Energy 63 (2013) 142e151

Contents lists available at ScienceDirect

Ene
journal homepage: www.elsevier.com/locate/energy

Thermo-economic environmental optimization of Organic Rankine

Cycle for diesel waste heat recovery
Zahra Hajabdollahi a, Farzaneh Hajabdollahi b, Mahdi Tehrani c, Hassan Hajabdollahi d, *
a
Mechanical Engineering Department, Shahid Bahonar University of Kerman, Kerman, Iran
b
Mechanical Engineering Department, Ferdowsi University of Mashhad, Mashhad, Iran
c
Mechanical Engineering Department, Iran University of Science and Technology, Tehran, Iran
d
Mechanical Engineering Department, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran

a r t i c l e i n f o a b s t r a c t

Article history: An Organic Rankine Cycle for diesel engine waste heat recovery is modeled and optimized. The design
Received 18 February 2013 parameters are nominal capacity of diesel engine, diesel operating partial load, evaporator pressure,
Received in revised form condenser pressure and refrigerant mass flow rate. In addition four refrigerants including R123, R134a,
22 August 2013
R245fa and R22 are selected and studied as working fluids. Then, the fast and elitist NSGA-II (Non-
Accepted 17 October 2013
Available online 14 November 2013
dominated Sorting Genetic Algorithm) is applied to maximize the thermal efficiency and minimize the
total annual cost (sum of investment cost, fuel cost and environmental cost) simultaneously. The results
of the optimal design are a set of multiple optimum solutions, called Pareto optimal solutions. The
Keywords:
Organic Rankine Cycle
optimization results show that the best working fluid is R123 in both of economical and thermo
Diesel engine dynamical view point for a specified value of output power. R245fa, R134a and R22 are placed in the next
Working fluid ranking, respectively. The optimum result of R123 shows the 0.01%, 4.39%, and 4.49% improvement for
Total annual cost the total annual cost in comparison with R245fa, R22, and R134a, respectively. The above values for
Thermal efficiency efficiency are obtained 1.01%, 12.79% and 10.57%, respectively. Furthermore R123 needs the highest in-
NSGA-II vestment cost while the environmental and fuel costs are the lowest.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction
ORC (Organic Rankine Cycle) enable efficient power generation
unit from low-grade heat sources by replacing water with organic
working fluids such as refrigerants or hydrocarbons. Najjar and
Radehwan recovered waste heat by combining a heat-exchanger
gas turbine cycle with closed Organic Rankine Cycle [1]. Some
authors investigate the effect of working fluids on Organic
Rankine Cycle for waste heat recovery [2e7]. Mago et al. pre-
sented an analysis of regenerative Organic Rankine Cycles using
dry organic fluids to convert waste energy to power from low-
grade heat sources [8]. Dai et al. described the Rankine cycles
for low grade waste heat recovery with different working fluids
[9]. Papadopoulos et al. presented the first approach to the sys-
tematic design and selection of optimal working fluids for ORCs
(Organic Rankine Cycles) based on CAMD (computer aided mo-
lecular design) and process optimization techniques [10]. The
results were compared in the regions when net power outputs
* Corresponding author. Tel.: þ98 913 2924318; fax: þ98 391 4221764.
E-mail addresses: Hajabdollahi@iust.ac.ir, hassan.hajabdollahi@gmail.com
(H. Hajabdollahi).
were fixed at 10 kW. The outcomes indicated that R11, R141b,
R113 and R123 manifested slightly higher thermodynamic per-
formances than the others. Some authors investigated the per-
formance of a low-temperature solar Rankine cycle system using
various working fluids [11e19]. Shengjun et al. presented an
investigation on the parameter optimization and performance
comparison of the fluids in subcritical ORC and transcritical power
cycle in low-temperature binary geothermal power system [20]. A
supercritical Rankine cycle using zeotropic mixture working fluids
for the conversion of low-grade heat into power was proposed
and analyzed by Chen et al. [21]. Unlike a conventional Organic
Rankine Cycle, a supercritical Rankine cycle does not go through
the two-phase region during the heating process. By adopting
zeotropic mixtures as the working fluids, the condensation pro-
cess also happens non-isothermally. Both of these features create
a potential for reducing the irreversibilities and improving the
system efficiency. Alessandro Franco analyzed and discussed the
exploitation of low temperature, water-dominated geothermal
fields with a specific attention to regenerative Organic Rankine
Cycles [22]. Yamada et al. proposed a new pump less Rankine-type
cycle for power generation from low-temperature heat sources
[23]. The new cycle mainly consists of an expander, two heat
exchangers, and switching valves for the expander and heat

0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2013.10.046
 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151 143

Nomenclature h efficiency (e)

a annual cost coefficient (e) Subscripts

A condenser heat transfer surface area (m2) a actual
C investment cost ($) D diesel
h enthalpy (kJ/kg K) i inlet
i interest rate (e) o outlet
LHV fuel lower heating value (kJ/kg) evap evaporator
m _ mass flow rate (kg/s) T turbine
p pressure (kPa) s isentropic
Q_ rate of heat transfer (kW) cond condenser
U overall heat transfer coefficient (W/m2 K) LMTD logarithmic mean temperature difference
W_ power (kW) CW cooling water
y depreciation time (year) p pump
env environment
Greek abbreviation inv investment
jem pollutant emission cost ($/kg) f fuel
jf fuel cost ($/kg) nom nominal
s hours of operation per year (h) PL partial load (%)
n specific volume (m3/kg) total total
ε total cycle thermal efficiency (e) wj water jacket

exchangers. Chen et al. studied transcritical Rankine cycles using
refrigerant R32 (CH2F2) and carbon dioxide (CO2) as the working
fluids for the conversion of low-grade heat into mechanical power
[24]. Wang et al. used waste heat from stationary and mobile
engine cycles to generate cooling for structures and vehicles [25].
It combined an ORC (Organic Rankine Cycle) with a conventional
vapor compression cycle. In order to maintain high system per-
formance while reducing size and weight for portable applica-
tions, micro channel based heat transfer components and scroll
based expansion and compression were used. Sun and Li pre-
sented a detailed analysis of an Organic Rankine Cycle heat re-
covery power plant using R134a as working fluid. Mathematical
models for the expander, evaporator, air cooled condenser and
pump were developed to evaluate and optimize the plant per-
formance [26]. Wagar et al. developed a model of an ammoniae
water Rankine heat engine and examined with the inclusion of a
two-phase expansion process. A general model for the optimal
cycle was developed based upon the maximum operating tem-
perature and the operating concentration [27]. Jing Li et al. pre-
sented a quantitative study on the convection, radiation, and
conduction heat transfer from a kW-scale expander. A mathe-
matical model was built and validated [28]. Xu and He proposed a
regenerative Organic Rankine Cycle that used a vapor injector as
the regenerator [29]. The thermal performance of both the novel
cycle and the basic ORC was calculated and compared by using
R123 as the working fluid. Invernizzi et al. investigated the pos-
sibility of enhancing the performances of micro-gas turbines
through the addition of a bottoming Organic Rankine Cycle [30].
They showed ORC cycles were particularly suitable for the re-
covery of heat from sources at variable temperatures. Quoilin and
et al. developed a thermodynamic model of a waste heat recovery
ORC in order to compare both the thermodynamic and the ther-
moeconomic performance of several typical working fluids for
low to medium temperature-range ORCs [31].
In this paper after thermo-economic modeling of ORCD (Organic
Rankine Cycle for Diesel) waste heat recovery, this equipment is
optimized by maximizing the thermal efficiency as well as mini-
mizing the total annual cost, simultaneously. nominal capacity of
diesel engine, diesel operating partial load, evaporator pressure,
condenser pressure and refrigerant mass flow rate are taken as five
design parameters and fast and elitist NSGA-II (Non-dominated
Sorting Genetic Algorithm) is applied to provide a set of Pareto
multiple optimum solutions.
As a summary, the followings are the contribution of this paper
into the subject:
Applying four simultaneous system analysis including energy,
efficiency, economic and environment (4E analysis) for equip-
ment selection.
Selecting the nominal capacity of diesel engine, diesel oper-
ating partial load, evaporator pressure, condenser pressure as
well as refrigerant mass flow rate as design parameters (not
selected as a group of variables in other available literature).
Performing the multi objective optimization of ORCD with ef-
ficiency and the total annual cost as two objectives (not selected
in other available literature).
Applying the optimization for four working fluids including
R123, R134a, R245fa and R22.
Sensitivity analysis of change in total annual cost when the
price of diesel fuel varies.
2. Thermal modeling
Schematic diagram of an ORCD (Organic Rankine Cycle for
Diesel) waste heat recovery is shown in Fig. 1. It mainly consists of
diesel engine and Rankine cycle including turbine, condenser,
pump and evaporator (heat exchanger). Refrigerant enters the
evaporator at a given pressure and temperature (state 4), where it is
vaporized by the absorbed heat energy from waste heat recovery in
diesel engine. The refrigerant exits the evaporator as superheated
vapor (state 1), and then passes through the expander (turbine).
The high quality refrigerant (state 2) enters the condenser and
transfers heat to the cooling tower. The condensed liquid refrig-
erant (state 3) is next pumped to the evaporating pressure and
enters directly to the evaporator (state 4). There are two sources of
power generation here, including the net power from diesel engine
and Rankine cycle.
In order to do the thermal modeling, mass and energy balances
on the system are required to determine the flow rates and energy
transfer rates at the control surface. Appling the first law of ther-
modynamic in the steady state, one can find the formula for mass
and energy balance as follow [32]:
144 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151

Fig. 1. Schematic diagram of Organic Rankine Cycle for diesel waste heat recovery.

Mass balance equation: _

W ni ðPo  Pi Þ
hp ¼ _ p;s ¼ (9)
X X W p;a ho  hi
_i ¼
m _o
m (1)

Energy balance equation: Diesel engine:

X X A part of input energy into the diesel engine is converted to
Q_  W
_ ¼ _ o ho 
m _ i hi
m (2) the power and remains transformed to the heat in exhaust, water
jacket, lube oil and radiation. Just the heat in exhaust and water
where subscripts i and o refer to streams entering and leaving the jacket are recoverable and useful. On the other hand, diesel en-
control volume, respectively. gine characteristics such as thermal efficiency or recoverable
The energy balance equations for the various parts of the tur- heat rate are a function of partial load (part load) which is
bine cycle as shown in Fig. 1 are as follow: defined as the percentage of nominal load. By increasing the
Evaporator: partial load, both power and recoverable heat rate increases but
fuel mass flow rate increases too. As a result, the optimum value
Q_ evap ¼ m
_ i ðho  hi Þ (3) of partial load should be determined in optimization process and
consequently the engine specifications as a function of partial
Turbine: load is needed too.
The graphical data of power and heat rate produced by diesel
_
W h h engine in various partial loads is shown in Fig. 2 [33]. The following
hT ¼ _ T;a ¼ i o;a
(4) relations are curve fitted to obtain the mathematical formulation
W T;s hi  ho;s

X X
_ T;a ¼
W _ i hi 
m _ o ho
m (5)

Condenser:
X X
Q_ cond ¼ _ i hi 
m _ o ho ¼ UAcond DTLMTD
m (6)

where U, Acond and DTLMTD are overall heat transfer coefficient,

condenser heat transfer surface area and logarithmic mean tem-
perature difference defined as follow:
   
DT1  DT2 T TCW;o  T3 TCW;i
DTLMTD;c ¼ ¼ 2   (7)
logðDT1 =DT2 Þ log T2 TCW;o T3 TCW;i

where CW (cooling water) indicates the cooling water recirculation

in condenser. The pressure drop in condenser/evaporator is
assumed to be Dp and as a result:

po =pi ¼ ð1  DpÞ (8)

Fig. 2. Power and heat produced by diesel engine versus partial load (points indicate
Pump: the actual data and lines indicate the curve fitting).
 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151 145

for power and heat rate as a function of partial load using data 3. Objective functions, design parameters and constraints
presented in Fig. 2:
Power as a function of partial load: In this study, the total cycle efficiency and total annual cost are
considered as two objective functions. The efficiency is defined in
W _ D;PL Equation (18) and the total annual cost includes investment cost
¼ 1:07 expð0:0005736ðPLÞÞ (capital cost of diesel and ORC), cost of diesel fuel as well as envi-
_ f;PL LHVf
m
ronmental cost (regarding the CO2 emission) is computed from:
 1:259 expð0:05367ðPLÞÞhD;nom (10)
Ctotal ¼ afCinv þ Cf þ Cenv (20)
recoverable energy from the water jacket and exhaust gas enthalpy
as a function of partial load: Cinv ¼ Cinv;D þCinv;T þCinv;cond þCinv;P
 d  d  d4
Q_ wj;PL ¼ b1 W _ D 1 þb W
2
_ T 2 þb ðA d3 _
3 cond Þ þb4 W p (21)
¼ 24:01 expð0:0248ðPLÞÞ
_ f;PL $LHVf
m
þ 15:35 expð0:002822ðPLÞÞ (11) _ f  jf  s
Cf ¼ 3600  m (22)

Cenv ¼ mtotal;CO2  jem (23)

Q_ o;PL
¼ 0:001016ðPLÞ2  0:1423ðPLÞ þ 31:72 (12)
_ f;PL $LHVf
m
here jf, jem and 4 are the fuel cost, pollutant emission cost and
maintenance factor, respectively. The constant values of b and
un-recoverable energy from oil and radiation (and etc) as a function
d coefficients are obtained based on the regional price of the
of partial load:
equipment. s and mtotal;CO2 are number of system operating hour in
    a year and total CO2 produced by fuel in a year. In addition a is the
Q_ oil
¼ 1:329  107 PL4  4:35  105 PL3 annual cost coefficient defined as:
_ f;PL $LHVf
m
i
þ 0:005631 PL2  0:3471 PL þ 11:81 (13) a ¼ (24)
1  ð1 þ iÞy

Q_ etc   where i and y are interest rate and depreciation time, respectively.
¼ 3:258  106 PL3 þ 0:001364 PL2
_ f;PL $LHVf
m In this study, nominal capacity of diesel engine, diesel operating
partial load, evaporator pressure, condenser pressure and refrig-
 0:1068 PL þ 15:64 (14)
erant mass flow rate are considered as design parameters.
The following constraints are introduced for the optimization
where hD,nom,m_ f and LHV (fuel lower heating value) are diesel
procedure:
nominal efficiency, fuel mass flow rate entered to diesel engine and
fuel lower heating value, respectively. p1 > p2 (25)
Furthermore, the fuel mass flow rate of prime movers is also
assumed to be a function of partial load as bellow [33]:
T2 > 40 (26)
m _ f;PL
¼ 0:02836 expð0:03254ðPLÞÞ the constraint, T2 > 40, is applied for keeping the condenser
_
mf;nom temperature above the ambient temperature for condensing
þ 0:2556 expð0:01912ðPLÞÞ (15) procedure.

_ f;nom is nominal diesel fuel consumption computed as:

where, m x2 > 0:95 (27)

_ this constraint is applied to avoid the turbine vane corrosion.

W D
_ f;nom ¼
m (16)
hD;nom LHVf T1 < Te;D (28)
Total cycle (ORC and diesel engine):
_ this is necessary for heat transfer from waste heat recovery to the
Moreover, W ORC is total output net power in ORC (turbine cycle)
refrigerant where Te,D is the diesel engine exhaust temperature. It is
estimated as follow:
also assumed that the stack temperature is not lower than 148.8  C
for diesel engine to avoid stack corrosion [33].
_ _ _
W ORC ¼ W T;a  W p;a (17)

In addition the total thermal efficiency of the cycle is computed 4. Genetic algorithm for multi-objective optimization
from:
4.1. Definition of multi-objective optimization

W_
total A multi-objective optimization problem requires the simulta-
ε ¼ (18)
_ f $LHV
m neous satisfaction of a number of different and often conflicting
objectives. It is required to mention that no combination of decision
_
where W total is total output net power estimated as follow: variables can optimize all objectives, simultaneously. Multi-
X X objective optimization problems generally show a possibly un-
_ _ Dþ _
W total ¼ W W ORC (19) countable set of solutions, whose evaluated vectors represent the
best possible trade-offs in the objective function space. Pareto
146 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151
optimality is the key concept to establish a hierarchy among the
solutions of a multi-objective optimization problem, in order to
determine whether a solution is really one of the best possible
trades-off [34]. A multi-objective problem consists of optimizing
(i.e., minimizing or maximizing) several objectives, simultaneously
with a number of inequality or equality constraints. The problem
can be formally written as follows:
Find x ¼ (xi)ci ¼ 1, 2 ,., Nparam such as
fi(x) is a minimum (respectively maximum) ci ¼ 1, 2,., Nobj
Subject to:
gj ðxÞ ¼ 0 cj ¼ 1; 2; .; M; (29)
hk ðxÞ  0 ck ¼ 1; 2; .; K; (30)
where x is a vector containing the Nparam design parameters,
(fi)i¼1,.,Nobj the objective functions and Nobj the number of ob-
jectives. The objective function (fi)i¼1,.,Nobj returns a vector con-
taining the set of Nobj values associated with the elementary
objectives to be optimized simultaneously. The first multi-
objective GA (Genetic Algorithm), called vector evaluated GA (or
VEGA), was proposed by Schaffer [35]. An algorithm based on
non-dominated sorting was proposed by Srinivas and Deb [36]
and called NSGA (Non-dominated Sorting Genetic-Algorithm)).
This was later modified by Deb et al. [37] which eliminated higher
computational complexity, lack of elitism and the need for spec-
ifying the sharing parameter. This algorithm is called NSGA-II
which is coupled with the objective functions developed in this
study for optimization.
4.2. Non-dominated sorting and Pareto front
(a)
As defined by Deb [38], an individual X is said to constrain-
dominate an individual X(b), if any of the following conditions are
true:
(1) X(a) and X(b) are feasible, with
(a) X(a) is no worse than X(b) in all objective, and
(b) X(a) is strictly better than X(b) in at least one objective.
(2) X(a) is feasible while individual X(b) is not.
(3) X(a) and X(b) are both infeasible, but X(a) has a smaller
constraint violation.
Here, the constraint violation [(X) of an individual X is defined to
be equal to the sum of the violated constraint function values [39],
X
B   hD,nom ¼ 0.35 for nominal efficiency of diesel engine by considering
[ðXÞ ¼ g gj ðXÞ gj ðXÞ; (31)
j¼1
where g is the Heaviside step function. A set of non-dominated
individuals is used to form a Pareto-optimal fronts.
4.3. Tournament selection
Each individual competes in exactly two tournaments with
randomly selected individuals, a procedure which imitates survival
of the fittest in nature.
4.4. Controlled elitism sorting
To preserve diversity, the influence of elitism is controlled by
choosing the number of individuals from each subpopulation, ac-
cording to the geometric distribution [39]:
1  c q1
Sq ¼ S c ; (32)
1  cw
To form a parent search population, Ptþ1 (t denotes the gener-
ation), of size S, where 0 < c < 1 and w is the total number of ranked
non-dominated.
4.5. Crowding distance
The crowding distance metric proposed by Deb [38] is utilized,
where the crowding distance of an individual is the perimeter of
the rectangle with its nearest neighbors at diagonally opposite
corners. So, if individual X(a) and individual X(b) have same rank,
each one has a larger crowding distance is better.
4.6. Crossover and mutation
Uniform crossover and random uniform mutation are employed
to obtain the offspring population,Qtþ1. The integer-based uniform
crossover operator takes two distinct parent individuals and in-
terchanges each corresponding binary bits with a probability,
0 < pc  1. Following crossover, the mutation operator changes each
of the binary bits with a mutation probability, 0 < pm < 0.5.
4.7. Historical archive
The NSGA-II algorithm has been modified to include an archive
of the historically non-dominated individuals, Ht. Archive is used to
update the data at each iteration.
5. Case study
In this study, four working fluids including R123, R134a, R245fa
and R22 are selected and an ORCD system is optimized for each of
them, separately. For this purpose, MATLAB and Refprop soft wares
are used to evaluate the thermo dynamical properties of the re-
frigerants. To minimize the total annual cost value and maximize
the total cycle efficiency, five design parameters including nominal
capacity of diesel engine, diesel operating partial load, evaporator
pressure, condenser pressure, and refrigerant mass flow rate are
selected. Design parameters and the range of their variations are
listed in Table 1. The ORCD system should deliver 200 kW output
net power operate at s ¼ 6000 hours in a year. System is optimized
for depreciation time y ¼ 20 years, interest rate i ¼ 0.12,
jem ¼ 0.02086 $/kg for CO2 pollutant emission cost and
5% pressure drop in evaporator, 0.9 as both turbine and pump
isentropic efficiency and 0.168 $/kg as diesel fuel cost [40]. More-
over, the constants of investment cost in relation (21) are taken
b ¼ [1763 4750 150 3500] and d ¼ [0.95 0.75 0.8 0.47] and 4 ¼ 1.05
is considered for the maintenance factor. In addition, evaporator
(heat recovery heat exchanger) is a part of diesel engine and cost of
it is considered with investment cost of diesel engine.
Table 1
The design parameters and their range of variation for the optimization procedure.
Case studies From To
Capacity of diesel engine (kW) 10 200
Partial load (%) 20 100
Evaporator pressure (kPa) 300 2000
Condenser pressure (kPa) 10 2000
Refrigerant mass flow rate (kg/s) 0.1 5
 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151
6. Results and discussion
6.1. The procedure of thermo-economic and environment analysis
By evaluating the nominal capacity of diesel engine and partial
load as design parameters, the power and recoverable heat from
diesel is specified. The turbine power, condenser required heat as
well as pump power is also computed by using the recovered heat
from diesel and other design parameters including evaporator
pressure, condenser pressure and refrigerant mass flow rate. By
knowing the diesel power, turbine power, pump power as well as
condenser heat transfer surface area (using Equation (6)), the in-
vestment cost of equipments are computed using relation (21). The
fuel consumption is also computed from relation (16) and conse-
quently the fuel and environmental cost (relations (22) and (23))
are determined. Considering the power generated by diesel engine
and net power output from ORC cycle, the total cycle efficiency is
also computed from Equation (18).
6.2. Optimization
The Genetic Algorithm Optimization is performed for 1000 gen-
erations, using a search population size of M ¼ 150 individuals,
crossover probability of pc ¼ 0.9, gene mutation probability of
pm ¼ 0.035 and controlled elitism value c ¼ 0.55, separately for four
working fluids including R123, R134a, R245fa and R22. The results of
optimum efficiency and total annual cost for four refrigerants are
depicted in Fig. 3. As it shown, the Pareto fronts suffer from diversity
and lower number of optimum points compared with Pareto fronts
presented in some other works reported in Refs. [41e44]. Actually the
diversity and number of points in the final Pareto front is highly
depended on searching area which is highly limited by five constraints
presented in Section 3. Moreover to preserve the number of optimum
points in evolution of algorithm, the historical archive (Section 4.7) as
well as high number of generation (1000) were used which guaran-
teed the optimum results. The Pareto optimum results reveal the
conflict between two objectives, the efficiency and the total annual
cost. Any change that increases the efficiency, leads to an increase in
the total annual cost and vice versa. It is observed that the best
refrigerant for this case study is R123. Their results are totally domi-
nated over the other refrigerants in both efficiency and total annual
cost. The refrigerants R245fa, R134a and R22 are respectively in the
Fig. 3. Pareto optimal front for four working fluids including R22, R123, R134a and
R245fa.
147
Table 2
The critical temperature and pressure of studied working fluids.
Working fluid R22 R123 R134a R245fa
Critical temperature (K) 369 457 374 427
Critical pressure (kPa) 4989 3668 4059 3639
next Ranking. The critical temperature and pressure of four studied
working fluids are listed in Table 2. It is worth mentioning that higher
ORC efficiency due to the higher working fluid critical temperature
was previously investigated by Aljundi [3]. Nevertheless, the effect of
critical points has been not reported in thermo-economic viewpoint.
Comparing the optimum results with critical points listed in Table 2
reveal that a working fluid with higher critical temperature and
lower critical pressure leads to the better thermo-economic and
environmental result. The distribution of variables for the optimal
points on Pareto front (Fig. 3) is shown in Fig. 4aee. The lower and
upper bounds of the variables are shown by dotted lines. The
following points for the optimal variables in Fig. 4 could be deduced:
1. A plant with higher value of efficiency needs a lower capacity of
diesel engine.
2. All the diesel engines should operate in the range of 90e100% of
their nominal capacity.
3. The turbine outlet pressure in the all plants is higher than
145 kPa. The corresponding value of this parameter in the
common steam cycle power plant is in the range of 10e20 kPa.
The higher selected turbine outlet pressure is due to the
constraint mentioned in the Section 3 (relation (26)).
The percent of power produced by ORC to total power generated
by ORCD for all the optimum points in the Pareto front are obtained
and depicted in Fig. 5. The power produced by ORC is about 31.5% of
total power generated by ORCD for R123 which has the maximum
efficiency, too. This value is approximately 30.8%, 23.2% and 21.5%
for R245fa, R134a and R22, respectively. Actually, the ratio of ORC to
ORCD power generation is corresponding with the total cycle effi-
ciency which is presented in Pareto fronts.
6.3. Final optimum solution
The selection of a single optimum point from existing points on
the Pareto front (boundary of infeasible region and feasible but non-
optimum region) needs a process of decision-making. In fact, this
process is mostly carried out based on engineering experiences and
importance of each objective for decision makers. The process of
final decision-making is usually performed with the aid of a hypo-
thetical point named as ideal point. If two objective functions would
be optimized individually, i.e. disregarding another objective func-
tion, the composition of these values represents the ideal point or
ideal objective point [45]. The typical Pareto front as well as ideal
point in maximizing the first objective and minimizing the second
objective is shown in Fig. 6. It is clear that it is impossible to have both
objectives at their optimum point, simultaneously. Since the ideal
point is not a solution located on the Pareto frontier, closest point of
Pareto frontier to the ideal point might be selected as final optimum
solution. Before it, the objectives should be non-dimensionalized. In
this paper, LINMAP (Linear programming techniques for multidi-
mensional analysis of preferences) method is used to non-
dimensionalize the objectives using the following relation [41]:
Fij
Fijn ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi (33)
2 Pm 
i¼1 ij F
148 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151

Fig. 4. Distribution of optimum values of design parameters for points in Pareto front. a. Diesel capacity, b. diesel partial load, c. turbine inlet pressure, d. turbine outlet pressure, e.
refrigerant mass flow rate.

where i is the index for each point on Pareto front, j is the index for where ideal is the index of ideal objective functions. Then the value
each objective and m denotes the number of points on the Pareto of distance for each point on the Pareto frontier is computed using
front. Then the distance of each point on Pareto front from the ideal the above relation and optimum point which leads to the minimum
point is obtained: d is selected as final optimum solution.
The final value of optimum objective functions, investment cost,
fuel cost and environmental cost along with corresponding design
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi parameters using the above procedure are obtained and listed in
u 2  2
uX Table 3.
di ¼ t Fijn  Fideal;j
n (34)
It is observed that the best refrigerant in economical view point
j¼1
is R123 with total annual cost of 86,253 $/year. The refrigerants
 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151 149

Table 3
The final optimum values of design parameters and objective functions for four
refrigerants.

Case studies R22 R123 R134a R245fa

Capacity of diesel 158.438 152.500 158.438 152.557

engine
(kW)
Partial load (%) 99.710 90.625 96.875 90.647
Evaporator pressure 1847.893 771.182 1851.596 525.654
(kPa)
Condenser pressure 887.280 146.480 1004.626 192.092
(kPa)
Refrigerant mass flow 2.550 2.564 4.693 3.785
rate (kg/s)
Thermal efficiency 0.4471 0.5127 0.4585 0.5075
()
Total annual cost 90,039 86,253 90,127 86,261
($/year)
Total investment 307,320 323,260 315,410 320,050
cost ($)
Environmental cost 12,911 11,323 12,563 11,330
($/year)
Cost of fuel ($/year) 34,147 29,948 33,228 29,966
Percent of 14.305 13.058 13.939 13.131
Fig. 5. Percent of ORC net power to total ORCD net power for optimum results pre-
Environmental
sented in Pareto front.
to total cost (%)

R245fa, R22, and R134a are in the next Rankine, respectively that
the total annul cost for them are 86,261 $/year, 90,039 $/year and
90,127 $/year, respectively. The optimum result of R123 is improved
0.01%, 4.39%, and 4.49% in comparison with R245fa, R22, and R134a
refrigerants, respectively. On the other hand the best refrigerant in
thermo dynamical efficiency view point is R123 with efficiency of
0.5127. The refrigerants R245fa, R134a and R22 are in the next
Rankine, respectively with efficiency of 0.5075, 0.4585 and 0.4471,
respectively. The optimum result of R123 is improved 1.01%, 10.57%
and 12.79% in comparison with R245fa, R134a and R22, respec-
tively. Furthermore, the plant with refrigerant R123 needs the
lower diesel capacity, diesel partial load, and refrigerant mass flow
rate in comparison with the plants with other studied refrigerants.
Moreover, plant with refrigerant R22 needs the lowest refrigerant
mass flow but higher partial load and diesel capacity in comparison
with the plants with the other refrigerants. It is also observed from
Table 3 that the R123 needs the highest investment cost while
lowest environmental and fuel costs.
6.4. Sensitivity analysis on diesel fuel price
In this study the diesel fuel price was considered 0.168 $/kg
based on local price of fuel [40]. However, the mentioned price is
significantly varied in geographical zone. For this purpose the
variation of total annual cost versus percent of variation in fuel
price for different working fluid at final optimum point are shown
in Fig. 7. The result demonstrates: for example by increase of 100%
in diesel fuel price, the optimum value of total annual cost increases
38.16%, 34.37%, 36.92% and 34.82% respectively for R22, R123,
R134a and R245fa. Furthermore, due to the higher investment cost
for plants with lower total annual cost, the optimum value of total
annual cost converged to a same point by decreasing the price of
diesel fuel.

Fig. 6. Concept of ideal point in the Pareto frontier for maximizing the objective 1 and Fig. 7. Variation of AAB (total annual cost) versus percent of variation in diesel fuel
minimizing the objective 2. price at final optimum point.
150 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151

by substituting the relations (10)e(14) into the above relation, the

energy conservation error is obtained based on partial load which is
depicted in Fig. 9. As it shown, the maximum deviation is in the
range of 1.5% which is shown the accuracy of the applied relations
for diesel engine.

7. Conclusions

An ORCD was optimally designed using multi objective opti-

_ T þ Q_
Fig. 8. Comparison of recovered heat from diesel engine and W cond  W p for
optimum results presented in Pareto front.
6.5. Model verification
Basically total recovered heat from diesel engine should be the
_ T þ Q_ _ R123 in both efficiency and total annual cost and it seems to be a
same with W cond  W p . These two values are obtained for all
optimum points in Pareto front and shown in Fig. 8. As it shown, the
maximum deviation is in the range of 2% which is acceptable for
engineering problem.
In addition, as it mentioned in thermal modeling section, the
graphical data regarding the power and heat produced by different
part of diesel engine in terms of partial load in the rang of 20e100%
were used and curve fitted using relations (10)e(14). To verify the
precision of applied relations and satisfying the first law of ther-
modynamic, the following error in terms of partial load is defined:
  
_ f;PL LHVf  W_ _ _ _ _
m D;PL þ Q wj;PL þ Q o;PL þ Q oil;PL þ Q etc;PL
ErPL ð%Þ ¼  
m_ f;PL LHVf
mization technique by considering the thermo-economic and
environmental aspects. The design parameters (decision variables)
were nominal capacity of diesel engine, diesel operating partial
load, evaporator pressure, condenser pressure and refrigerant mass
flow rate. In the presented optimization problem, the efficiency and
total annual cost were two objective functions (the efficiency was
maximized and total annual cost was minimized). The above pro-
cedure was performed for four working fluids including R123,
R134a, R245fa and R22. Based on the studied system the following
conclusions can be inferred:
_
1. R123 is found to be the best working fluid while R22 is the
worst. Beside the worst results, R22 has ozone depletion and
high global warming potential which is not recommended as
working fluid in our studied case.
2. R245fa has approximately the same results in comparison with
good replacement for R123.
3. The optimum result of R123 was improved 0.01%, 4.39%, and
4.49% for total annual cost in comparison with R245fa, R22, and
R134a refrigerants, respectively. The above values for efficiency
were found to be 1.01%, 12.79% and 10.57%, respectively.
4. A plant with lower total annual cost needs the higher invest-
ment cost.
5. Selecting a working fluid with higher critical temperature and
lower critical pressure leads to the better thermo-economic and
environment result in this case.

6. By decreasing the diesel fuel price, the optimum value of
total annual cost for different working fluid leads to the
same result.

100 References
(35)
[1] Najjar YSH, Radhwan AM. Cogeneration by combining gas turbine engine with
organic Rankine cycle. Appl Therm Eng 1988;8:211e9.
[2] TauLiu B, Chien KH, Wang CC. Effect of working fluids on Organic Rankine
Cycle for waste heat recovery. Energy 2004;29:1207e17.
[3] Aljundi IH. Effect of dry hydrocarbons and critical point temperature on the
efficiencies of organic Rankine cycle. Renew Energy 2011;36:1196e202.
[4] Chen Y, Lundqvist P, Johansson A, Platell P. A comparative study of the carbon
dioxide transcritical power cycle compared with an Organic Rankine Cycle
with R123 as working fluid in waste heat recovery. Appl Therm Eng 2006;26:
2142e7.
[5] Desai NB, Bandyopadhyay S. Process integration of organic Rankine cycle.
Energy 2009;34:1674e86.
[6] Wang EH, Zhang HG, Fan BY, Ouyang MG, Zhao Y, Mu QH. Study of working
fluid selection of Organic Rankine Cycle (ORC) for engine waste heat recovery.
Energy 2011;36:3406e18.
[7] Roy JP, Mishra MK, Misra A. Parametric optimization and performance anal-
ysis of a waste heat recovery system using Organic Rankine Cycle. Energy
2010;35:5049e62.
[8] Mago PJ, Chamra LM, Srinivasan K, Somayaji C. An examination of regenera-
tive organic Rankine cycles using dry fluids. Appl Therm Eng 2008;28:998e
1007.
[9] Dai Y, Wang J, Gao L. Parametric optimization and comparative study of
Organic Rankine Cycle (ORC) for low grade waste heat recovery. Energy
Convers Manag 2009;50:576e82.
[10] Papadopoulos AI, Stijepovic M, Linke P. On the systematic design and selection
of optimal working fluids for Organic Rankine Cycles. Appl Therm Eng
2010;30:760e9.
[11] Wang XD, Zhao L, Wang JL. Experimental investigation on the low-
temperature solar Rankine cycle system using R245fa. Energy Convers
Fig. 9. Energy conservation error for curve fitted relations in various partial load. Manag 2011;52:946e52.
 Z. Hajabdollahi et al. / Energy 63 (2013) 142e151
[12] Manolakosa D, Papadakisa G, Mohameda ES, Kyritsisa S, Bouzianasb K. Design
of an autonomous low-temperature solar Rankine cycle system for reverse
osmosis desalination. Desalination 2005;183:73e80.
[13] Jing L, Gang P, Jie J. Optimization of low temperature solar thermal electric
generation with Organic Rankine Cycle in different areas. Appl Energy
2010;87:3355e65.
[14] Nafey AS, Sharaf MA, Rodríguez LG. Thermo-economic analysis of a combined
solar organic Rankine cycle-reverse osmosis desalination process with
different energy recovery configurations. Desalination 2010;261:138e47.
[15] Gang P, Jing L, Jie J. Analysis of low temperature solar thermal electric gen-
eration using regenerative Organic Rankine Cycle. Appl Therm Eng 2010;30:
998e1004.
[16] Tchanche BF, Papadakis G, Lambrinos G, Frangoudakis A. Fluid selection for a
low-temperature solar organic Rankine cycle. Appl Therm Eng 2009;29:
2468e76.
[17] Astolfi M, Xodo L, Romano MC, Macchi E. Technical and economical analysis of
a solaregeothermal hybrid plant based on an Organic Rankine Cycle. Geo-
thermics 2011;40:58e68.
[18] Wang JL, Zhao L, Wang XD. A comparative study of pure and zeotropic mix-
tures in low-temperature solar Rankine cycle. Appl Energy 2010;87:3366e73.
[19] Rayegan R, Tao YX. A procedure to select working fluids for solar Organic
Rankine Cycles (ORCs). Renew Energy 2011;36:659e70.
[20] Shengjun Z, Huaixin W, Tao G. Performance comparison and parametric
optimization of subcritical Organic Rankine Cycle and transcritical power
cycle system for low-temperature geothermal power generation. Appl Energy
2011;88:2740e54.
[21] Chen H, Goswami DY, Rahman MM, Stefanakos EK. A supercritical Rankine
cycle using zeotropic mixture working fluids for the conversion of low-grade
heat into power. Energy 2011;36:549e55.
[22] Franco A. Power production from a moderate temperature geothermal
resource with regenerative Organic Rankine Cycles. Energy Sustain Dev
2011;15:411e9.
[23] Yamada N, Minami T, Mohamad MNA. Fundamental experiment of pumpless
Rankine-type cycle for low-temperature heat recovery. Energy 2011;36:
1010e7.
[24] Chen H, Goswami DY, Rahman MM, Stefanakos EK. Energetic and exergetic
analysis of CO2- and R32-based transcritical Rankine cycles for low-grade heat
conversion. Appl Energy 2011;88:2802e8.
[25] Wang H, Peterson R, Harada K, Miller E, Goble RI, Fisher L, et al. Performance
of a combined Organic Rankine Cycle and vapor compression cycle for heat
activated cooling. Energy 2011;36:447e58.
[26] Sun J, Li W. Operation optimization of an Organic Rankine Cycle (ORC) heat
recovery power plant. Appl Therm Eng 2011;31:2032e41.
151
[27] Wagar WR, Zamfirescu C, Dincer I. Thermodynamic performance assessment
of an ammoniaewater Rankine cycle for power and heat production. Energy
Convers Manag 2010;51:2501e9.
[28] Li J, Pei G, Li Y, Ji J. Evaluation of external heat loss from a small-scale expander
used in organic Rankine cycle. Appl Therm Eng 2011;31:2694e701.
[29] Xu RJ, He YL. A vapor injector-based novel regenerative organic Rankine cycle.
Appl Therm Eng 2011;31:1238e43.
[30] Invernizzi C, Iora P, Silva P. Bottoming micro-Rankine cycles for micro-gas
turbines. Appl Therm Eng 2007;27:100e10.
[31] Quoilin S, Declaye S, Tchanche BF, Lemort V. Thermo-Economic optimization
of waste heat recovery Organic Rankine Cycles. Appl Therm Eng 2011;31:
2885e93.
[32] El-Wakil MM. Powerplant technology. McGraw-Hill; 2002.
[33] ASHRAE handbookCogeneration systems and engine and turbine drives; 1999
[Chapter S7].
[34] Fonseca CM, Fleming PJ, Back T, Fogel DB, Michalewicz Z. Handbook of
evolutionary computation. Oxford University Press; 1997.
[35] Schaffer JD. Multiple objective optimization with vector evaluated genetic
algorithms. In: Proceedings of the international conference on genetic algo-
rithm and their applications 1985.
[36] Srinivas N, Deb K. Multi-objective optimization using non-dominated sorting
in genetic algorithms. J Evol Comput 1994;2(3):221e48.
[37] Deb K, Pratap A, Agarwal S, Meyarivan T. A fast and elitist multi-objective
genetic algorithm: NSGA-II. IEEE Trans Evol Comput 2002;6(2):182e97.
[38] Deb K, Goel T. Controlled elitist non-dominated sorting genetic algorithms for
better convergence. In: Proceedings of the first international conference on
evolutionary multi-criterion optimization Zurich 2001. p. 385e99.
[39] Deb K. Multi-objective optimization using evolutionary algorithms. Chi-
chester: John Wiley and Sons Ltd,; 2001.
[40] www.ifco.org, [Web Site of Iranian Fuel Conservation Organization].
[41] Hajabdollahi H, Hajabdollahi F, Hajabdollahi Z. Soft computing based multi-
objective optimization of steam cycle power plant using NSGA-II and ANN.
Appl Soft Comput 2012;12:3648e55.
[42] Hajabdollahi H, Ahmadi P, Dincer I. Multi-objective optimization of plain fin-
and-tube heat exchanger using evolutionary algorithm. J Thermphys Heat
Transf 2011;25:424e31.
[43] Hajabdollahi H, Ahmadi P, Dincer I. Exergetic optimization of shell-and-tube
heat exchangers using NSGA-II. Heat Transf Eng 2012;33:618e28.
[44] Hajabdollahi H, Ahmadi P, Dincer I. Cost and entropy generation minimization
of a cross-flow plate fin heat exchanger using multi-objective genetic algo-
rithm. J Heat Transf Trans ASME 2011:133.
[45] Branke J, Deb K, Miettinen K, Slowinski R. Multiobjective optimization:
interactive and evolutionary approaches, vol. 5252. Springer; 2008.