Alkanes as Working Fluids for High-temperature Exhaust Heat Recovery of Diesel Engine Using Organic Rankine Cycle

ra

Gequn Shu a, Xiaoning Li a, Hua Ta State Key Laboratory of Engines, Tianjin University, Wb School of Aerospace, Mechanical and Manufacturing En

h i g h l i g h t s

o theire mostC is obtctive th

gy destruction factor (EDF), turbine size parameter (SP), total exergy destruc-

The primary driving force in seeking sustainable and economi-cally viable technologies is escalating fuel prices, national energysecurity and stringent regulations for environmental emissions.Clean and effective ways for energy conversion and utilizationhave to be found in order to tackle these situations. Diesel engine(DE) is the prime mover often chosen for its reliability, exible

low speciAlthough DE efciency can be improved by such advanced teogies as electronic control high-pressure common rainjection, homogeneous charge compression ignition (HCCIhaust-gas turbocharging, it is difcult to achieve a peak efciencyof DE higher than 45% [1]. Except for energy conversion to power,in regard to the balance of DE energy, exhaust energy accounts forabout one-thirds of fuel chemical energy, which is exhausted to theenvironment by engine in several forms.

Recently, researches on heat recovery from engine exhaustgases have become hot topics in waste heat recovery (WHR) Corresponding author. Tel./fax: +86 22 27891285.

Applied Energy 119 (2014) 204217

Contents lists availab

lseE-mail address: [email protected] (H. Wei).1. Introduction layout, high power density and efciency and0306-2619/$ - see front matter 2014 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.apenergy.2013.12.056c cost.chnol-il fuel) or ex-High-temperature exhaust gasAlkanesWorking uidsOrganic Rankine cycle (ORC)

tion rate (IORC), turbine volume ow ratio (VFR) and net power output per unit mass ow rate of exhaust(Pnet). Afterwards, the impact of molecular complexity on the indicators of VFR and SP is analyzed. Fur-thermore, the energy distribution of engine exhaust gases and the improvement of fuel economy, afterintegrating the bottoming ORC with diesel engine, are also discussed. Finally, the performance compar-ison between Cyclohexane-based ORC and steam cycle with relative pressure is carried out. The resultsshow that optimized working uids are not always constant subject to different indicators and operationparameters. However, cyclic Alkanes, Cyclohexane and Cyclopentane are considered as the most suitableworking uids when taking into account of all comprehensive indicators. The maximum improvement of10% in brake specic fuel consumption (BSFC) is obtained for DE-ORC combined systems with Cyclohex-ane used as working uid. In addition, although steam has more advantages in thermal efciency in thecurrent conditions, from a technical and economic point of view, Alkane-based ORCs may be more attrac-tive than conventional steam cycles, specically for DE waste gas heat recovery.

2014 Elsevier Ltd. All rights reserved.Keywords:Waste heat recovery (WHR)

thermal efciency (g), exer Less complex uids are preferred due t The cyclic Alkanes are considered as th Maximum improvement of 10% in BSF Alkane-based ORCs may be more attra

a r t i c l e i n f o

Article history:Received 19 August 2013Received in revised form 26 December 2013Accepted 29 December 2013Available online 25 January 2014ian a, Xingyu Liang a, Haiqiao Wei a,, Xu Wang beijin Road 92, Nankai District, Tianjin 300072, Chinagineering, RMIT University, Australia

excellent performances.promising candidate.ained by DE-ORC combined systems.an steam cycle for exhaust heat recovery.

a b s t r a c t

Study on recovering waste heat of engine exhaust gas using organic Rankine cycle (ORC) has continuouslyincreased in recent years. However, it is difcult to nd out appropriate working uids to match withexhaust gas waste heat due to high temperature. In this work, several tentative attempts and explora-tions are made in selecting Alkanes as working uid owing to their excellent thermo-physical and envi-ronmental characteristics. Parameters optimization of the combined system of diesel engine withbottoming ORC (DE-ORC) is performed on Alkane-based working uids with six indicators, includingof diesel engine using organic Rankine cycle

Alkanes as working uids for high-tempe

Applied

journal homepage: www.eture exhaust heat recovery

le at ScienceDirect

Energy

vier .com/ locate/apenergy

nergNomenclature

Cp isobaric ideal gas heat capacity (J/(mol K))cp mean specic heat at constant pressure (J/(mol K))_E exergy ow rate (kW)h specic enthalpy (kJ/kg)_I irreversibility (kW)_m mass ow rate (kg/s)n number of carbon atomsM molecular weight (g/mol)P pressure (kPa)_Q heat ow rate (kW)R universal gas constant (J/(mol K))s specic entropy (kJ/(kg K))T temperature (C)_V volumetric ow rate (m3/s)_W power (kW)DHis isentropic enthalpy difference (kW)

Greek symbolsn inverse of slope of saturated vapor curveg thermal efciencyr molecular complexity

Subscripts and superscriptsc condenser

G. Shu et al. / Applied E[26]. For example, Horst et al. [2] developed a dynamic model ofevaporator used in Rankine cycle for exhaust WHR in passengercar. Yun et al. [3] established a power generation and heat recoverymodel for internal combustion engines (ICEs). Fu et al. [4] obtainedthe mapping characteristics of naturally aspirated engine by exper-iments, and the characteristics of exhaust waste heat energy wereanalyzed using energy and exergy analysis method. Love et al. [5]investigated thermoelectric devices as a means of improving fueleconomy for engines through converting exhaust waste heat tousable electricity. Pandiyarajan et al. [6] investigated in detail theintegration of a n-tube heat exchanger with an IC engine setupto extract exhaust gas heat energy, and excessive energy wasstored in a thermal energy storage tank.

ORC processes, with similar structure as typical Rankine cycle,use organic uids instead of water as working uids. In ORC pro-cess, turbines are usually less complex as the organic uids havethe relatively lower enthalpy drop, and higher efciencies can beobtained at part loads as well [7]. In addition, a number of advan-tages can be attributed to ORC systems, such as simple structure,component availability, and the ease of application to localsmall-scale power generation systems. ORC as bottoming cycle tocouple with engines to recover waste heat energy results in com-bined power unit with enhanced overall system efciency, and isconsidered as one of the most promising technologies. ORC sys-tems consume no additional fuel for extra power. Moreover, com-pared with the original engine, the combined system has thepotential to greatly reduce specic pollutant emissions.

Many researchers have performed studies on this subject fromthe perspectives of both theoretical analysis and experimental

cond condensingcrit criticalcw cooling watere evaporatorevap evaporatingf working uidfuel diesel fuelg exhaust gasi each state pointin inletloss lossmax maximalmin minimalnet netout outletp pumps ideal statesv saturated vaport turbine0 standard condition14 state points2s, 4s, 4s0 stat points for the ideal case

AbbreviationsBSFC brake specic fuel consumptionDE diesel engineDE-ORC diesel engine-organic Rankine cycleEDF exergy destruction factorest estimated valueGWP global warming potentialHCCI homogeneous charge compression ignitionICE internal combustion engineODP ozone depletion potentialORC organic Rankine cycle

y 119 (2014) 204217 205investigation [813]. Zhang et al. [8] analyzed the characteristicsof a dual loop ORC, in which the coolant, intake air and engine ex-haust was used as waste heat source, integrated with a light-dutydiesel vehicle engine using R245fa and R134a as working uids.The optimization of the system by combining a heavy duty diesel(HDD) engine with a bottoming cycle that was presented (Ref.[9]) where R245fa and water were used as working uids to recov-ers waste heat energy from the sources. Based on various refriger-ant-based uids, Tian et al. [10] performed a study on uids andparameters optimization of ORCs integrated with stationary ICEswhere the engine exhaust gases were proposed to be a thermalsource and used in their solutions. Yu et al. [11] presented anORC simulation model, which had similar structure to actual bot-toming system, using R245fa as working uid to recover wasteheat energy of cooling water and engine exhaust gas. Wang et al.[12] performed a study of working uid selection of ORCs for en-gine WHR, and the refrigerants were used as working uids. Vajaand Gambarotta [13] coupled ORCs with a stationary ICE to estab-lish a combined power generation model, in which the overall ef-ciency was improved by 12% compared with the original engine,and three different working uids including benzene, R11 andR134a were employed for three different cycle setups.

Based upon the above-mentioned researches, it is found thatrefrigerants are most selected as working uid candidates in previ-ous studies, while critical temperature and thermal decompositiontemperature are relatively low for most refrigerants. A great tem-perature difference occurs in heat transfer between high-tempera-ture exhaust gases and uids, which inevitably results in highirreversible rate. Furthermore, there are some environmental

PPTD pinch point temperature differenceRD relative deviationref reference valueSP size parameterVFR volumetric ow ratioWHR waste heat recovery

zero ozone depletion potential (ODP) and a relatively low globalwarming potential (GWP) value [25]. Moreover, Alkanes have

ergissues for the applications of some refrigerants; so it is imperativeto nd appropriate uids to match with engine high-temperatureexhaust gases to obtain the best performance.

The availability of waste heat energy at different temperaturelevels is important to choose a suitable organic uid for the appli-cation of ORCs. Roy et al. [14] presented performance analysis of anORC, in which four refrigerant-based uids were used as workinguids, using the heat source at different temperature levels undersuperheated conditions. A detailed analysis was performed toinvestigate CO2-based transcritical power cycle in which low-gradewaste gases from industrial process were used as a heat source[15]. Performance was compared for both transcritical power cycleand subcritical ORC using various kinds of refrigerant-based uidsto exploit low-temperature geothermal energy [16]. Wang et al.[17] conducted an experimental study on recuperative Rankine cy-cle system in order to exploit low-temperature solar energy usingR245fa as working uid. Generally, low-temperature heat sourceshave a temperature lower than 230 C, and different working uidscan be used in ORC power generation systems to achieve outputpower.

However, high-temperature applications have attractedintensive interests in recent years. In contrast, there is less knowl-edge about high-temperature applications, in which parametersand working uids are greatly different from the low-tempera-ture applications. Typical uses include the combined cycles forbiomass applications [7], combustion fuel gas WHR [19], bottom-ing cycle for molten carbonate fuel cells [20], among others. Now-adays, Alkane-based working uids have been employed by ORCmanufacturers in order to achieve better system performance insome high-temperature applications as higher working tempera-tures can be reached by these compounds.

High-temperature ORCs based on Alkane-based working uidsare available in the open literature [2024]. Algieri and Morrone[21] evaluated cyclic Alkanes and linear Alkanes as workinguids in a biomass application. Lai et al. [22] proposed Alkane-based high-temperature ORCs in which thermodynamic proper-ties were obtained through computational program based onthe law of corresponding states. Siddiqi and Atakan [23]conducted a detailed performance comparison analysis and dis-cussion between Alkanes and other uids used in Rankine cyclesfor the waste heat sources with different temperature ranges.Linear Alkanes were adopted in a bottoming ORC in order to en-hance the micro-gas turbine performances [24]. From the aboveliteratures, it is evident that only few Alkane-based workinguids have been investigated. Moreover, most of the above-mentioned researches were performed based on the rst law ofthermodynamics and thermal efciency was assigned as the onlyperformance indicator. Up to now, however, there are few papersin which a systematic study was carried out on the applicationof Alkane-based working uids based on various indicators forWHR from DE exhaust gases.

The present paper focuses on evaluating the improvement ofBSFC by using DE exhaust gases as a heat source for saturated va-por power cycle, thus with a rather simplied design of heatrecovering unit, under the hypothesis of operating the engine atfull load while considering the cycles using Alkane-based uidswhere there are no more than 10 carbon atoms in the carbonbackbone. The aim is to investigate the performances of theAlkane-based working uids used in ORCs to recover engineexhaust gases with a high thermal level. Here, the exhaust gasesof DE with an outlet temperature of 519 C under a rated operat-ing condition are used as a heat source. A thermodynamic modelis established and six indicators, including thermal efciency (g),

206 G. Shu et al. / Applied Ennet power output per unit mass ow rate of hot exhaust (Pnet),total exergy destruction rate (IORC), exergy destruction factorbeen widely used in ORC process, in which good performancecan be achieved.

The positive slope of the saturated vapor entropy line in Tsdiagram is one of the main advantages of using organic uid asworking uid of Rankine cycle. The water steam cycles will be inthe two-phase region at the outlet of the turbine if a saturated cy-cle is performed, while organic uids are used instead of watersteam, with regard to the majority of the organic uids, the outletof the turbine will be in the superheated vapor region, which de-pends on the shape of the saturated vapor curve. The organic work-(EDF), turbine volume ow ratio (VFR) and turbine size parameter(SP), are proposed and analyzed. In addition, the inuence ofmolecular complexity on the indicators of VFR and SP is analyzed.Energy distribution of engine exhaust gases, as well as the BSFCimprovement after the integration of the bottoming ORC withthe DE, is discussed. The optimal value of each performanceindicator for different Alkane-based working uids is also beencompared respectively. Finally, the performance comparisonis made between Cyclohexane-based ORC and steam cycle withrelative pressure.

2. Alkanes

2.1. Nomenclature and taxonomy

Alkanes are one kind of simple chemical compounds only com-posed of hydrogen and carbon atoms. Since only single bonds arecontained in the compounds, they are often known as saturatedhydrocarbons. The saturated hydrocarbons can be categorized onthe basis of their molecular structure:

Linear Alkanes, having a general formula CnH2n+2, such as Pen-tane and Hexane.

Branched Alkanes, having a general formula CnH2n+2 (n > 3),such as Isopentane and Isohexane.

Cyclic Alkanes, having a general formula CnH2n (n > 2), such asCyclopentane and Cyclohexane.

Following standard rules, saturated straight-chain Alkanestake the sufx -ane, and the rst portion of the name is derivedfrom the Greek numeric prex that cites the number of carbons inthe Alkane, with the exceptions of Nonane, which has a Latin pre-x. Straight-chain Alkanes are sometimes indicated by the prexn- (for normal) where a non-linear isomer exists. While cyclicAlkanes are simply prexed with cyclo- before the correspond-ing straight-chain hydrocarbons. For example, C5H10 is Cyclopen-tane and C6H12 is Cyclohexane. The name rules of branchedAlkanes become relatively complex compared with straight-chainand cyclic Akanes, which are named as a straight-chain Alkanewith attached alkyl groups. Simple branched Alkanes, however,often have a common name using a prex iso- to distinguishthem from linear Alkanes. For example, C5H12 is Isopentane andC6H14 is Isohexane.

The thermodynamic characteristics of some selected Alkanesare shown in Table 1. These hydrocarbons are attractive sincesome of them have near-ambient boiling points to enable con-densation near atmospheric pressure. For the high-temperatureORC, Alkanes are selected as working uids because of theirappropriate critical temperature and pressure. In addition,Alkanes are also environmentally friendly working uids with a

y 119 (2014) 204217ing uids can be classied according to the shape of the saturatedvapor curves, which can be characterized by the slope of the

saturated vapor curve on a Ts diagram. The inverse of slope of thesaturated vapor curve can be expressed as

n ds

1

In subcritical cycles, the evaporation temperature is limited bythe critical temperature, and the uids can be divided into threecategories based on the critical temperature [18]:

Table 1Thermodynamic characteristics of Alkanes.

Name n Tcrit (C) Pcrit (kPa) n (J/kg K2) Type M (g/mol) ODP GWP

Pentane 5 196.55 3370 1.7151 Dry 72.149 0 Very lowHexane 6 234.67 3034 1.9675 Dry 86.175 0 Very lowHeptane 7 266.98 2736 2.1141 Dry 100.2 0 Very lowOctane 8 296.17 2497 2.1943 Dry 114.23 0 Very lowNonane 9 321.4 2281 2.2500 Dry 128.26 0 Very lowDecane 10 344.55 2103 2.2765 Dry 142.28 0 Very lowIsopentane 5 187.2 3378 1.8098 Dry 72.149 0 Very lowIsohexane 6 224.55 3040 2.1227 Dry 86.175 0 Very lowCyclopentane 5 238.54 4515 0.1902 Isentropic 70.133 0 Very lowCyclohexane 6 280.49 4075 0.7740 Isentropic 84.161 0 Very low

G. Shu et al. / Applied Energy 119 (2014) 204217 207dT sv

Eq. (1) can be used to predict the types of working uids, that is, awet uid (n < 0), an isentropic uid (n 0) and a dry uid (n > 0). Fora non-regenerative cycle, in order to achieve the highest efciency,the uid should be expanded directly from saturated vapor line toavoid unnecessary superheating [7]. For a wet uid (e.g. water),liquid droplets may be formed during expansion in turbine if theuid is expanded directly from saturated vapor line [14]. Dry andisentropic uids, however, which take the vast majority of organicuids used in ORCs, may remain in the state of saturated or super-heated vapor even after the expansion from the turbine [16]. Thus,dry and isentropic uids exhibit better thermodynamic efcienciesof expansion devices, as there are no liquid droplets in the turbine,contrary to wet uids.

Fig. 1 presents the inverse of the slope of the saturated vaporcurve (n) determined by Eq. (1) with different induced tempera-tures, and the reference temperature is set as 60 C (the conden-sation temperature of ORC) for the working uids considered. Asseen in Fig. 1, the values of n are all higher than zero for theinvestigated working uids at reference temperatures, and theuids with this behavior are known as dry uids or isentropicuids according to the aforementioned classication. The cate-gory results for the selected working uids can be seen inTable 1.Fig. 1. The inverse of slope of the saturated vapor curve as a function of reducedtemperature. High temperature uids, wherein the critical temperatures areabove 250 C.

Medium temperature uids, wherein the critical temperaturesare in the range 150250 C.

Low temperature uids, wherein the critical temperatures arebelow 150 C.

According to the critical temperatures as displayed in Table 1,the selected working uids belong to the category of medium-hightemperature.

2.2. Molecular complexity

Molecular structures have great inuence on thermo-physicalproperties of uids. The difference in molecular structure of differ-ent uids can be characterized as molecular complexity. Harincket al. [26] dened the complexity of the uids as follows:

r 2 M CpTcrit RR

2

wherein Tcrit means the critical temperature of the uid (K), Cp is theheat capacity under constant pressure (J/mol/K), and R denotes theuniversal gas constant (J/mol/K).

Molecular complexity has positive inuence on the specicisentropic work in accordance with the literature [26]. For adense-gas, the more complex the molecular structure is, the closerFig. 2. Molecular complexity of selected working uids.

to an ideal gas behavior, which results in higher specic isentropic

mended to be added to improve ORC performance in some cases.

complexity and only minor performance improvement [16]. Also,it has been certied that signicant increase in system irreversibil-ity will be achieved if superheating and subcooling of the workinguids are employed [27].

Fig. 4 shows Ts process diagrams of ORC for engine exhaustgases heat recovery. After condensation, the saturated workinguid is pressurized through pump to the set-pressure in the ORCprocess (12). Saturated vapor is generated in the evaporator, inwhich the heat from engine exhaust gas is absorbed by the cold

Fig. 3. Schematic diagram of DE-ORC combined system.

208 G. Shu et al. / Applied Energy 119 (2014) 204217In the present work, a regenerator will not be included as it willlead to considerable increase of heat exchanger cost and systemwork output. The molecular complexity of selected working uidscan be seen in Fig. 2.

3. System modeling

3.1. Description of ORC system

As shown in Fig. 3, the combined system is composed of DE andits bottoming cycle. Here, only subcritical saturation steam cycle isemployed in the proposed solutions. A regenerator is recom-ge of the efciencies of turbine and pump are from 75 % to 85 %

Fig. 4. Ts diagram of the ORC process.according to literature [8,16,2729]. For this study, the efcienciesof turbine and pump have been imposed as 0.8, which is reason-able and close to the actual value for ORC applications [7,15,21].The pinch-point temperature difference (PPTD) is the smallesttemperature difference in the ORC heat exchanger, establishingthe maximum allowable evaporation pressure and, thus, limitingthe ORC performance. For the evaporator, the PPTD (DTg) at pinchpoint, to meet the gas-uid heat exchanger performance, is consid-ered to be 30 C [10,13]. Note that the outlet exhaust temperature(Tg,out) depends on the characteristics of ORC heat exchanger. Theminimum exhaust gas temperature at the evaporator outlet shouldbe above 90 C. Otherwise, low temperature corrosion will occur if

Table 2The main parameters of the diesel engine under rated conditions.

Parameter Values Units

Power output 235.8 kWTorque 1500 N mExhaust temperature 519 CRotate speed 1500 rpmFuel consumption 47.79 kg/hBrake fuel consumption rate 202.7 g/kW hworking uid (23). And then expansion work is produced by tur-bine after the saturated vapor produced in the evaporation processenters the turbine (34). After leaving the turbine, the exhaustedworking uid vapor enters the condenser, in which saturated li-quid is generated after condensation process, and then a cyclecompletes (41). The energy and exergy analysis will be performedin the following section after choosing each system or componentin Fig. 3 at control volume. In addition, proper simplication ismade to simplify computing and complexity without losing com-putational accuracy. The cycle is assumed to work in steady statein this work, and the heat loss by radiation and pressure drop inthe pipes, condenser and evaporator can be neglected.

3.2. Initial conditions and boundary conditions

In the present study, a commercial 6-cylinder, 4-stroke super-charged diesel generator set is employed as topping system, andthe main parameters of the DE are listed in Table 2. Among them,the available exhaust gas temperature is obtained after the turbo-charger of the DE. ORC will be employed in the study as bottomingcycle to recover the engine exhaust waste heat. The gas propertiescan be evaluated by the composition of the exhaust gases, includ-ing N2 = 73.04%, H2O = 5.37%, O2 = 6.49%, CO2 = 15.10% based onmass. In this work, the rated operating conditions are assumedfor the adopted engine, and the most appropriate organic workinguids that would better t the given heat source conditions will bedetermined.

As shown in Fig. 4, evaporators used in the ORCs are usuallysimple components designed as heat exchanger to directly usehot gases released from the thermal source, and intermediary u-ids (e.g. heat conducting oil) can be excluded [13]. The typical ran-Exhaust mass ow 990.79 kg/hSmoke intensity 0.55 FSN

t of t

nergFig. 5. The cycle upper limi

Table 3Parameters and boundary conditions of the ORC model.

Parameter Value Units

Inlet temperature of exhaust gases 519 CInlet temperature of cooling water 25 CPPTD (condensation) 10 CPPTD (evaporation) 30 CCondensation temperature 60 CIsentropic efciency (turbine/pump) 0.8

G. Shu et al. / Applied Ethe temperature is below this threshold [27]. Condensing temper-ature has been set to 60 C in this study. The corresponding con-denser pressures are above 5 kPa, as suggested in literature [19],but a lower condensing pressure will be chosen for Decane andNonane (1.52 kPa and 3.98 kPa respectively) in order to keep con-densing temperature consistent with that of other working uids.The lower pressure and temperature limit of the cycle can be seenin Fig. 5 for the selected working uids. A water-cooled condenserwill be adopted in this study. And it is very easy to replace thewater cooling type condenser with an air-cooled type condenserfor WHR application at the condensing temperature consideredabove. Other parameters and boundary conditions of the ORC mod-el can be seen in Table 3.

The actual temperature and pressure limits of the adopted cyclewill be determined in the next step according to cycle characteris-tics of each working uid. Generally speaking, the higher the pres-sure ratio it has, the more mechanical energy will be extractedfrom a given mass of working uid due to its higher thermal ef-ciency. However, some practical restrictions have to be taken intoaccount when attempting to increase operation pressure towardthe critical pressure.

The systemmay become unsteady near the critical region as thepressure has a high sensitivity to minor temperature change at -this time. Namely, great change in pressure will take place whentemperature varies within a very small range. Hence, an appropri-ate higher limit of the cycle should be considered based on the crit-ical point of the uid. However, there is no uniform view onchoosing a rational gap between the cycle upper limit and the crit-ical point of the uid.

Schuster and Heberle [28,29] suggested setting critical pressuremultiplied by a certain coefcient (e.g. 0.7) as the pressure upperlimit of the cycle. Since the differences of critical properties amongdifferent working uids are evident, a xed ratio used to determine

Generator efciency 1the pressure limit may not be a very unanimous way to identifypressure interval between the pressure upper limit and the criticalpressure.

In the present work, the upper limit of the adopted cycle will bedetermined with the slope of the saturated vapor entropy line inthe Ts diagram. Maximum live vapor pressure of the turbineshould not exceed the pressure upper limit to eliminate theimpingement of liquid droplets on turbine blades. The pressureupper limit can be determined with the point (point 3 inFig. 4) at which the slope of the saturated vapor entropy line inthe Ts diagram is equal to innity, and beyond which condensa-tion will take place within the turbine (point "30 0" in Fig. 4). Theupper cycle limits have been calculated based on this criterion,as shown in Fig. 5.

3.3. Mathematical model

3.3.1. Energy analysisAn evaluation model based on the rst and the second laws of

thermodynamics is introduced to describe the energy behavior ofdifferent working uids under different working conditions. These

he selected working uids.

y 119 (2014) 204217 209formulas introduce the equations to perform the thermodynamiccomparative analysis.

For the pump:

_Wp _Wp;s=gp _mfh2s h1=gp 3For the evaporator:

_Qe _mghg;in hg;out _mfh4 h2 4For the turbine:

_W t _W t;sgt _mfh3 h3sgt 5For the condenser:

_Q c _mcwhc;out hc;in _mf h4 h1 6For the net output:

_Wnet _W t _Wp 7Thermal efciency of the cycle can be calculated as:

g _Wnet=Qe _W t _Wp=Qe 8The net power output per unit mass ow rate of exhaust is given by

Pnet _Wnet= _mg 9

used to compare different turbine sizes and is an appropriate indi-

wherein ref denotes the reference value, est means the estimatedvalue.

According to the developed model, a simulation program inwhich the process units are modeled is developed using Matlablanguage. Under the conditions of the same operation, the obtainedresults have been compared with those of Vaja and Gambarotta[13] to validate the present model. The results of the mass ow

Table 4The results have been compared for the present work and reference [13].

Parameter Wnet (kW) Mass ow rate (kg/s)

Source Ref. [13] Present RDa Ref. [13] Present RDa

Benezene 349.3 351.2 0.54% 2.737 2.743 0.22%R11 290.3 292.4 0.72% 7.487 7.514 0.36%R134a 147.5 148.7 0.81% 8.9667 9.013 0.52%

a RD: relative difference, see Eq. (19).

ergy 119 (2014) 204217cator of its relative cost. Larger size parameter means bulkier andmore expensive turbines.

3.4. Validation of the model

The RD (relative deviation) is given by3.3.3. Turbine designVolumetric ow ratio (VFR), accounting for the compressibility

effect through the expansion, is dened as:

VFR _V4= _V3 17wherein _V3 and _V4 are the volume ows at inlet and outlet of theturbine respectively.

According to Ref. [30], lower values of VFR deliver higherturbine efciency. In addition, according to the viewpoint in Ref.[24] the turbine efciency over 80% can be achieved only whenVFR is below 50.

Turbine size parameter (SP), taking volumetric ow in turbinesoutlet and isentropic enthalpy difference throughout the turbineinto account, is expressed as

SP _V4

q=DH1=4is 18

wherein _V4 stands for volumetric ow in the turbines outlet,DHis isthe isentropic enthalpy difference throughout the turbine. Theparameter of turbine size is shown as SP here with meter as unit.

Since there is certain proportion between the SP and the actualturbine size, SP, being regarded as an indicator of turbine dimen-sions, given by Eq. (18), can be used to evaluate the actual turbinesize in place of a detailed design calculation [24,30]. This can beThe BSFC of the engine and the DE-ORC conguration can beestimated as:

BSFCDE _mfuel= _WDE 10

BSFCDE-ORC _mfuel= _WDE _Wnet 11

3.3.2. Exergy analysisThe thermodynamic irreversibility, occurring generally on each

component, can be determined with exergy analysis method. Atany point of the ORC, the exergy ow rate of working uid canbe calculated as:

_Ei _mf hi h0 T0si s0 12The 0 subscripts are used to denote the specied dead reference

state under ambient pressure and temperature conditions. In thepresent work, the dead state is specied by T0 = 25 C andP0 = 1 atm.The irreversibilityof eachcomponentcanbeexpressedas

_Ii X

_Ein X

_Eout _W i 13The total irreversibility of the cycle can be calculated as:

_IORC X

_Ii 14The exergy loss of the exhaust gases leaving the ORC can be calcu-lated as:_Iloss _mgcp;gTg;out T0 T0 logTg;out=T0 15Exergy destruction factor (EDF) can be determined as:

EDF _IORC= _Wnet 16

210 G. Shu et al. / Applied EnRD ref -estref

100% 19rate and net power output have been compared between this workand the reference as shown in Table 4. A corresponding result withreference is obtained by the developed model and the relative dif-ference is less than 1%. Therefore, the model has been validated.

4. Results and discussion

The calculated conditions of the present system are given inTable 3 along with the characteristics of the turbine, pump, etc.,and the main parameters of an actual DE are expressed in Table 2,which provides the heat source (exhaust gases) conditions of thedeveloped model. The thermodynamic properties of uids arecalculated with REFPROP 8.0 software [31]. The results of a com-parative study for different working uid-based ORC systems willbe presented in the following paragraphs, and different workinguids can be equally evaluated. Based on six different indicatorsconsidered in this analysis, including g, Pnet, IORC, EDF, VFR andSP, the performances of different working uids investigated forthe bottoming ORC system are shown in Figs. 612. For the conve-nience of description, relative pressure, which is the ratio ofevaporating pressure (Pevap) and the maximum allowableevaporating pressure (Pmax), is introduced.

4.1. The inuence of evaporating pressure

Volume ow ratio (VFR) is the ratio between outgoing isentro-pic and incoming volumetric ows, showing how much the uidvolume increases through expansion. VFR is a particularlyFig. 6. Turbine volume ow ratio (VFR) with relative pressure.

nergG. Shu et al. / Applied Esignicant parameter since it is related to turbine efciency andwhether a single stage expander could be employed. Accordingto Invernizzi et al. [24] a VFR of above 50 is not considered, since

Fig. 7. The turbine size parameter (SP) with relative pressure.

Fig. 8. Thermal efciency with relative pressure.

Fig. 9. Net power output per unit mass ow rate of hot exhaust with relativepressure.y 119 (2014) 204217 211in this case turbine efciency will be less than 80%. This is mainlytrue for axial single-stage turbines. As shown in Fig. 6, the curvesascend with the increase of evaporating pressures for all uids.

Fig. 10. Total exergy destruction rate with relative pressure.

Fig. 11. Exergy destruction factor (EDF) with relative pressure.

Fig. 12. Turbine size parameter (SP) and turbine volume ow ratio (VFR) withmolecular complexity.

ergVFR values increase with the increase of the numbers of carbonatom in chain for the three Alkane categories considered underthe same relative pressure. Cyclic Alkanes show higher values thancorresponding linear Akanes with the same number of carbonatom numbers in chain, whereas branched Alkanes present lowervalues than corresponding linear Akanes. Based on the demarca-tion line of VFR = 50, the VFR values of the working uids, includ-ing Isopentane, Pentane, Cyclopentane and Isohexane are below 50under all evaporating pressures, and the VFR values of the rest arebelow 50 but only under a portion of evaporating pressures. Specif-ically, in addition to the aforementioned four working uids, sin-gle-stage turbines cannot be adopted for the rest working uidsunder some evaporation pressures due to relatively low turbineefciency. For the working uids with high carbon atom numbersin chain, taking Decane as an example, VFR value is close to 1200under the maximum evaporating pressure, so multi-stage turbineshave to be adopted, which increases turbine size and relative cost.

Turbine size parameter (SP) given by Eq. (18) is an indicator ofturbine size. SP is proportional to actual turbine size, which can beused to compare different turbine sizes, thus is an appropriate indi-cator of its relative cost. Larger size parameters mean that bulkierand more expensive turbines will be employed. As shown in Fig. 7,it is evident that SP initially drops rapidly with the increase ofevaporating pressure, and then decreases more slowly under high-er evaporating pressures for all considered working uids. SP val-ues increase with the increase of the numbers of carbon atom inchain under certain relative pressures for the considered three cat-egories of working uids. There is a similar trend of the workinguids with the same carbon atom number in chain. SP values of lin-ear Alkanes are between that of corresponding cyclic Alkanes andbranched Alkanes with the same numbers of carbon atom underhigh relative pressure. Curve intersections appear only under verylow relative pressure. It is evident that a larger turbine size will berequired for the working uids with more numbers of carbonatoms in chain. It can be observed that a maximum SP value of1.168 m is achieved by Decane under the maximum evaporatingpressure, while Isopentane achieves the minimum value of0.128 m among all working uids. The working uids includingIsopentane (SP = 0.128 m), Pentane (SP = 0.142 m), Cyclopentane(SP = 0.165 m) and Isohexane (SP = 0.189 m) require lower SP val-ues, which is in agreement with the above-mentioned observationsregarding the inuence of the volumetric ow rate on the cost andsize of the system. In addition, for some uids with high molecularcomplexity r, the demarcation points of VFR = 50 have beenmarked in Fig. 7. For these uids, the minimum values of SP havebeen extended in different degrees (e.g. SP = 1.257 m for Decane)according to this constraint condition. For the uids consideredin this paper, the optimal values of SP have been listed in Table 6after the isentropic turbine efciencies are conned to an accept-able range.

Fig. 8 shows the variation of thermal efciency with relativepressure for Alkane-based working uids considered. Although lin-ear Alkanes and branched Alkanes have similar variation with rel-ative pressure, branched Alkanes have a lower thermal efciencythan corresponding linear Alkanes containing the same car-bon atom number. The thermal efciency increases with the num-bers of carbon atom in chain for linear Alkanes and branchedAlkanes. However, a rapid increasing trend can be found for cyclicAlkanes, which is obviously different from linear Alkanes andbranched Alkanes. This is mainly due to the facts that higher evap-oration pressures can be obtained by cyclic Alkanes under thesame relative pressure, which contribute to achieving a highercompression ratio compared to other uids. In addition, the cyclic

212 G. Shu et al. / Applied EnAlkanes also have relatively high levels of evaporation tempera-ture, which will contribute to achieving better temperature match-ing between heat carrier (exhaust gas) and uids to reduce the losscaused by heat transfer between them. The highest thermal ef-ciency of about 19.3% is achieved by Cyclohexane among all uidsconsidered under maximum relative pressure. Cyclohexane is fol-lowed by Decane. Decane exhibits a thermal efciency about18.1% that is about 6% lower than Cyclohexane. Moreover, Cyclo-pentane also shows high thermal efciency of about 17.8%, whichis about 8% lower than Cyclohexane. For some uids with highmolecular complexity r, however, the thermal efciencies havebeen diminished to various extents by conning the isentropic tur-bine efciencies within a limited range. Taking Cyclohexane as anexample, the thermal efciency is reduced from 19.3% to 18.2%.The maximum thermal efciencies of uids considered within anacceptable isentropic turbine efciencies range have been shownin Table 6.

Net power output per unit mass ow rate of hot exhaust can beused to denote the power output performance of the exhaust gases.Fig. 9 presents variation of Pnet values for various working uidsunder different relative pressures. Higher Pnet value means thatmore power output can be obtained with the same heat sourcemass ow rate. As shown in Fig. 9, Pnet value increases with the rel-ative pressures. The increase trend of various working uids isobvious under low relative pressures, and becomes smooth nearmaximum live vapor pressure. In addition, Pnet values ascend withthe increase of the numbers of carbon atom in chain under any rel-ative pressure for linear Alkanes, and branched Alkanes show asimilar trend as the linear Alkanes do. However, cyclic Alkanesshow different trends, and Pnet values are signicantly higher thanthat of the corresponding linear Alkanes and branched Alkanescontaining the same number of carbon atom for all admissible livevapor pressures. For cyclic Alkanes, the variation trends of Pnet aresimilar to that of thermal efciency (g) in Fig. 8, which are causedby the same reasons as mentioned above. Among all the consideredworking uids, Cyclohexane shows the highest Pnet value of about91.7 kJ/kg under maximum allowable evaporating pressure3529.8 kPa. Decane also presents higher Pnet value of about86.2 kJ/kg under maximum allowable evaporating pressure1898.3 kPa, which is approximately 6% lower than Cyclohexane.Moreover, Cyclopentane also presents high Pnet value of about84.3 kJ/kg (at 3313.6 kPa), which is 8% lower than that of Cyclohex-ane. However, for some uids with high molecular complexity r,the maximum Pnet values have been decreased at different levelswithin the scope of the isentropic turbine efciencies restriction,among which the Pnet value of Cyclohexane has fallen to86.47 kJ/kg from 91.7 kJ/kg. The maximum Pnet values for the uidsconsidered within an acceptable isentropic turbine efcienciesrange have been presented in Table 6.

The behavior of the exergy destruction rate is opposite. It de-creases with relative pressure until a minimum value is obtained,as shown in Fig. 10. This is mainly due to the entropy generationdecrease in evaporation process-the main contribution to the exer-gy destruction rate due to a great temperature difference existingbetween the heat sources and the working uids considered inthe systems. As shown in Fig. 10, for linear Alkanes, exergydestruction rate decreases with both the increase of the numbersof carbon atom in chain and the rise of relative pressure, whichis also applicable to branched Alkanes and cyclic Alkanes. Differentfrom branched Alkanes, cyclic Alkanes present lower IORC value of29.5 kW (at 3529.8 kPa) and 32.3 kW (at 2379.5 kPa) for Cyclohex-ane and Cyclopentane, which are 4.6% and 14.5% higher than thatof Decane respectively. Decane presents the lowest value of28.2 kW (at 1898.3 kPa) among all the working uids considered.However, the cases are quite different after the constraint is givento limit the isentropic turbine efciencies to an acceptable range.

y 119 (2014) 204217The minimum IORC values for some uids with high molecular com-plexity r have been increased to different extents. Although theminimum IORC value is also achieved by Cyclohexane after the

isentropic turbine efciencies are conned within an acceptablerange, it has increased from 29.5 kW to 31.11 kW. The minimumvalues of IORC for uids considered within an acceptable isentropicturbine efciencies range have been summarized in Table 6.

The exergy destruction rate reects, to a certain extent, energyconversion performance for the systems considered, but the poweroutput performance cannot be covered. Exergy destruction factor(EDF) is proposed in order to evaluate the energy conversion andpower output performance of the systems considered comprehen-sively. As shown in Fig. 11, EDF decreases with both the increase ofthe numbers of carbon atom in chain and the rise of relative pres-

to discuss the performances when the values of VFR are no morethan 50. Performance sequence of the working uids at VFR = 50has been given in Table 6. Compared with linear Alkanes andbranched Alkanes, cyclic Alkanes have obvious advantage in ther-mal efciency (g), Pnet, IORC and EDF. The maximum thermal ef-ciency (g) and Pnet value are 18.20% and 86.47 kJ/kg respectivelyfor Cyclohexane. And the minimum IORC value of 31.11 kW andEDF value of 1.296 are also obtained by Cyclohexane. However,the performances of linear Alkanes, especially for the uids of highmolecular complexity r, have been signicantly degraded whenthe isentropic turbine efciencies are conned to the accept-

vap/P

G. Shu et al. / Applied Energy 119 (2014) 204217 213sure for the considered three categories of working uids. The de-crease trend of various working uids is obvious under low relativepressures and becomes smooth near maximum live vapor pres-sure. There are only minor differences among them when the min-imum values are reached for the majority of the working uids.Cyclohexane presents the lowest EDF value of 1.16 (at3529.8 kPa), owing to a relatively high power output performanceand low exergy destruction rate. In addition, Decane and Nonanealso present lower EDF values of 1.18 (at 1898.3 kPa) and 1.22(at 2041.6 kPa), which are 1.7% and 5.2% higher than that of Cyclo-hexane respectively. In addition, after conning the isentropic tur-bine efciencies to an acceptable range, the distributions of EDFare similar to that of IORC in Fig 10. The minimum EDF value is stillachieved by Cyclohexane, but it has been increased from 1.16 -to 1.296. The optimal values of EDF for uids considered withinacceptable isentropic turbine efciencies range have been givenin Table 6.

Through the above analysis, it is evident that the maximum va-lue of Pnet and VFR, and the minimum value of IORC, EDF and SP ap-pear when evaporating pressures reach the maximum values(Pevap/Pmax = 1) for selected working uids. Table 5 reports themain performance sequence of the uids under maximum relativepressure (Pevap/Pmax = 1). It can be observed that the maximumthermal efciency (g) of 19.32% and Pnet value of 91.68 kJ/kg andthe minimum EDF value of 1.157 are achieved by Cyclohexane,while the maximum thermal efciency (g) and Pnet value decreaseto 18.13% and 86.17 kJ/kg respectively for Decane. From Table 5 itcan also be observed that minimum values are obtained by Decanein the case of IORC = 28.24 kW. However, it should be notedthat Decane reaches the maximum values in the cases ofVFR = 2590.73 and SP = 1.168 m respectively. The minimum values(the optimal value in these cases) of 14.55 for VFR and 0.128 m forSP are achieved by Isopentane. Simultaneously, a large deviationcan be observed in VFR and SP among different working uids.Whether a working uid is the optimized one or not depends onthe trade-off between different indicators.

In practice, the optimal value under the maximum relativepressure Pevap/Pmax = 1 may fail to realize for some uids as theVFR values are more than 50 under this pressure. The isentropicturbine efciencies will be lower than 80% at this time if a singlestage expander is employed. So it is more rational and practical

Table 5Performance sequence of the working uids under the maximum relative pressure Pe

Working uids g (%) Pnet (kJ/kg) IORC (kW) EDF ()

Decane 18.13 86.17 28.24 1.180Nonane 17.95 85.34 29.01 1.224Octane 17.67 83.97 30.01 1.287Heptane 17.10 81.22 31.40 1.392Cyclohexane 19.32 91.68 29.46 1.157Hexane 16.09 76.38 33.38 1.574Isohexane 15.31 72.68 34.33 1.701Cyclopentane 17.78 84.34 32.30 1.379

Pentane 14.08 66.82 36.64 1.974Isopentane 13.32 63.22 37.60 2.142able range. For SP, cyclic Alkanes are wholly at medium or lowerlevel while linear Alkanes reach the medium and even high levelfor some uids with high molecular complexity r. On the whole,cyclic Alkanes have more outstanding performance than other u-ids after the isentropic turbine efciencies are limited to an accept-able range.

4.2. The inuence of molecular complexity

Fig. 12 shows SP and VFR of the ORC with molecular complexityof working uids considered, where Tevap = 170 C and Tcond = 60 Care assumed for WHR cycle. Working uids are arranged in thetransverse coordinate axis in ascending order of r in the gure. Itcan be seen that SP and VFR are quite variable and strictly depen-dent on the working uids considered. In the cases analyzed here,SP varies from a minimum of about 0.13, in correspondence withhigher values of r, to about 1.3 and the variety of VFR takes placewithin a wide range from about 13 for Cyclopentane to about 58for Decane. Based on all performance factors discussed in this sec-tion, higher r results in higher SP and VFR values. The two excep-tions to this rule are Cyclohexane and Cyclopentane. This meansthat large turbine size and volume ow ratio will be required inORC system employing the working uids of high r if the workinguids are not cyclic Alkanes.

It has been made clear from above-mentioned standpoint thatefciencies higher than 80% can only be achieved when VFR is lessthan 50 for single stage axial turbines. For given Fig. 12, from theVFR respect, target efciency (higher than 80%) of turbine can onlybe realized when r is below 92. Among the working uids given inTable 1, only Decane (VFR = 58) dissatises this limit under theaforementioned conditions. Summarizing the impact of r on SPand VFR, less complex uids are preferred, since such uids resultin acceptable turbine efciency and low SP and VFR values, so thatthe use of simple expanders such as single-stage turbines will beallowed.

4.3. The energy and exergy distributions

The major purpose to investigate the energy distribution of en-gine exhaust gases is to determine energy grade and energy distri-bution. Energy grade, on which energy conversion performance

max = 1.

VFR () SP (m) Pevap (kPa) Tevap (C) Pcond (kPa)

2590.73 1.168 1898.3 337.1 1.521012.62 0.761 2041.6 313.4 3.98404.61 0.497 2200.3 287.0 10.45157.44 0.325 2379.5 257.1 28.04122.46 0.250 3529.8 269.0 51.9054.18 0.213 2499.0 221.3 76.4242.95 0.189 2526.7 212.0 100.6432.76 0.165 3313.6 215.3 142.44

17.98 0.142 2581.4 179.3 214.5414.55 0.128 2632.8 171.1 273.13

depends, is determined by exergy content in the energy with theabove analysis. Fig. 13 reveals the energy distribution of the hotsource (engine exhaust gases) and the improvement of BSFC beforeand after the integration of the bottoming ORC with the DE for theworking uids considered. It is evident that the heat energy ab-sorbed by the ORC system accounts for about 73.5% from exhaustgases and, in which, the exergy occupies about 46.2% of this partof the energy. The above data indicate that the energy absorbedby an ORC system takes on a high energy grade. The recovery sys-tem converts a part of this exergy into turbine work, with conver-sion quantity depending on the adopted working uids. The

Table 6Performance sequence of the working uids at VFR = 50.

Working uids g (%) Pnet (kJ/kg) IORC (kW) EDF ()

Decane 13.73 65.36 36.6 2.017Nonane 14.49 68.97 35.45 1.851Octane 15.28 72.70 34.26 1.700Heptane 15.87 75.29 33.47 1.600Cyclohexane 18.20 86.47 31.11 1.296Hexane 16.00 75.99 33.49 1.587Isohexaneb 15.31 72.68 34.33 1.701Cyclopentaneb 17.78 84.34 32.30 1.379Pentaneb 14.08 66.82 36.64 1.974Isopentaneb 13.32 63.22 37.60 2.142

b VFR < 50, performance sequence of the working uid at maximum achievable value

214 G. Shu et al. / Applied Energconversion capacities (the ratio of turbine work to the exergy in ex-haust gases) of cyclic Alkanes (43.8% for Cyclohexane, 40.7% forCyclopentane) are higher than that of linear Alkanes (40.1% forDecane, 39.9% for Nonane) and branched Alkanes (35% for Isohex-ane, 31.2% for Isopentane). The pump power occupies only a smallportion of turbine work depending on the adopted working uids.It is evident that high pump work will be required for the workinguids with less numbers of carbon atoms. After the heat is ab-sorbed by the ORC system, exhaust gases still contained a portionof available energy, accounting for about 25% of the total energy.However, this part of the energy represents a very low energyFig. 13. The energy ow distribution and BSFC with different working uids wherePevap/Pmax = 1.grade since the exergy occupies only about 4% of the energy, andits utilization value is still low at present.

As shown in Fig. 13, the BSFC improvement for DE-ORC com-bined systems is signicant compared with the original engine,and there are different improvement degrees depending on theadopted working uids. Cyclohexane presents the greatest BSFCimprovement, about 10%, due to high power output. In addition,due to relatively high power output and low pump power con-sumption, Decane and Nonane also present higher BSFC improve-ment, about 9.2% for Decane and 9.1% for Nonane. Isopentaneshows the lowest BSFC improvement, about 6.9%, because of lowpower output and high pump power consumption. However, thecases are quite different after the limited condition is given to limitthe isentropic turbine efciencies to an acceptable range. The max-imum BSFC improvements of DE-ORC combined systems for someuids with high molecular complexity r have been decreased atdifferent levels, among which the maximum BSFC improvementof Cyclohexane has fallen to 9.3% from 10%. This is mainly due tothe facts that the rapid decrease of pressure ratios will be observedwith the sharp drop from the peak of evaporation pressures, espe-cially for some uids with high molecular complexity r, after theisentropic turbine efciencies are conned to an acceptable range.

The performance sequence of different working uids consid-ered in the present paper based on four different screening criteriaare shown in Table 7. The four different screening criteria aredened as follows:

RD1 Pnet;max Pnet=Pnet;max 100%RD2 IORC;min IORC=IORC;min 100%RD3 EDFmin EDF=EDFmin 100%RD4 SPmin SP=SPmin 100%

20

As shown in Table 7, each uid has great discrepancies in the four

VFR () SP (m) Pevap (kPa) Tevap (C) Pcond (kPa)

50 1.257 80.0 165.1 1.5250 0.806 199.0 178.2 3.9850 0.518 492.6 194.0 10.4550 0.332 1142.8 209.1 28.0450 0.255 2389.1 228.0 51.9050 0.214 2045.6 218.3 76.4242.95 0.189 2526.7 212.0 100.6432.76 0.165 3313.6 215.3 142.4417.98 0.142 2581.4 179.3 214.5414.55 0.128 2632.8 171.1 273.13

of VFR.

y 119 (2014) 204217screening criteria. There is relatively minor deviation for net poweroutput per unit mass ow rate of hot exhaust for the working uidsunder each optimized operating condition. The greatest deviationvalue is 31%. Cyclohexane performs best among all the uidsaccording to the rst screening criterion. Moreover, the greatestdeviation value of EDF value of the working uids under each opti-mized operation condition is 85.1%, and Cyclohexane shows betterperformance than the other uids. With regard to the secondscreening criterion, the greatest deviation is 33.1% in terms of thetotal exergy destruction rate of uids under each optimized operat-ing condition, and Decane shows better performance than the otheruids in this case. However, Decane shows the worst performancein terms of SP under optimized operating condition, and the great-est deviation value is as high as 812.2%. Furthermore, the condens-ing pressures are less than 5 kPa for Decane and Nonane at thecorresponding condensing temperature of 60 C. Based on thefourth screening criterion, Isopentane, which contains the leastnumbers of carbon atoms in chain, presents the optimized value

in this case. But Isopentane shows the worst performance in othersituations. Although the optimized uids are not the same accord-ing to different screening criteria, the cyclic Alkanes Cyclohexaneand Cyclopentane are considered the most suitable working uidsdue to relatively high power output, reasonable condensingpressure, relatively low irreversibility and SP based on the afore-

cle is made under the same initial and boundary conditions (sameheat source/sink temperatures and other setting conditions). The

working uids to be compared with water owing to its good over-all performance.

As shown in Fig. 14, the steamoffers better efciency than that ofCyclohexane over the whole relative pressure range. This is main-ly because the higher level of temperature (highest temperature of482 C for steam compared to 269 C for Cyclohexane) can be

Table 7Performance sequence of different working uids according to four different screening criteria.

Pnet Pe (kPa) RD1 IORC Pe (kPa) RD2 EDF Pe (kPa) RD3 SP Pe (kPa) RD4

Cyclohexane 3529.8 0.0 Decane 1898.3 0.0 Cyclohexane 3529.8 0 Isopentane 2632.8 0Decane 1898.3 6.0 Nonane 2041.6 2.7 Decane 1898.3 2.0 Pentane 2581.4 10.5Nonane 2041.6 6.9 Cyclohexane 3529.8 4.3 Nonane 2041.6 5.8 Cyclopentane 3313.6 28.7Cyclopentane 3313.6 8.0 Octane 2200.3 6.3 Octane 2200.3 11.2 Isohexane 2526.7 48.0Octane 2200.3 8.4 Heptane 2379.5 11.2 Cyclopentane 3313.6 19.2 Hexane 2499 66.6Heptane 2379.5 11.4 Cyclopentane 3313.6 14.4 Heptane 2379.5 20.3 Cyclohexane 3529.8 95.3Hexane 2499 16.7 Hexane 2499 18.2 Hexane 2499 36.0 Heptane 2379.5 153.5Isohexane 2526.7 20.7 Isohexane 2526.7 21.6 Isohexane 2526.7 47.0 Octane 2200.3 288.1Pentane 2581.4 27.1 Pentane 2581.4 29.7 Pentane 2581.4 70.6 Nonane 2041.6 494.1Isopentane 2632.8 31.0 Isopentane 2632.8 33.1 Isopentane 2632.8 85.1 Decane 1898.3 812.2

G. Shu et al. / Applied Energy 119 (2014) 204217 215difference is that the essential condition of the superheating mustbe achieved to keep the steam in the superheated vapor regionafter expansion. The Cyclohexane is selected from Alkane-basedmentioned results. Also, Heptane and Hexane can be consideredas the suitable working uids when considering all indicatorscomprehensively.

4.4. Comparison of ORC and steam cycle

Water is a typical wet uid that is characterized by a negativeslope of the saturated vapor curve on a Ts diagram. This impliesthat the superheating of the vapor is required in a steam cycle toavoid moisture formation in the steam turbine. Therefore, a muchhigher evaporation temperature must be achieved in a conven-tional steam cycle. Much higher requirements are imposed to tem-perature level of the waste heat source that limits the maximumsuperheating temperature and the evaporation pressure of thesteam cycle. In this study, the waste heat temperature (exhausttemperature) is over 500 C. Thus the steam cycle is availablein a certain range. A contrastive analysis of ORC and steam cy-Fig. 14. Performance comparison between Cyclachieved by steam due to superheating. The total irreversibility ofsteam has a similar variation trend to that of Cyclohexane with theincrease of relative pressure. But the steam can provide less totalirreversibility than that of Cyclohexane under higher relative pres-sure. With the increasing of superheating temperature, bettertemperaturematchingbetweenheat carrier (exhaustgas) andsteamcan be achieved. There is a very similar trend of SP between Cyclo-hexane and steam with relative pressure. However, the SP of steamis about two times higher than that of the Cyclohexane. This meansa larger turbine will be required for steam cycle.

The above analysis indicates that steam has more advantages inthermal efciency than Cyclohexane in the current conditions, butbulkier and more expensive turbines must be employed for steamat the same time. The higher thermal efciency of steam is main-ly caused by the higher superheating temperature. So it has a highdemand for the quantity and quality of energy, which makes itmore suitable for high-temperature large-scale applications. How-ever, since the superheated vapor has the low heat exchange coef-cients, very large and expensive heat exchangers must beemployed in this case [28]. Furthermore, the rated operating con-ditions are assumed for the adopted engine in this work, but thediesel engine more often operates in transient and variableworking conditions. At some engine operating points, the exhaustohexane and water with relative pressure.

Alkanes show favorable performance based on the above anal-

ergysis. However, such drawbacks as higher ammability of the uidsand their toxicity also have to be taken into account when Alkane-based working uids are used. The two necessary conditions ofgood sealing and excellent ventilation must be satised in orderto ensure safe operation for Alkane-based WHR systems if the pro-posed plants are applied in practical engineering. In future works,heat transfer oil may be used to operate between engine exhausttemperature is below 500 C, or even lower than 300 C. Under thissituation, the waste heat (exhaust gas) temperature is not suit-able for operating the steam cycle. For Alkane-based ORCs, how-ever, good performances still can be achieved by them. From atechnical and economic point of view, Alkane-based ORCs maybe more attractive than conventional steam cycles, specicallyfor DE waste gas heat recovery.

5. Conclusions

In this study, Alkane-based working uids are employed for theORC processes. The processes operate at different evaporating tem-peratures and under different pressures to recover waste heat en-ergy from high-temperature exhaust gases where a commercialdiesel generator is operated under a rating condition. The systemparameters of the ORC using different working uids are optimizedwith six indicators in this paper. Moreover, the inuence of molec-ular complexity on VFR and SP are analyzed. And the energy distri-butions of engine exhaust gases and BSFC improvement aftercombining a DE with a bottoming ORC are also discussed. Thefollowing conclusions can be drawn based on the performanceinvestigation and parametric analysis carried out in the presentwork:

1. The optimized uids are not the same in terms of different indi-cators. However, on the whole, the uids with more number ofcarbon atoms in chain (e.g., Decane or Nonane) have betterthermal efciency, power output performance and lower exergydestruction rate. The exceptions to this rule are cyclic Alkanes.They show good performance in thermal efciency, power out-put performance and exergy destruction rate, while requiringsmall turbine sizes.

2. Molecular complexity has signicant and positive impact on SPand VFR. That is, greater molecular complexity results in highervalues of SP and VFR if they are not cyclic Alkanes. Less complexuids are preferred in this case, since they result in acceptableturbine efciency with relatively low values of SP and VFR.Therefore, using simple expanders such as single-stage turbinesis allowed.

3. Heat energy absorbed by the ORC system from exhaust gaseshas a high energy grade because of its high content of exergy.The highest power output is obtained by Cyclohexane, whichresults in prominent BSFC improvement, about 10% for DE-ORC combined systems, compared with the original engine.

4. In comprehensive consideration, cyclic Alkanes Cyclohexaneand Cyclopentane are considered as the most suitable workinguids due to relatively high power output, reasonable condens-ing pressure, relatively low irreversibility and SP values.

5. Although steam has more advantages in thermal efciency inthe current conditions, from a technical and economic pointof view, Alkane-based ORCs may be more attractive than con-ventional steam cycles, specically for DE waste gas heatrecovery.

216 G. Shu et al. / Applied Engases and Alkane-based working uids in order to improve systemstability and security. However, additional irreversibility of heatexchange process, which makes inferior system efciency, willfollow if heat conduction oil is introduced. The entity of theselosses will be numerically quantied by a detailed exergy analysis.A detailed economy analysis model for Alkane-based power gener-ation system will be established in future work. In summary, dis-cussions and conclusions of this paper are helpful in selectingproper working uids for the DE-ORC system and in determiningoptimal system operation conditions.

Acknowledgements

The authors would like to acknowledge the National NaturalScience Foundation of China (No. 51206117), the National Basic Re-search Program of China (973 Program) (No. 2011CB707201), andthe Natural Science Foundation of Tianjin (No. 12JCQNJC04400)for Grants and supports.

References

[1] United Stated Department of Energy, Ofce of Energy Efciency andRenewable Energy. Vehicle technologies multi-year prog-ram plan 2011-2015; 2010 December. 59 (last visited 01.03.13). .

[2] Horst TA, Rottengruber HS, Seifert M. Dynamic heat exchanger model forperformance prediction and control system design of automotive waste heatrecovery systems. Appl Energy 2013;105:293303.

[3] Yun KT, Cho H, Luck R, Mago PJ. Modeling of reciprocating internal combustionengines for power generation and heat recovery. Appl Energy2013;102:32735.

[4] Fu JQ, Liu JP, Feng RH, g Yang YP, Wang LJ, Wang Yong. Energy and exergyanalysis on gasoline engine based on mapping characteristics experiment.Appl Energy 2013;102:62230.

[5] Love ND, Szybist JP, Sluder CS. Effect of heat exchanger material and fouling onthermoelectric exhaust heat recovery. Appl Energy 2012;89:3228.

[6] Pandiyarajan V, Pandian MC, Malan E, Velraj R, Seeniraj RV. Experimentalinvestigation on heat recovery from diesel engine exhaust using nned shelland tube heat exchanger and thermal storage system. Appl Energy2011;88:7787.

[7] Drescher U, Brggermann D. Fluid selection for the organic Rankine cycle(ORC) in biomass power and heat plants. Appl Therm Eng 2007;27:2238.

[8] Zhang HG, Wang EH, Fan BY. A performance analysis of a novel system of adual loop bottoming organic Rankine cycle (ORC) with a light-duty dieselengine. Appl Energy 2013;102:150413.

[9] Macin V, Serrano JR, Dolz V, Snchez J. Methodology to design a bottomingRankine cycle, as a waste energy recovering system in vehicles. Study in a HDDengine. Appl Energy 2013;104:75871.

[10] Tian H, Shu GQ, Wei HQ, Liang XY, Liu LN. Fluids and parameters optimizationfor the organic Rankine cycles (ORCs) used in exhaust heat recovery of InternalCombustion Engine (ICE). Energy 2012;47:12536.

[11] Yu GP, Shu GQ, Tian H, Wei HQ, Liu LN. Simulation and thermodynamicanalysis of a bottoming organic Rankine cycle (ORC) of diesel engine (DE).Energy 2013:110.

[12] Wang EH, Zhang HG, Fan BY, Ouyang MG, Zhao Y, Mu QH. Study of workinguid selection of organic Rankine cycle (ORC) for engine waste heat recovery.Energy 2011;36:340618.

[13] Vaja I, Gambarotta A. Internal combustion engine (ICE) bottoming withorganic Rankine cycles (ORCs). Energy 2010;35:108493.

[14] Roy JP, Mishra MK, Misra A. Performance analysis of an organic Rankine cyclewith superheating under different heat source temperature conditions. ApplEnergy 2011;88:29953004.

[15] Cayer E, Galanis N, Desilets M, Nesreddine H, Roy P. Analysis of a carbondioxide transcritical power cycle using a low temperature source. Appl Energy2009;86:105563.

[16] Zhang SJ, Wang HX, Guo T. Performance comparison and parametricoptimization of subcritical organic Rankine cycle (ORC) and transcriticalpower cycle system for low-temperature geothermal power generation. ApplEnergy 2011;88:274054.

[17] Wang JL, Zhao L, Wang XD. An experimental study on the recuperative lowtemperature solar Rankine cycle using R245fa. Appl Energy 2012;94:3440.

[18] Tchanche BF, Lambrinos G, Frangoudakis A, Papadakis G. Low-grade heatconversion into power using organic Rankine cycles a review of variousapplications. Renew Sust Energy Rev 2011;15:396379.

[19] Fernndez FJ, Prieto MM, Surez I. Thermodynamic analysis of high-temperature regenerative organic Rankine cycles using siloxanes as workinguids. Energy 2011;36:523949.

[20] Snchez D, Muoz de Escalona JM, Monje B, Chacartegui R, Snchez T.Preliminary analysis of compound systems based on high temperature fuelcell, gas turbine and organic Rankine cycle. J Power Energy 2011;196:435563.

y 119 (2014) 204217[21] Algieri A, Morrone P. Comparative energetic analysis of high-temperaturesubcritical and transcritical organic Rankine cycle (ORC). A biomassapplication in the Sibari district. Appl Therm Eng 2012;36:23644.

[22] Lai NA, Wendland M, Fischer J. Working uids for high-temperature organicRankine cycles. Energy 2011;36:199211.

[23] Siddiqi MA, Atakan B. Alkanes as uids in Rankine cycles in comparison towater, benzene and toluene. Energy 2012;45:25663.

[24] Invernizzi C, Iora P, Silva P. Bottoming micro-Rankine cycles for micro-gasturbines. Appl Therm Eng 2007;27:10010.

[25] Desai NB, Bandyopadhyay S. Process integration of organic Rankine cycle.Energy 2009;34:167486.

[26] Harinck J, Guardone A, Colonna P. The inuence of molecular complexity onexpanding ows of ideal and dense gas. Phys. Fluids 2009;21. 086101/1-14.

[27] Li YR, Wang JN, Du MT. Inuence of coupled pinch point temperaturedifference and evaporation temperature on performance of organic Rankinecycle. Energy 2012;42:5039.

[28] Schuster A, Karellas S, Aumann R. Efciency optimization potential insupercritical organic Rankine cycles. Energy 2010;35:10339.

[29] Heberle F, Brggemann D. Exergy based uid selection for a geothermalorganic Rankine cycle for combined heat and power generation. Appl ThermEng 2010;30:132632.

[30] Macchi E, Perdichizzi A. Efciency prediction for axial-ow turbines operatingwith non conventional uids. Trans ASME, J Eng Power 1981;103:71824.

[31] Lemmon EW, Huber ML, McLinden MO. NIST reference uid thermodynamicand transport properties-REFPROP. NIST standard reference database 23,Version 8.0; 2007.

G. Shu et al. / Applied Energy 119 (2014) 204217 217

Alkanes as working fluids for high-temperature exhaust heat recovery of diesel engine using organic Rankine cycle1 Introduction2 Alkanes2.1 Nomenclature and taxonomy2.2 Molecular complexity

3 System modeling3.1 Description of ORC system3.2 Initial conditions and boundary conditions3.3 Mathematical model3.3.1 Energy analysis3.3.2 Exergy analysis3.3.3 Turbine design

3.4 Validation of the model

4 Results and discussion4.1 The influence of evaporating pressure4.2 The influence of molecular complexity4.3 The energy and exergy distributions4.4 Comparison of ORC and steam cycle

5 ConclusionsAcknowledgementsReferences