Performance and design optimization of a low-cost solar

organic Rankine cycle for remote power generation

S. Quoilin
a,
, M. Orosz
b
, H. Hemond
b
, V. Lemort
a
a
Thermodynamics Laboratory, University of Lie` ge, Campus du Sart Tilman, B49, B-4000 Lie` ge, Belgium
b
Department of Civil and Environmental Engineering, Massachusetts Institute of Technology (MIT), 15 Vassar St. 48-216, Cambridge, MA 02139, USA
Received 7 December 2010; received in revised form 10 February 2011; accepted 15 February 2011
Available online 16 March 2011
Communicated by: Associate Editor R. Petela
Abstract
Recent interest in small-scale solar thermal combined heat and power (CHP) power systems has coincided with demand growth for
distributed electricity supplies in areas poorly served by centralized power stations. One potential technical approach to meeting this
demand is the parabolic trough solar thermal collector coupled with an organic Rankine cycle (ORC) heat engine.
The paper describes the design of a solar organic Rankine cycle being installed in Lesotho for rural electrication purpose. The system
consists of parabolic though collectors, a storages tank, and a small-scale ORC engine using scroll expanders.
A model of each component is developed taking into account the main physical and mechanical phenomena occurring in the cycle and
based on experimental data for the main key components.
The model allows sizing the dierent components of the cycle and evaluates the performance of the system. Dierent working uids
are compared, and two dierent expansion machine congurations are simulated (single and double stage).
2011 Elsevier Ltd. All rights reserved.
Keywords: Organic Rankine cycle; ORC; Solar concentrator; Parabolic trough; Rural electrication; Concentrating solar power
1. Introduction
Concentrating Solar Power (CSP) systems have been
implemented with a variety of collector systems such as the
parabolic trough, the solar dish, the solar tower or the Fres-
nel linear collector. However, most of the currently installed
CSP plants use a steam Rankine cycle in the power block.
This technology requires a minimum power of a few MWe
in order to be competitive and involves high collector
temperatures.
Particularly in the case of small-scale systems, an organic
Rankine cycle (i.e. a Rankine cycle using an organic uid
instead of water) may show a number of advantages over
the steamcycle. These include a lower working temperature,
the absence of droplets during the expansion, the low main-
tenance requirements and the simplicity (fewer compo-
nents). According to McMahan (2006), those advantages
make the ORC technology more economically attractive
when used at small and medium power scales.
Solar ORCs have been studied both theoretically
(Davidson, 1977; Probert et al., 1983) and experimentally
(Monahan, 1976) as early as in the 1970s and with reported
overall eciencies varying between 2.52% and 7%. Experi-
mental studies usually involved the use of vane expanders
(Badr et al., 1985 multi-vane expanders, Davidson, 1977),
and high Ozone Depleting Potential (ODP) refrigerants
such as R11 or R13 were often used. Recent studies have
tended to emphasize optimization of uid selection for
dierent cycle architectures and collecting temperatures
(Wolpert and Riat, 1996; McMahan, 2006; Delgado-
Torres and Garcia-Rodriguez, 2007, 2010; Bruno et al.,
0038-092X/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2011.02.010

Corresponding author. Tel.: +1 32 4 366 48 22; fax: +1 32 4 366 48 12.

E-mail address: squoilin@ulg.ac.be (S. Quoilin).
www.elsevier.com/locate/solener
Available online at www.sciencedirect.com
Solar Energy 85 (2011) 955966
2008; Wang et al., 2010a). It is interesting to note, however,
that no single uid has been identied as optimal for the
ORC. This is mainly due to the strong interdependence
between the optimal working uid, the working conditions
and the cycle architecture. It follows that the study of the
working uid candidates should be integrated into the
design process of any ORC system.
Few studies have provided experimental data from oper-
ational solar ORC systems: Kane et al. (2003) studied the
coupling of linear Fresnel collectors with a cascaded 9-
kWe ORC, using R123 and R134a as working uids. An
overall eciency (solar to electricity) of 7.74% was
obtained, with a collector eciency of 57%. Manolakos
et al. (2007) studied a 2 kWe low-temperature solar ORC
with R134a as working uid and evacuated tube collectors:
an overall eciency below 4% was obtained. Wang et al.
(2010b) studied a 1.6 kWe solar ORC using a rolling piston
expander. An overall eciency of 4.2% was obtained with
evacuated tube collectors and 3.2% with at-plate collec-
tors. The dierence in terms of eciency was explained
by lower collector eciency (71% for the evacuated tube
vs. 55% for the plate technology) and lower collection
temperature.
Detailed models of such systems are also scarce in the
scientic literature: McMahan (2006) proposed a detailed
model and an optimization of the ORC cycle for solar
applications, but this model was not coupled to a solar col-
lector model; Forristall (2003) proposed a model of the
solar collectors validated with the SEGS plants data, inde-
pendent of a power cycle model. Jing et al. (2010) devel-
oped a model of an ORC cycle using R123 as working
uid and coupled to CPC collectors: the predicted overall
eciency was about 7.9% for a solar insolation of
800 W/m
2
and an evaporating temperature of 147 C.
Kane (2002) developed a model of a cascaded ORC using
scroll expanders and coupled to a collector model. This
model was used to conduct a thermoeconomic optimiza-
tion on the system.
Most of the above mentioned studies show that the
ORC eciency is signicantly improved by inclusion of a
Nomenclature
A area, m
2
c specic heat, J/(kg K)
D diameter (m)
FF
lling factor,
h heat transfer coecient, W/(m
2
K)
h specic enthalpy, J/(kg K)
k conductivity, W/m K
L length, m
M mass, kg
M mass ow rate, kg/s
n number of nodes
N
p
number of plates
N
rot
rotating speed, rpm
p pressure, Pa
pinch pinch point value, K
Q heat power, W
q linear heat ux, W/m
r ratio,
r
v,in
Internal built-in volume ratio,
S
beam
beam solar insolation, W/m
2
T temperature, C
U heat transfer coecient, W/(m
2
K)
v specic volume, m
3
/kg
V velocity, m/s
V
s
swept volume, m
3
V volume ow rate, m
3
/s
w specic work, J/kg
W width, m
Greek symbols
a absorptivity
e eectiveness
e emissivity
g eciency
q density, kg/m
3
q reectivity,
s transmittance
Subscripts and superscripts
abs absorber
amb ambient
cd condenser
col collector
eV evaporator
ex exhaust
exp expander
i relative to cell i
htf heat transfer uid
hx heat exchanger
l liquid
opt optical
p pressure
pp pump
rec recuperator
su supply
sf secondary uid
tp two-phase
tot total
v vapor
v volume
956 S. Quoilin et al. / Solar Energy 85 (2011) 955966
recuperator, of cascaded cycles, or of reheating (McMa-
han, 2006; Kane et al., 2003; Prabhu, 2006).
At present, only one commercial solar ORC power plant
is reported in the technical literature: the 1 MWe Saguaro
Solar ORC plant in Arizona, USA. This plant uses
n-pentane as working uid and shows an overall eciency
of 12.1%, for a collector eciency of 59% (Canada et al.,
2004).
If medium-scale solar ORCs are already commercially
available, work remains to be done for very small-scale
units (a few kWe), especially to reduce the specic invest-
ment costs and to control the system in order to avoid
the need of an on-site operator.
2. System description
Researchers at MIT and University of Lie`ge have col-
laborated with the non-governmental organization STG
International for the purpose of developing and imple-
menting a small-scale solar thermal technology utilizing
medium temperature collectors and an ORC. A rst unit
was installed by STG in 2007, and is shown in Fig. 1.
The goal is to provide rural areas of developing coun-
tries with a system that can be manufactured and assem-
bled locally (unlike PV collectors) and can replace or
supplement Diesel generators in o-grid areas, by generat-
ing clean power at a lower levelized cost (Orosz et al.,
2010).
At the core of this technology is a solar thermal power
plant consisting of a eld of parabolic solar concentrating
collectors and a vapor expansion power block for generat-
ing electricity. An electronic control unit is added for
autonomous operation as sub-megawatt scale plants can-
not justify the stang of operating personnel. Operating
at a lower cycle temperatures (<200 C) and Carnot e-
ciency is an example of a design tradeo for maintaining
low cost at small scales. For a given level of output power,
lower temperatures enable cost savings in the materials and
manufacture of the absorber units, heat exchangers, uid
manifolds and parabolic troughs.
Because no thermal power blocks are currently manu-
factured in the kilowatt range a small-scale ORC has to
be designed for this application. The design is based on
modied commercially available components e.g. HVAC
scroll compressors (for the expander), and industrial
pumps and heat exchangers. It should be noted that the
main challenge for ORC development is the high cost of
specially designed expandergenerator equipment. At pres-
ent, no volumetric expander is available on the market. In
order to reduce the cost of a practicable system, the expan-
der is obtained by adapting an o-the-shelf hermetic scroll
compressor to run in reverse, as proposed and successfully
tested by Lemort et al. (2009a). Scroll machines show the
advantage of being widely available, reliable and with a
limited number of moving parts (Zanelli and Favrat, 1994).
The goal of this paper is to design and dimension an
improved solar ORC unit to be installed in a rural clinic
in Berea District of Lesotho and to evaluate its perfor-
mance with dierent working uids. The main characteris-
tics of this unit are the following:
v Target net output power: 3 kWe
v Collector eld: 75 m
2
single-axis parabolic trough, using
Miro aluminum reectors and a Heat Collection Ele-
ment (HCE) with selective coating and air-lled annulus
between absorber pipe and glazing.
v ORC: One or two-stage expansion of R245fa using mod-
ied commercial HVAC compressors, brazed plate heat
exchangers for high pressure heat transfer, and commer-
cial HVAC tubes-and-ns air condenser for heat
rejection.
v Heat transfer uid (HTF): Monoethylene glycol (MEG)
with thermal buering in a thermal storage tank with a
2 m
3
packed bed of 19 mm quartzite.
The heat transfer uid is heated up in the collector eld
and driven to the evaporator by the heat transfer uid
pump. A thermal storage is installed in order to attenuate
the fast uctuations of solar irradiation during the day
and to maintain stable operation of the ORC engine (see
Fig. 2).
The main components in the classical Rankine cycle
include an evaporator, expander, a condenser and a recir-
culation pump. In an ORC with a dry uid, recuperation
from the superheated exhaust to the subcooled liquid is
typically achieved with a heat exchanger interposed
between the expander exhaust and the pump outlet. This
superheated exhaust is also readily exploitable for cogene-
ration, requiring an additional heat exchanger which can
be positioned in series with or parallel to the recuperator.
In the proposed system the cycle heat exchangers (evap-
orator, recuperator and condenser) are sized in order to
obtain the required pinch point and pressure drop. The
working uid is condensed in an air condenser in order
to avoid unnecessary water consumption (but at the Fig. 1. Solar ORC prototype installed by STG in Lesotho.
S. Quoilin et al. / Solar Energy 85 (2011) 955966 957
expense of non-negligible fan consumption) and then
repressurized in a piston pump The expansion process is
performed by one or two modied HVAC scroll machines
congured in series. Cogeneration is obtained with the
additional plate heat exchanger installed between the
expander and the recuperator in order to produce hot
water in addition to electricity, depending on the local
demand.
3. Modeling
In this section, a steady-state model of the system pre-
sented in Fig. 2 is developed, for the rating and sizing of
the dierent components and to optimize the working con-
ditions on a nominal point. The transient behavior of the
solar source is not taken into account here and an average
insolation is utilized. It is assumed that the storage is sized
in such a way to maintain almost constant heat transfer
uid ow rate and temperature during the operating time
of the system: the ORC engine is assumed to stand by in
case of insucient solar insolation for meeting temperature
requirements and in order to avoid part-load conditions
that might reduce the cycle eciency. In practice this
means that during periods of low insolation, the time to
charge the storage to the set operational point is longer
than the operating time of the ORC engine. In light of this
steady state hypothesis, the storage tank is not modeled.
The water heating heat exchanger is also neglected, since
the main goal of the model is to evaluate the electricity gen-
eration potential of the system.
The solar ORC is model within the EES environment
(Klein, 2010): a model is developed for each subcomponent
and included into a module. These modules are further
interconnected to obtain the global model of the system.
The proposed global model cannot be validated experi-
mentally because of the lack of experimental data. However,
it was shown in a previous publication that connecting vali-
dated component submodels in order to build a global ORC
model can lead to an acceptable overall error lower than 10%
compared to experimental data (Quoilin et al., 2010).
Since a nominal size (or power) must be set, the pro-
posed model is a hybrid between a simulation model and
a sizing model: on one hand, the design, the size and the
parameters of the collector are set according to the collec-
tor technology developed by STG International and
installed in Lesotho. On the other hand, the size of the
ORC cycle and of its components is recalculated by the
model in order to obtain a good match between collector
power and ORC engine power.
3.1. Parabolic trough model
The trough module, largely adapted from Forristall
(2003), is a one-dimensional energy balance model around
a Heat Collection Element (HCE) of user specied dimen-
sions and materials: Radiation impinges on a reector ele-
ment with user-input focal length, reective coecient, and
aperture. The energy is correspondingly reduced (e.g. due
to a reective coecient <1) and concentrated onto a nodal
area of the HCE, where it is transmitted through a glass
envelope and a gas annulus, and nally absorbed or
reected at the surface of the HCE.
Depending on the absorptivity and emissivity character-
istics of the selective coating and the temperature of the
HTF owing through the HCE at a given node, some
amount of absorbed energy is transferred through the
HCE wall into the HTF (process 21 in Fig. 3) with a heat
exchange coecient calculated from the uid thermal prop-
erties and ow regime parameters. The remaining absorbed
heat is lost at the HCE outer surface, via convection and
radiation back through the annulus (process 34), conduc-
tion through the envelope (process 45), radiation between
Fig. 2. Conceptual scheme of the solar ORC.
Fig. 3. Heat transfer in the absorber.
958 S. Quoilin et al. / Solar Energy 85 (2011) 955966
the envelope and the sky (process 57) and convection to
the ambient air (process 56).
This process is repeated for each node, where the input
for each node is the output of the previous node, resulting
in an overall enthalpy and temperature gain for the focal
line length specied by the user. The collector module thus
derives a thermal eciency and outputs a heat ux and
temperature gain for the HTF at the user specied ow rate
and initial temperature.
The dierent heat transfer relations used to compute the
heat ows are provided in Table 1.
The amounts of radiation absorbed by the collector and
by the glass enveloped are respectively given by (Forristall,
2003):
q
sun;3
= q
sun
g
opt
s
env
a
coating
(1)
q
sun;5
= q
sun
g
opt
a
env
(2)
where q
sun
= S W
col
is the linear beam insolation.
And
g
opt
= q
mirror
g
shadowing
g
geomery
g
unaccounted
(3)
g
opt
is the optical eciency, lumping dierent sources of
losses such as mirror reectivity (q
mirror
), tracking losses
(g
tracking
), shadowing (g
shadowing
), geometrical eects
(g
geometry
and other unaccounted losses (g
unaccounted
). The
values of the dierent eciencies are the ones recom-
mended by Forristall (2003), except for g
tracking
and
g
geometry
, where a signicantly lower eciency is selected.
This conservative hypothesis is made in order to account
for the relatively lower optical intercept factor (~0.9)
resulting from the low-cost design of the collector, which
could reduce the performance of the system.
In order to reduce the magnitude of q
34,rad
, a selective
coating is applied on the collector tube, maximizing the
solar absorptivity and minimizing the infra-red emissivity.
This emissivity is calculated according to Forristall recom-
mendation for a Solar UAV cermet coating:
e
coating
= 2:249 10
7
T
2
3
1:039 10
4
T
3
5:599 10
2
(4)
For each cell of the discretized collector (having a length
equal to the total focal line length divided by the number
of nodes) the dierent energy balance equations can be
applied:
T
ex;cell
= T
su;cell

q
12
Dx
M
htf
cp
htf
(5)
q
45
= q
34;conv
q
34;rad
(6)
q
56;conv
= q
45
q
sun;5
q
57
(7)
q
12;conv
= q
23
(8)
q
23
q
34;conv
q
34;rad
q
sun;3
= 0 (9)
Moreover, a pressure drop in the heat transfer uid can be
computed in each cell using the following equation:
Dp
cell
=
f DL G
2
htf
2 D
2
q
(10)
where f is the friction factor, calculated with the Gnielinski
correlation (Incropera and Dewitt, 2002).
The dierent parameters used for the modeling of the
solar collected are summarized in Table 2.
3.2. Evaporator model
The evaporator is modeled by means of the Logarithmic
Mean Temperature Dierence (LMTD) method for
Table 1
Absorber heat transfer models.
Heat transfer type Heat transfer law
12 Convection Gnielinski correlation for turbulent heat transfer in pipes (Incropera and Dewitt, 2002)
q
23
= 2pk
23
T2T3
lnD3=D2
23 Conduction k
23
is the assumed-to-be-constant steel conductivity: k
23
= 50 W/mK
Convection Hollands correlation for natural convection in an annular space between horizontal cylinders (Hollands et al., 1975).
34 Radiation q
34;rad
=
rpD3(T
4
3
T
4
4
)
1
e
3

1e
4
D
3
e
4
D
4
45 Conduction q
45
= 2pk
45
T4T5
ln D5=D4
56 Free convection Churchill and Chu correlation for laminar convection from a horizontal cylinder (Incropera and Dewitt, 2002).
56 Forced convection Zhukauskas correlation for external forced convection ow normal to an isothermal cylinder (Incropera and Dewitt, 2002)
57 Radiation q
57;rad
= rpD
5
T
4
5
T
4
7
T
7
is taken 8 below the ambient temperature.
Table 2
Collector model parameters.
Parameter Description Value
q
mirror
Mirror reectivity 0.94
g
tracking
Tracking error 0.92
g
shadowing
Shadowing error 0.98
g
geometry
Geometry error 0.93
g
unaccounted
Unaccounted losses 0.96
D
2
Absorber tube inner diameter 66 mm
D
3
Absorber tube outer diameter 70 mm
D
4
Envelope inner diameter 80 mm
D
5
Envelope outer diameter 88 mm
L
col
Total collector length 46 m
W
col
Collector width 2.5 m
s
env
Envelope transmissivity 0.96
a
env
Envelope absorptivity 0.04
a
coating
Coating absorptivity 0.96
e
env
Envelope emissivity 0.86
N Number of node 15
S. Quoilin et al. / Solar Energy 85 (2011) 955966 959
counter-ow heat exchangers. The heat exchanger is subdi-
vided into three moving-boundaries zones, each of them
being characterized by a heat transfer area A and a heat
transfer coecient U (Quoilin et al., 2010).
The heat transfer coecient U is calculated by consider-
ing two convective heat transfer resistances in series (sec-
ondary uid and refrigerant sides).
1
U
=
1
h
r

1
h
sf
(11)
The total heat transfer area of the heat exchanger is given
by:
A
tot
= A
l
A
tp
A
v
= (N
p
2) L W (12)
N
p
being the number of plates, L the plate length and W
the plate width.
3.2.1. Single-phase
Forced convection heat transfer coecients are
evaluated by means of the non-dimensional relationship:
Nu = CRe
m
Pr
n
(13)
where the inuence of temperature-dependent viscosity is
neglected.
The parameters C, m and n are set according to Tho-
nons correlation for corrugated plate heat exchangers
(Thonon et al., 1995).
The pressure drops are computed with the following
relation:
Dp =
2 f G
2
q D
h
L (14)
where f is the friction factor, calculated with the Thonon cor-
relation, G is the mass velocity (kg/s m
2
), q is the mean uid
density, D
h
is the hydraulic diameter and Lis the plate length.
3.2.2. Boiling heat transfer coecient
The overall boiling heat transfer coecient is estimated
by the Hsieh correlation, established for the boiling of
refrigerant R410a in a vertical plate heat exchanger. This
heat exchange coecient is considered as constant during
the whole evaporation process and is calculated by (Hsieh
and Lin, 2003):
h
tp
= Ch
l
Bo
0:5
(15)
where Bo is the boiling number and h
l
is the all-liquid non-
boiling heat transfer coecient.
The pressure drops are calculated in the same manner as
in Eq. (14), using the Hsieh correlation for the calculation
of the friction factor.
3.2.3. Heat exchanger sizing
For a given corrugation pattern (amplitude, chevron
angle, and enlargement factor), two degrees of freedom
are available when sizing a plate heat exchanger: the length
and the total ow width. The total ow width is given by
the plate width multiplied by the number of channels:
W
tot
= W
hx

N
p
1
2
(16)
The two degrees of freedom are xed by the heat exchange
area requirement and the limitation on the pressure drop
on the working uid side:
v Increasing the total width decreases the Reynolds num-
ber. This leads to a lower pressure drop and to a higher
required heat transfer area, since the heat transfer coef-
cient is also decreased.
v Increasing the plate length leads to a higher pressure
drop.
Therefore, by imposing a pinch point and a pressure
drop, it is possible to dene the total width and the length
of the plate heat exchanger. The ow chart of the sizing
process is shown in Fig. 4.
The imposed parameters of the evaporator model are
presented in Table 3.
3.3. Recuperator model
The recuperator model is similar to the evaporator
model, with one zone (single-phase) instead of three. The
Fig. 4. Plate heat exchangers sizing process.
960 S. Quoilin et al. / Solar Energy 85 (2011) 955966
inputs of the model are the maximum pressure drop and
the heat exchanger eciency, which allows sizing the
exchanger in terms of total width and length.
3.4. Expander model
Volumetric expanders, such as the scroll, screw or recip-
rocating technologies present an internal built-in volume
ratio (r
v,in
) corresponding to the ratio between the inlet
pocket volume and the outlet pocket volume. This can gen-
erate two types of losses if the system specic volume ratio
is not equal to the expander nominal volume ratio:
Under-expansion occurs when the internal volume ratio
of the expander is lower than the system specic volume
ratio. In that case, the specic volume in the expansion
chambers at the end of the expansion process (P
in
) is lower
than the specic volume in the discharge line.
Likewise over-expansion occurs when the internal vol-
ume ratio imposed by the expander is higher than the sys-
tem specic volume ratio.
These two eects can considerably reduce the eciency
of the expansion process. Other sources of losses include
friction losses, supply pressure drop, internal leakage and
heat transfers (Lemort et al., 2009a,b).
Since no o-the-shelf small-scale expansion machine is
currently available on the market, the expander was
obtained by modifying a scroll compressor to make it run
in reverse. This allows keeping the expander cost low, her-
metic compressors being very common components in
HVAC applications. On the other hand, since the device
is not optimized for expander applications, experimental
results by Lemort et al. showed that the eciency is
reduced by about 10% when working in expander mode
(about 60% eciency) compared to the compressor mode
(typically 70%). In this work, the considered expander is
a hermetic scroll expander tested and modeled by Lemort
et al. (2009b).
A semi-empirical thermodynamic model such as the one
proposed by Lemort et al. is not suitable for the purpose of
this work since it was developed for one machine in partic-
ular. Here, a sizing model is needed, that can predict the
performance of scroll expanders with very dierent swept
volumes.
If ambient heat losses are neglected, scroll expanders can
be modeled by their isentropic eciency and by their lling
factor, respectively dened by (Lemort et al., 2009b):
e
exp
=
W
exp
M (h
su; exp
h
ex; exp;s
)
(17)
and
FF =
60 M
q
su
V
s
N
rot
(18)
where V
s
is the swept volume of the expander and N
rot
its
rotational speed (assumed-to-be-constant at 3000 rpm).
where W
exp
is the electrical power generated by the expan-
der and h
ex,exp,s
is the isentropic exhaust enthalpy. It should
be noted that the given eciency is an electrical and not a
mechanical isentropic eciency (i.e. electromechanical
losses in the generator are accounted for).
In order to simulate realistic performance close to the
actual experimental data, the model developed by Lemort
et al. (2009b) is used to express e and FF as a polynomial
law of the main working conditions. The two selected
working conditions are the uid inlet density q
su
and pres-
sure ratio over the expander r
p
since they turned out to be
the two main representative variables of the working con-
ditions. The polynomial ts are expressed in the following
form:
e =
X
n1
i=0
X
n1
j=0
a
ij
ln(r
p
)
i
q
j
su
a
0n
ln(r
p
)
n
a
0n
q
n
1
= f (r
p
; q
su
) (19)
For e, a 4th-order (n = 4) polynomial t is used, while for
FF a second-order (n = 2) polynomial t turned out to be
sucient. The correlations have been established on the ba-
sis of the validated model for 800 dierent working points
inside of the following operating conditions:
30 < q
su
< 200; 1:2 < r
p
< 12
The values of e and FF were respectively predicted by the
polynomial t with R
2
= 99.98% and R
2
= 99.96%.
It is assumed that, when changing the scale of the expan-
der (and thus the swept volume), the isentropic eciency
and the lling factor remain similar if the pressure ratio
and the inlet density are kept equal.
3.4.1. Double-stage expander
As mentioned above, volumetric expanders are opti-
mized for a given specic volume ratio. It appears that
the specic volume ratios involved in refrigeration for
which the scroll compressors are designed is typically much
lower than the specic volume ratios involved in ORC
cycles. When the expander is obtained from a modied
scroll compressor, under-expansion losses can therefore
become prohibitive for high specic volume ratios (typi-
cally higher than 10). A possible solution consists in using
two expanders assembled in series, as shown in Fig. 5.
When sizing a double-stage expander, it is important to
dene carefully the two swept volumes in order to optimize
the intermediate pressure p
2
. If the expander eciency was
only dependent on the pressure ratio, the optimal single-
stage pressure ratio would be dened as the square root
of the overall pressure ratio (r
p;1
= r
p;2
=

r
p
_
) However,
the eciency also depends on the ow rate going through
Table 3
Evaporator model parameters.
Parameter Description Value
D
h
Hydraulic diameter 2 mm
b
chevron
Chevron angle 45
S. Quoilin et al. / Solar Energy 85 (2011) 955966 961
the expander because a higher ow rate entails a higher
output power and makes the constant losses (e.g. friction
losses) relatively smaller. In the polynomial correlations,
the inuence of the ow rate is reected by the dependence
in terms of supply vapor density.
In order to determine the optimal rst-stage pressure
ratio, the overall isentropic eciency is maximized using
the following equation:
de
dr
p;1
=
d
dr
p;1
h
1
h
3
h
1
h
3s

= 0 (20)
This can be done numerically or analytically. For the latter
solution, e must be expressed in terms of r
p,1
, which can be
achieved using the ideal gas hypothesis.
3.5. Condenser model
Since air condensers are well-known components in
HVAC applications, a simplied model based on manufac-
turer data (Witt, 2004) is used to compute the condenser
performance and fan consumption.
The two inputs are the pinch point, dened as the dier-
ence between the condensing temperature and the ambient
temperature, and the condensing power.
Special attention is paid to the fan consumption since it
can amount for a non-negligible share of the generated
power. The fan consumption is computed as a function
of the heat transfer power and of the pinch point with
the following relation:
W
fan;cd
= 54:5 0:0185 Q
cd

8:333
pinch
cd
(21)
3.6. Pumps model
Two pump consumptions are taken into account: the
heat transfer uid pump and the working uid pump. They
are modeled by their isentropic eciency, dened by
(Quoilin et al., 2010):
e
pp
=
v
su;pp
(p
ex;pp
p
su;pp
)
h
ex;pp
h
su;pp
(22)
For the HTF pump, the pressure dierence is given by the
sum of the pressure drops in the evaporator and in the
collector.
A constant, realistic value of 70% is assumed for the
pump eciency (Lin, 2008).
3.7. Cycle model
The global model of the system is obtained by intercon-
necting each subcomponent model according to Fig. 6.
Several performance indicators can be dened.
The overall collector eciency:
g
col
=
M
htf
cp
htf
(T
htf;ex;col
T
htf;su;col
)
S
beam
L
col
W
col
(23)
The net electrical output power:
W
net
= W
exp
W
pp
W
fans
W
pp;htf
(24)
The ORC cycle eciency:
g
ORC
=
W
net
Q
eV
(25)
The overall system eciency:
g
overall
=
W
net
S
beam
L
col
W
col
= g
col
g
ORC
(26)
In this work, the pinch points are set to 8 K, the super-
heating at the expander inlet is set to 10 K, the subcooling
at the condenser outlet is set to 5 K, and the maximum
pressure drop on the refrigerant side of each heat exchan-
ger is set to 75 mbar.
4. System performance and uid comparison
This section aims at understanding the inuence of
dierent cycle parameters on the system and to compare
Fig. 5. Two-stage expander.
Fig. 6. Global model parameters, inputs and outputs.
962 S. Quoilin et al. / Solar Energy 85 (2011) 955966
several working uids and cycle architectures. For that
purpose, nominal ambient conditions are imposed and
are kept constant for all the simulations performed below:
T
amp
= 15

C
p
amp
= 0:83 bars
S
beam
= 800 W=m
2
V
wind
= 2 m=s
These conditions are typical of the mid-season or winter
time conditions in the highlands of Lesotho.
Three main degrees of freedom are available to control
the working conditions of the cycle: the heat transfer uid
ow rate, the working uid ow rate and the expander
swept volume or rotational speed. Setting the working uid
ow rate and the expander speed allows dening the evap-
orating temperature and the superheating (Quoilin et al.,
2010). Setting the heat transfer uid ow rate allows den-
ing the temperature glide in the collector. According to
Yamamoto et al. (2001), the superheating should be main-
tained as low as possible when using high molecular weight
working uids. The two remaining degrees of freedom
(evaporating temperature and collector temperature glide)
can be determined optimally, as shown in the next sections.
4.1. Inuence of the temperature glide in the collector
Modifying the heat transfer uid ow rate entails two
main antagonist eects: on the one hand, the overall tem-
perature level in the collector is modied (Fig. 7), which
will impact its thermal eciency via the various heat loss
mechanisms.
On the other hand, changing the uid ow rate aects
the heat transfer coecient between the heat transfer uid
and the absorber, which also impacts the collector
eciency.
Fig. 8 shows that the second eect is predominant: For
very low temperature glides, the overall eciency is
reduced because of the very high HTF uid pumping
consumption. For high temperature glides, the overall
eciency is lowered by the low heat transfer coecient in
the collector. An optimum is obtained for a temperature
glide of 15 K, corresponding to a heat transfer uid ow
rate of 1.2 kg/s.
In the following parametric studies, the temperature
glide value will always be set to its optimal value for each
computed working point. This value is obtained with the
Golden Section Search algorithm.
4.2. Inuence of the evaporating pressure
The selection of the optimal evaporating temperature
results in a tradeo between collector eciency and cycle
eciency. In the particular case of an ORC using volumet-
ric expanders, increasing the evaporation temperature also
increases the under-expansion losses and reduces the cycle
eciency, which constitutes an additional inuence.
The goal of this section is to illustrate the inuence of
the evaporating temperature on dierent cycle parameters
and performance indicators. For that purpose, an arbitrary
working point is selected: the selected working uid is
R245fa, with a two-stage expander, an optimized heat
transfer uid temperature glide, a superheating of 5 K,
and a subcooling of 5 K.
Fig. 7. Ts diagram of the ORC process.
Fig. 8. Inuence of the HTF temperature glide.
Fig. 9. Inuence the evaporation temperature on the performance.
S. Quoilin et al. / Solar Energy 85 (2011) 955966 963
As shown in Fig. 9, increasing the evaporating tempera-
ture leads to higher cycle eciency and to lower collector
eciency. An optimal overall eciency is stated around
150 C, which is just below the critical point (154 C for
R245fa).
The evaporating temperature also has an impact on the
size of the dierent components. Fig. 10 shows that with
high evaporating temperature levels, smaller swept volumes
are needed for both expanders since the inlet densities are
higher. This is an appreciable advantage since the cost of
the expanders is reduced.
A similar eect is stated for the heat transfer area of the
evaporator (Fig. 11): for a given pressure drop, a higher
vapor density allows reducing the passage area, which in
turn reduces the required area. Fig. 11 also shows that a
modication of the evaporating temperature has a very
limited eect on the required recuperator area. Although
those calculations were performed for R245fa, a similar
behavior is stated for alternative working uids.
4.3. Working uid and architecture comparison
In order to compare a reasonable amount of working
uids, a pre-screening is performed with the following
conditions:
v The critical point of the working uid should be similar
to the target temperature range (100200 C).
v The working uid should be a well known-uid in state-
of-the-art ORC applications or in the scientic literature
dealing with working uid selection (see Quoilin and
Lemort (2009) for a review). This criterion ensures that
several conditions such as the toxicity, the cost or the
ammability are fullled.
v The working uid should have a null Ozone Depleting
Potential (ODP) in order to avoid the phasing-out of
the Montreal Protocol.
Four uids have been selected using this methodology:
R134a, R245fa, Solkatherm (SES36) and n-pentane.
One of the main inuences of the working uid on the
cycle architecture lies in the specic volume ratio: generally
speaking, the higher the critical temperature, the higher the
specic volume ratio over the expander. The scroll expand-
ers considered in this work are designed for a volume ratio
close to 3. As discussed above, when used at much higher
Fig. 10. Required swepts volumes vs. evaporating temperature.
Fig. 11. Required heat transfer area vs. evaporation temperature.
Fig. 12. Specic volume ratio vs. evaporation temperature.
Fig. 13. Overall eciency for dierent working uids.
964 S. Quoilin et al. / Solar Energy 85 (2011) 955966
specic volume ratios, their eectiveness is reduced.
Fig. 12 shows that the specic volume ratio remains accept-
able for single-stage expansion only for R134a and R245fa
(at low evaporating temperature). In the simulations,
R134a is therefore used with single-stage expansion archi-
tecture. Solkatherm and n-pentane are simulated with a
double-stage expansion. R245fa is simulated with both
architectures.
Fig. 13 shows the overall eciency of the system with
the four dierent uids. A maximum appears in terms of
evaporating temperature when the single-stage architecture
is selected. This is explained by the very high under-
expansion losses that reduce the expander eectiveness at
high evaporating temperature. On the contrary, when using
a two-stage expansion, the eciency is limited by the crit-
ical temperature or by unrealistic working conditions such
as very high specic volume ratios.
Solkatherm is the uid showing the highest eciency,
with a maximum close to 8%. It should however be noted
that refrigeration compressors are not designed for temper-
atures higher than 150 C, which might reduce their life-
time. If this limit is applied, the maximum overall
eciency is 7.5% for Solkatherm and 7% for R245fa.
Additional parameters to the sole eciency must be taken
into account when comparing working uids. Table 4 shows
the more relevant cycle parameters for a fewselected optimal
points. The bold characters indicate the most advantageous
value for each column. If Solkatherm is the most ecient
uid, it is also the one requiring the biggest expander, with
a suction swept volume (in expander mode) of 180.7 cm
3
for the second stage of expansion and an evaporating tem-
perature of 150 C. R245fa on the contrary shows very
advantageous swept volumes, which could reduce the cost
of the system. n-Pentane must be run at very high tempera-
ture to show a good eciency. Its required evaporator area
is advantageous, but the required recuperator area is very
high due to the low density and the very high pressure drops
in the low-pressure vapor.
4.4. Inuence of the working conditions
The developments proposed above were conducted for
nominal conditions, dened in Section 4. However, the
selection of these working conditions can have a non-neg-
ligible inuence on the simulation results.
A parametric study is therefore performed to evaluate
the inuence of the nominal working conditions on the
overall eciency: this study is performed for the SES36
working uid and an evaporating temperature imposed at
150 C (third line in Table 4). Fig. 14 shows the inuence
of the wind speed, of the ambient temperature and of the
solar beam insolation on the system performance. The
inuence of the wind speed is straightforward: the higher
the speed, the lower the overall eciency since the heat
transfer coecient from the collector to the ambient is
increased. The same trend is stated for the beam insolation:
a higher value makes the ambient losses of the collector
proportionally smaller, and the overall eciency is
increased. The ambient temperature inuences the cycle
performance in two dierent ways: the ambient heat losses
of the collector are increased with a lower ambient temper-
ature, and the cycle eciency is increased because of a
lower condensing temperature. Fig. 14 shows that this sec-
ond inuence is predominant: for a 330 C evolution of
the ambient temperature, the collector eciency is
increased by 2%, while the ORC cycle eciency is
decreased by 15%, resulting in a 13% decrease of the overall
eciency.
5. Conclusions
Small-scale solar Organic Cycles are well adapted for
remote o-grid areas of developing countries. Compared
to the main competitive technology, the PV collector, Solar
Table 4
Simulations results for the dierent working uids.
T
ev
(C) DT
HTF
(C) g
col
(%) g
ORC
(%) g
overall
(%) e
exp
(%) V
s,1
(cm
3
) V
s,2
(cm
3
) A
ev
(m
2
) A
rec
(m
2
)
n-Pentane 189 31.6 59.1 11.9 7.0 47.2 22.2 98.9 0.95 3.53
SES36 169 19.5 60.4 13.1 7.9 54.9 27.1 137.3 1.1 1.24
SES36 150 14.0 61.6 12.3 7.5 55.0 44.9 180.7 1.71 1.29
R245fa 150 22.9 61.6 11.2 6.9 58.7 20.8 92.8 1.48 2.54
R245fa 109 12.8 63.9 7.7 4.9 50.3 59.9 0 4.02 2.73
R134a 85 17.1 65.1 5.6 3.6 59.7 37.2 0 2.7 1.53
Fig. 14. Inuence of the working conditions on the eciency.
S. Quoilin et al. / Solar Energy 85 (2011) 955966 965
ORCs have an advantage of being manufacturable locally.
They are also more exible and allow the production of hot
water as a by-product.
This work focused on the evaluation of the thermody-
namic performance of the system. With conservative
hypotheses, and real expander eciency curves, it was
shown that an overall electrical eciency between 7%
and 8% can be reached. This eciency is a steady-state e-
ciency at a nominal working point. In order to evaluate the
yearly energy output, a dynamic model is needed. In partic-
ular, the behavior of the storage tank should be modeled to
perform a 1-year simulation.
It should be noted that these calculations were per-
formed for o-the-shelf components, especially the expan-
der, whose combined electro-mechanical eciency did not
exceed 60%. Components specically developed for the tar-
get applications (e.g. a high volume ratio expander, opti-
mized for the ORC working uid) could signicantly
increase the system performance.
The comparison between working uids showed that the
most ecient uid is Solkatherm. However, it is also the
uid requiring the highest expander swept volumes, which
increases the cost of the system. R245fa also shows a good
eciency and has the advantage of requiring much smaller
equipment.
Even though part-load conditions were not simulated in
the present work, the proposed model allows computing
the performance of the system for a wide range of working
and ambient conditions.
References
Badr, O., Probert, S.D., OCallaghan, P., 1985. Performances of multi-
vane expanders. Applied Energy 20, 207234.
Bruno, J.C., Lopez-Villada, J., Letelier, E., Romera, S., Coronas, A.,
2008. Modelling and optimisation of solar organic rankine cycle
engines for reverse osmosis desalination. Applied Thermal Engineering
28, 22122226.
Canada, S., Cohen, G., Cable, R., Brosseau, D., Price, H., 2004. Parabolic
Trough Organic Rankine Cycle Solar Power Plant. In: DOE Solar
Energy Technologies, Denver, Colorado.
Davidson, T.A., 1977. Design and Analysis of a 1 kw Rankine Power
Cycle, Employing a Multi-vane Expander, For Use with a Low
Temperature Solar Collector, 61 leaves (61 fold.).
Delgado-Torres, A.M., Garcia-Rodriguez, L., 2007. Comparison of solar
technologies for driving a desalination system by means of an organic
Rankine cycle. Desalination 216, 276291.
Delgado-Torres, A.M., Garcia-Rodriguez, L., 2010. Analysis and optimi-
zation of the low-temperature solar organic Rankine cycle (ORC).
Energy Conversion and Management 51, 28462856.
Forristall, R., 2003. Heat Transfer Analysis and Modeling of a Parabolic
Trough Solar Receiver Implemented in Engineering Equation Solver.
NREL Technical Reports.
Hollands, K.G.T., Raithby, G.D., Konicek, L., 1975. Correlation
equations for free convection heat transfer in horizontal layers of air
and water. International Journal of Heat and Mass Transfer 18, 879
884.
Hsieh, Y.Y., Lin, T.F., 2003. Evaporation heat transfer and pressure drop
of refrigerant R-410A ow in a vertical plate heat exchanger. Journal
of Heat Transfer Transactions of the ASME 125, 852857.
Incropera, F.P., DeWitt, D.P., 2002. Fundamentals of Heat and Mass
Transfer, fth ed. Wiley, New York.
Jing, L., Gang, P., Jie, J., 2010. Optimization of low temperature solar
thermal electric generation with organic Rankine cycle in dierent
areas. Applied Energy 87, 33553365.
Kane, M., 2002. Integration et optimisation thermoeconomique &
environomique de centrales thermiques solaires hybrides. Ecole
polytechnique Federale de Lausanne, Lausanne.
Kane, M., Larrain, D., Favrat, D., Allani, Y., 2003. Small hybrid solar
power system. Energy 28, 14271443.
Klein, S.A., 2010. Engineering Equation Solver, F-Chart Software,
Middleton, WI.
Lemort, V., Quoilin, S., Cuevas, C., Lebrun, J., 2009a. Testing and
modeling a scroll expander integrated into an organic Rankine Cycle.
Applied Thermal Engineering 29, 30943102.
Lemort, V., Quoilin, S., Pire, C., 2009b. Experimental investigation on a
hermetic scroll expander. In: 7th International Conference on Com-
pressors, Papiernicka.
Lin, C., 2008. Feasibility of using power steering pumps in small-scale
solar thermal electric power systems, S.B. Thesis, Dept. of Mechanical
Engineering, MIT.
Manolakos, D., Papadakis, G., Kyritsis, S., Bouzianas, K., 2007.
Experimental evaluation of an autonomous low-temperature solar
Rankine cycle system for reverse osmosis desalination. Desalination
203, 366374.
McMahan, A., 2006. Design and Optimization of Organic Rankine Cycle
Solar Thermal Powerplants. MSc. Thesis, University of Wisconsin,
Madison.
Monahan J., 1976. Development of a 1-kW, Organic Rankine Cycle
Power Plant for remote applications. In: Intersociety Energy Conver-
sion Engineering Conference, New York.
Orosz, M.S., Quoilin, S., Hemond, H., 2010. SORCE: A design tool for
solar organic Rankine cycle systems in distributed generation appli-
cations. In: International Scientic Conference on Solar Heating,
Cooling and Buildings.
Prabhu, E., National Renewable Energy Laboratory (US), 2006. Solar
Trough Organic Rankine Electricity System (STORES). Stage 1.
Power Plant Optimization and Economics November 2000May 2005.
National Renewable Energy Laboratory, Golden, Colo.
Probert, S.D.H.M., OCallaghan, P.W., Bala, E., 1983. Design optimisa-
tion of a solar-energy harnessing system for stimulating an irrigation
pump. Applied Energy, 15.
Quoilin, S., Lemort, V., 2009. Technological and economical survey of
organic Rankine cycle systems. In: 5th European Conference on
Economics and Management of Energy in Industry, Vilamoura,
Portugal.
Quoilin, S., Lemort, V., Lebrun, J., 2010. Experimental study and
modeling of an organic Rankine cycle using scroll expander. Applied
Energy 87, 12601268.
Thonon, B., Vidil, R., Marvillet, C., 1995. Recent research and develop-
ments in plate heat-exchangers. Journal of Enhanced Heat Transfer 2,
149155.
Wang, J.L., Zhao, L., Wang, X.D., 2010a. A comparative study of pure
and zeotropic mixtures in low-temperature solar Rankine cycle.
Applied Energy 87, 33663373.
Wang, X.D., Zhao, L., Wang, J.L., Zhang, W.Z., Zhao, X.Z., Wu, W.,
2010b. Performance evaluation of a low-temperature solar Rankine
cycle system utilizing R245fa. Solar Energy 84, 353364.
Witt, 2004. Air Cooled Condenser 550 RPM/Flying Bird 2. <http://
www.witthtp.com/literature.html> (consulted on 2.12.10).
Wolpert, J.L., Riat, S.B., 1996. Solar-powered Rankine system for
domestic applications. Applied Thermal Engineering 16, 281289.
Yamamoto, T., Furuhata, T., Arai, N., Mori, K., 2001. Design and testing
of the organic Rankine cycle. Energy 26, 239251.
Zanelli, R., Favrat, D., 1994. Experimental investigation of a hermetic
scroll expander-generator. In: International Compressor Engineering
Conference, Purdue.
966 S. Quoilin et al. / Solar Energy 85 (2011) 955966