Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ENRICO SCIUBBA
University of Roma 1 ‘‘La Sapienza’’
Rome, Italy
Encyclopedia of Energy, Volume 6. r 2004 Elsevier Inc. All rights reserved. 231
232 Turbines, Steam
design procedures to grant separate treatment of gas rotational momentum. At rotor exit (station 2), the
and steam turbines. steam is at a lower pressure and energy level than at
station 0 and is discharged through the outlet scroll
or diffuser channel. Station ‘‘out’’ is the outlet section
of the stage. The purpose of the inlet channel is to
1. SOME BASIC CONCEPTS guide the flow from the turbine inlet port to the
statoric vanes with the minimum possible loss in
1.1 The Flow through a Turbine total pressure. If the first stage after the admission
Figure 1 shows a schematic cross section of an axial throttle is axial, the inlet is usually shaped as a
(Fig. 1A) stage and a radial stage (Fig. 1B) of a pseudo-conical transition piece between the inlet
modern steam turbine. High-pressure steam enters duct and the (larger in diameter) stator. In this case,
the machine at station ‘‘in’’ (throttle valve exit no flow deviation is imposed and only a modest
section), passes through the inlet channel, and acceleration is attained. For intermediate axial
reaches the stator inlet (station ‘‘0’’). The stator stages, the inlet is missing and the flow from the exit
accelerates the flow by expanding it by a fraction of the previous stage flows directly onto the stator of
bst (1Zbst40) of the total stage pressure drop and the following stage. Radial turbines can be centripe-
introduces a tangential velocity component (inlet tal (flow inward) or centrifugal (flow outward).
swirl). The steam jets exiting the stator (station ‘‘1’’) Because the only significant centrifugal application
impinge on the rotoric blades that deviate the flow so is the Ljüngstrm turbine, shown in Fig. 2 and
as to decrease this inlet swirl and, thus, absorb discussed at the end of this section, ‘‘radial turbine’’
A
1 2 3 α1
Tip C1
Cα1
Annulus
area A N R
Cw 2
Root
U
Nozzle blades V2 Cα 2
α2 β2
C2
rr 3 rt3
Cw3
U
Rotor blades
Cα 3 C3 β3
V3
α3
R¹
FD
R²
FIGURE 1 (A) Axial turbine stage. (B) Radial turbine stage. Notice the two-stage rotor (R1 and R2) and the flow deviator
(FD).
Turbines, Steam 233
A Exhaust to B
condenser
Incoming steam
Concentric
labyrinth rings (to
reduce leakage)
FIGURE 2 The Ljüngstrm radial turbine. (A) Meridional section through turbine. (B) Blading arrangement and directions
of rotation.
A B V0
V1
W1
ωr 1
V2 W2
ωr 2
r1 r2 r 1
r2
Center of rotation
+ +
Closed
FIGURE 3 (A) Axial stage geometry. (B) Radial stage geometry. (C) Variable geometry axial stator. (D) Variable geometry
radial stator.
Figs 1A and 1B). If the area increase is obtained by and provides it to the shaft, and is a stator in the sense
increasing the average radius, there is the additional that it introduces a (relative) tangential velocity com-
advantage that the exit swirl is recovered as well. ponent for the next blade row. Ljüngstrm turbines are
Ljüngstrm turbines (Fig. 2), are atypical in more used when axial size constraints are very stringent
than one sense. The steam enters the machine radially because their power density is higher than that of other
from slots in the casing or in the shaft and proceeds types of turbines. However, their use is restricted to
radially outward. The wing-like blades are inserted in lower mass flow rates due to blade length limitations.
two counterrotating disks, with the arrangement being
such that each blade row rotates between two rows 1.1.1 Partial Admission
attached to the opposite disk. Each blade row is a If the throttle pressure is very high, the incoming
rotor in the sense that it extracts energy from the flow steam into the first stage has a very high density;
Turbines, Steam 235
1 Nozzle
2 Rotor
therefore, even for large flow rates, the required much older. In the first century bc, Hero of
passage area may be relatively small. Because the Alexandria, in his treaty ‘‘De Re Pneumatica’’
external diameter (both in axial and radial stages) is (literally ‘‘On Fluid Flow’’), described what in
selected on the basis of the best attainable efficiency modern terms would be defined as a ‘‘single-stage,
and provides an optimal ds, this may lead to pure reaction turbine’’ (Fig. 6). Hero, who seemed to
extremely low blade heights, which increase the imply that the ‘‘turbine’’ had been actually built prior
end wall losses and decrease the practical stage to his description of it, provided no indication about
efficiency. To uncouple (at least to a certain extent) practical uses for this device, although other sources
the blade height from the external diameter, the indicate that similar machines were used to power
statoric nozzles of the first stage of class 4 and 5 the opening mechanisms of the heavy bronze doors
turbines are so constructed as to limit the steam of some Egyptian or Babylonian temples. (This
injection angle to a value ao3601 (Fig. 4). Thus, only indicates that the turbine provided B0.5 kW with a
the statoric channels within this angle a will be steam mass flow rate of B0.01 kg/s, which seems
active, and for a given external diameter, the blade unfeasible for those times because it implies a nozzle
height can be increased of 1/a without efficiency diameter of B15 cm.) Be that as it may, the
penalties. This configuration is called partial admis- technological level of those times clearly had no use
sion and can be introduced only in pure impulse for such machines, and for centuries hydraulic and
stages, where the steam in the rotoric passages is not wind turbines were the only turbomachines em-
subject to a streamwise Dp (otherwise, massive ployed in power generation. Much later, there were
leakage into the inactive adjacent channels would isolated attempts to convert the motive energy of
destroy the regularity of the flow). steam (now called stagnation exergy) to mechanical
shaft power. Giovanni Branca’s machine (in Italy in
1629 [Fig. 7A]) was actually inspired by an earlier
1.2 Brief Historical Background
illustration (from 1551) contained in a sort of
The word ‘‘turbine’’ was first proposed in its modern technological almanac compiled by the Arab histor-
usage by the French Claude Burdin in 1820, with ian Taqi-al-Din. John Dumbell’s proposal (in Eng-
reference to hydraulic machines. Although the first land in 1808 [Fig. 7B]), although in essence based on
steam turbines in commercial service were those the same principle, was technologically much more
installed in the northern United States by William advanced. However, neither machine ever produced
Avery in 1831 to power some sawmills (Fig. 5), the any net power. In 1848, the Scot Robert Wilson
idea of exploiting the ‘‘motive power’’ of live steam is patented a series of steam turbines, one of which
236 Turbines, Steam
Most of the so-called cogeneration units in use by the class 5 turbines, the size of the exhaust end limits the
industry are of the noncondensing (or NC) type. acceptable steam volumetric flow through the final
For the same rated power output and inlet stages because at the usual rotational regimes (1500/
conditions, NC turbines are much more compact 1800 or 3000/3600 rpm) ‘‘long’’ blades (Z1 m) are
than C turbines because the steam at the former’s subject to excessive centrifugal stresses. Therefore,
outlet has a much lower specific volume. For large the largest units (especially in nuclear plants) are
built in a double-flow configuration where there are
two low-pressure turbine bodies for each medium-
TABLE I
pressure one (as shown in Fig. 1A).
Typical Values for Steam Admission Conditions in Modern
Turbines
1.3.3 Extraction Turbines
Throttle
For the overall economy of the steam turbine cycle, it
steam Throttle steam Net power
Type of pressure temperature putput per is often convenient to extract a portion of the steam
application (bars) (1C) unit (MW) evolving in the turbine before it reaches the turbine
exhaust. This involves an extraction turbine. The
Geothermal 5–15 150–250 0.5–10 extraction can be automatic or controlled. In both
Nuclear 45–70 200–300 50–250 cases, steam is extracted through openings properly
Industrial 40–60 400–450 1–50 located on the turbine casing between one stage and
cogeneration
the other. However, in the automatic extraction case
Electrical 120–180 510–550 50–250
power
(Fig. 13A), there is no means of controlling the
generation pressure of the extracted stream, whereas in the
Supercritical 4 220 550–600 50–100 controlled extraction case (Fig. 13B), a system of
throttling valves can modify the pressure drop
TABLE II
Some General Classification Criteria for Steam Turbines
Inlet steam Subcritical: pino220 bars and Tin usually o5501C Design and Process Engineering
Conditions
Supercritical: pin4220 bars and Tin usually 45501C Industry
Exhaust pressure ‘‘C’’, condensing if pexhaustopatmospheric and the exhaust steam is condensed Design and Process Engineering
‘‘NC’’, noncondensing otherwise (also called back-pressure)
Main steam flow Axial if the main steam flow path is parallel to the turbine axis Design
path
Radial otherwise
Unit power Class 1: 15–400 kW, noncondensing Industry
Class 2: 450–1500 kW
Class 3: 2000–15,000 kW
Class 4: 15,000–100,000 kW
Class 5: 4 100,000 kW, reheat and condensing
Steam extraction Extraction if a portion of the steam is extracted between stages for process Design and Process Engineering
purposes other than power generation
Non-extraction otherwise Industry
Inlet steam High pressure, above 40 bars Design
pressure
Intermediate pressure, 5–40 bars Industry
Low pressure, o 5 bars
Blade design Impulse if the stage pressure drop takes place only in the nozzles Research
Reaction otherwise Design
Turbines, Steam 239
4000
Thermodynamic Properties of
Water 3900
3800
96 00
10
0
0 3700
92
8 80 3600
0
84
800 3500
760
3400
0.04
100
g
m /k
720
v=1 3
50
0.4
0.1
20
3300
10
680
10
2
1
0.5
640 3200
a
0.2
MP
40
h (kJ/kg)
600
0.1
3100
100
5
0.0
P=
2
560
0.0
1
3000
0.0
05
520
0.0
02
T= 480 K 2900
0.0
01
0.0
440 2800
400
0.9
8
2700
0.9 360
6
0.9 2600
4 320
0.9
2
0.9
0.8 0 2500
ne
or li
8
0.8
vap
6
.
2400
atm
0.8
0.8 4
ted
~1
0. 2
ura
80 2300
a
Sat
MP
0.
78 6
0.1
0.7 74
2200
P=
0. .72
0
2100
0.7
id
0.62
0.66
0.54
0.58
0
Liqu
2000
4 5 6 7 8 9 10
s (kJ/[kg.K])
FIGURE 11 The Mollier steam chart. From Stanford University, Department of Mechanical Engineering.
between the extraction point and final use, thereby the fluid with the statoric and rotoric blades is very
actively regulating the flow rate of the extracted short (typically 0.1–1.0 ms), so that the amount of
steam. Extractions are commonly used to preheat the energy exchanged between the fluid and blade under
feed water in modern steam cycle-based power plants the form of heat is entirely negligible with respect to
(class 5 C-type turbines) and to supply process steam that exchanged under the form of work. In the ideal
in industrial plants (classes 3, 4, and 5, usually case (no internal friction), the expansion is also
C-type turbines). isentropic (Fig. 14). At the inlet, the steam possesses a
static enthalpy h0(p0,T0) and an absolute velocity V0,
so that its total (stagnation) enthalpy content is
2. ENERGY TRANSFER h0,tot ¼ h0 þ 0.5V20. At stage exit, the static enthalpy
IN A TURBINE has dropped to h2i(p2i,T2i) and the velocity is V2i, so
that the stagnation enthalpy is now h2i,tot ¼
2.1 Expansion of a Fluid through a h2i þ 0.5 V22i. The maximum specific energy that can
be extracted from the fluid under these conditions is
Turbine: Stage Efficiency
called the isentropic (or adiabatic) work:
The flow through a turbine stage is usually modeled
as an adiabatic expansion because the contact time of wad ¼ Dhad ¼ h0;tot h2i;tot ð1Þ
240 Turbines, Steam
Throttling
T1
Steam inlet
P4
P1
P3
P2
tr 1
ex
P
Enthalpy (kJ/kg)
tr 2
ex
P
Condensate outlet
tr 3
expansion line
ex
P
A
r
ex
P
Turbine
discharge
and
2.3 From Zeroth-Order Approximation to
V22 ¼ V2i
2
ð 1 aÞ where ao1:
Three-Dimensional Design
The following is obtained:
The thermodynamic stage efficiency constitutes a
T0 Ds useful global description of the expansion process.
Zis;tt ¼1 ð5Þ
h0;tot h2i;tot However, the design of a turbine must be based on a
detailed knowledge of the actual coupling between
1 the working fluid and the machine channels. The
Zis;ts ¼ Ztt : ð6Þ
V 2 ð 1 aÞ ‘‘real’’ flow is fully three-dimensional, turbulent, and
1 þ Ztt 2i
2 h0;tot h2i;tot unsteady, and this makes a complete description
242 Turbines, Steam
extremely difficult. As in many engineering tasks, we ratios between corresponding forces (or force-related
proceed by successive approximations: quantities) are the same (i.e., if Dpstator,A/Dpsta-
tor,B ¼ scentrifugal,tip,A/scentrifugal,tip,B). Notice that si-
1. First, we consider the so-called zero-dimen-
milarity considers ideal flow concepts only. Strictly
sional problem in which the flow is steady and the
speaking, two geometrically and kinematically simi-
fluid–machine interaction is expressed by global
lar machines are dynamically similar only if the
parameters such as mass (or volume) flow rate,
dissipative features of the flow (leakages, boundary
expansion ratio, average values for the value and
layers, and viscous and inviscid losses) are the same.
direction of velocity at the inlet and outlet of the
With specific reference to a turbine stage, let us
blades, angular velocity of the rotor, and average
consider the following question: can the global design
physical properties of the fluid. We derive, via
parameters of the stage be reduced to a small set of
dimensional analysis, a set of general relations that
fundamental dimensional variables that can be used
constitute a basic design tool.
to construct an even smaller set of dimensionless
2. We then turn our attention to the ideal, steady,
indexes that describe stage performance? A brief
quasi-one-dimensional flow, devoid of boundary
reflection indicates that a stage is described by the
layers on either moving or fixed boundaries, and
mass flow rate it elaborates, the work it extracts, the
derive first-order expressions for the most important
conditions of the fluid and the flow passage area at
characteristic quantities. The flow rate and the
inlet and outlet, a proper measure of its size, and the
specific work are expressed in terms of average
angular velocity of its rotor. A possible set of relevant
velocities, and the stage performance is linked
design parameters is listed in Table III. A fundamental
explicitly to the average flow field parameters.
set of dimensionless parameters that can be con-
3. The effects of boundary layers and other
structed with these quantities is reported in Table IV.
viscous and compressibility phenomena are then
Turbines with equal Reynolds numbers, specific
taken into account, and the results previously
speeds, and specific diameters are dynamically
obtained are corrected to include the effects of
similar (because the losses scale, albeit only approxi-
losses, secondary flows, flow nonuniformities, sonic
mately, with Reynolds numbers). This result, which
phenomena, and so on.
can be derived theoretically and justified on the basis
4. The fully unsteady, three-dimensional flow is
of first- and second-order considerations, has been
studied. As of today, this is possible only under
confirmed by an extraordinarily large and compre-
stringent simplifying assumptions and at the expense
hensive series of experimental results. It has several
of substantial computational resources. Usually,
important consequences. First, the efficiencies of the
cascade or stage simulations are performed to
two machines are also equal. Second, if we know the
investigate particular characteristics of the flow field
performance (Q2,A, oA, and Dhad,A) and the scale
and are validated by comparing their results with
(D2,A) of the first machine, we can derive the
experimental investigations.
corresponding quantities for the second machine
once we fix either its dimensions (D2,B) or one of its
2.3.1 Zeroth-Order Approximation: design specifications (Q2,B or Dhad,B, or oB). The
Dimensional Analysis
Dimensional analysis is a branch of applied mathe-
TABLE III
matics that investigates the relationship between
A Fundamental Set of Design Parameters for a Turbine
global characteristics of a physical phenomenon and
a possibly small set of likewise global parameters Physical phenomena to which
that uniquely define it. This article adopts here a Parameter it relates
simple (but theoretically sound) approach. In turbo-
machinery design, dimensional analysis is essentially Fluid enthalpy drop (Dhtot) Specific work
linked to the concept of similarity. Two machines, A Fluid flow rate (mo) Machine size, power
and B, are said to be geometrically similar if the Fluid density (r(r,T)) Machine size, losses
ratios between homologous dimensions all are equal Rotor and stator dimensions Machine size, mechanical
stresses, flow kinematics
(i.e., if D2,A/D2,B ¼ ctip,A/ctip,B). The same two
Stage efficiency (Zis) Irreversible dissipation in
machines are also said to be kinematically similar if
blade rows
the ratios between corresponding velocities are the
Rotor angular velocity (o) Mechanical stresses, flow
same (i.e., if U2,tip,A/U2,tip,B ¼ V1,A/V1,B). Finally, two kinematics
machines are said to be dynamically similar if the
Turbines, Steam 243
TABLE IV 0.006
1000
1100
1200
1300
200
300
400
500
600
700
800
900
Specific speed o Q2
ns ¼ 3=4
ðDhad Þ
Specific diameter ds ¼
Dtip ðDhad Þ1=4
pffiffiffiffiffi Temperature (°C)
Q2
Flow coefficient j ¼ UQ
tip A
(A is local flow
passage area) losses. For two turbines with the same ns and ds (or
Stage efficiency Zis ¼ wreal
wad
Dc and j, which is the same), a widely used formula,
Degree of reaction h1 h2 originally from Staufer in 1925, states that
Rr ¼ h0;tot h2;tot
1 Zis;tt;A ReB n
Tip clearance/blade height h
gap
s ¼ hblade (h is average height or ¼ C1 þ C2 ; ð15Þ
ratio 1 Zis;tt;B ReA
clearance)
Rotor blade aspect ratio a ¼ hcblade
blade
(h and c are average where C1E0.30.4, C2E0.70.6, and nE
representative values) 0.20.25. Equation (15) implies that not all losses
are of viscous origin (C1 is the percentage of inviscid
losses on the total).
following relations hold: On the basis of similarity analysis, universal
oA D2;A oB D2;B performance charts can be compiled. If we have
1=2
¼ 1=2
ð10Þ one verified performance map of a given turbine
Dhad;A Dhad;B
stage, similarity theory states (and experimental
oA Q2;A oB Q2;B results confirm) that such a map applies to all stages
¼ ð11Þ with the same pair (ns,ds)—or (c,j), which is the
DA DhA DB DhB
same—so long as their operation implies Reynolds
oA D22;A ¼ oB D22;B ; ð12Þ numbers on the order of the one in the test. The
admissible range is somewhat application dependent,
where Eq. (10) derives from the Reynolds numbers
but for turbines of commercial standard, a variation
equality, assuming that the steam viscosity at the exit
between Re ¼ 105 and Re ¼ 107 leads to a total-to-
of the two machines is the same (see Fig. 15 for the
total efficiency change of approximately 0.05 (from
dependence of m on temperature).
0.87 to 0.92). Two widely used performance maps,
As can be seen in the following, the flow and head
known as the Balje and Smith charts, are shown in
coefficient are also fundamental design parameters.
Figs. 16 and 17, respectively.
They are related to ns and ds by
In spite of the formal identity, there is a substantial
1 practical difference in taking the sets (Re, ns, ds) or
Dc ¼ ð13Þ
ðns ds Þ2 (Re, Dc, j) to represent the variation of the stage
efficiency at different design conditions. By defini-
ns Dc
j ¼ Kshape ; ð14Þ tion, ns/ds maps are dependent on the stage outlet
ds temperature and pressure and, therefore, do not
pdb ð1w Þ
2 b
where Kshape is equal to 2 and 2pdb D for apply strictly to similar machines working under
axial and radial turbines, respectively. The Reynolds different operating conditions. To eliminate this
number definition is clearly conventional. In most inconvenience, ‘‘corrected’’ quantities are sometime
real cases, nowhere in the real flow field does the used, especially in industrial references. If for a
Reynolds number assume the value given by the turbine stage mr is defined as the reference mass flow
formula in Table IV. Nevertheless, if used consis- rate evolving in the stage when the inlet steam is at Tr
tently, it allows a direct relative estimate of global and pr and the rotor is rotating at a reference angular
244 Turbines, Steam
Choked
flow
4
Expansion ratio
η = 90%
(1/πt = Pt5/Pt4 )
t
3 86%
Operating
line
83%
2 80%
1
50 60 70 80 90 100 110 120
Percentage design corrected mass flow times corrected rpm
FIGURE 18 Corrected turbine performance map.
2 2
Vx 2
¼ Uj jj þ 1 cj ð25Þ
_
V2
Wj;t ¼ Vj;t Uj ð26Þ
Vθ2
Vj;x ¼ Wj;x ð27Þ
ω Uj2 þ Vj2 Wj2
Uj Vj;t ¼ : ð28Þ
2
Relative total temperatures (Trel ) are defined as well:
Wj2
Tj;tot;rel ¼ Tj þ
2cp
Wj2 Vj2
¼ Tj;tot þ
2cp
FIGURE 19 Schematics of the flow in an axial turbine rotor.
Uj2 ð1 2cÞ
¼ Tj;tot þ : ð29Þ
the inlet (r1). These triangles can also be defined in 2cp
terms of one velocity and two angles: the relative Ttot,rel is the stagnation temperature in the rotating
flow angle b and the absolute flow angle a. The frame of reference and, therefore, is the temperature
cascade nomenclature provided in Fig. 1 is not to which the rotor material is subjected.
universally adopted, and sometimes (especially in It is customary to classify turbine stages on the
English monographs) the values b901 and 901a basis of four characteristic indexes identifying their
are used instead. The following relations hold velocity triangles: the degree of reaction Rr, the exit
between the modules of the velocity vectors (‘‘j’’ is swirl o2, the stage loading coefficient Dc ¼ c1e2c2,
a running index denoting the station to which the and the flow coefficient j. The respective values must
triangle refers): be labeled with the radius at which the index is
computed (in the absence of another indication,
Vj;x ¼ Vj sin a ¼ jj Uj ð22Þ blade midspan is to be assumed). The degree of
246 Turbines, Steam
reaction is defined as the ratio of the static enthalpy jets—one for each statoric passage). In the rotor,
drop across the rotor to the decrease of the total these jets flow through a blade with a (nearly)
enthalpy across the stage: constant relative velocity along the rotoric passage.
h1 h2 Therefore, the pressure drop in the rotor is zero
Rr ¼ (Eq. 30). The stage work is produced in the rotor at
h0;tot h2;tot
2 the expense of the absolute velocity of the steam.
U1 U22 W12 W22 Assuming zero losses in the stator,
¼ 2 : ð30Þ 2
U1 U22 þ V12 V22 W12 W22
U1 U22 þ V12 V22
h0;tot h2;tot ¼ ; ð31Þ
A stage in which Rr ¼ 0 is said to be an impulse 2
stage. For axial turbines, Eq. (30) implies that in such we see the following:
a stage there is no pressure drop across the rotor
(W1 ¼ W2); that is, the entire stage pressure drop is 1. The stage work increases with the expansion
attained in the inlet nozzle (the stator), and the rotor ratio p0/p2 until V1 becomes sonic and the stage is
absorbs only the fluid’s kinetic energy. All stages for ‘‘choked.’’
which Rr40 on a portion of the blade are said to be 2. With the stage work being maximum for
reaction stages. The value of Rr is, whenever V2,t ¼ 0, it can be easily shown that the optimal U/
possible, limited to approximately 0.5 to limit the V1 ratio for axial turbines is approximately 0.5. In
absolute kinetic energy at stage outlet. centripetal turbines, experience shows that (U/
A zero exit swirl corresponds to the minimum V1)optimalB0.7/0.8.
absolute kinetic energy at stage exit. For a single 3. In centripetal turbines, the flow configuration is
stage and for the end stages of all machines, this is a such that even if V2 ¼ V1 (which is never realized in
desirable condition. For an intermediate stage, where practice because it would lead to high rotoric losses),
the exit kinetic energy can be recovered by the next the stage would deliver a positive work (U21–U22)/2.
(bottoming) stage, work maximization often leads to 4. Radial centrifugal turbines deliver, for the same
a negative c2. steam conditions and peripheral velocity, a lower
The loading coefficient is chosen as large as stage work. In fact, they have been abandoned except
possible, compatible with the desired value of the for the Ljüngstrm turbine (described in the first
efficiency (Fig. 20). Common values for axial stages section of the article).
range from 1 to nearly 3. For radial stages, DcD1. Therefore, in an impulse stage, the rotor is likely
to be unshrouded and devoid of the leakage seals that
2.3.2.1 Impulse versus Reaction Stages: Blade abound in the stator. A ‘‘typical’’ axial impulse stage
Twist It is useful to provide a more detailed is depicted in Fig. 21. Notice that for today’s higher
discussion of impulse and reaction stages. In an throttle conditions (400–5601C and 30–220 bars),
impulse stage, the steam experiences the entire stage the sonic velocity in the first nozzle is so high that the
pressure drop in the statoric channels and is ejected ratio U/V1 would be much lower than 0.5, leading to
from the stator as a jet (or, better, as a number of a substantial efficiency penalty. An ingenious solu-
tion (first suggested by Curtis) is that of absorbing
the absolute kinetic energy drop in two separate
rotors ‘‘fed’’ by a single stator. The intermediate fixed
0.9 blading is simply a deflector (Fig. 21B). This config-
Cx /U = 0.4
∆ψ = 1 uration is often adopted in the first stage(s) of large
H/b = 3.0 steam turbines because, for the same peripheral
0.8 Re = 105 velocity U, the velocity V1 is higher than 2U (its
η
∆ψ = 2
optimum is atB4U); therefore, the enthalpy (and
pressure) drop per stage is increased. This means not
0.7
only a lower overall number of stages but also a
∆ψ = 3 smaller high-pressure casing, and this results in a
0.6 lessening of the installation costs. The price is a
1.0 0.5 0 somewhat lower stage efficiency.
Reaction (R ρ ) Because the ratio of the blade height to the
FIGURE 20 Loading coefficient versus degree of reaction and external diameter is small only for the very first
stage efficiency. high-pressure stages, and because the requirement of
Turbines, Steam 247
A B
P1 Velocity P1 Velocity
Pressure
Ce Pressure Ce
P2 P2
Moving
blades Moving
blades
Blades or
or buckets
Nozzles buckets Nozzles
Fixed
blades
Fixed
blade
Nozzle Nozzle
Blade Moving
blades
Wheel Wheel
Shaft
Shaft
FIGURE 21 Impulse stages: (A) simple and (B) velocity compounded (‘‘Curtis’’).
constant total enthalpy along the radius (to avoid Dw ¼ Dwad Dhtot;inlet Dhtot;stator Dhtot;rotor
excessive radial motion in the nonrotating channels)
Dhtot;diffuser Dhtot;exit lmechanical : ð32Þ
demands that the Euler work given by Eq. (20) be
also constant along the radius, it follows that neither Notice that fluid-related losses have been expressed
Vt nor Rr can be constant along the blade. Therefore, in terms of total enthalpy drop. According to the
the term reaction stage implies that the degree of Gouy–Stodola theorem, this may be also written
reaction at its midspan diameter is greater than zero. Dhtot;j ¼ T0 Dsj ¼ Dej ; ð33Þ
Actually, there are cases in which Rr ¼ 0 at the hub
where T0Dsj is the irreversible entropy generation for
and reaches values higher than 0.5 at the tip. For h/
the jth loss and Dej is the corresponding exergy
D40.02, the velocity triangles differ markedly at the
destruction. In what follows, a concise and general
hub, midspan, and tip, and the blade must accom-
modate to radially varying V1t and V2t. Several description of how to include these losses in design
calculations is provided.
solutions are possible, with the one called free vortex
(Vt*r ¼ constant) being used most frequently. The Inlet losses. These are due to frictional and flow
resulting rotation of the blade profile at various radii turning effects (which cause flow separation and
around its central radial axis is called ‘‘twisting.’’ An originate recirculation regions).
example of a twisted blade is shown in Fig. 22. Partial admission losses. In partial admission
stages, there are two additional losses caused by the
drag experienced by the blades in their inactive arc
(‘‘ventilation’’ losses) and by the fact that a portion of
2.3.3 Second-Order Approximation: the energy of the statoric jet is perforce lost when the
Loss Correlations blade first enters its active arc and when it exits it
The difference between the ideal turbine work Dwad (‘‘sector’’ losses).
and the actual shaft work is due to the losses in the Statoric losses. These are due to friction within
turbine. It is common design practice to apportion the blade row, to secondary flows caused by turning
these losses to each component and to write effects (which cause flow separation and originate
248 Turbines, Steam
A
A B
Laminar separation
bubble
Stagnation point
vortex
~ 440
Passage vortex
B Stream surface
Inlet boundary
layer
Endwall
FIGURE 22 Untwisted (A) and twisted (B) steam turbine Passage vortex
blades.
Endwall Counter vortex
crossflow
recirculation regions), to boundary layer growth
FIGURE 23 Vortical structures along the blade channel.
along the blade (‘‘blockage’’), and to flow compres-
sibility effects. These losses are usually split into their
individual contributions, which are usually consid-
ered separately during design. Profile losses are
Rotoric losses. These are modeled in a manner
originated by a complex system of vortical structures
similar to that of stators. However, here the relevant
along the blade channels that strongly interact with
velocity is the relative inlet velocity W1:
each other and cannot be conveniently treated
separately (Fig. 23). These losses are usually evalu- Dhtot;rotor ¼ Dhtot;profile þ Dhtot;secondary þ Dhtot;blockage
ated by computing or measuring the so-called lift and þ Dhtot;shock þ Dhtot;leakage þ Dhtot;windage
drag coefficients, defined as the ratios of the lift and ¼ 0:5 Krotor W12 ; ð34Þ
drag force, respectively, to the kinetic energy of the
oncoming flow (Fig. 24A). A large database exists for where Krotor is a weight-averaged loss coefficient.
several blade profiles, and designers make use of Diffuser losses. These are also due to frictional
‘‘cascade performance charts’’ like the one depicted and recirculation effects. The diffuser is a component
in Fig. 24B. If the velocity triangles are known, so is in which the fluid flows against a pressure gradient
the cascade efficiency. Sometimes an additional loss, (velocity is recovered into pressure); therefore, it is
the wake loss, is included in the profile loss. intrinsically affected by the highest losses (in
Secondary and blockage losses are similarly modeled percentage). In steam turbines, the importance of
in terms of the square of the average meridional diffusers is normally negligible.
velocity, whereas shock losses, originated by sonic Mechanical losses. These are due to frictional
and supersonic shocks interacting with the boundary dissipation in bearings and seals. Customarily, they
layers and with the wake, are computed on the basis are accounted for introducing an organic efficiency:
of semi-empirical charts. Leakage flows (usually lmechanical ¼ 1 Zorganic Dwad ; ð35Þ
negligible in stators) are due to the unavoidable
leaks around seals and gaps. They depend on the with Zorganic ranging from 0.95 to 0.98 for the entire
blade aspect ratio (h=l; height over chord) and gap shaft line (if a shaft supports N stages, the loss
size (g/h). apportioned to each stage is 1 Zorganic =N).
Turbines, Steam 249
s/c
statoric and rotoric blade rows been simulated in full.
1.5
All of these studies strongly depend on the turbulence
1.0 1.0 modeling adopted (‘‘full’’ calculations, or the so-
called DNSs [direct numerical simulations], are still
0.5 impossible on the complex geometries of a turbine
stage and in their operational high Reynolds num-
0.5 bers), and some of the effects can be assessed only
under the simplifying assumption of inviscid flow.
Nevertheless, reliable work has been performed to
0 study both inviscid (i.e., purely kinematical) and
10 0 10 20 30 40 50 60 70 viscous (i.e., including boundary layers and wake
Air outlet angle (α2 [degrees]) interactions) flow features, and the results compare
well with global and local experimental measure-
FIGURE 24 (A) Lift and drag coefficient for a cascade of fixed
geothermal form. (B) Design lift coefficients. ments. A situation is now being approached where
blade design can be performed on the basis of
successive approximations, by iterative adjustment
Once the total stage losses have been computed, of several design parameters, to determine the best
they are used to blade profile for a specified performance. However, at
the industrial level, most of the design activity is still
1. Correct the ideal Euler work (Eq. 20). and based on proprietary loss correlations, and fully
derive the actual work delivered by each stage three-dimensional simulations are still considered a
(Eq. 31). helpful tool but not a completely reliable design
2. Recompute the fluid thermodynamic conditions procedure. With the advance of computer hardware
at the exit of each stage and draw the actual (which leads to higher computational speeds and
expansion line of Fig. 14. reduced simulation time), computational fluid dy-
namic research (which provides a better understand-
ing of local flow features), and available software
2.3.4 Three-Dimensional Calculation of the Real (which improves the quality of the commercially
Flow in a Complete Turbine Stage available flow models), it is clear that in the near
The real flow in a turbine stage is highly turbulent, future, inverse design problems will be solved by
fully three-dimensional, and intrinsically unsteady. means of standard, well-understood, and relatively
Although its general features are well understood easy-to-use procedures.
250 Turbines, Steam
1.2
81 70.5
80.2 80.8
76
84
80
85.2 84
1.0 86
3.7 Preliminary Blade Design and Costing Numerical experiments, which now constitute a
substantial part of research and development
Once all geometrical dimensions are known, the
(R&D) programs in the field, are attractive because,
blades and all other components can be preliminarly
if properly conceived and performed, they share most
sized. The choice of materials is imposed in practice
of the properties of physical experiments but can be
by the environment within which the design activity
carried out at a fraction of the cost and possess an
takes place (with industrial and commercial strate-
intrinsic degree of flexibility that makes their repeti-
gies usually conflicting with purely technical con-
tion under slightly different conditions relatively easy.
siderations). A preliminary costing assessment of the To be useful for design purposes, numerical simula-
machine is developed.
tions must have the following requisites:
/ h
ds at
po
u n
io
n A Although only marginal further increases in their
ct IG
0 00 t ra PS performance can be expected to come from design
0, ex 0
15 s/
h 25
20 C u nd n
at improvements, a more substantial gain may be
po tio
,0
00 tra
c obtained by increasing the temperature and pressure
0 ex IG
10
d s/
h
0
PS of the feed steam. Steam turbines for motive
un 25
0
po
n
at applications (marine and terrestrial) are rapidly
, 00 tio
50 t r ac becoming obsolete because in this field they are
ex
10 s /h being replaced by the lighter and more flexible diesel
u nd
0
po engines and (especially for the higher power outputs)
gas turbines.
Turbine design is also a very mature engineering
B field. Current procedures are quite well defined and
0
10,000 20,000 make extensive use of the automated optimal design
Shaft power (kW) methods. In this respect, a quantitative leap is
necessary in computer speed and numerical methods
FIGURE 27 A typical extraction turbine performance map
showing the Willans lines.
(especially turbulent modeling) to attain an appreci-
able advantage over the current state of the art.
output/mass flow rate dependence is no longer
linear.
SEE ALSO THE
FOLLOWING ARTICLES
5. CONCLUSIONS
Cogeneration Mechanical Energy Turbines, Gas
Steam turbines are the most important machines in
the energy conversion field. Their cumulative me-
chanical and electrical installed power exceeds that Further Reading
of any other type of unit by far. Their design and
Balje, O. (1981). ‘‘Turbomachines: A Guide to Design, Selection,
operational procedures are very advanced, and they and Theory.’’ John Wiley, New York.
can be regarded as mature technological compo- Demeulenaere, A., and van den Braembusche, R. A. (1998). Three
nents. They are characterized by the highest stage dimensional inverse method for turbomachinery blading design.
efficiency among all ‘‘prime movers’’ (with Zstage ASME J. Turbomachinery 120, 247–255.
ranging from 0.80 to 0.95), by the highest power Lakshminarayana, B. (1996). ‘‘Fluid Dynamics and Heat Transfer
of Turbomachinery.’’ John Wiley, New York.
output per unit (up to 300 MW for a single body), by Mathis, D. M. (1994). Fundamentals of turbine design. In
one of the highest power densities (up to 700 kW/cm2 ‘‘Handbook of Turbomachinery’’ (E. Logan, Jr., Ed.). Marcel
of active flow area), and by one of the highest mean Dekker, New York.
times between failure (4100,000 h between major Peltier, R. V. (1994). Steam turbines. In ‘‘Handbook of Turbo-
machinery’’ (E. Logan, Jr., Ed.). Marcel Dekker, New York.
overhauls) and also cover an extremely wide range of
Stodola, A. (1945). ‘‘Steam and Gas Turbines.’’ Peter Smith, New
applications (from a few kilowatts as auxiliary York.
power units to more than 1000 MW in major nuclear Wislicenus, G. F. (1947). ‘‘Fluid Mechanics of Turbomachinery.’’
plants). Their installed cost is relatively modest in McGraw–Hill, New York.