Professional Documents
Culture Documents
1 s2.0 S0306261914011027 Main
1 s2.0 S0306261914011027 Main
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
h i g h l i g h t s
We propose an accurate 0-D model to design low size radial ORC turboexpanders.
Different methods for radial turbine design are discussed and analyzed.
The model thoroughly estimates the different losses contributing to efficiency drop.
Real equations of state for the different ORC fluids are adopted.
An off design model is built to apply to the designed turbines far from design point.
a r t i c l e i n f o a b s t r a c t
Article history: In this paper, a zero-dimensional model for the design of radial turbo-expanders for ORC applications is
Received 18 March 2014 discussed, with special reference to the estimation of losses and efficiency; a comparison between differ-
Received in revised form 6 October 2014 ent fluids (R134a, R1234yf, R236fa, R245fa, Cyclohexane, N-Pentane) is presented and discussed, refer-
Accepted 21 October 2014
ring to a typical small-size application (50 kW). In the model, different methods for the design of
Available online 19 November 2014
radial turbines are screened, with special attention to the estimation of losses, for which correlations
from literature are used. Real Equations Of State (EOS) are applied to the expansion process in place of
Keywords:
the traditionally adopted Mach relationships for ideal gas, which is a significant advancement for mod-
Radial turbine design
Expansion efficiency losses
eling organic fluids in ORC, often operating near to critical conditions. The results show that the total to
Off design total efficiency of the designed machines range between 0.72 and 0.80, depending on the considered
Micro-ORC fluid. Generally, higher efficiency (1.5–2.5% points) can be achieved adopting backswept-bladed rotors.
The most significant losses come from the rotor secondary flows, due to the high curvature of blade pro-
files combined to the large pressure gradient. The best performing fluids are R236fa and R245fa, followed
by R134a and R1234yf.
Finally, starting from the developed design tool, an off-design analysis of turbo-expanders is presented.
Once the design data are available, the characteristic curves of the expander at variable temperature,
pressure and fluid mass flowrate at the expander inlet for different values of the specific speed are built.
It is thus possible to evaluate the performance of the radial expanders when working far from design
point. This analysis, demonstrated for R134a, shows that the total to static efficiency has a relatively mod-
est sensitivity to the off design of the expansion ratio, especially at corrected speed below the design
value.
Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.apenergy.2014.10.052
0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
518 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
Nomenclature
large enthalpy drops with relatively low peripheral speeds, possi- axial expanders, whereas much less data are available for radial
bility of adopting a single-stage design. On the whole, this results turbines. Most data available refer to ideal gas and make use of
in good performance and affordable price. The optimization of the Mach compressibility relations [15–22]. This may not be a sat-
the thermodynamic cycle, with special reference to fluid selection, isfactory approximation when dealing with ORC turbines, which
has been studied widely in the last years. Fluid-dynamic design of operate near the saturation line or close to the critical point. In
turbo-expanders can take advantage of the availability of modern the model hereafter proposed, correlations from literature are used
CFD methods [13,14]; however, there is a need for preliminary [15–32], but real Equations Of State (EOS) are applied to the expan-
design methods, and of modeling tools capable of predicting the sion process (static and total variables) in place of ideal gas rela-
off-design performance (which is determined, for example, by tions. This is an important feature, allowing to preliminary
the variation of the resource for solar-driven EGS, or by variation design and performance prediction of turbo-expanders that work
of ambient temperature in geothermal or heat recovery applica- with real substances. The correlations can be refined progressively
tions). In the field of low power output (i.e. up to 100–150 kW), as more data on ORC expanders become available, either from field
radial expanders are almost the only choice, with the eventual operation, or from specific test arrangements. The thermodynamic
competition of screw expanders. Generally, literature is rich of the- properties of the working fluids are calculated using the libraries of
oretical–experimental correlations for the estimation of losses in the EES software [33], which is the programming environment
D. Fiaschi et al. / Applied Energy 138 (2015) 517–532 519
adopted in this work. Making use of the design model, a sensitivity presented and discussed in books and scientific papers, most of
analysis investigating the effects of the different design parameters which are based on experimental work performed at NASA
on the expander performance is presented. Finally, an off-design between 1965 and 1975 [18–20,22,32]. At that time radial tur-
model has been developed and some results are discussed, in order bines, working with ideal gases (air, helium), were designed and
to assess the behavior and estimate the performance of the turbo- tested for aerospace applications. The high values of centrifugal
expanders when working out of nominal conditions. This often stresses on rotor blades and the limits on materials performance
happens when the ORC high-temperature resource is a time- and production technology led to the design of the ideal 90° IFR
dependent energy source, like solar; but also, with seasonal change turbine: that is, a rotor having radial blades at inlet. Some years
of condenser conditions, on account of the heat/mass transfer later (1983) NASA researchers [20] studied and developed a
performance of the cooling system (condenser/cooling tower/air radial-inflow turbine with a more performing rotor design, charac-
cooler). terized by a blade sweep angle b2 at rotor inlet (IFG, Fig. 1). The
typical velocity triangles are shown in Fig. 2; in both cases, the
2. Fundamental design concepts and parameters for absolute velocity c3 is assumed to result axial at rotor outlet, in
Radial-Inflow Turbines (IFR) order to guarantee good diffuser performance (Fig. 2). Generally,
nominal design conditions are referred to zero-incidence at rotor
Radial-inflow turbines have been less studied than the axial inlet and zero-deviation at rotor outlet; however, in radial tur-
ones, and have been manufactured by a limited number of compa- bines, the best efficiency values are obtained when incidence is
nies [34–36]. The fundamental design of radial-inflow turbines is non-zero (Fig. 3): this is a consequence of the rotational flow,
Fig. 1. General schematic of Radial inflow turbine: 90° IFR (shaded), and general rotor shape (IFG, unshaded).
Fig. 2. Velocity triangles – nominal conditions; (a) 90° IFR and (b) IFG.
Fig. 3. Velocity triangles – optimal incidence conditions; (a) 90° IFR and (b) IFG.
520 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
which displaces the tangential component of the relative velocity 3. Design guidelines
(wu) in the opposite direction with respect to the peripheral rotor
velocity. From the technical literature [15,16] it appears that the 3.1. Input data and expected process output
best values of rotor inlet angle in the relative flow are between
20° and 40°, referred to the camber line direction (the positive The input data consist in the expander rated power output, in
sign is conventionally assumed toward the direction of peripheral the thermo-fluid dynamic variables determined by the thermody-
velocity u). namic cycle [37], and in a set of dimensional and non-dimensional
In radial turbine design, some non-dimensional parameters parameters chosen by the designer (Table 1). The outputs of the
help designers to select a geometry optimizing efficiency using a calculations are the basic geometry with the related velocity trian-
limited set of variables; the following parameters are recom- gles and the efficiency of the designed expander. The model is also
mended in the literature [15,16,19,21,22]: able to calculate the turbine losses and their relative share in the
resulting inefficiency.
d3h
¼ 0:40 ð1Þ
d3sh 3.2. Preliminary sizing
d3sh
¼ 0:70 ð2Þ The first step is the preliminary calculation of geometry, using
d2 the non-dimensional parameters listed in Table 1, which
Another important parameter is the isentropic velocity ratio determine also the isentropic nozzle and rotor enthalpy varia-
(u2/cs), which maximizes the efficiency in the range 0.69–0.71 tions. The load and flow coefficients (W ¼ Dh0 =u22 and
[15,16,19,21,22]. cs is the spouting velocity, defined as the velocity U ¼ cm2 =u2 respectively) are adjusted to calculate the peripheral
at which the kinetic energy of the flow is equal to the isentropic velocity (u2), specific speed (ns), speed of revolution (x) and
enthalpy drop from turbine inlet stagnation pressure p01 to the meridional component of absolute velocity at nozzle exit/rotor
final exhaust pressure [16]. The definition of spouting velocity is inlet (cm2). These data are used for the evaluation of the mass
different depending on (I) whether a diffuser is present or not flow rate m:_
downstream the turbine, see relationships (3)–(5) respectively,
_ ¼ q2 cm2 b2 d2 pð1 BK 2 Þ
m ð6Þ
and (II) if total (4) or static (5) conditions are considered at the
turbine exit (Fig. 4): The meridional component of the absolute velocity at nozzle
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi inlet (cm1) – which is the same as absolute velocity (c1) as the flow
cs ¼ 2ðh01 h4ss Þ ð3Þ at the IGV nozzle inlet is assumed to be radial – is given by the
application of mass balance in section 1 (Fig. 1):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cs ¼ 2ðh01 h03ss Þ ð4Þ _
m
cm1 ¼ ð7Þ
q1 pb1 d1 ð1 BK 1 Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cs ¼ 2ðh01 h3ss Þ ð5Þ Consequently, knowing the inlet enthalpy h1(p1, T1), it is possi-
ble to calculate the total enthalpy (h01). The thermodynamic vari-
u2 w2
ables at point 2 (p2, T2, h2, s2, q2, Ma2, Mau2 ¼ VS2
, Mar2 ¼ VS2
) and
the velocity triangles at rotor inlet are determined calculating first
the nozzle isentropic expansion and then the real transformation
using the nozzle loss coefficient nN (Fig. 2):
h
h2 ¼ h2s þ 0:5nN c22 ð8Þ
1
p0
2
p0
Once the nozzle exit/rotor inlet conditions are known, the thermo-
dynamic variables at point 3 (rotor exit/diffuser inlet, Fig. 1) are cal-
01
p1
4
p0
03 04
p3
2 2
1/2c42
d3sh d3h
_ ¼ q3 cm3
m pð1 BK 3 Þ ð10Þ
2
4
1/2c3
4
03ss
Finally, using the rotor loss coefficient and assuming an axial
04ss 3 discharge at rotor outlet, the static enthalpy and velocity triangles
4ss at rotor outlet can be calculated:
3s
3ss
h3 ¼ h3s þ 0:5nR w23 ð11Þ
Table 1
Input data.
3.3. Geometry of the stator (IGV) which determines an exceedingly large number of rotor blades,
the formulation proposed by Glassman [18] is followed:
The prediction of the angle of flow leaving the bladed nozzle of p
a radial turbine, discussed in the previous section, is a fundamental ZR ¼ ð110 a2 Þ tan a2 ð13Þ
30
design topic. The next step is the calculation of the angles of blades,
which are radial at inlet for no pre swirled IGVs. At outlet, while Once the number of blades is determined, the optimum rotor inci-
leaving the nozzle, the flow does not follow the vanes completely dence angle can be calculated. As previously discussed, better effi-
but it turns toward the meridional direction of an angle known ciency values are achieved when incidence is non-zero, due to the
as deviation, due to the combined effects of boundary layer growth pressure gradient within the blade channel which tends to move
(limited by the accelerating flow) and the subsequent abrupt the flow from the pressure to the suction side. Referring to the cam-
expansion due to the trailing edge thickness. As a simple design ber line direction, the best values of rotor inlet angle in relative flow
approach, the actual angle of blades at nozzle outlet is calculated are between 20° and 40°, in the opposite direction with respect to
interpolating data from Hiett and Johnston [15,31], which have a the peripheral velocity u2.
rather linear behavior, approximated by the interpolating function Referring to IFR rotors (radial blades at rotor inlet), the follow-
ab2 ¼ 0:884 ab2 þ 4:56. ing are the recommended correlations [15,16]:
Another important parameter is the number of stator blades, 2 u2
tan b2opt ¼ ð14Þ
which directly influences the losses. Increasing the number of Z R cm2
blades leads to better flow guidance at the price of higher frictional
0:73p
losses. A general design approach to define the stator number of tan b2opt ¼ 1 ð15Þ
blades is that it should be a prime number compared to the rotor ZR
one. In addition to this basic criterion, the criterion of Zweifel on
1:98 tanða2 Þ
the optimal ratio between the chord and the blade spacing can tan b2opt ¼ ð16Þ
be adopted [16,25]. It suggests that, in order to minimize the Z R 1 1:98
ZR
losses, the ratio between the tangential load of an actual to that
of an ideal blade (WT) should be about 0.80. Knowing the absolute The above Eqs. (14)–(16) provide similar results, even though (15)
flow angles at nozzle inlet and outlet, and after calculating the has been obtained considering minimum Mach number conditions
chord length, from the expression of blade pitch Z s = p d2, the at the rotor inlet [15].
following Eq. (12) gives the optimum blade spacing, from which In the case of the IFG design (non-radial blades at rotor inlet),
the number of blades can be easily determined: the procedure suggested by Meitner and Glassman [20] can be fol-
lowed by calculating first the optimal value of the peripheral com-
s
ponent of the absolute velocity [20]:
WT ¼ 2 cos2 a2 ðtan a1 þ tan a2 Þ ð12Þ 8
x pffiffiffiffiffiffiffiffiffiffiffiffiffi !
>
> 1 cos ðab3 Þ=ðZ R Þ0;7 f1½ðr3 =r2 elim Þ=ð1elim Þ3 g
>
> u if r3
> elim ðaÞ
>
< 2 1
tg ðab3 Þ r2
tg ða2 Þ
cu2opt ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi !
3.4. Optimal incidence at rotor inlet and number of rotor blades >
> cos ðab3 Þ=ðZ R Þ0;7
>
> 1 r3
6 elim ðbÞ
: u2
>
1
tg ðab3 Þ
if r2
tg ða2 Þ
Calculating the number of rotor blades is a fundamental issue in
the design of radial turbines, because it defines the basic structure ð17Þ
of the machine and has a primary role in the estimation of losses.
The parameter elim can be determined by [20]:
There are no absolute criteria allowing an univocal evaluation of
number of blades; the design guidelines tend to avoid very low 1
elim ¼ ð18Þ
local velocities near the blade surface in the inlet region of the e8:16 cos ab3 =Zb
rotor, with a consequent tendency to early separation [15]. On Once cu2opt has been calculated, it is possible to calculate the rel-
the other hand, the adoption of a high number of blades is not so ative velocity and blade angle as follows [20]:
convenient, especially for small rotors: the blockage effects, the
weight and inertia of the rotor become very high. Moreover, a large wu2opt ¼ cu2;opt u2 ð19Þ
number of blades is also responsible for a large wetted surface
area, which increases the friction losses. In the present model, wu2;opt
b2opt ¼ tan1 ð20Þ
rather than using the formulation proposed by Jamieson [16,17], cm2
522 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
3.5. Expander efficiency, power output, degree of reaction and specific h0j h0j;s
Dqj ¼ ð32Þ
speed u2j
The calculation of the expander efficiency, power output, design This is the approach followed in the present model: starting from
degree of reaction and specific speed is done following the guide- non-dimensional loss coefficients obtained from the correlations,
lines in [38], which are briefly recalled for completeness. it calculates n and the related efficiency drop (Dg), for each kind
The efficiency can be either referred to turbine discharge or of loss. It allows, in all the typologies of expanders and operating
including also the diffuser; in the first case, the total-to-total and conditions, to analyze the distribution of losses and to investigate
total-to-static efficiency are given by: how do they affect the overall performance. The so built model pro-
h01 h03 vides a reliable basis to improve the design of different kinds of
gtt ¼ ð21Þ rotors with several possible working fluids.
h01 h03ss
NQ 1=2
3
4.2. Rotor losses
Ns ¼ 3=4
ð28Þ
Dh0s
The flow in the rotor of a radial turbine is subject to a rapid
where acceleration in the flow direction, and to a turn both in the merid-
ional plane and along the camberline. These effects give rise to a
Dh0s ¼ h01 h03ss ð29Þ
complex pattern of secondary flows. The flow in the rotor of a
radial turbine does not result into a high growth of the boundary
4. Calculation of losses layer and separation, even though, due to the three dimensional
behavior, it develops a significant non-uniformity of the total pres-
In order to evaluate the actual performance of a turbomachine, sure inside the flow channel, which can lead to generation of
the contributions of different losses must be calculated. This calcu- losses. The probability of this occurrence increases when the blade
lation cannot be substituted by advanced CFD methods, because it loading is augmented. In the present model, the losses are divided
provides vital information to the designer, about the process of loss in different contributions:
buildup determining the final turbomachine performance. On the
other hand, advanced CFD is very useful for cross-checking the – Rotor incidence losses.
overall results of the efficiency/loss model, and often to supple- – Skin friction losses.
ment data which would require very detailed (often impossible) – Tip clearance losses.
measurements. – Blade loading losses.
Generally, the first step of the design procedure considers con- – Disk friction losses.
sequently the effects of losses through appropriate dimensionless
coefficients. The related efficiency drop Dg is then subtracted from 4.2.1. Rotor incidence loss
the isentropic value to calculate the actual efficiency [15]: In the actual working conditions of the expander, the incidence
gact ¼ gs Dg ð30Þ angle of the relative flow at rotor inlet is rarely at the optimal value
(Eqs. (14)–(16)). For this reason, the incidence loss appears. The
In this model, the overall loss of the turbine is obtained by the recommended models for the estimation of rotor incidence loss
sum of several contributions, each one estimated through correla- were developed at NASA [15,16,26]. The general approach is to
tions which depend on kinematic and geometric parameters. The assume that the kinetic energy associated with the variation of
dimensionless loss coefficients are defined in several ways, and it the tangential component of the relative velocity with respect to
is important to merge them to a common basis, in order to apply the design value, which is the result of the fluid–blade impact, is
them within the same model and do some reliable comparisons converted in internal energy of the fluid, which leads to an increase
with results from literature. Referring to the jth loss [15]: of entropy. The detailed calculation procedure is reported in [16].
hj hj;s The incidence losses may also be calculated using alternative
nj ¼ 1 2
ð31Þ approaches [15], obtained with the same conceptual assumptions
V
2 j
and therefore formally similar [15]:
D. Fiaschi et al. / Applied Energy 138 (2015) 517–532 523
2 2
w22 sin b2 b2;opt e cu2
dh0;i ¼ ð35Þ DqR;cl ¼ 0:4 ð39Þ
2 b2 ut;le
2 When the values of radial clearance are adopted, an acceptable
w2 sin b2 w2 sin b2;opt
dhi ¼ ð36Þ agreement with literature results are achieved [32] for equation
2 (39).
The above discussed methods provide similar results, consistent
with the literature. Thus, any of the two proposed correlations may 4.2.4. Blade loading losses (including secondary flows)
be adopted leading to negligible differences. Blade loading losses, including secondary flows, are caused by
the high curvature of the profile and the pressure gradient in the
4.2.2. Skin friction losses rotor vanes. They give the largest contribution to the reduction
Friction losses can be estimated referring to a rotor-equivalent of the expander efficiency. However, they are not extensively
duct working on the same flow rate [26]: reported and discussed in literature, often because they are treated
kR LR in combination with other losses using experimental coefficients
nR;f ¼ ð37Þ [20,26,30]. The model here proposed evaluates the secondary flow
DR
losses through correlations, as functions of kinematic and geomet-
Details about the calculation of the characteristic diameter and ric parameters. For the calculation of blade loading losses, the fol-
length can be found in [26] or [28]. As for the stator, the friction lowing correlation proposed by Rodgers can be used [21,41]:
factor can be determined using Moody’s diagram. The relative
roughness and Reynolds number are estimated referring to the 2
cu2
rotor-equivalent duct. Alternatively to Eq. (37), the friction losses ut;le
dqR;bl ¼ 2 ð40Þ
may be calculated using the following expression [15]: Z R rz2
h i
4kR ðw2 =VS01 Þ2 þ ðw3 =VS01 Þ2 where z/r2 is the ratio between the expander axial length and rotor
DqR;f ¼ ð38Þ inlet radius [21].
4 DR =LR ðu2 =VS01 Þ2
For the calculation of profile losses, another correlation pro-
This correlation tends, generally, to overestimate the friction posed by Rodgers [41] and suitably revised by Whitfield [21]
losses by 70–80%. may be adopted:
0 1
b3 b2
! !
4.2.3. Tip clearance losses B þ 2
r 2 C w2 þ w3
r2
2
VS201
Tip clearance losses are due to the fluid leaking through the dqR;p ¼ 0:5@ 2 A ð41Þ
1 r3 2VS201 u22
clearance gaps between the blade tips and the shroud. With refer- r2
ut;le r 22 kv
0:25q
DqR;df ¼ ð42Þ
m_
where:
8 0:1
>
> 3:7 erax
>
>
>
<
2
Re0:5
; Re < 3 105
kv ¼ 0:1 ð43Þ
>
>
>
> 0:102 erax
>
: 2
Re0:2
; Re > 3 105
u2 r 2
Re ¼ ð44Þ
v2
As a possible alternative, providing similar results (in the range of
1%), the correlations proposed by NASA [16,19,20] can be
Fig. 5. Tip clearance along the development of a rotor blade. recommended.
524 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
the fluid. The calculation of the diffuser losses follow the standard 0,6
IFG
d3h/d3s
procedure described in [42,43] for conical diffuser with non-uni-
0,55
form flow at inlet, which are the sum of two contributions: (1)
enlargement of the section nenl and (2) flow friction against the 0,5
0,35
where k is a coefficient accounting for the non-uniformity of inlet
flow. nenl and nf can be expressed as follows: 0,3
0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09
rffiffiffiffiffiffiffiffiffiffiffi 2 b 2/d 2
cD 4 cD A3
nenl ¼ 3:2 tg tg 1 ð46Þ
2 2 A4 Fig. 7. Ratio of rotor exit diameter at hub and shroud as a function of ratio between
blade height and diameter at rotor inlet b2/d2 (IFR and IFG).
" 2 #
kD A3
nf ¼ cD 1 ð47Þ
8 sin 2 A4 blade height at the inlet and outlet of the stator. The absolute value
1 is strongly dependent on the considered fluid: for example, the
The friction factor kD can be expressed as kD ¼ ð1:8logðRe3 Þ1:64Þ2
.
cyclohexane flow rate is much lower than that of the other working
fluids, because of the larger specific enthalpy drop. Among the flu-
5. Results and parametric analysis (design process) ids here considered, R1234yf shows the highest flowrate, due to
the lowest specific enthalpy drop. Generally, with the reduction
Making use of the above-described loss correlations, a paramet- of density and velocity of fluids, the required blade height ratio
ric analysis has been run to assess the behavior of losses against increases to achieve the fixed expander power output. It is also
the main design parameters and input data: important to remark that, for a fixed flowrate, expanders with
backswept blades (IFG) require higher b2/d2 to achieve the same
a. Blade height – inlet rotor diameter ratio (b2/d2). flow rate as for the IFR design. Generally, the rotor outlet blade
b. Flow coefficient (U). height increases with increasing b2/d2 (i.e. the ratio between hub
c. Load coefficient (W). and shroud diameters at rotor outlet, d3h/d3s decreases, Fig. 7). In
d. Isentropic degree of reaction (Rs). IFR configurations, R134a and R1234yf show a substantially differ-
ent trend compared to the other fluids. In IFG designs, the most
The reference case is a 50 kW turboexpander operating with a remarkable different behaviors are shown by R134a, R236fa and
saturated or superheated vapor ORC between the upper/lower cyclohexane. It is due to the fact that the blade height at rotor out-
temperature levels of 147/95 °C (referred to R134a [38]). The main let has not constraints and is determined in order to match the
operating data are reported in Table 1. required mass flowrate while satisfying the other geometric
parameters, which are constrained by input data and non-dimen-
5.1. Blade height – inlet rotor diameter ratio (b2/d2) sional fixed amounts (Fig. 7). It because the outlet rotor blade
height is determined from the mass flowrate balance, with the con-
This parameter influences the power output, mass flow rate, straint of axial flow. Consequently, the meridional component of
efficiency, blade shape and flow conditions, especially at rotor out- the outlet rotor velocity increases with the reduction of blade
let (Figs. 6 and 7) for both radial (IFR) and backswept (IFG) rotor height. Fig. 7 reflects a widely different design geometry of
geometries. The mass flowrate (and thus the power output) expanders working with different fluids.
increases linearly when b2/d2 is augmented, due to the increased
5.2. Flow coefficient (U)
2,5
U defines the velocity triangle at rotor inlet. The flow coefficient
2,3
is one of the main parameters in the design of turboexpanders, as it
2,1
directly influences the mass flow rate, the performance (power
1,9 output and efficiency), the geometry and the rotor number of
1,7 blades. To give an idea of how U influences the flow, the variation
m [kg/s]
1,5 of shape of the velocity triangle at rotor inlet at two different val-
IFR ues of U is shown in Fig. 8(a) (radial rotor blades IFR) and (b) (back-
1,3
IFG
1,1
swept rotor blades IFG). Keeping constant the other non-
dimensional design parameters, the meridional component of the
0,9 R134a
R1234yf
inlet absolute velocity (cm2) increases with increasing U, whereas
0,7
Cyclohexane the peripheral velocity remains almost unchanged, as it mainly
0,5 R236fa
R245fa depends on the load coefficient. This results in an increase of abso-
0,3 lute and relative velocity at rotor inlet (c2 and w2 respectively) and
0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09
in a reduction of the related flow angles (a2 and b2). The change in
b2/d2
b2 directly affects the incidence losses, whereas a2 influences the
Fig. 6. Mass flow rate as a function of ratio between blade height and diameter at number of rotor blades, according to Eq. (13). As a2 decreases with
rotor inlet b2/d2 (IFR and IFG). increasing flow coefficient, the number of rotor blades is reduced,
D. Fiaschi et al. / Applied Energy 138 (2015) 517–532 525
α 2' β 2'
α2' c 2'
' β2 ' w2 '
c2
α2
α2 β2
c2 w 2'
c2 w2 β2
cm2
w2
u2,u2' u2,u2'
(a) (b)
Fig. 8. Variation of velocity triangles with increasing flow coefficient (from solid black to dashed green, (a) IFR and (b) IFG. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
η ts
0,72 IFR
sidered field of flow coefficient (typical of radial turboexpanders, IFG
0.08 < U < 0.22), no significant differences in the ‘‘optimal’’ number 0,7
of blades was found for the different investigated fluids (R134a
shows the lowest optimal number of blades, whereas Cyclohexane 0,68
R134a
R1234yf
shows the highest one, for both IFR and IFG geometries). From R245fa
0,66 Cyclohexane
Fig. 10, it is evident that backswept bladed (IFG) expanders have
a total to static efficiency (gts) 1.5–2 points higher than radial 0,08 0,1 0,12 0,14 0,16 0,18 0,2 0,22
22 IFR
20 IFG Anyway, when W is within the range 1.05–1.15, b2 values are at
the same levels for the different fluids in the case of backswept
18
bladed rotors (IFG).
16
14
5.4. Isentropic degree of reaction (Rs)
12
0,08 0,1 0,12 0,14 0,16 0,18 0,2 0,22
φ
Rs is defined as the ratio between the static isentropic enthalpy
drop through the rotor and that of the overall stage. It strongly
Fig. 9. Number of rotor blades as a function of flow coefficient (IFR and IFG). affects the performance of the expander. When combined with
526 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
α2
α2
β2 c2 c2 '
c2 c 2' w2 α2'
α 2' cm2,cm2'
β 2' β2'
cm2,cm2' w 2' β2
w2' w2
u2,u2' u2,u2'
(a) (b)
Fig. 11. Variation of velocity triangles at rotor inlet with increasing load coefficient (from black solid to green dashed) (a) IFR and (b) IFG. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this article.)
46000 65
R134a 60
44000 R1234yf
55
R245fa
42000 50
Rotational speed [ rpm]
Ciclohexane
40000 45
40
38000
35
36000 30
34000 25
β 2 [°]
20
32000 15
30000 10 IFR
5 IFG
28000
0
26000 -5
IFR -10 R134a
24000
IFG -15 R1234yf
22000 -20 R245fa
-25 Ciclohexane
20000
0,9 0,95 1 1,05 1,1 1,15 1,2 1,25 1,3 -30
0,9 0,95 1 1,05 1,1 1,15 1,2 1,25 1,3
Ψ
Ψ
Fig. 12. Speed of revolution vs. load coefficient (IFR and IFG).
Fig. 14. Angle of relative flow velocity (b2) at nozzle outlet/rotor inlet vs. load
coefficient (IFR and IFG).
84
level and thus the lowest reduction of power output. Moreover, IFR
83
expanders with radial blades have higher Rs than those with back-
swept blades. The variation of Rs has strong effects on the rotor
82
peripheral speed. Referring to Fig. 17, the remarkable difference
in peripheral speed for the different investigated fluids and rotors
81 can be noticed. Specifically, R245fa shows the highest values,
α 2 [°]
whereas the lowest are shown by R1234yf, due to its low value
IFR
80 of the total enthalpy drop. Finally, the backswept rotors (IFG) have
IFG
a lower peripheral speed compared to the radial bladed ones. This
79 is due to the higher load coefficient W (with fixed total enthalpy
R134a drop) which characterizes the backswept geometry.
78 R1234yf
R245fa
Ciclohexane
6. Interpretation of results – design process
77
0,9 0,95 1 1,05 1,1 1,15 1,2 1,25
Ψ 6.1. IFR vs. IFG design
Fig. 13. Angle of absolute flow velocity (a2) at nozzle outlet/rotor inlet vs. load
Moving from the radial (IFR) to backswept blades (IFG) config-
coefficient (IFR and IFG).
uration, the load coefficient increases. Thus, in the backswept
(IFG) design, the meridional component of velocity at rotor inlet
the other main parameters U and W, it completes the definition of is reduced to maintain the fixed flow coefficient. For this reason,
the design geometry. When the other design parameters are fixed, in order to achieve the target 50 kW power output, the IFG design
with increasing Rs the isentropic stator static enthalpy drop is shows higher b2/d2 ratios. An additional consequence of the higher
reduced. For this reason, the pressure at the stator outlet is higher load coefficient of backswept bladed rotors is their lower periphe-
and the related fluid density is increased. With fixed stator outlet ral velocity (and rotational speed) for a given rotor size. Finally,
cross sectional area, given the relatively limited change in mass backswept machines show a lower degree of reaction than the cor-
flow rate, the meridional component of the absolute velocity is responding radial bladed ones. As remarked in the parametric anal-
reduced, which offsets the increase in static pressure. Keeping con- ysis, when the overall enthalpy drop (stator + rotor) is fixed, the
stant the flow coefficient, the rotational speed is reduced. The reduction of the degree of reaction implies a lower stator outlet
related velocity triangle is modified as shown in Fig. 15. The backpressure and, consequently, a lower fluid density in this sec-
expander power output decreases with increasing Rs of an amount tion. As the outlet conditions are fixed, and because the mass flow-
variable with the different investigated fluids, as shown in Fig. 16. rate undergoes only limited variations, the meridional component
R245fa is the least sensitive to Rs because it has the lowest flowrate of the absolute velocity increases to counterbalance the reduction
D. Fiaschi et al. / Applied Energy 138 (2015) 517–532 527
(a) (b)
Fig. 15. Variation of velocity triangles with increasing in isentropic degree of reaction Rs (from black to green, (a) IFR and (b) IFG). (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this article.)
51000 ing fluids. All those here considered are good candidates for the
power cycle specifications (power output, temperature levels).
50000 On the other hand, large differences in kinematic, geometric and
performance characteristics are found between the different fluids.
49000
Power output [W]
165 speed. Generally, high nozzle exit angles are found (77–83.5°),
with larger values for backswept rotor design (IFG). When consid-
155 ering the flow within the rotor, a high deflection level has to be
145
remarked, which is in the range 40–90°, generally higher for back-
swept configurations. Due to the shape of velocity triangles (see
R134a
135 IFR
R1234yf velocities and angles in Table 2), cyclohexane shows the highest
IFG R245fa deflection level for the IFG configurations and the lowest for the
125
0,54 0,56 0,58 0,6 0,62 0,64 0,6 6 IFR ones. The hub to tip diameter ratio at rotor exit (d3h/d3s) is in
R is the 0.39–0.52 range and agrees with literature data [16,19]. At
rotor exit, the adoption of a diffuser for the partial recuperation
Fig. 17. Expander peripheral velocity at rotor inlet vs. isentropic degree of reaction
of the kinetic energy may be important only for fluids like cyclo-
(IFR and IFG).
hexane, which have high values of absolute Mach number in this
section. The total- to-total efficiency of the investigated expanders
and fluids ranges between 0.72 and 0.80. Generally, higher efficien-
of fluid density. Thus, in order to maintain the flow coefficient cies are achieved with expanders and fluids having lower velocities
unchanged, the peripheral velocity at rotor inlet increases. Finally, and deflections, higher expansion average density and lower
in order to keep the load coefficient constant, the total enthalpy expansion ratios, like R134a and R1234yf (see Table 2 and
drop of the expander is increased and, consequently, the related Fig. 18). The lowest values are instead found for hydrocarbons like
power output. In this way, by tuning b2/d2, the load coefficient cyclohexane and N-Pentane, which combine low average density
and the reaction degree, it is possible to move across the two dif- and high expansion ratios. The combined effects of these two
ferent configurations. Generally, backswept bladed expanders parameters lead to larger expanders, higher velocities and deflec-
show 1.5–2% better efficiency levels than the corresponding radial tions of flows, which, on the whole, increase the losses. Average
bladed ones. values of efficiency are found for R236fa. The comparison between
the velocity triangles of three representative different fluids is
6.2. Different working fluids shown in Fig. 18, for both IFG and IFR configurations. Finally, it is
important to remark that, for the specific size here considered,
A specific interpretation of the results is needed when consider- sub-atmospheric values of total pressure at rotor exit are not rec-
ing the important matter of expander design with different work- ommendable. Thus, cyclohexane and N-Pentane are critical from
528 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
Table 2
Design results of the ORC expander for different working fluids (IFR and IFG).
this point of view, in spite of their interesting efficiency levels in efficiency (0.4% < Dgts,v < 0.8%). The highest values are shown by
ORC cycles. On the basis of expander design features and cycle per- the cyclohexane, mainly due to the large rotor diameter and
formance, R245fa and R236fa represent thus the most interesting peripheral velocity.
options. The tip clearance losses, here referred to an average 3% clearance
fraction of blade height and to backswept configuration, give a rel-
6.3. Distribution of losses and efficiency evant contribution on the total losses, ranging from 10% of R1234yf
to 21% of Cyclohexane. Generally, higher tip clearance losses are
It is also important to analyze the distribution of the different shown by fluids having higher ratio between radial clearance and
losses through the expander, whose contribution to the overall inlet blade height (e/b2), in agreement with Eq. (39). The related
reduction of efficiency (Dg) is shown in Table 2. It is practically overall efficiency drop (Dgts,cl) is within the 2–6.5% range and
the same for the different investigated fluids and configurations. agrees with literature data [32,40].
The contribution of the stator losses to the overall losses ranges The highest relative contribution to the losses (60–70%) is given
between 2% of R134a and 12% of cyclohexane, both referred to by secondary flows in the rotor, which are due to the high blade cur-
radial geometry. Especially in IFR configurations, the stator losses vature and pressure gradient through the blade vanes. As sug-
have the highest incidence for hydrocarbons (mainly cyclohexane) gested by Rodgers [41], this contribution is shared between blade
and R245fa compared to the other investigated fluids. It is mainly loading and profile curvature. The former represent the highest
due to the high velocity in the nozzle (Table 2, see also the high contribution to overall losses, ranging from 27% to about 50% of
values of M2). Literature data [15] report that the stator losses gen- total. The largest values are found for R134a and 1234yf, which
erally range between 5% and 15% of the overall, which is in line have the lowest number of blades and thus the highest blade load-
with the results achieved with the here proposed model. ing. For this reason, cyclehexane shows the lowest relative contri-
Under design conditions, the incidence losses are negligible, as bution of blade loading losses, in agreement with Eq. (40). Their
the relative velocity angle at rotor inlet is the optimizing value cal- share to the overall efficiency reduction (Dgts,cp) is variable
culated by the (14)–(16) and (20) relationships. Their contribution between 8% and 13%.
to the overall reduction of efficiency Dgts,i ranges between 0.03% The contribution of profile losses on the entire turbine losses
and 0.2%. ranges from 12% of R134a to 27% of R1234yf. It is generally higher
The disk friction losses give a contribution within 2–4% to the for hydrocarbons, ranging from 20% to 26% and is attributable to
total and have a reduced relative influence on the expander the combined effects of rotor geometry (i.e. higher d3/d2), fluid
D. Fiaschi et al. / Applied Energy 138 (2015) 517–532 529
Fig. 18. Comparison of velocity triangles at rotor inlet/outlet for three different investigated fluids in IFR and IFG configurations.
properties (i.e. sound speed) and kinematic conditions (i.e. higher 0,9
Rohlik
relative velocities w2 and w3), in agreement with Eq. (41). These
0,85
losses are responsible for a total to static efficiency drop (Dgts,p)
variable from 3% to more than 8%.
0,8 R1234yf
The friction losses in the rotor also represent an important contri-
bution to the overall efficiency drop, reducing its value (Dgts,a) 0,75 R134a N-Pentane
from 0.9% to about 2.3% points. They are generally higher in cases
R236fa
ηts
mc/(mc)des
NC /(Nc)des=0,85
0,96
flowrate at the turbine inlet (the latter depends on pressure drop 0,94
through the expander), for different values of the corrected speed:
0,92
Fig. 21. Corrected flow rate–expansion ratio characteristic curves of the expander
– Blade height. (Referred to R134a).
– Stator and rotor inlet/outlet diameters.
– Passage section areas.
– Number of rotor and stator blades. 1,04
– Blade metal angles. 1,02 N /(N ) =0,70
C c des
1
Moreover, in the present case the rotor outlet total pressure is
maintained at the design value, because it is fixed by the condi- (η ts)/( η ts)des 0,98
NC /(Nc)des =0,85
tions at the condenser to satisfy the cogeneration conditions. Any- 0,96
way, with the developed model it is possible to let it be variable, in 0,94
order to take into account of the variable conditions at the con-
0,92
denser due to the timely change of the environmental conditions.
The loss correlations under off design conditions are the same 0,9
adopted in the expander design, as well as those relating flow 0,88 IFR
and metal angles. These allow the determination of the actual flow 0,86
NC /(Nc)des =1.0
IFG
angles, given the blade metal angles and the size of the expander,
0,84
both coming from the design procedure. 0,5 0,6 0,7 0,8 0,9 1 1,1
In order to analyze the off-design behavior of the turbo (p01/p03)/(p01/p03)des
expander, the inlet total pressure and temperature are changed.
Consequently, the mass flowrate is determined from the inlet– Fig. 22. Total to static–expansion ratio characteristic curves of the expander
(Referred to R134a).
outlet pressure drop. Another possible way of estimating the off-
design behavior is to fix the expander inlet mass flowrate and total
ity, the off design analysis is carried out for R134a only, without
temperature and calculate the total pressure. Thus, the perfor-
losing in generality, as the behavior is qualitatively similar for all
mance maps of the expander (i.e. characteristic curves) are
the investigated fluids. From Fig. 20, it is clear that the power out-
achieved by building the related curves under design conditions
put is reduced when the expansion ratio falls below the design
for different values of Nc. They may be directly adopted within
value. Moreover, it is important to notice that the power output
the thermodynamic code for the analysis of the ORC.
is reduced with reducing the corrected speed: this is the result of
The curves of power output, mass flowrate and efficiency vs. off
the lower enthalpy drop, due to the lower rotational speed, which
design pressure ratio at different values of corrected speed Nc are
is also responsible for the reduced variation of peripheral velocity
shown on Figs. 20–22 respectively, each one reporting the results
(i.e. the effective component for momentum). Finally, it should be
for both radial and backswept configurations. For the sake of brev-
remarked that the reduction of power output with respect to the
1,2 nominal value is larger than the reduction of the pressure ratio,
NC /(Nc)des =1.0
when Nc is at the design value: for example, when the expansion
1,1
ratio is at 90% of the design value, the power output is reduced
1 to about 88% of the design value. When the expansion ratio is at
0,9
NC /(Nc)des =0,85 80% of design value, the power output is about 75% of the nominal.
Power/Powerdes
0,8
The overall behavior is found for both IFG and IFR geometries. The
NC /(Nc)des =0,70 other typical turbomachinery off design parameter is the corrected
0,7
mass flowrate, defined as
0,6
0,5
pffiffiffiffiffiffiffi
_ T 01
m
_c¼
m ð49Þ
0,4 p01
IFR
0,3
IFG Starting from the values of the total inlet pressure and temper-
0,2 ature, from the maps of Fig. 21 it is possible to determine the mass
0,5 0,6 0,7 0,8 0,9 1 1,1 1,2
flow rate with variable expansion ratio for the different values of
(P01/P03)/(P01/P03)des
the ratio Nc/(Nc)des. It can be noticed that the radial bladed rotors
Fig. 20. Power output ratio–expansion ratio characteristic curve of the expander (IFR) are more influenced by operation under off-design pressure
(Referred to R134a). ratio than the backswept ones (IFG).
D. Fiaschi et al. / Applied Energy 138 (2015) 517–532 531
N/(T01)1/2
N/(T01)1/2
β2
β2
1 0.85 0.70 1 0.85 0.70
(a) (b)
Fig. 23. Velocity triangles at rotor inlet at variable corrected speed (a) radial blades and (b) backswept blades.
3,5 1,04
IFR
IFG
3 1,03
Incidence 1,02
2,5
(η ts)/( η ts)des
Skin friction
1,01
2 Stator
Δη ts %
Blade curvature
1
1,5
0,99
1
IFR
0,98
IFG
0,5
0,97
0,6 0,65 0,7 0,75 0,8 0,85 0,9 0,95 1 1,05
0 NC /(Nc)des
0,6 0,65 0,7 0,75 0,8 0,85 0,9 0,95 1 1,05
NC /(Nc)des
Fig. 25. Total-to-static efficiency vs. Nc under off design conditions.
Fig. 24. behavior of total to static efficiency losses (Dgts) with variable Nc under off
design conditions. Fig. 23). Hence, the shape of the velocity triangles is modified
towards the backswept configuration. This modification affects the
loss coefficients of the expander. Specifically, a reduction of Nc down
Fig. 22 shows the behavior of total to static efficiency
to 60–65% of the design value leads to an increase of the incidence
(expressed by the off-design to design ratio) as a function of the
losses and a reduction of the stator, friction and secondary flow
variable (off-design) expansion ratio. It is important to remark
losses, as reported in Fig. 24. This effect is due to the change of veloc-
the relatively low sensitivity to the expansion ratio for corrected
ity profiles within the rotor vanes. These considerations explain the
speeds below the design value, due to the typical accelerating
behavior of the ratio of the off-design to design total to static effi-
behavior of fluids into the expanders, which allows to work into
ciency (gts/(gts)des) vs. the corrected speed ratio (Nc/(Nc)des), shown
a relatively wide range of incidence angles with only a limited
in Fig. 25 for both radial and backswept configurations, at fixed
increase of loss coefficients. Hence, the reduction of Nc below the
design inlet total pressure and temperature. In both cases, gts is max-
design value could be regarded as a way to reduce the efficiency
imized at values of Nc/(Nc)des where the sum of losses is minimum.
drop of the expander at reduced values of the expansion ratio. As
it is seen, the investigated expanders show a relatively limited sen-
sitivity of the efficiency to off design expansion ratio, unless it is 8. Conclusions
reduced to very low values. It is interesting to notice the increasing
efficiency (1.5–2%) at corrected speed lower than the design value. The paper describes the features and analyzes the results of a
To better understand this apparently anomalous behavior, it is zero dimensional model for the design of high efficiency small size
important to analyze the variability of velocity triangles under ORC expanders. Two basic rotor blade geometries (radial IFR and
the three different values of corrected speed (Fig. 23), by consider- backswept IFG) and six different possible organic working fluids
ing a reduction of its value at fixed thermodynamic conditions of (R134a, R1234yf, R236fa, R245fa, Ciclohexane, N-Penthane) have
the inlet fluid. As the inlet–outlet expander pressure drop is fixed, been analyzed and discussed. In all cases, the rated power output
the flow rate remains practically unchanged. The lower variation of has been fixed at 50 kW. The reference thermodynamic data for
peripheral velocity leads to a reduction of stage and nozzle the specific application are taken from [37]. The relationships for
enthalpy drop (notice that R – 0). For this reason, the pressure at the estimation of the expander losses, as well as the main design
nozzle exit is higher, which leads to an increase of fluid density. parameters, have been collected by an extensive investigation of
As the mass flow rate is constant, the meridional component of models and experimental data available in literature for radial
absolute velocity must be reduced, thus the related velocity trian- turbo-expanders. Generally, literature for radial expanders is much
gle height decreases. When Nc is reduced, at fixed inlet total tem- less rich than for axial turbines and, often, data and models are
perature (T01), the rotational speed of the expander decreases derived from these last and from centrifugal compressor applica-
largely, and thus also the rotor peripheral velocity. The latter tions, with limited adaptments. Moreover, these relationships
entails an increase of relative flow angle at rotor inlet (b2) in the and models are referred to ideal gases and Mach relationships,
same direction (see the modified red1 and green triangles of which is also the approach often applied in many CFD calculation
codes. In the present work, the model applies the most recent,
1
For interpretation of color in Fig. 23, the reader is referred to the web version of currently available, equations of state of the investigated real fluids
this article. expanding into the ORC turbine.
532 D. Fiaschi et al. / Applied Energy 138 (2015) 517–532
For the investigated fluids, the rotor diameters are in the [9] Fiaschi D, Lifshitz A, Manfrida G. Fuel-assisted solar thermal power plants with
supercritical ORC cycle. In: Proceedings of ECOS 2010. Lausanne.
80–110 mm range and the rotational speed is variable between
[10] Di Pippo R. Geothermal power plants: principles, applications and case
30,000 and 50,000 rpm (specific speed is always below 0.1). The only studies. London, UK: Elsevier Advanced Technology; 2006.
exception is cyclohexane, which needs higher rotor diameters (190– [11] Heberle F, Brüggemann D. Exergy based fluid selection for a geothermal
200 mm) and a lower rotational speed. The designed expanders are organic rankine cycle for combined heat and power generation. Appl Therm
Eng 2006;30:1326–32.
mostly supersonic and have high values of nozzle exit angles [12] Lentz A, Almanza R. Solar–geothermal hybrid system. Appl Therm Eng
(a2 = 77–83.5°), whereas the deflection of flow in the rotor is 2006;26:1537–44.
between 40° and 90°. The outlet rotor hub to tip diameter ratio [13] Mueller L, Alsalihi Z, Verstraete T. Multidisciplinary optimization of a
turbocharger radial turbine. J Turbomach 2013;135(2). art. no. 021022.
resulting from the developed calculation code is within the 0.39– [14] Rahbar K, Mahmoud S, Al-dadah RK, Elsayed A. Modeling and CFD analysis of a
0.52 range, which agrees with literature data. In order to partially miniature radial turbine for distributed power. In: 5th International
recover the outlet kinetic energy of flow, the adoption of a diffuser conference on applied energy, 2013, South Africa; 2013.
[15] Whitfield A, Baines NC. Design of radial turbomachines. New York: Longman;
at rotor outlet is proposed, which is particularly recommendable 1990.
for the expansion of the cyclohexane, having the highest value of [16] Dixon SL. Fluid mechanics and thermodynamics of turbomachinery. Butterworth;
Mach at rotor discharge. 1998.
[17] Cohen H, Rogers GFC, Saravanamuttoo HIH. Gas turbine theory. New
The expected total-to-total efficiency of the designed units York: Longman; 1996.
ranges between 0.72 and 0.80, depending on the considered fluid [18] Glassman AJ. Turbine design and application. In: Scientifical and technical
and geometry configuration. The highest contribution to the information program-1994. Washington, DC: National Aeronautics and Space
Administration (NASA); 1994.
expander efficiency losses is given by the secondary flows within
[19] Rohlik Harold E. Analytical determination of radial inflow turbine design
the rotor (blade loading and profile curvature), due to the high cur- geometry for maximum efficiency. Washington, DC: National Aeronautics and
vature of blade profiles and the high pressure gradient. The best Space Administration (NASA); 1968.
expander efficiencies at design conditions are achieved with fluids [20] Meitner Peter L, Glassman Arthur J, Computer code for off-design performance
analysis of radial-inflow turbines with rotor blade sweep. National
coupling higher average density and lower volumetric expansion Aeronautics and Space Administration (NASA), Scientifical and Technical
ratios like R134a and R1234yf, followed by R236fa. The worst effi- Information Branch; 1983.
ciency is achieved by the hydrocarbons. [21] Whitfield A. The preliminary design of radial inflow turbines. ASME J
Turbomach 1990;112:51–7.
Generally, backswept bladed rotors (IFG) show 1.5–2.5% higher [22] Rohlik HE. Radial inflow turbines. NASA SP 290, vol. 3; 1975 [Chapter 10].
efficiencies. Moreover, they have larger values of the number of [23] Ainley DG, Mathieson GCR. A method of performance estimation for axial flow
blades, load coefficient, nozzle exit angle and rotor deflection angle turbines. ARC.R. and M.2974; 1951.
[24] Wiesner FJ. A review of slip factors in centrifugal impellers. J Eng Power
than the corresponding radial bladed ones. The peripheral speed 1967;89(4):558–72.
and the reaction degree of the backswept configurations are [25] Zweifel O. The spacing of turbomachine blading, especially with large angular
instead lower. Generally, the results of the design and parametric deflection. Brown Boveri Rev 1945;32:12.
[26] Rowland Benson S. A review of methods for assessing loss coefficients in radial
analysis are in agreement with literature data. The proposed calcu- gas turbines. Int J Mech Sci 1970;12:905–32. Pergamon Press.
lation model has been successively used to predict the off-design [27] Hiett GF, Johnston IH. In: Proceedings of the institution of mechanical
performance, through the construction of the expander character- engineers, vol.178(part 31); 1963–64.
[28] Balje OE. A contribution to the problem of designing radial turbomachines.
istic curves (corrected mass flowrate, power output and efficiency
Trans ASME 1952;741:451.
as functions of expansion ratio and corrected speed Nc). These last [29] Bridle EA, Boulter RA. A simple theory for the prediction of losses in the
can be directly introduced into the thermodynamic ORC calcula- rotors of inward radial flow turbines. In: Proceedings of the institution of
tion code to evaluate the off design behavior of the expander under mechanical engineers, conference proceedings 1964–1970 (vols. 178–184);
1967.
variable thermodynamic inlet conditions (total temperature and [30] Futral Samuel M, Jr., Wasserbauer Charles A. Off design performance
pressure and mass flow rate for different values of corrected prediction with experimental verification for a radial-inflow turbine. Ohio,
speed), once the geometry of the expander is defined. Washington, DC: Lewis Research Center Cleveland, National Aeronautics and
Space Administration; February 1965.
The calculation tool is open to improvements by the use of [31] Hiett GF, Johnston IH. Experiments concerning the aerodynamic performance
2D–3D CFD models and experimental tests on existing rotors, of inward flow radial turbines. In: Paper 7, Thermodynamics and fluid
which potentially could allow the improvement of the less conval- mechanics conv., proceedings of the institution of mechanical engineers, vol.
178(Pt 3 I (ii)), 28; 1963–64.
idated correlations, making it effective and reliable for the design [32] Futral Samuel M, Jr., Holeski Donald E. Experimental results of varying the
and off-design analysis of radial turbo-expanders for small-size blade-shroud clearance in a 6.02-inch radial-inflow turbine. Ohio,
ORC powerplants. Washington, DC: Lewis Research Center Cleveland, National Aeronautics and
Space Administration; January 1970.
[33] F-Chart Software. http://www.fchart.com/ees/; 2014.
References [34] Atlas Copco Gas and Process Division. Driving expander technology <http://
www.atlascopco-gap.com/download_file.php?id=457>; 2009.
[1] Invernizzi C, Iora P, Silva P. Bottoming micro-Rankine cycles for micro-gas [35] GE Energy Oil & Gas. turboexpanders/generators for naturalgas applications,
turbines. Appl Therm Eng 2007;27:100–10. <http://www.geoilandgas.com/businesses/ge_oilandgas/en/prod_serv/prod/
[2] Chacartegui R, Sánchez D, Muñoz JM, Sánchez T. Alternative ORC bottoming turboexpanders/en/downloads/turboexpanders.pdf>; 2005.
cycles FOR combined cycle power plants. Appl Energy 2009;86:2162–70. [36] Pratt&Whitney Power Systems. Organic Rankine Cycle Technology. <http://
[3] Al-Sulaiman FA, Dincer I, Hamdullahpur F. Exergy analysis of an integrated www.pw.utc.com/StaticFiles/Pratt%20&%20Whitney%20New/Media%20Center/
solid oxide fuel cell and organic Rankine cycle for cooling, heating and power Assets/1%20Static%20Files/Docs/pwps_orc_brochure_2010.pdf>; 2010.
production. J Power Sources 2010;195:2346–54. [37] Manfrida G, Tempesti D. Maximum exergy control of micro-CHP systems
[4] Vaja I, Gambarotta A. Internal combustion engine (ICE) bottoming with organic operating with geothermal and solar energy. In: ICAE 2011 conference,
rankine cycles (ORCs). Energy 2010;35:1084–93. Perugia; 2011.
[5] GE energy announces industrial waste-heat recovery innovation for onsite [38] Daniele Fiaschi, Giampaolo Manfrida, Francesco Maraschiello. Thermo-fluid
power plants <http://www.genewscenter.com/content/detail.aspx?releaseid= dynamics preliminary design of turbo-expanders for ORC cycles. Appl Energy
7229&newsareaid=2>, 6 July; 2009. 2012;97(September):601–8.
[6] Turboden combined heat and power orc units for the pellet industries; 2008 [39] Balje OE. Turbomachines. A guide to design, selection and theory. Wiley; 1981.
<http://www.turboden.eu/en/public/press/Turboden_ORC_for_pellets_english. [40] Deng QingHua, Niu JiuFang, Feng ZhenPing. Study on leakage flow
pdf>. characteristics of radial inflow turbines at rotor tip clearance. Science in
[7] Schuster A, Karellas S, Kakaras E, Spliethoff H. Energetic and economic China Series E, Technological Sciences, Springer; 2008.
investigation of organic rankine cycle applications. Appl Therm Eng [41] Rodgers C, Geiser T. Performance of a high-efficiency radial/axial turbine.
2009;29:1809–17. ASME J Turbomach 1987;109:151–4.
[8] Zhai H, Dai YJ, Wu JY, Wang RZ. Energy and exergy analyses on a novel hybrid [42] Idelc’ijk. Memento des pertes de charge. Eyrolles; 1978.
solar heating, cooling and power generation system for remote areas. Appl [43] Runstadler PW, Dolan FX, Dean RC. Diffuser data book. CREARE Inc; 1974.
Energy 2009;86:1395–404.