
constant 35
This shows that in the (M
i
, P
i
) range where
Equation (31) is valid, Equation (35) becomes
a bM
i
cP
i
P
i
constant 36
If the effect of P
i
on Z
i
is negligible, as in the
case of the HP and IP data of Figure 3, then
Equation (36) reduces to
a bM
i
P
i
constant 37
The mass of the individual turbine should increase
linearly with the pressure level of that turbine.
In particular, if the linear approximation (31)
reveals that a/booM
i
, as in the calculations shown
under Equation (14), then Equation (37) states
that the M
i
s must be distributed in proportion
with the P
i
s, and, in view of Equation (34), in
proportion with the DP
i
s:
M
i
P
i
DP
i
38
Equation (38) indicates that more mass should be
placed in the expanders at higher pressures.
DISTRIBUTION OF SIZE IN STEAM TURBINE POWER PLANTS 995
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er
4. RESULTS AND DISCUSSION
These is a considerable volume of work on turbine
design, where the size is selected from other
considerations such as the distribution of maximal
stresses due to centrifugal forces, and the aring
out of the ow passages to accommodate the
expansion of the steam along the turbine train [15].
The work reported in the present paper suggests
that the design of future concepts of turbine train
conguration must combine the traditional con-
siderations [16,17] with size allocation principle
illustrated in this paper.
This paper also highlights the need for more
extensive and more exact information on how the
size of each turbine affects its thermodynamic
performance. The data that we used (Figure 3) are
few and provide a narrow view of the size effect
that is needed for future design. These data also
require an understanding of how the multiple
turbines are arranged in the power plant.
As the power generation capacity of the plant
increases, the boiler and the turbine also increase
in size. It is easer to increase the size of the boiler
than the size of the turbine. Today, in a 1000 MW
power plant there is still a single boiler, while the
number of turbines is ve or six. Each pressure
stage employs one or several (two or three)
turbines in parallel. For example, in power plants
with more than 500 MW capacity, the LP stage
consistently employs two or more turbines.
Several physical limitations require the use of
multiple turbines in parallel at a single pressure
stage. A major limitation posed by centrifugal
stresses is the length of the last blade. The current
maximum length is approximately 1.3 m. The data
of Figure 3 come from a current design for a
1000 MW power plant with an LP stage consisting
of four turbines in parallel, each turbine with last
blades 1.14 m long. The ordinate of Figure 3
indicates the efciency of the single LP turbine,
while the abscissa represents the total mass
employed for each pressure stage. If the mass of
the LP turbines is divided by 4, then the LP data of
Figure 3 move closer to the HP and IP data.
Note also that the efciency of one turbine is
affected by the operation of the turbine upstream
of it. The irreversibility of a turbine is due to six
losses: deviation from the ideal velocity ratio,
rotational loss, diaphragm-packing leakage loss,
nozzle end loss, moisture and supersaturating loss
and exhaust loss [15]. The last two losses are
present only in the LP turbines, not upstream,
therefore the efciency of the single LP turbine
should be lower than the efciency of HP and IP
turbines, cf. Figure 3. Furthermore, because the
HP turbine is installed at the head of the train, its
efciency should be lower than the IP turbine
efciency because of entrance losses due to the
conguration of the steam passage. This too is
conrmed by the data of Figure 3.
The optimal IP Equations (7)(9) may not be
attainable in a design with two turbines model
because of the properties of water. For example,
the IP in a 500 MW power plant is approximately
4 MPa, while the IP based on Equation (9) is
0.35 MPa. The discrepancy between the IPs is due
to the ideal gas model used for steam. If we model
as isentropic the expansion through the HP
turbine, because of the saturated steam curve of
water the attainable IP pressure is 3.15 MPa. The
expansion cannot proceed beyond 3.15 MPa
because of engineering limitations such as water
droplet impingement on the blades. This limitation
impacts on the mass distribution of turbines for
two turbines model. When the number of pressure
stages increase to three or more, the expansion
limitation due to steam properties diminishes in
importance.
5. CONCLUSION
The main conclusion is that the total turbine mass
must be divided in a certain way when two or more
turbines are used in series. The allocating of mass
is synonymous with the discovery of the cong-
uration of the turbine train. Important is that in
this article the distribution of size along the
turbine train came from the pursuit of global
thermodynamic performance. The allocation of
mass is driven by the size effect on turbine
efciency: larger turbines operate closer to the
reversible limit (e.g. Figure 3).
Future extensions of this work should take into
account the physical limitations that make
Y. S. KIM, S. LORENTE AND A. BEJAN 996
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er
necessary the use of several turbines in parallel at a
pressure stage. Such work could address the
distribution of turbine mass per pressure stage,
instead of per turbine. This work would be assisted
further by the availability of more extensive data
of the type sampled in Figure 3. To that end, it
would be useful to construct models that account
for the trend exhibited by the data. In other words,
the analytical form of Z(M, P), which was
approximated here in Equations (10) and (31),
should be derived from a model that accounts for
irreversibility and nite size.
NOMENCLATURE
a
1,2
5constants
b 5constant
b
1,2
5constants
c 5constant
c
p
5specic heat at constant pressure
(J kg
1
K
1
)
F 5aggregate function, Equation (28)
h 5enthalpy (J kg
1
)
HP 5high pressure
IP 5intermediate pressure
LP 5low pressure
_ m 5mass ow rate (kg s
1
)
M 5mass (kg)
N 5number of turbines
P 5pressure (Pa)
Q 5heat transfer rate (W)
Q
B
5heat transfer rate to the
boiler (W)
Q
R
5heat transfer rate to the
reheater (W)
R 5ideal gas constant (J kg
1
K
1
)
s 5entropy (J kg
1
K
1
)
T
H
5highest allowable temperature
(K)
W 5power (W)
Greek letters
DP 5pressure drop (Pa)
Z 5turbine isentropic efciency
l, m 5Lagrange multipliers
F 5auxiliary function, Equation (13)
Subscripts
H 5high
i 5ith turbine
i 5intermediate
in 5inlet
L 5low
max 5maximum
out 5outlet
rev 5reversible
T 5total
1 5inlet of the rst turbine
1
0
5outlet of the rst turbine
2 5inlet of the second turbine
2
0
5outlet of the second turbine
ACKNOWLEDGEMENTS
This research was supported by Doosan Heavy
Industries & Construction Co., Ltd., Changwon, South
Korea.
REFERENCES
1. Dincer I, Al-Muslim H. Thermodynamic analysis of reheat
cycle steam power plants. International Journal of Energy
Research 2001; 25:727739.
2. Ran J, Yang L, Zhang L. A method of economical load
dispatch for steam turbine unit of thermal power plant.
International Journal of Energy Research 2008; 32:752764.
3. Farhad S, Saffar-Avval M, Younessi-Sinaki M. Efcient
design of feedwater heaters network in steam power plants
using pinch technology and exergy analysis. International
Journal of Energy Research 2008; 32:111.
4. Mago PJ, Srinivasan KK, Chamra LM, Somayaji C. An
examination of exergy destruction in organic Rankine
cycles. International Journal of Energy Research 2008;
32:926938.
5. Yoon J. Doosan Heavy Industries, Private communication,
June 2008.
6. Ghoniem AF. Energy and the entropy challenge, meeting
the entropy challenge. AIP Conference Proceedings, New
York, 2008.
7. EG&G Technical Services. Fuel Cell Handbook (7th edn).
U.S. DOE, Ofce of Fossil Energy, NETL: Morgantown,
West Virginia, 2004; 891.
8. Singer JG (ed.). Combustion Engineering: A Reference Book
on Fuel Burning and Steam Generation (revised edn).
Combustion Engineering, Inc.: New York, 1991.
9. United States Environmental Protection Agency. Light-
duty automotive technology and fuel economy trends: 1975
through 2008. EPA420-S-08-003, United States Environ-
mental Protection Agency, 2008.
DISTRIBUTION OF SIZE IN STEAM TURBINE POWER PLANTS 997
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er
10. Bejan A. Advanced Engineering Thermodynamics (3rd edn).
Wiley: Hoboken, 2006; 534537.
11. Strobridge TR. Cryogenic refrigerators: an updated
survey. NBS Technical Note (U.S.), No. 655, June
1974.
12. Bejan A, Lorente S. Thermodynamic optimization of
ow architecture: dendritic structures and optimal sizes
of components. ASME Paper IMECE 200233158,
presented at the International Mechanical Engineering
Congress and Exposition, New Orleans, LA, November
1722, 2002.
13. Provided by Doosan Heavy Industries & Construction Co.,
Ltd., June 2008.
14. Bejan A, Lorente S. Design with Constructal Theory, section
11.3. Wiley: Hoboken, NJ, 2008.
15. Salisbury JK. Steam Turbines and Their Cycles. Wiley:
New York, 1950.
16. Sanders WP. Turbine Steam Path. Pennwell: Tulsa, 2004.
17. Day IJ, Hodson HP, Longley JP, Denton JD, Gallimore S,
Came P, Robinson C, Young JB. Cambridge Turbomachin-
ery Course, Whittle Laboratory, Department of Engineer-
ing, University of Cambridge, 2529 June 2001.
Y. S. KIM, S. LORENTE AND A. BEJAN 998
Copyright r 2009 John Wiley & Sons, Ltd. Int. J. Energy Res. 2009; 33:989998
DOI: 10.1002/er