- Email: [email protected]

Contents lists available at SciVerse ScienceDirect

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

Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform Leonardo Pierobon*, Tuong-Van Nguyen, Ulrik Larsen, Fredrik Haglind, Brian Elmegaard Technical University of Denmark, Department of Mechanical Engineering, Building 403, DK-2800 Kgs. Lyngby, Denmark

a r t i c l e i n f o

a b s t r a c t

Article history: Received 11 February 2013 Received in revised form 7 May 2013 Accepted 11 May 2013 Available online xxx

This paper aims at ﬁnding the optimal design of MW-size organic Rankine cycles by employing the multi-objective optimization with the genetic algorithm as the optimizer. We consider three objective functions: thermal efﬁciency, total volume of the system and net present value. The optimization variables are the working ﬂuid, the turbine inlet pressure and temperature, the condensing temperature, the pinch points and the ﬂuid velocities in the heat exchangers. The optimization process also includes the complete design of the shell and tube heat exchangers utilized in the organic Rankine cycle. The methodology is applied to recover the waste heat from the SGT-500 gas turbine installed on the Draugen off-shore oil and gas platform in the North Sea. Results suggest two optimal working ﬂuids, i.e. acetone and cyclopentane. Thermal efﬁciency and net present value are higher for cyclopentane than for acetone. Other promising working ﬂuids are cyclohexane, hexane and isohexane. The present methodology can be utilized in waste heat recovery applications where a compromise between performance, compactness and economic revenue is required. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: Organic Rankine cycle Multi-objective optimization Shell and tube heat exchanger Off-shore platform Gas turbine

1. Introduction In March 2007 the Commission of the European Communities [1] set the “20-20-20” targets as the three key objectives for 2020. The requirements are, namely, a 20% reduction in greenhouse gas emission from the 1990 levels, a 20% share of renewable sources in the energy demand and a 20% improvement in the energy efﬁciency. De Paepe [2] considers the annual recoverable industrial waste heat potential to be about 140 TWh in Europe, corresponding to a CO2 reduction of about 14 Mt/y. As surveyed by Colonna [3], this waste heat is available at high (350e250 C), medium (250e 150 C) and low temperatures (150e90 C) and the power capacity may range from large to small in size (15 MWe3 kW). With increasing incentives for reducing the CO2 emissions off-shore, waste heat recovery on off-shore platforms has become a focus area. In off-shore applications, the key selection criteria for the waste heat recovery unit supporting the electrical demand on the platform are high efﬁciency, fuel ﬂexibility, compactness and low weight. Single and dual-pressure steam Rankine cycles are established and reliable solutions for high-temperature waste heat recovery as discussed, for example, in Gewald et al. [4], Rokni [5] and

* Corresponding author. Tel.: þ45 45 25 41 29; fax: þ45 45 88 43 25. E-mail address: [email protected] (L. Pierobon).

Domingues et al. [6]. However, the moisture content at the turbine outlet and the limits on the turbine blade height in practice restrict the application to MW-size power units. The ORC (organic Rankine cycle) is a technology that is receiving more and more attention from the academic world, companies and research institutes. Major ORC advantages are the simplicity of the cycle and the possibility of tailoring the working ﬂuid to the speciﬁc temperature proﬁle of the heat source. Furthermore, the ORC eliminates the problem of turbine blade erosion due to the liquid droplet formation by utilizing a “dry” ﬂuid as the working ﬂuid. Vélez et al. [7] provide an ample review of existing and possible applications of the ORC technology. A crucial aspect in the design of an ORC is the selection of the working ﬂuid. Moreover, the thermal efﬁciency, compactness, weight, availability and cost are among the most important concerns of a complete design process. Additionally, the operating ﬂuid should be chemically stable, environmentally friendly and safe in terms of toxicity and ﬂammability. As emphasized in Velez et al. [7], no ﬂuid satisﬁes all these aspects; therefore, the selection is a compromise between the different possibilities. In the past, much research was conducted to develop optimization algorithms to adapt the ORC to the speciﬁc heat source and to address different objectives. Sun and Li [8] implement the ROSENB optimization algorithm to search the optimal set of operating variables to maximize either the system net power generation or the system thermal efﬁciency. Roy et al. [9] carry out a parametric optimization and performance analysis of an organic

0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.energy.2013.05.039

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

2

L. Pierobon et al. / Energy xxx (2013) 1e12

Rankine cycle where the heat source is the ﬂue gas at a temperature of 140 C exiting the discharged fans of a coal power plant. Hettiarachchi et al. [10] use as the objective function the ratio of total heat exchange area to net power output. Quoilin et al. [11] optimize a small-scale ORC for waste heat recovery applications; economic proﬁtability and thermodynamic efﬁciency are the objective functions. Baik et al. [12] employ the pattern search algorithm to maximize the net power output considering the overall heat transfer conductance and turbine inlet pressure and temperature as optimization variables. In Wang et al. [13] and in Dai et al. [14], the GA (genetic algorithm) is used as the optimization method for a comparative study of ORCs for low-temperature waste heat recovery. Cayer et al. [15] present a parametric study of a CO2 supercritical power cycle using six performance indicators: thermal efﬁciency, speciﬁc net output, exergetic efﬁciency, total UA-value and surface of the heat exchangers, and the relative cost of the system. The concept is extended by Shengjun et al. [16] to subcritical and supercritical ORCs minimizing the levelized energy cost and heat exchanger area per unit power output. Salcedo et al. [17] apply the multi-objective optimization to solar Rankine cycles coupled with reverse osmosis desalination considering the speciﬁc total cost and the environmental impact of the plant. Wang et al. [18] perform a parametric optimization using a multi-objective optimization to design ORCs for low temperature waste heat. The screening criteria include heat exchanger area per unit power output and heat recovery efﬁciency. The present paper aims at presenting a generic methodology to design and optimize ORCs where shell and tube heat exchangers are used. In order to assess the compactness of the system, a detailed dimensioning of the shell and tube heat exchanger is carried out considering both the heat transfer coefﬁcients and the pressure drops on the shell and tube sides. We use a multi-objective optimization modeled by the genetic algorithm using the following objective functions: thermal efﬁciency, total volume of the ORC and net present value. We apply the methodology to recuperate the waste heat from the SGT-500 gas turbine installed on the Draugen platform (Kristiansund, The North Sea). Compared with previous works [8e18], the approach in this paper is novel in the sense that it includes the total volume of the organic Rankine cycle and the net present value as objective functions. Furthermore, in contrast to previous works, the geometry of the shell-and-tube heat exchanger is included in the optimization procedure. The novel set of optimization variables includes 109 working ﬂuids, turbine inlet pressure and temperature, pinch points of condenser, internal recuperator and evaporator/economizer and ﬂuid velocities in the tubes and on the shell side, respectively, of all heat exchangers. Other characteristics of the working ﬂuids, such as health, ﬁre and physical hazards [19], and the GWP (global warming potential) are, to some extent, also considered. The methodology presented in this paper can be applied in waste heat recovery applications where the ORC design is the result of a compromise between performance, compactness and economic revenue. We describe the shell and tube design process, the ORC’s governing equations and the multi-objective optimization in Section 2. The case study is also presented in Section 2. Results of the multiobjective optimization are reported in Section 3 and discussed in detail in Section 4. Finally, we state the main conclusions in Section 5.

heat transfer ﬂuid, Dowtherm Q, is introduced in Subsection 2.3. Finally, Subsections 2.4 and 2.5 introduce the multi-objective optimization and the case study where the methodology is applied. 2.1. Shell and tube heat exchangers The basic design procedure requires determining the surface area that is needed using the available temperature difference. The governing equation for the heat transfer across a surface is

Q_ ¼ UAFt DTlm

(1)

where Q_ is the heat rate, U is the overall heat transfer coefﬁcient, A is the heat transfer area, DTlm is the logarithm mean temperature difference and Ft is the temperature correction factor which accounts for co-current and cross-ﬂow. We compute the correction factor in Eq. (1) utilizing the method proposed by Fakheri [20]. The overall heat transfer coefﬁcient can be regarded as the sum of the following ﬁve different items: the outside ﬂuid ﬁlm coefﬁcient ho, the inside ﬂuid ﬁlm coefﬁcient hi, the outside dirt coefﬁcient (fouling factor) hod, the inside dirt coefﬁcient hid and the thermal conductivity of the tube wall material lw. The overall coefﬁcient based on the outside area of the tube can be calculated as follows:

d0 ln dd0 d0 1 1 1 1 d 1 i ¼ þ þ þ þ 0 Uo ho hod di hid di hi 2lw

(2)

where d0 and di are the outer and inner diameter of the tubes. Fig. 1 shows the generic geometry of the shell and tube heat exchanger. The geometrical parameters considered in the design are the following: the inner and outer diameters of the tubes, the tube length lt, the distance between the tube centers (pitch) pt, the number of tube passes Nt and the bafﬂe spacing lb. Based on the well-established design procedure outlined in Richardson and Peacock [21], the geometry of the heat exchanger and the ﬂuid velocity in the tubes and on the shell side can be calculated. We evaluate the volume Vht of the shell and tube heat exchanger assuming a cylindrical shape:

p

Vht ¼ Fts d2s lt 4

(3)

The shell diameter ds and the tube length are the diameter of the base and the height of the cylinder (see Eq. (3)). A correction factor Fts is applied to account for the space occupied by the shell and tube inlet and outlet ducts. As formulated in Hall [22], the purchasedequipment cost PECht . is a function of the heat exchanger area A and it can be computed as follows:

PECht ¼ 10; 000 þ 324 A0:91

(4)

2. Methodology We outline the features and details of the new methodology in this section. The modeling of shell and tube heat exchangers is described introducing the equations computing the heat transfer coefﬁcients, the geometry and the investment cost. In Subsection 2.2, we present the modeling of the other ORC components. The

Fig. 1. Shell and tube heat exchanger geometry and tube pattern.

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

L. Pierobon et al. / Energy xxx (2013) 1e12

The heat transfer coefﬁcient ht and the pressure drop Dpt on the tube side in the subcooled liquid and superheated vapor regions are related the Reynolds number Re, Prandtl number Pr and velocity in the tubes ut and are evaluated using the following methodology [21]:

ht ¼ jth

lt di

RePr0:33

mt mtw

0:14 (5)

4200ð1:35 þ 0:02tÞu0:8 t d0:2 i

(7)

where t is the average temperature of the water in the tube. The calculation of the heat transfer coefﬁcient on the shell side hs is based on the experimental work carried out by Kern [23] on commercial exchangers with standard tolerances. Richardson and Peacock [21] state that the methodology gives a satisfactory prediction of the heat-transfer coefﬁcient. The heat transfer coefﬁcient hs and the pressure drop Dps on the shell side in the subcooled liquid and superheated vapor regions are related to the Reynolds number Re, Prandtl number Pr and velocity on the shell side us and are evaluated using the following methodology [21]:

ls de

Dps ¼ 8jsf

RePr1=3

ms msw

0:14 (8)

ds lt ms 0:14 rs u2s de lb msw 2

G ¼

1=3

rl rl rg g ml G

(11)

_ m Nt lt

(12)

where g is the standard gravity, G is the tube loading and rl, ll, ml are the density, thermal conductivity and dynamic viscosity at the saturated liquid state, while rg is the density at the saturated vapor condition. As suggested by Kern [23], the pressure drop on the condensing side is quantiﬁed as half of the pressure drop (Eq. (9)) based on the vapor inlet conditions. 2.2. Organic Rankine cycle modeling As shown in Fig. 2, the ORC components are the TUR (turbine), connected through the shaft to the GEN (electric generator), the liquid pump and ﬁve different heat exchangers: ECO (economizer), EVA (evaporator), SUP (superheater), IR (internal recuperator) and condenser. Each component is modeled at steady state conditions. The outlet enthalpy h2 and the power consumption P_ p of the pump are calculated as follows:

h2 ¼ h1 þ

p2 p1

(13)

r

_ 2 h1 Þ mðh P_ p ¼

(14)

hm

where p1 and p2 are the pressure at the inlet and at the outlet of the pump, h1 is the inlet enthalpy, r is the density of the working ﬂuid,

13

12

11

4

5

10

(9)

where de, rs, ls and ms are the equivalent shell diameter, the density, the thermal conductivity and viscosity calculated with the average temperature between the inlet and the outlet conditions of the shell; msw is the viscosity of the ﬂuid calculated with the temperature of the outer wall of the tube. The quantities jsh and jsf are the heat transfer and friction factor of the shell and are evaluated as reported in Ref. [21]. Assuming that the evaporator operates in the nucleate boiling region, we evaluate the heat transfer coefﬁcient with the Cooper correlation [24]. The nucleate boiling heat transfer coefﬁcient hnb is a function of the reduced pressure pr, the molecular weight of the ﬂuid M, the speciﬁc heat rate Q_ =A and the surface roughness Rp (assumed to be 1 mm [23]) and it can be expressed by mathematically as 0:120:4343ln Rp hnb ¼ 55pr ð0:4343lnpr Þ0:55 M 0:5

(6)

where rt, lt and mt are the density, the thermal conductivity and the dynamic viscosity calculated at the average temperature between the inlet and the outlet conditions of the tube; mtw is the dynamic viscosity of the ﬂuid calculated at the temperature of the inner wall of the tube. The quantities jth and jtf are the heat transfer and friction factor of the tubes and are evaluated as reported in Ref. [21]. The coefﬁcient m is equal to 0.25 for laminar ﬂow (Re < 2100) and 0.14 for turbulent ﬂow (Re > 2100). The following equation gives a more accurate estimate of the heat transfer coefﬁcient of water, utilized as the cooling ﬂuid in the condenser tubes [21]:

hs ¼ jsh

states. In ORCs the heat rejection starts in the superheated region. Hence, the calculation of the heat transfer coefﬁcient is split into two processes: de-superheating and condensation. Since it is assumed that the condensation takes place on the shell side, Eq. (8) is utilized. In the second step, assuming condensation outside the horizontal tubes, the following equations are utilized as suggested by Richardson and Peacock [21]:

hcond ¼ 0:95ll

mt m rt u2t l Dpt ¼ Nt 8jtf t þ 2:5 di mtw 2

ht ¼

3

Q_ A

ECO

EVA

SUP

TUR GEN

3 7 IR

!0:67

Condenser

(10)

The pressure drops are evaluated as the average between the pressure drops calculated with the thermodynamic properties and the speed computed at the saturated liquid and saturated vapor

6

ORC pump 2

8 2w Water 1w

1

Fig. 2. Organic Rankine cycle layout.

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

4

L. Pierobon et al. / Energy xxx (2013) 1e12

_ is the mass ﬂow circulating in the organic cycle and hm is the m mechanical efﬁciency of the pump.

hth ¼

1 0:8 0:71 1:41 þ 1:41 PECp ¼ 422 P_ p 1 hm

where h6 and h3 are the enthalpies at the outlet of the superheater and at the inlet of the economizer.

(15)

The purchased-equipment cost PECp is evaluated with Eq. (15) which was utilized by Arsalis et al. [25] for water pumps. The volume of the pump is considered negligible. The turbine is modeled by using the polytropic efﬁciency. The purchased-equipment cost of the turbine PECtur .is assumed to be comparable with the purchase price of conventional steam axial turbines. In this paper we use the analytical expression proposed by Lian et al. [26] which _ tur : depends on the power output W

_ 0:7 PECtur ¼ 6000 W tur

(16)

Since the methodology is applied to MW-size ORCs, it is assumed that the expander is an axial turbine with a unique stage (nozzle and rotor). As shown in Fig. 3, the volume Vtur is modeled as a cylindrical trapezoid. We evaluate the inlet and outlet ﬂow areas Ain and Aout through the continuity equation considering an inlet Mach number of 0.3 [27]. The external inlet and outlet diameter din, e and dout, e are calculated assuming a tip to hub ratio of 1.43 [28] and an axial length lx of 0.3 m. A correction factor Ftv of 1.2 is applied to account for the space required by the inlet and outlet ducts and the electric generator.

Vtur ¼ Ftv

hp 4

d2out; e

þ

d2in; e

i l x þ dout; e din; e 4 3

p

PECgen

0:95 ¼ 60P_ gen

(17)

2.3. Dowtherm Q thermodynamic and physical properties The ORC working ﬂuid is typically a carbon-based or hydrogenbased ﬂuid, and the combustion products of a biomass plant or the exhaust gases exiting gas turbines, diesel and gas engines have high oxygen content. Hence, for safety reasons, an intermediate loop is placed between the ORC and the heat source. As suggested by Pierobon et al. [29], Dowtherm Q is selected as an intermediate heat carrier. We calculate the thermodynamic and physical properties by ﬁtting the experimental data released by DOW Chemical Company [30] and assuming an incompressible liquid with a high density:

r ¼ 1187 206:51

T 273:15

cp ¼ 0:7702 0:8264

(22)

T 273:15

(23)

ZT h ¼

cp ðTÞdT

(24)

l ¼ 0:1651 0:0398

T 273:15

(25)

m ¼ exp 4:053994 þ 6:0844339= (18)

(19)

where hel is the electric efﬁciency of the generator. According to the previous equations, the net power output P_ net and the thermal efﬁciency hth of the ORC are deﬁned as

P_ net ¼ P_ gen P_ p

(21)

T0

We calculate the electric power output P_ gen and the PEC (purchased-equipment cost) of the electric generator PECgen as follows:

_ tur P_ gen ¼ hel W

P_ net _ 6 h3 Þ mðh

(20)

Fig. 3. Geometric representation (cylindrical trapezoid) of the axial turbine in the organic Rankine cycle.

1:5 T 273:15

(26)

where cp is the speciﬁc heat capacity at constant pressure and T is the Dowtherm Q temperature in Kelvin. 2.4. The multi-objective optimization A multi-objective optimization involves minimizing or maximizing simultaneously two or more functions subjected to a set of constraints. In contrast to single-objective optimization, a solution to a multi-objective problem is a range of optimal points, the socalled Pareto front [31]. We use the genetic algorithm [32] for the beneﬁts of avoiding the calculation of derivatives and enabling the search of global optima. The genetic algorithm parameters are speciﬁed according to the following values: population size 1000, generation size 200, crossover fraction 0.8 and migration fraction 0.2. These numerical values are selected in order to ensure the repeatability of the solution when different simulations are performed. The genetic algorithm stops when the maximum number of generations is reached or when the average change in the spread of the Pareto front is lower than the speciﬁed tolerance (which in this paper is 103). Fig. 4 shows the structure of the algorithm. As indicated in the ﬁgure, the multi-objective optimization governs a global routine developed in MATLAB 2012a. The thermodynamic properties of the ORC working ﬂuid are acquired by linking MATLAB and the commercial software REFPROP 9 [33]. First, in the pre-screening process the number of available working ﬂuids is restricted by discarding the ﬂuids for which the physical, health, ﬁre hazard or GWP exceed the maximum allowable values set by the user. This is carried out by linking mathematically the working ﬂuids to a unique integer number (from 1 to

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

L. Pierobon et al. / Energy xxx (2013) 1e12

5

Fig. 4. Structure of the multi-objective algorithm. The optimization routine involves the pre-screening process, the ORC solver, the Nelder-Mead direct search optimizer, the shell and tube designer, and the volume and economic evaluations.

109) corresponding to a speciﬁc ﬂuid in REFPROP. Consequently, the algorithm can interpret and operate numerically on this optimization variable. The listed ﬂuids, their GWPs in 100 years and their hazards, are taken from Pierobon et al. [34]. After this preliminary step, the optimization variables are passed to the GA. In the present paper the following variables (see Fig. 4) are optimized: ORC working ﬂuid Condenser outlet temperature T1 Condenser pinch point (located at the saturated vapor state) Internal recuperator pinch point T2 T8 Minimum temperature difference (pinch point) in the economizer or vaporizer f) Turbine inlet pressure p6 g) Superheating temperature difference T10 T6 h) Target velocities in the tubes and on the shell side for each heat exchanger utt, ust a) b) c) d) e)

The ORC solver then acquires these values and computes the thermodynamic properties at each state, the thermal efﬁciency and the net power output. During the ﬁrst run of the ORC solver, the pressure drops in the heat exchangers are set to zero. The solution is then utilized as the design condition for the heat exchangers and the expander. The heat rate, the mass ﬂow, the inlet and outlet temperatures, and the ﬂuid velocities in the tubes and on the shell side are passed to the shell and tube designer. At this point the constrained Nelder-Mead optimizer [35] is employed to select the tube and shell geometry that gives the speciﬁed velocity in the tubes and on the shell side. In order to reduce the computational time required by the sub-optimization, we use the tube length to obtain the speciﬁed velocity in the tubes; thereby, the function to be minimized can be expressed by the following equation:

f ðdo ; utt ; pt ; lb Þ ¼

juts us j uts

(27)

where juts us j is the absolute difference between the targeted shell speed uts and the shell speed calculated in the shell and tube design process. Table 1 lists the lower and upper bounds of the geometric variables and the design parameters which are kept constant in the shell and tube heat exchanger design. If the heat transfer process occurs in the two-phase region, we employ an average velocity evaluated at the saturated liquid and saturated vapor condition. The outcomes of the heat exchanger design are the overall heat transfer coefﬁcient, the surface area, the volume and the pressure drops. The pressure drops are set as inputs to the ORC solver, and a new thermal efﬁciency and net power output are computed. We then check the consistency of the results. For each heat exchanger we verify that the value of the function, calculated with Eq. (27), is lower than the required accuracy, which we specify to be 102.

Table 1 Lower and upper bounds of the geometric variables utilized in the shell and tube heat exchanger design process. Variable

Lower bound

Upper bound

Outer diameter do [21] Tube pitch pt [36] Bafﬂe spacing lb [21] Parameters [21] Bafﬂe cut bc Conﬁguration Outside dirt coefﬁcient hod Inside dirt coefﬁcient hid Thermal conductivity (tube wall) Correction factor Fts

16 mm 1.15d0 0.2ds

50 mm 2.25d0 1.0ds

25% Triangular 6000 W/(m2 C) 6000 W/(m2 C) 50 W/(m C) 1.2

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

6

L. Pierobon et al. / Energy xxx (2013) 1e12

Furthermore, in case the inlet temperature difference T13 T3 of the economizer or T12 T4 of the vaporizer is lower than the speciﬁed minimum temperature difference, we discard the solution. The three objective functions are expressed below. The function f1 aims at maximizing the thermal efﬁciency. To increase the ORC compactness, the sum of the volumes of the ORC components is minimized (function f2) and to analyze the proﬁtability of the investment, the net present value is maximized (function f3).

As suggested by Bejan et al. [37], the total investment cost of the ORC can be regarded by evaluating the direct and indirect costs. Using the values reported in Table 2, we calculate the total investment cost of the ORC as follows:

has three SGT-500 engines to provide the normal total electric load. The SGT-500 is an industrial twin-spool gas turbine, and the engine model is the C-version launched in the beginning of the 1980s. Table 3 reports the design point speciﬁcations of the SGT-500. We apply the multi-objective optimization described in Section 2.4 to design the ORC for recovering the waste heat from the SGT500 gas turbine. Fig. 5 shows the plant layout including the SGT500 twin-spool gas turbine, the intermediate loop and the ORC. The design point parameters of the ORC are listed in Table 4 and are maintained constant. The terminal temperature of the off-gases exiting the waste heat recovery unit is ﬁxed to 145 C [29]. A prudential value of 335 C [29] is assumed for the maximum temperature. We evaluate the cold temperature (130 C) and the mass ﬂow (49.5 kg/s) of Dowtherm Q in the intermediate loop by applying an energy balance in the internal heat exchanger. Next, we must decide the necessary ﬂuid allocation in order to dimension the shell and tube heat exchangers. Where a phase change occurs (evaporator and condenser) the ORC working ﬂuid is placed on the shell side, since it is the most common conﬁguration [21]. For the economizer and superheater, the ﬂuid with the greatest tendency to foul (which is Dowtherm Q) is allocated on the tube side; while for the internal recuperator, the ﬂuid with the highest temperature (working ﬂuid exiting the ORC turbine) is placed on the tube side to reduce heat loss and meet safety conditions [21]. We expect two major sources of income with the installation of the waste heat recovery unit. The ﬁrst is associated with the fuel savings and the second with the CO2 taxes. In fact, the power produced by the ORC enables a reduction of the load of the other gas turbines operating on the platform. Consequently, the saved natural gas can be exported and sold to the market. The income related to the saved natural gas RNG is estimated as follows:

ITOT ¼ 3:7 PECtur þ PECeco þ PECeva þ PECsup þ PECIR þ PECp þ PECcond þ PECgen

RNG ¼ 3:6

f1 ¼ hth

(28)

f2 ¼ Veco þ Veva þ Vsup þ VIR þ Vcond þ Vtur ¼ Vtot

(29)

f3 ¼ NPV

(30)

According to Bejan et al. [37], the NPV (net present value) can be calculated considering the equipment lifespan n, the interest factor q, the total capital investment ITOT and the annual income Ri:

NPV ¼

n X

Ri

i¼1

ð1 þ qÞi

ITOT

(31)

The discounted payback period DPB (discounted payback period) that estimates the time required to recover the principal amount of an investment is mathematically deﬁned as the minimum year at which the NPV is greater than zero:

DPB ¼ minfn : NPVðnÞ > 0g

(32)

(33) 2.5. Case study e the Draugen platform We applied this methodology to recover the waste heat from the Siemens SGT-500 gas turbine employed on the Draugen off-shore platform, located 150 km from Kristiansund, in the Norwegian Sea. The platform, operated by A/S Norske Shell, produces gas and oil. Gas is exported via the Åsgard gas pipeline to Kårstø (Norway). Oil is ﬁrst stored in storage cells at the bottom of the sea and then exported via a shuttle tanker (once every 1e2 weeks). The platform Table 2 Estimate of total capital investment based on direct and indirect costs as suggested by Bejan et al. [37]. Total capital investment A. Direct costs 1. Onsite costs a) PEC (Purchased-equipment costs) b) Purchased-equipment installation c) Piping d) Instrumentation þ controls e) Electrical equipment þ materials 2. Offsite costs f) Civil, structural þ architectural work g) Service facilities B. Indirect costs i) Engineering þ supervision j) Construction costs þ constructors proﬁt k) Contingency

45 35 20 11

%PEC %PEC %PEC %PEC

30 %PEC 50 %PEC 8 %DC 15 %DC 15% (of i and j)

P_ net pNG huV_ NG P_

(34)

GT

where P_ GT is the gas turbine net power output, hu is the utilization factor, pNG is the price of natural gas and V_ NG is the volumetric ﬂow of natural gas. The volumetric ﬂow V_ NG of natural gas is calculated as follows:

P_ net HR V_ NG ¼ vNG 3600 LHV

(35)

where vNG is the speciﬁc volume calculated at 15 C and 1.013 bar, HR (heat rate) is the heat rate of the gas turbine and LHV (lower heating value) is the low heat value of natural gas. The second major come is due to the CO2 tax. Since 1991 Norway levies carbon tax on petroleum, mineral fuel and natural gas the rates based on the fuel’s carbon content [39]. Thus, the new method alleviates the carbon tax cost associated with the combustion of natural gas. The income RCO2 related to the CO2 savings is computed as follows: Table 3 Design point speciﬁcations for the Siemens SGT-500 twin-spool gas turbine [29]. Model

Siemens SGT-500

Turbine inlet temperature Exhaust gas temperature Exhaust gas mass ﬂow Net power output Heat rate Fuel

850 C 376 C 93.5 kg/s 17.014 MW 11,312 kJ/kWh Naphtha, crude oil, heavy fuel oil, bio oil, natural gas, syngas

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

L. Pierobon et al. / Energy xxx (2013) 1e12 Table 5 Parameters assumed for the economic analysis.

Natural gas CC

LPC

HPC

HPT

LPT

PT GEN1

INH Exhaust gases

Ex,out

Ex,in

Air 13

12

11

4

5

ECO

EVA

SUP

6

GEN2

3 7 IR

Condenser ORC pump 2

8 2w Water 1w

1

Fig. 5. Combined cycle layout including the SGT-500 Siemens twin-spool gas turbine, the intermediate loop and the organic Rankine cycle.

P_ net _ CO2 pCO2 hum P_

(36)

GT

_ CO2 is the mass ﬂow of the where pCO2 is the carbon tax and m avoided carbon dioxide. The net present value of the SGT-500 and ORC combined cycle can be rewritten as

NPV ¼

n X Ma RNG þ RCO2 i¼1

ð1 þ qÞi

Parameter

Value

Natural gas price pNG Utilization factor hu Low heating value (natural gas) Speciﬁc volume (natural gas) Carbon tax Carbon dioxide emission rate Maintenance Equipment lifespan n Interest factor q Conversion factor

681.65 NOK/ton m3st [38] 7000 h/y [26] 48,530 kJ/kg 1.3139 m3/kg 410 NOK/t of CO2 [40] 2.75 kg(CO2)/kg(NG) [41] 0.9 [26] 20 y 10% 0.18 NOK/$

10

TUR

RCO2 ¼ 3:6

7

ITOT

(37)

where Ma is a non-dimensional factor that accounts for the operating and maintenance costs. The numerical values assumed in Eqs. (34)e(37) are reported in Table 5. 3. Results Table 6 lists the variables, and the lower and upper bounds set in the multi-objective optimization. The velocities of Dowtherm Q in Table 4 Organic Rankine cycle parameters assumed in the multi-objective optimization. Parameter

Value

Exhaust gas inlet temperature tex, in Exhaust gas outlet temperature tex, out Exhaust gas mass ﬂow Dowtherm Q inlet temperature t10 Dowtherm Q outlet temperature t13 Dowtherm Q mass ﬂow Pump efﬁciency hm Turbine polytropic efﬁciency Generator efﬁciency hel Cooling water t1w

376 C 145 C [29] 93.5 kg/s 335 C [29] 130 C 49.5 kg/s 80% 80% 98% 5 C

the economizer, evaporator and superheater, and the velocity of the water in the condenser are not included in Table 6. These are all set to 1 m/s. We assume that the pump work in the intermediate loop is negligible and that the inlet pressure of the water is sufﬁcient to overcome the pressure drops associated with the ﬂow through the condenser. We set a maximum GWP of 50 and a maximum physical, ﬁre and health hazards level of 3 in the pre-screening process of the available working ﬂuids. The pre-screening process decreases the number of available working ﬂuids from 109 down to 20. The GWP boundary excludes all the refrigerants (namely R113, R114, RC318 and R134A) except for R1234YF and R1234ZE. Gaseous ﬂuids, such as air, were eliminated because their critical temperatures are below 25 C (Table 6). The limit on the ﬁre hazard eliminates ﬂuids that may ignite spontaneously with air (e.g. ethane, butane, pentane, propane and methane). Among the available ﬂuids, seven (hexane, i-hexane, heptane, decane, nonane, octane, dodecane) belong to the alkane class and ﬁve (methylcyclohexane, propylcyclohexane, cyclohexane, cyclopentane and cyclopropane) to the cycloalkane group. Other possible ﬂuids are acetone, ammonia, benzene, ethanol, methanol, carbon dioxide, toluene and triﬂuoroiodomethane. The results suggest that variations of the GWP limits and of the hazard levels do not change the results considerably. The multi-objective optimization provides a threedimensional Pareto front composed of 350 points. The GA selects two ﬂuids, cyclopentane and acetone, as optimal. In general cyclopentane shows the highest efﬁciency, but at the cost of greater total volumes and investment. However, regarding the net present value, cyclopentane is the best choice. Fig. 6 depicts the total volume of the ORC plant versus the thermal efﬁciency. Acetone exhibits a Pareto front ranging from a thermal efﬁciency of 23.7e27.0% and from a volume of 14.5e 57.5 m3. For cyclopentane, the thermal efﬁciency varies from 27.0 to 28.1% and the volume ranges from 55.0 to 98.4 m3.

Table 6 Lower and upper bounds speciﬁed for the variables included in the multi-objective optimization (bounds for all velocities are taken from Richardson and Peacock [21]). Parameter

Lower bound

Upper bound

Outlet condenser temperature T1 Condenser pinch point Internal recuperator pinch point T2 T8 Evaporator or economizer pinch point Turbine inlet pressure p6 Superheating temperature difference T10 T6 Economizer velocity on the shell side Evaporator velocity on the shell side Superheater velocity on the shell side Internal recuperator velocity on the shell side Internal recuperator velocity in the tubes Condenser velocity on the shell side

25 C 10 C 15 C 10 C 10 bar 50 C 0.3 m/s 5 m/s 5 m/s 0.3 m/s 10 m/s 10 m/s

35 C 25 C 40 C 30 C 40 bar 130 C 1 m/s 10 m/s 10 m/s 1 m/s 30 m/s 30 m/s

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

8

L. Pierobon et al. / Energy xxx (2013) 1e12

120

60 cyclopentane

i-hexane

acetone

100

hexane

50 Volume [m 3]

Volume [m 3]

cyclohexane 80 60

40 30

40

20

20

10

0 0.23

0.24

0.25 0.26 0.27 Thermal efficiency [-]

0.28

0 0.14

0.29

Fig. 6. Pareto fronts (acetone and cyclopentane) obtained through the multi-objective optimization representing the total volume of the ORC versus thermal efﬁciency.

The net present value is given as a function of the total volume in Fig. 7. The NPV increases from 17.7 up to 19.8 M$ for acetone and from 19.7 up to 20.1 M$ for cyclopentane. It can be noticed that the trend of the two Pareto fronts initially increases and subsequently ﬂattens out. Since the net present value is a function of the total investment cost and the yearly incomes (dependent on the thermal efﬁciency of the ORC), an optimum is reached at a volume of 49.8 m3 for acetone and 86.2 m3 for cyclopentane. After this maximum, increasing the performance of the ORC by increasing the volume diminishes the economic revenue since the total investment cost becomes too high. To illustrate how other working ﬂuids perform, we run the multi-objective optimization excluding acetone and cyclopentane from the candidates. The results are shown in Figs. 8 and 9, where the Pareto fronts of the three optimal ﬂuids (cyclohexane, hexane and i-hexane) are reported. The population size in the GA is limited to 500. The trend of the thermal efﬁciency versus total volume and of the total volume versus net present value for both acetone and cyclopentane can be ﬁtted by interpolating the results shown in Figs. 6 and 7 using the commercial software TableCurve 2D v5.01 [42].

1 ¼ a þ bh3th Vtot

(38)

2 NPV2 ¼ c þ dVtot þ eVtot

(39)

0.16

0.18 0.2 Thermal efficiency [-]

0.22

0.24

Fig. 8. Pareto fronts (cyclohexane, hexane and i-hexane) obtained through the multiobjective optimization representing the total volume of the ORC versus thermal efﬁciency. Cyclopentane and acetone are excluded.

Table 7 reports the coefﬁcients a, b, c, d, e and the coefﬁcient of determination for acetone and cyclopentane. Regarding the optimization variables, the pinch point is located in the evaporator rather than in the economizer and it ranges between 13.9 and 19.6 C. The minimum temperature difference in the condenser is constant along the Pareto front (20.0 C in average). The optimal turbine inlet pressure varies between 37.2 and 39.7 bar, and it almost reaches the upper bound of 40 bar (see Table 6). The proﬁle of the outlet temperature of the condenser throughout the Pareto front ranges from 25.9 up to 29.5 C for acetone and from 25.9 up to 27.1 C for cyclopentane. Figs. 6 and 7 enable the selection of the design point of the ORC employed as the waste heat recovery unit for the SGT-500 gas turbine. If an upper limit for the total volume is speciﬁed and the NPV increases at greater volumes, the optimal solution corresponds to that of the maximal acceptable total volume. On the contrary, if the NPV decreases or if the speciﬁed volume is greater than the maximum volume in the Pareto front, the optimum is located where the NPV is maximized. For example, if the available volume is lower than 30 m3, the optimal solution falls in the acetone Pareto front at a total volume of 29.9 m3; the thermal efﬁciency and net present value are 26.1% and 19.4 M$ (see Figs. 6 and 7). If the available volume is greater than 100 m3, cyclopentane is the most suitable working ﬂuid and an optimum is set where the NPV reaches the maximum (20.1 M$). This corresponds to a total volume of 86.2 m3 and a thermal efﬁciency of 27.8%. The discounted payback period for both cases is estimated to be around 5 years. Table 8 lists

18 21

i-hexane cyclopentane

Net present value [M$]

Net present value [M$]

acetone 20

19

18

hexane

16

cyclohexane 14 12 10 8 0

17 0

20

40

60 Volume [m 3]

80

100

120

Fig. 7. Pareto fronts (acetone and cyclopentane) obtained through the multi-objective optimization representing the net present value versus the total volume of the ORC.

10

20

30 Volume [m 3]

40

50

60

Fig. 9. Pareto fronts (cyclohexane, hexane and i-hexane) obtained through the multiobjective optimization representing the net present value versus the total volume of the ORC. Cyclopentane and acetone are excluded.

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

L. Pierobon et al. / Energy xxx (2013) 1e12 Table 7 Coefﬁcients a, b, c, d, e and coefﬁcient of determination for acetone and cyclopentane obtained by interpolating in the Pareto fronts in Figs. 7 and 8. Coefﬁcient

Acetone

Cyclopentane

a b Coefﬁcient of determination for Eq. (37) c d e Coefﬁcient of determination for Eq. (38)

0.1629 7.3035 0.9580 230.9653 7.0789 0.0759 0.9905

0.0802 -3.1202 0.8858 277.1201 3.0718 0.0187 0.8197

the geometry, the volume and the investment cost of the heat exchangers and axial turbine for the two alternatives. It can be noted that the largest components are the internal recuperator, the economizer and the condenser. In the internal recuperator, the heat is exchanged between liquid and vapor, thus the heat transfer coefﬁcient of the vapor side is relatively low. The economizer and the condenser are associated with a high overall heat transfer coefﬁcient. However, a large volume is required since a large heat rate is exchanged.

Table 8 Geometry, investment cost and volume for the proposed optimal solution selected from the Pareto front of acetone and cyclopentane.

ORC thermal efﬁciency ORC net power output Total volume Net present value Discounted payback time Economizer Tube length Tube diameter Bafﬂe spacing Pitch ratio Volume Purchased-equipment cost Evaporator Tube length Tube diameter Bafﬂe spacing Pitch ratio Volume Purchased-equipment cost Superheater Tube length Tube diameter Bafﬂe spacing Pitch ratio Volume Purchased-equipment cost Internal recuperator Tube length Tube diameter Bafﬂe spacing Pitch ratio Volume Purchased-equipment cost Condenser Tube length Tube diameter Bafﬂe spacing Pitch ratio Volume Purchased-equipment cost Axial turbine Axial speed Volume Purchased-equipment cost

Case a)

Case b)

26.1% 6.04 MW 29.9 m3 19.4 M$ 4.9 year

27.8% 6.43 MW 86.2 m3 20.1 M$ 5.2 year

5.44 m 17.3 mm 0.29 1.36 5.77 m3 0.1195 M$

17.51 m 37.3 mm 0.28 1.13 32.8 m3 0.2137 M$

7.16 m 25.8 mm 0.27 1.41 1.84 m3 0.0335 M$

6.66 m 21.7 mm 0.28 1.40 1.8 m3 0.0378 M$

4.19 m 16.0 mm 0.32 1.55 1.44 m3 0.0415 M$

8.6 m 16.3 mm 0.36 1.9 2.8 m3 0.0679 M$

1.33 m 41.1 mm 0.29 1.02 12.0 m3 0.1157 M$

2.24 m 45.9 mm 0.29 1.01 37.6 m3 0.2316 M$

1.19 m 16.0 mm 0.31 1.55 8.2 m3 0.2003 M$

1.59 m 16.9 mm 0.31 1.52 10.4 m3 0.2609 M$

66.3 m/s 0.5 m3 2.7656 M$

58.3 m/s 0.5 m3 2.9181 M$

9

4. Discussion We compare the shell and tube heat exchanger model, outlined in Subsection 2.1, using an example outlined in Richardson and Peacock [21], comprising the design of a heat exchanger to sub-cool condensate from a methanol condenser with the use of liquid water as coolant. The results indicate differences less than 1% in overall heat transfer coefﬁcients and pressure drops between the models derived here and the results provided in Richardson and Peacock [21]. These small differences are expected to be caused by differences in the calculations of thermodynamic properties of ﬂuids, suggesting that the shell and tube heat exchanger model derived here gives reasonable results. The total investment cost for case a) is 13.1 M$ and 15.0 M$ for case b). The axial turbine is the major contributor to the total investment cost of the ORC, namely, 78.6% and 72.5% for case a) and b). In Gonçalves et al. [43], the price of the expander and generator represents 48.4% of the total cost of a 385 kW ORC fueled by sawmill wastes. However, the presence of a boiler increases the total expense by around 24.2%. Assuming the component prices reported in Gonçalves et al. [43] and a conversion factor of 1.31 $/V, the total investment cost calculated with Eq. (33) is 18.4% lower for case a) and 10.1% lower for case b). The total investment cost calculated with Eq. (33) is also compared with the results of an equation derived in Ghirardo et al. [44] utilizing the cost of the Turboden T1100-CHP ORC and a scale factor coefﬁcient equal to 0.867. The investment costs computed with Eq. (33) are 22.8% and 28.8% greater than the values calculated with Ghirardo’s approach. Findings suggest that the pay-back time is within the same range (4e6 years) as the one reported by Wang et al. [18] for heat source temperatures of 120 C and 140 C. The results of these comparisons suggest that the economic analysis in this paper is in accordance with the results available in the literature. In accordance with the results presented in Sun and Li [8] and Roy et al. [9], higher expander inlet pressures provide a higher net power generation and a higher compactness of the economizer, evaporator and superheater. Similarly to the works carried out by Roy et al. [9], Baik et al. [12] and Dai et al. [14], an optimal value for the turbine inlet temperature can be found. In this paper average turbine inlet temperatures of 320 C (cyclopentane) and of 300 C (acetone) are found to be the optimal compromise between system performance, compactness and economic revenue. As is also suggested in Baik et al. [12], increasing the size of the evaporator and condenser enhances the thermal efﬁciency of the ORC (see Figs. 6 and 8). However, this results in a higher investment cost of the heat transfer equipment. Hence, an optimal volume is found where the net present value reaches the maximum (see Figs. 7 and 9). As reported in Section 3, the multi-objective process indicates that acetone and cyclopentane are the optimal working ﬂuids in terms of efﬁciency, compactness and economy. The optimal working ﬂuids differ from the ones suggested in previous works Refs. [8e18] since in this paper the heat source is at around 370 C while in Refs. [8e18] the heat source is at lower temperature level (100e200 C). However, acetone and cyclopentane are suggested to be suitable working ﬂuids for ORC applications also in other works. In He et al. [45], cyclopentane provides the highest thermal efﬁciency (20.8%) in an ORC used for waste heat recovery of an internal combustion engine. Lai et al. [46] rank cyclopentane as the third best working ﬂuid: ORC net power output is 1 MW and the heat carrier inlet and outlet temperatures are 280 C and 350 C. Ginosar et al. [47] assess the thermal stability of cyclopentane. The authors measure a decomposition rate up to 1500 ppm at 350 C [47]. The maximum temperature in the ORC should then be lower than 300 C where the decomposition rate is in the order of 270 ppm [47]. In the present paper, the highest turbine inlet temperature is

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

10

L. Pierobon et al. / Energy xxx (2013) 1e12

276.9 C. Wang et al. [48] show that acetone exhibits the lowest exergy destruction in the overall ORC for low-temperature waste heat recovery. However, the authors discard the ﬂuid, since the condensing pressure is less than the atmospheric pressure leading to the inﬁltration of ambient air into the loop. As reported by Dai et al. [14], low speciﬁc volumes are crucial to decrease the dimensions of the expander. As surveyed by Nouman [49], acetone presents the lowest volumetric ﬂow rate and expansion ratio in ORC applications. Therefore, the ﬂuid is proposed for the design of compact ORCs, since the size and material cost of the system are reduced. The present methodology can be implemented to design ORC units converting heat at different temperature levels into electric or mechanical power. At each temperature level, the threedimensional Pareto front for each optimal working ﬂuid is identiﬁed. In the multi-objective optimization method proposed in Salcedo et al. [17], the objective functions are the environmental impact and the speciﬁc cost of a solar steam Rankine cycle. Wang et al. [18] employs the heat exchanger area per unit power output and heat recovery efﬁciency as targets. In this paper the desired compactness and economic revenue can be selected from the optimal front (see Figs. 6 and 7). As an improvement of the work presented by Wang et al. [18], in this work the optimal pinch points and the ﬂuid velocities in the shell and on the tubes side of economizer, evaporator, superheater, and internal recuperator can be identiﬁed. In contrast with Wang et al. [18] where simple heat transfer and pressure drop correlations for horizontal tubes are employed, this paper introduces speciﬁc equations (see Eqs. (5)e (11)) for the shell and tube heat exchanger in the multi-objective optimization. Thus, the geometry of the heat transfer equipment can be assessed and utilized to select available components on the market. However, since the heat transfer equipment considered in this analysis is the shell and tube heat exchanger, the ﬁeld of application is directed towards MW-size systems with high temperature heat sources (350 Ce250 C). In fact, shell and tube heat exchangers are normally employed for high temperature and pressure processes. At a maximum operating temperature and pressure in the ORC lower than 250 C and 30 bar and at mass ﬂow rates lower than 2500 m3/h [21], the plate heat exchanger are the preferable heat transfer equipment due to its ﬂexibility and compactness. The algorithm provides also the geometry of the economizer, evaporator, superheater, internal recuperator and condenser. Hence, the standard dimensions of the tubes (outer diameter and length), the shell diameter, the bafﬂe spacing and the pitch ratio can be selected directly from the outcomes of the shell and tube design process. However, for a more accurate estimation of the overall heat transfer coefﬁcient and of the pressure drop, speciﬁc correlations for the selected working ﬂuid should replace the more generic approach presented in Section 2.1.

on the shell side of the heat exchangers, and the temperature at the outlet of the condenser. We apply the methodology to recover the waste heat from the SGT-500 gas turbine utilized to support the power demand in the Draugen off-shore platform. The results suggest that the two most suitable working ﬂuids are acetone and cyclopentane. For acetone, the thermal efﬁciency ranges from 23.7 to 27.0%, in volume from 14.5 to 57.5 m3, and in net present value from 17.7 to 19.8 M$. Cyclopentane performs better both in terms of thermal efﬁciency (27.0e28.1%) and of net present value (19.7e20.1 M$). However, the usable volume becomes larger (55.0e98.4 m3). Based on the volume versus thermal efﬁciency and NPV versus volume curves, we propose two possible solutions and list the geometry of the economizer, evaporator, superheater, internal recuperator and condenser. Acknowledgments The funding from the Norwegian Research Council through Petromaks with project number 203404/E30 is acknowledged. We also acknowledge the kind support from Siemens Industrial Turbomachinery AB, Finspång, Sweden for providing necessary technical documentation. Nomenclature

Abbreviations CC combustion chamber DC direct cost DPB discounted payback period [year] ECO economizer EVA evaporator GA genetic algorithm GEN electric generator GT gas turbine GWP global warming potential HPC high-pressure compressor HPT high-pressure turbine INH intermediate heat exchanger IR internal recuperator LHV lower heating value [kJ/kg] LPC low-pressure compressor LPT low-pressure turbine NG natural gas NPV net present value ORC organic Rankine cycle PEC purchased-equipment cost PT power turbine SUP superheater TUR turbine

5. Conclusions We propose a multi-objective optimization that considers the thermal efﬁciency, compactness and net present value, by employing the genetic algorithm to design organic Rankine cycles. The shell and tube heat exchangers are the heat transfer equipment. The space requirement of the ORC is assessed by calculating the geometry (tube diameter and length, pitch and bafﬂe spacing) of the shell and tube heat exchanger following a well-established design procedure. We employ different heat transfer correlations depending on the ﬂuid phase, and we also quantify pressure drops within the cycle. The variables considered in the optimization routine are the turbine inlet pressure, pinch points of the evaporator, superheater, internal recuperator and condenser, the velocity

Notations A area [m2] a; b; c; d; ecoefﬁcients in Eqs. (38) and (39) bc bafﬂe cut c speed [m/s] cp speciﬁc heat capacity [kJ/(kg C)] d diameter [m] f function Ft temperature correction factor Fts heat exchanger correction factor Ftv turbine correction factor g acceleration of gravity [m/s2] h heat transfer coefﬁcient [W/(m2 C)] or enthalpy [kJ/kg]

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

L. Pierobon et al. / Energy xxx (2013) 1e12

hu HR ITOT i jth jtf jsh jsf l M Ma m _ m Nt n NPV Nu P_ p pt PEC Pr q Q_ R Rp Re T t U u V v V_ _ W

utilization factor [h/y] heat rate [kJ/kWh] total capital investment [$] year tube side heat transfer factor tube side friction factor shell side heat transfer factor shell side friction factor length [m] molecular mass [kg/kmol] operating and maintenance factor exponent in Eq. (6) mass ﬂow [kg/s] number of tubes number of years net present value Nusselt number electric power [kW] pressure [Pa] or price [NOK/tons] tube pitch [m] purchased-equipment cost Prandtl number interest factor heat rate [kW] yearly income [$/y] surface roughness [mm] Reynolds number temperature [K] temperature [ C] overall heat transfer coefﬁcient [kW/(m2 K)] ﬂuid velocity [m/s] volume [m3] speciﬁc volume [m3/kg] volume rate [m3/s] mechanical power [kW]

Greek symbols D difference r density [kg/m3] l thermal conductivity [kW/(m C)] m dynamic viscosity [N s/m2] G tube loading [kg/(m s)] h efﬁciency Subscripts b bafﬂe CO2 carbon dioxide cond condenser e equivalent eco economizer el electric eva evaporator g gas gen generator ht heat exchanger i inside id inside dirt coefﬁcient in inlet l liquid lm logarithmic mean m mechanical nb nucleate boiling net net o outside od outside dirt coefﬁcient

out p r s sw sup t th tot ts tt tur tw w x

11

outlet pump reduced shell shell wall superheater tube thermal total target shell target tube turbine tube wall wall or water axial

References [1] Commission of the European Communities. Europe’s climate change opportunity. Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, Brussels, Belgium 2008. [2] De Paepe M. Situation of the ORCNext project. In: Proceedings of the International Symposium on Advanced Waste Heat Valorization Technologies (September 13, Lecture n 1), September 13e14, Kortrijk, Belgium 2012. [3] Colonna P. Development from early days, current status and an outlook on relevant research topics and new applications. In: Proceedings of the International Symposium on Advanced Waste Heat Valorization Technologies (September 13, Lecture n 2), September 13e14, Kortrijk, Belgium 2012. [4] Gewald D, Karellas S, Schuster A, Spliethoff H. Integrated system approach for increase of engine combined cycle efﬁciency. Energy Conversion and Management 2012;60:36e44. [5] Rokni M. Plant characteristics of an integrated solid oxide fuel cell cycle and a steam cycle. Energy 2010;35:4691e9. [6] Domingues A, Santos H, Costa M. Analysis of vehicle exhaust waste heat recovery potential using a Rankine cycle. Energy 2013;49:71e85. [7] Vélez F, Segovia JJ, Martín CM, Antolín G, Chejne F, Quijano A. A technical, economical and market review of organic Rankine cycles for the conversion of low-grade heat for power generation. Renewable and Sustainable Energy Reviews 2012;16:4175e89. [8] Sun J, Li W. Operation optimization of an organic rankine cycle (ORC) heat recovery power plant. Applied Thermal Engineering 2011;31:2032e41. [9] Roy JP, Mishra MK, Misra A. Parametric optimization and performance analysis of a waste heat recovery system using Organic Rankine Cycle. Energy 2010;35:5049e62. [10] Hettiarachchi HDM, Golubovic M, Worek WM, Ikegami Y. Optimum design criteria for an organic Rankine cycle using low-temperature geothermal sources. Energy 2007;32:1698e706. [11] Quoilin S, Declaye S, Tchanche BF, Lemort V. Thermo-economic optimization of waste heat recovery Organic Rankine Cycles. Applied Thermal Engineering 2011;31:2885e93. [12] Baik Y-J, Kim M, Chang K-C, Lee Y-S, Yoon H-K. A comparative study of power optimization in low- temperature geothermal heat source driven R125 transcritical cycle and HFC organic Rankine cycles. Renewable Energy 2013;54: 78e84. [13] Wang J, Yan Z, Wang M, Ma S, Dai Y. Thermodynamic analysis and optimization of an (organic Rankine cycle) ORC using low grade heat source. Energy 2013;49:356e65. [14] Dai Y, Wang J, Gao L. Parametric optimization and comparative study of organic Rankine cycle (ORC) for low grade waste heat recovery. Energy Conversion and Management 2009;50:576e82. [15] Cayer E, Galanis N, Nesreddine H. Parametric study and optimization of a transcritical power cycle using a low temperature source. Applied Energy 2010;87(4):1349e57. [16] Shengjun Z, Huaixin W, Tao G. Performance comparison and parametric optimization of subcritical organic Rankine cycle (ORC) and transcritical power cycle system for low-temperature geothermal power generation. Applied Energy 2011;88:2740e54. [17] Salcedo R, Antipova E, Boer D, Jiménez L, Guillén-Gosálbez G. Multi-objective optimization of solar Rankine cycles coupled with reverse osmosis desalination considering economic and life cycle environmental concerns. Desalination 2012;286:358e71. [18] Wang ZQ, Zhou NJ, Guo J, Wang XY. Fluid selection and parametric optimization of organic Rankine cycle using low temperature waste heat. Energy 2012;40:107e15. [19] American coatings association. HMISÒ implementation manual, 3rd ed. www. paint.org/programs/hmis.html [accessed 07.02.13].

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

12

L. Pierobon et al. / Energy xxx (2013) 1e12

[20] Fakheri A. A General expression for the determination of the log mean temperature correction factor for shell and tube heat exchangers. Journal of Heat Transfer 2003;125(3):527e30. [21] Richardson JF, Peacock DG. Chemical and biochemical reactors and process control (Coulson and Richardson’s Chemical Engineering; volume 3). 3rd ed. Oxford, UK: Elsevier, ISBN 978-0-08-041003-6; 1994. [22] Hall SG. Capital cost targets for heat exchanger networks comprising mixed materials of construction, pressure ratings and exchanger types. Chemical Engineering 1990;14(3):319e35. [23] Kern DQ. Process heat transfer. Singapore: McGraw-Hill; 1950. ISBN 0-07Y85353-3. [24] Cooper MG. Heat ﬂow rates in saturated nucleate pool boiling e a wideranging examination using reduced properties. Advances in Heat Transfer 1984;16:157e239. [25] Arsalis A, Spakovsky von RM, Calise F. Thermoeconomic modeling and parametric study of hybrid solid oxide fuel cell-gas turbine-steam turbine power plants ranging from 1.5 MWe to 10 MWe. Journal of Fuel Cell Science and Technology 2009;6. 011015e1. [26] Lian ZT, Chua KJ, Chou SK. A thermoeconomic analysis of biomass energy for trigeneration. Applied Energy 2010;87:84e95. [27] Colonna P, Rebay S, Harinck J, Guardone A. Real-gas effects in ORC turbine ﬂow simulations: inﬂuence of thermodynamic models on ﬂow ﬁelds and performance parameters. In: Wesseling P, Oñate E, Périaux J, editors. Proceedings of the European Conference on Computational Fluid Dynamics, ECCOMAS CFD 2006. Delft, The Netherlands: TU Delft; 2006. Paper 602. [28] Saravanamuttoo HIH, Rogers GFC, Cohen H, Straznicky PV. Gas turbine theory. 6th ed. Dorset, UK: PEARSON Prentice Hall, ISBN 978-0-13-2224376; 2009. [29] Pierobon L, Rambabu K, Haglind F. Waste heat recovery for off-shore applications. In: Proceedings of the ASME 2012 International Mechanical Engineering Congress & Exposition, IMECE2012, November 9e15, Houston, Texas, USA 2012. Paper 86254. [30] DOW Chemical Company. Dowtherm Q heat transfer ﬂuid [accessed 07.02.13], msdssearch.dow.com/PublishedLiteratureDOWCOM/dh_005f/0901b8038005f2c 1.pdf?ﬁlepath¼heattrans/pdfs/noreg/176-01467.pdf&fromPage¼GetDoc; June 1997. [31] Marler RT, Arora JS. Survey of multi-objective optimization methods for engineering. Structural and Multidisciplinary Optimization 2004;26(6):369e95. [32] Falkenauer E. Genetic algorithms and grouping problems. Chichester, UK: John Wiley & Sons Ltd., ISBN 978-0-471-97150-4; 1997. [33] Lemmon WE, Huber LM, McLinden OM. NIST reference ﬂuid thermodynamic and transport properties-REFPROP, version 9.0, user’s guide. Boulder, Colorado, USA: Thermophysical Properties Division, National Institute of Standards and Technology; 2010.

[34] Pierobon L, Nguyen T-V, Larsen U, Haglind F. Optimization of organic Rankine cycles for off-shore applications. In: Proceedings of the ASME Turbo Expo 2013, June 3e7, San Antonio, Texas, USA 2013. [Final version available online from 27.05.13]. [35] Lagarias JC, Reeds JA, Wright MH, Wright PE. Convergence properties of the Nelder-Mead Simplex method in low dimensions. SIAM Journal of Optimization 1998;9(1):112e47. [36] Tahseen AT, Ishak M, Rahman MM. A numerical study laminar forced convection of air for in-line bundle of cylinders crossﬂow. Asian Journal of Scientiﬁc Research 2013;6:217e26. [37] Bejan A, Tsatsaronis G, Moran M. Thermal design and optimization. Canada: John Wiley & Sons, Inc., ISBN 0-471-58467-3; 1995. [38] Index Mundi. Natural gas monthly price e Norwegian Krone per thousands of cubic meters [accessed October 2012], www.indexmundi.com/commodities/? commodity¼natural-gas&;currency¼nok; Apr 2012eOct 2012. [39] Lin B, Li X. The effect of carbon tax on per capita CO2 emissions. Energy Policy 2011;39:5137e46. [40] Ministry of the Environment. The government is following up on the climate Agreement. Norway: Press release, Oslo [accessed 07.02.13], www. regjeringen.no/en/dep/md/press-centre/Press-releases/2012/thegovernment-is-following-up-on-the-cl.html?id¼704137; 08.10.2012. [41] Rypdal K. Anthropogenic emissions of the greenhouse gases CO2, CH4 and N2O in Norway. Report No.93/24. Oslo, Norway: Central Bureau of Statistics of Norway, ISBN 82-537-3917-6; 1993. [42] SYSTAT Software Inc. TableCurve 2D v. 5.01. www.sigmaplot.com/products/ tablecurve2d [accessed 07.02.13]. [43] Gonçalves N, Faias S, Sousa de J. Biomass CHP technical and economic assessment applied to a Sawmill plant. In: Proceedings of the International Conference on Renewable Energies and Power Quality (ICREPQ’12) Santiago de Compostela, Spain, 28e30 March 2012. Paper 638. [44] Ghirardo F, Santin M, Traverso A, Massardo A. Heat recovery options for onboard fuel cell systems. International Journal of Hydrogen Energy 2011;36:8134e42. [45] He M, Zhang X, Zeng K, Gao K. A combined thermodynamic cycle used for waste heat recovery of internal combustion engine. Energy 2011;36:6821e9. [46] Lai NA, Wendland M, Fischer J. Working ﬂuids for high-temperature organic Rankine cycles. Energy 2011;36:199e211. [47] Ginosar DM, Petkovic LM, Guillen DP. Thermal stability of cyclopentane as an organic rankine cycle working ﬂuid. Energy & Fuels 2011;25:4138e44. [48] Wang D, Ling X, Peng H. Performance analysis of double organic Rankine cycle for discontinuous low temperature waste heat recovery. Applied Thermal Engineering 2012;48:63e71. [49] Nouman J. Comparative studies and analysis of working ﬂuids for organic Rankine cycles e ORC [Master of Science Thesis]. Stockholm, Sweden: KTH School of Industrial Engineering and Management; 2012.

Please cite this article in press as: Pierobon L, et al., Multi-objective optimization of organic Rankine cycles for waste heat recovery: Application in an offshore platform, Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.05.039

Copyright © 2021 COEK.INFO. All rights reserved.