Applied Thermal Engineering 36 (2012) 307e313

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Thermal-hydraulic design of fan-supplied tube-n condensers for refrigeration cassettes aimed at minimum entropy generation
Christian J.L. Hermes*, Waldyr de Lima e Silva Jr., Felipe A.G. de Castro
Center for Applied Thermodynamics, Department of Mechanical Engineering, Federal University of Paran, PO Box 19011, 81531990 Curitiba-PR, Brazil

a r t i c l e i n f o
Article history: Received 6 September 2011 Accepted 19 October 2011 Available online 29 October 2011 Keywords: Tubeen condenser Entropy generation minimization Simulation Optimization Thermal design Refrigeration cassettes

a b s t r a c t
The method of minimum entropy generation is used to assess the effects of several design parameters on the performance of fan-supplied tube-n condensers for light commercial refrigeration applications (heat transfer duty w1 kW). A simplied mathematical model is put forward to simulate the thermalhydraulic behavior of the condenser, and validated against experimental data obtained elsewhere, showing a good agreement between calculated and measured counterparts. The dimensionless rate of entropy generation due to heat ow across nonzero temperature difference and to viscous uid ow is calculated for different condenser designs (number of ns and tubes, tube spacing and outer diameter) at xed operating conditions (heat transfer duty, ow rates, inlet temperatures). It is shown that there do exist optima values for the face velocity, n density, tube diameter and heat exchanger effectiveness that minimize the production of entropy, so that some design guidelines are proposed. 2011 Elsevier Ltd. All rights reserved.

1. Introduction In general, tube-n heat exchangers consist of one or more rows of copper tubes with external aluminum ns through which air ows either by free convection or supplied by a fan. This type of heat exchanger has been widely used in many HVAC-R engineering applications, remarkably in light commercial refrigeration appliances (cooling capacities w1 kW), which consume around 1% of the electricity produced worldwide [1]. In such applications, the refrigerant ows inside the tubes whereas air ows externally over the bundle of tubes and ns. Given the importance of small-capacity refrigerators for the global energy matrix, efforts have been devoted to reduce the energy consumption of such systems. For instance, Waltrich et al. [2] analyzed the impact of the components (compressor, condenser and evaporator) on both overall system cost and COP, concluding that the heat exchangers have a wide margin for performance improvement with little or no cost penalty. The design of tube-n heat exchangers involves both geometric (e.g. tube diameter, tube spacing, number of tube rows, n thickness, shape and spacing) and operational parameters (e.g. ow rates and temperatures of the air and refrigerant streams) in order to accomplish a certain heat transfer duty at the penalty of fan pumping power. Some of these parameters affect quite signicantly

* Corresponding author. Tel.: 55 41 3361 3239; fax: 55 41 3361 3129. E-mail address: chermes@ufpr.br (C.J.L. Hermes). 1359-4311/$ e see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.applthermaleng.2011.10.038

the thermo-hydraulic performance of the heat exchanger whilst others may be relaxed. In general, performance indicators (e.g. heat transfer rate and pumping power) respond similarly to changes in design parameters, i.e. a heat transfer enhancement (desired effect) is usually followed by an increase in pressure drop and then in pumping power (undesirable effects). Therefore, the heat exchanger should be designed in such a way as to balance the heat transfer enhancement and the pumping power increase trade-offs. A method devised for this purpose consists of counteracting the thermodynamic losses associated with irreversible heat transfer across a nite temperature difference with the irreversibilities associated with viscous uid ow [3]. As their effects on the rate of entropy generation oppose each other, there does exist a design (geometry and running conditions) that yields the overall entropy generation towards a minimum. This method, named Entropy Generation Minimization (EGM) after Bejan [3], has been widely used in geometric optimization of various types of heat exchangers [4e9]. However, studies of the application of the EGM to tube-n heat exchanger design are scarce. For instance, Saechan and Wongwises [10] applied the EGM as the objective function to conduct a computational optimization of a tube-n condenser for air conditioning applications (cooling capacity w10 kW, air ow rate w10,000 m3/h). Recently, Pussoli et al. [11] employed the EGM to size peripheral n heat exchangers, a new heat exchanger concept for household refrigeration applications (cooling capacity w0.1 kW, air ow rate w100 m3/h). Nevertheless, it has not been found in the open literature studies focusing on the design and optimization of

308

C.J.L. Hermes et al. / Applied Thermal Engineering 36 (2012) 307e313

Nomenclature Roman A C cp Dt f G h L m Nf Nlo NS Ntr Ntu p Plo Ptr Q Re s Sgen t UA V heat transfer surface, m2 thermal capacity, W/K isobaric specic heat, J/kg K tube outer diameter, m friction factor mass ux, kg/m2s specic enthalpy, J/kg length, m mass ow rate, kg/s number of ns number of longitudinal tubes dimensionless entropy generation number of transversal tubes number of transfer units pressure, Pa longitudinal tube spacing, m transversal tube spacing, m heat transfer rate, W Reynolds number (raVfaceDt/ma) specic entropy, J/kg K entropy generation rate, W/K temperature, K conductance, W/K velocity, m/s

Greek

a r m df dt DS

heat exchanger effectiveness heat transfer coefcient, W/m2 K specic mass, kg/m3 viscosity, Pa s n thickness, m tube thickness, m entropy variation rate, W/K

Subscripts a air side cond condensing temperature evap evaporating temperature f n face heat exchanger face area i inlet max maximum min minimum o outlet r refrigerant side sat saturated sub subcooled sup superheated t tube w wall

tube-n heat exchangers for light commercial refrigeration appliances (cooling capacity w1 kW, air ow rate w1000 m3/h). In this context, this paper is aimed at investigating the thermalhydraulic performance of tube-n condensers for light commercial applications with focus on the entropy generation minimization. For doing so, a mathematical model to simulate the thermalhydraulic performance of tube-n condensers operating with refrigerant R-134a as working uid was developed and validated against experimental data obtained in the literature [12]. The model was then used to explore the effects of various geometric and operating parameters on the condenser performance. 2. Mathematical model In tube-n condensers, the refrigerant ows inside the tubes whereas air ows transversally through the bundle of nned-tubes, receiving heat from the refrigerant, as depicted in Fig. 1. The refrigerant enters the condenser as superheated vapor, it is then condensed and exits as subcooled liquid, as illustrated in Fig. 2. In both the refrigerant and air streams, irreversibilities due to heat transfer with nite temperature difference and uid friction take place, so that the condenser model was divided into two submodels, namely thermal and hydraulic, in order to account for both effects. 2.1. Thermal model The thermal model is concerned with the calculations of the heat transfer rate, and the air and refrigerant temperatures at the outlet ports. The former is calculated separately for each ow region (superheated, saturated and subcooled), so that the overall heat transfer rate is obtained from:

Gonalves et al. [13] showed that for high air ow rates, such as the gures found in refrigeration cassettes (w1000 m3/h), the three-dimensional inuence of the refrigerant circuit on the thermal performance (i.e. heat exchanger effectiveness) can be neglected. This means that the heat exchanger can be linearized for sake of heat transfer calculations, as depicted in Fig. 2. Thus, the heat transfer rates in each ow region can be calculated from:

Qsup mr hr;i hv sup mr cp;r;v tr;i ta;v

(2)

Q Qsup Qsat Qsub mr hr;i hr;o

(1)

Fig. 1. Schematic representation of a tube-n condenser for refrigeration cassetes.

C.J.L. Hermes et al. / Applied Thermal Engineering 36 (2012) 307e313

309

where Amin is the minimum free ow passage, and Ga is the air mass ux dened as the ratio between the air mass ow rate, ma, and the face area, Aface. The friction factor, f, is obtained from the correlation of Wang et al. [15]. The pressure drop on the refrigerant side was neglected, keeping in mind that the irreversibilities due to the refrigerant are very small compared to those found on the air side [10]. 2.3. Entropy generation rate The entropy generation in a heat exchanger takes place in both air and refrigerant sides due to two main sources of irreversibilities: heat transfer with nite temperature difference and pressure drop due to viscous ow. Like Saechan and Wongwises [10], the total entropy generation, Sgen, is obtained adding the rates of entropy variation (reversible and irreversible terms) in both air and refrigerant sides, so that the reversible terms (Q/tw) are canceled out:

Fig. 2. Schematic representation of the heat exchanger physical model.

Qsat mr hv hl sat ma cp;a tr;sat ta;l Qsub mr hl hr;o sub mr cp;r;l tr;sat ta;i

(3) (4)

Sgen DSa DSr

 Sgen;a

   Q Q Sgen;r Sgen;a Sgen;r tw tw (9)

where is the heat exchanger effectiveness for each ow region, ta,i, ta,l and ta,v are the air temperatures at the heat exchanger inlet, at the entrance of the saturated region and at the entrance of superheated region, respectively (see Fig. 2). Assuming an uniform refrigerant temperature in the two-phase ow region, the heat exchanger effectiveness in this region can be calculated from [14]:

Assuming the air behaves as an ideal gas, the entropy variation on the air side can be obtained from the integration of the Gibbs equation, dH tdSVdp, yielding

sat 1 expNtu

DSa ma cp;a ln

(5)

ta;o ta;i

! ma Ra ln 1

Dpa
pa;i

! (10)

The effectivenesses in the single-phase ow regions, on the other hand, can be obtained from the following -Ntu relationship for cross ow heat exchangers [14]:

sup=sub 1 exp

   0:22 Ntu  0:78 exp Cratio Ntu 1 Cratio

(6)

where the rst term on the right-hand side represents the entropy variation due to temperature difference, whereas the second term stands for the entropy variation due to pressure drop. Note that the air temperature at the heat exchanger outlet is calculated from:

where Ntu is the number of transfer units (UA/Cmin), Cratio mrcp,r/ macp,a and UA is the thermal conductance of each ow region. Disregarding the presence of incrustations in both refrigerant and air sides, and neglecting the thermal resistance of the tube wall, the heat exchanger conductance can be calculated from:

ta;o ta;i

Q ma cp;a

(11)

For the refrigerant side, the entropy variation is calculated from:

DSr mr sr;o sr;i

(12)

UA1

hf aa Ao

1

ar Ai 1

(7)

where hf is the efciency of the nned surface, assumed to be equal to the unity for this type of heat exchanger [12], Ao At Af is the heat transfer surface of the nned region, and Ai is the heat transfer surface inside the tube for each ow region, so that the total heat transfer surface is A Asup Asat Asub. The overall air-side heat transfer coefcient, aa, was obtained from the correlation of Wang et al. [15]. For the refrigerant single-phase ow regions, the heat transfer coefcient, ar, was calculated from the correlation of Dittus and Boelter [16], whereas an innite heat transfer coefcient was assumed for the two-phase ow region. 2.2. Hydrodynamic model The hydrodynamic model is aimed at determining the pressure drops in both the air and refrigerant streams, which are related to the pumping power consumed by the fan and the refrigerant compressor, respectively. On the air side, the pressure loss is obtained from the following equation [14]:

In general, the rate of entropy generation is written in a dimensionless form, the so-called number of entropy generation, NS, as follows:

NS

Sgen ma cp;a

(13)

2.4. Solution scheme The model was implemented using the EES platform [17] which has built-in routines for calculating the thermodynamic properties of air and refrigerant R-134a. The equation set of 140 nonlinear equations is solved simultaneously through the NewtoneRaphson technique. The input data are the heat exchanger geometry (Ptr, Plo, Ntr, Nlo, Nf, L, Dt, dt, df) and the operating conditions (ma, ta,i, mr, tr,i, psat or Q). The solution algorithm allows simulations to be performed either with a prescribed refrigerant pressure (as used in the model validation exercise) or with a prescribed heat transfer duty (as adopted in the optimization tasks). 3. Model validation The mathematical model predictions were compared against 48 experimental data points obtained from Waltrich et al. [12] for

Dpa 2f

G2 Ao a

ra;i Amin

(8)

310

C.J.L. Hermes et al. / Applied Thermal Engineering 36 (2012) 307e313

condensers with two and three staggered rows of tubes and at ns, running with refrigerant R-134a. The common parameters to all condensers tested are as follows: Dt 9.5 mm, df 0.14 mm, dt 0.6 mm, Plo 22 mm and Ptr 26 mm. The following parameters were varied during the experiments: 304 L 380 mm, 153, 2 Nlo 3, 6 Ntr 10, 20 ta,i 30  C, 68 Nf tsat 42  C, 260 Va 1130 m3/h and 40 tr,i 54  C, 30 5 mr 26 kg/h. The experimental data span a range of heat transfer rate from 270 to 1360 W. Fig. 3 compares the experimental results with the model predictions for the heat exchange rate and air-side pressure drop. It can be noted that, for the heat transfer rate, the model results showed an excellent agreement with the experimental data, with all errors within a 5% band (Fig. 3.a). For the air-side pressure drop,

the differences between the numerical and experimental results showed errors ranging from 25% to 15% (Fig. 3.b). It should be noted that the errors observed are of the same order of magnitude as those reported in [15]. 4. Optimization results The following parameters were considered during the optimization exercise: number of tubes in the air ow direction (Nlo), number of ns (Nf), tube diameter (Dt), air face velocity (Vface) and, therefore, the Reynolds number (Re raVfaceDt/ma). It can be noted that the condenser face area was held xed (Ntr 10, L 350 mm) as this parameter is usually taken as a constraint for the refrigeration cassette design [2]. In all cases, the transversal and longitudinal tube spacings were held constrained as the experimental data used for the model validation exercise are restricted to Plo 22 mm and Ptr 26 mm. In addition, the optimization was conducted for an inlet air temperature of 32  C, which is a standardized condition for light commercial refrigeration appliances [2]. Moreover, the heat transfer duty was held constrained at 1.0 kW. The other boundary conditions are the refrigerant mass ow rate (18 kg/h), and the refrigerant condition at the condenser inlet, which was calculated considering the condensing pressure (which is kept free to vary and, therefore, shall be determined by the model) and the specic entropy of refrigerant at the condenser inlet (1.0 kJ/kg K), calculated from an isentropic compression process from an evaporating temperature of 10  C and an evaporator superheating of 20  C, as illustrated in Fig. 4. One should note that for a refrigeration system, COP w tevap/ (tcond e tevap), in such a way that a lower condensing pressure leads to a better overall system performance. Since the condensing pressure is treated as a oating parameter during the optimization task, the heat exchanger design that presents the best local (component-level) performance in terms of pressure drop and temperature difference will lead to the best system performance. On one hand, the optimization for heat transfer enhancement leads to lower average temperature differences and, therefore, to a lower condensing pressure. On the other hand, the optimization for pressure drop reduction yields a higher air ow rate for the same pumping power, which improves the air-side heat transfer coefcient that also reduces the condensing pressure.
10
5

1400

1200

Calculated heat transfer rate [W]

+5%

1000
-5%

800

600

400

200

400

600

800

1000

1200

1400

Measured heat transfer rate [W]

b
Calculated air-side pressure drop [Pa]

100

80

+15%

1 kW

constrained heat duty

J/k gK

60

40

Pressure [kPa]

10

outlet

pressure free to vary

inlet

20
10
2

0 0

20

40

60

80

100

Measured air-side pressure drop [Pa]

Fig. 3. Comparison between calculated and measured (a) Heat transfer rates and (b) Airside pressure drops.

10 -50

50

100

150

200

250

300

s=

350

1k

-25%

10

condenser

400

Enthalpy [kJ/kg]
Fig. 4. Schematic representation of the optimization procedure.

C.J.L. Hermes et al. / Applied Thermal Engineering 36 (2012) 307e313

311

Fig. 5 explores the inuence of the number of tubes in the air ow direction on the entropy generation number, NS, for different Reynolds numbers, Re, considering a constrained number of ns (Nf 80) and tube diameter (Dt 9.5 mm). The minima observed in Fig. 5.a are due to the fact that the temperature difference between the streams decreases with an increasing Reynolds number (region of the curves on the left of the points of minimum), while the pressure drop and so the pumping power both increase with an increasing Re (region on the right of the points of minimum), as can be seen in Fig. 5.b. One should note that the optima are observed for Re w 4250 (Nlo 2), Re w 3500 (Nlo 3) and Re w 2750 (Nlo 4), i.e. the points of minimum move towards lower Re values for an increasing number of tubes. An increasing number of tube rows, on one hand, augments the heat transfer surface and reduces the temperature difference, and, on the other hand, increases the pressure drop. In addition, an

0.0011
4

0.0010 0.0009 0.0008

3

0.0007 0.0006 0.0005 0.0004 0.0003 2500

Nlo = 2

3500

4500

5500

6500

7500

inspection of Fig. 5.b reveals that for low Re the pressure drops are of the same magnitude, but for high Re, the pressure drop for 4-tube rows are much higher than that observed for 2-tube rows. This explains the reason why the curves cross each other. A similar behavior can be observed in Fig. 6, where the number of ns was varied from 80 to 140. In this case, the points of minimum entropy generation occurred at Rew4250 (Nf 80), Rew3750 (Nf 110) and Rew3500 (Nf 140), with the entropy generation ruled by heat transfer with nite temperature difference for low Re, and by the viscous ow for high Re. Again, the curves cross each other: as the number of ns increases, the optima move to the left, indicating that the effect of pressure drop becomes dominant for low Re values. Despite the face area has been kept constrained in the previous analyses, a study of the effect of the total number of tubes (Nlo Ntr) on the entropy generation rate was also carried out. As can be seen in Fig. 7, the arrangement with a larger face area and a smaller number of tube rows (i.e. the slender heat exchanger, with 2 10 tubes) shows an entropy generation rate dramatically smaller than the other arrangements (3 7 and 4 5 tubes, having both a heat transfer surface of the same order of that observed for the 2 x 10 conguration), with NS / min for Re w 4000. This is so as the ow passage is larger and the length is smaller for the 2 10 arrangement, which reduce the pressure drop dramatically for the same heat transfer duty. An additional effect, not accounted for the model, relies on the fact that the local heat transfer coefcient is higher for the rst tube rows [18], where more heat transfer area should be allocated for sake of heat transfer enhancement. So far the tube spacings in both transversal and longitudinal directions have been kept constrained at Plo 22 mm and Ptr 26 mm due to manufacture restrictions. Fig. 8 assesses the effects of transversal and longitudinal tube spacings on the dimensionless entropy generation when the number of tubes are xed at No 2 and Ntr 10, and Re 4000. One can see that the rate of entropy generation can be diminished by 15% if Plo and Ptr are increased to 25 and 34 mm, respectively. The former increases the heat exchanger surface and the pressure drop, whilst the latter increases the face area and the ow passage. These competing trade-offs are addressed by the EGM approach, which provides

NS

Re

400 350 300 250

3 4

0.00080

0.00070

pa [Pa]

0.00060

200 150 100

Nlo = 2

NS

Nf = 80

0.00050
110 140

0.00040

50 0 2500
0.00030 2500

3500

4500

5500

6500

7500

3500

4500

5500

6500

7500

Re
Fig. 5. Effect of number of longitudinal tubes on the (a) Entropy generation number and (b) Air-side pressure drop (Nf 80, Dt 9.5 mm).

Re
Fig. 6. Effect of number of ns on the entropy generation number (Nlo 2, Dt 9.5 mm).

312

C.J.L. Hermes et al. / Applied Thermal Engineering 36 (2012) 307e313

0.0013 0.0012 0.0011 0.0010 0.0009 0.0008 0.0007 0.0006 0.0005 0.0004 2500
Nlo x Ntr = 2 x 10 4 x5

280 5.8 5.2 240 4.6 4.3 4.0 200

3 x7

0.65 0.6 0.55 0.5 0.45 0.4 3.7

NS

Nf
160 0.35 0.3 120 0.25
3500 4500 5500 6500 7500

0.2 80 0.008 0.009 0.010 0.011

NS x104
0.012

Re
Fig. 7. Effect of face area (Nlo Ntr) on the entropy generation number (Nf 80, Dt 9.5 mm).

D t [mm]
Fig. 9. Effect of tube diameter and number of ns on (a) Entropy generation number and (b) Heat exchanger effectiveness (Nlo 2, Ntr 10, Re 4000).

a clear indication of the optimum tube spacing. It should be mention that an additional effect, not accounted for the model, relies on the inuence of the tube spacing on the n efciency, which is reduced inasmuch as the tube spacing is increased for a xed number of tubes. Fig. 9 explores the combined effect of the number of ns and tube diameter on the entropy generation rate for Re 4000. The original tube spacings were considered in this analysis. A region of minimum entropy generation is observed for a tube diameter of w11 mm and w180 ns. A smaller number of ns or a smaller diameter provides, on one hand, a larger ow passage area and, on the other hand, a smaller heat transfer surface. In contrast, a larger n number associated with larger tube diameters restrict the free ow passage and increase the heat transfer surface, thereby
0.00060

increasing the pressure drop and pumping power. These competing trade-offs explain the minimum observed. It is worthy of mention that a similar behavior was reported by Saechan and Wongwises [10] for air conditioning condensers, although the heat transfer duty was not held constrained in their analysis. Fig. 9 also shows the parametric curves for the heat exchange effectiveness, calculated as follows:

Q ma cp;a tr;sat ta;i

(14)

It can be observed that the lowest entropy generation number does not represent the highest effectiveness, which increases
0.00045
80

Ptr [mm]

0.00055
0.022

0.00043

105

0.00050

0.00041

Nf=280 205 155 255 180 230

NS

0.00045
0.026

NS
0.00039

130

0.00040
0.030 0.034

0.00037
global minimum

0.00035 0.018

0.020

0.022

0.024

0.026

0.00035 0.10

0.20

0.30

0.40

0.50

0.60

0.70

Plo [m]
Fig. 8. Effect of tube pitches (Plo and Ptr) on the entropy generation number (Nf 80, Dt 9.5 mm Re 4000).

Fig. 10. Entropy generation number as a function of heat exchanger effectiveness (Nlo 2, Ntr 10, Re 4000.

C.J.L. Hermes et al. / Applied Thermal Engineering 36 (2012) 307e313

313

monotonically with both the tube diameter and number of ns, revealing that there is an effectiveness that minimizes the entropy generation rate, as illustrated in Fig. 10. Nevertheless, it should be kept in mind that the effectiveness is a performance evaluation parameter based uniquely on the energy transfer (1st-law), and does not account for the losses caused by the viscous uid ow that in tube-n heat exchangers plays a major role on the rate of entropy generation (see Fig. 5.b). Since both the hydraulic and thermal behavior are taken into account when the minimum entropy generation criterion is adopted (2nd-law), this approach seems to be more suitable for the optimization of this type of heat exchanger. 5. Summary and conclusions This work employed the method of entropy generation minimization to investigate some design aspects of tube-n condensers for refrigeration cassettes. A simplied model to simulate the thermal-hydraulic performance of tube-n condensers was put forward and validated against experimental data obtained from the literature, showing a good agreement between the model predictions for the heat transfer rate and air-side pressure drop and the measured counterparts. The optimization exercise was carried out letting the condensing pressure free to vary so that the heat transfer duty (a design constraint) was held xed. It was found that there do exist entropy generation minima for the number of ns, the number of tubes rows, the total number of tubes, the tube diameter and the air face velocity. In addition, it was also observed that the values of tube diameter and number of ns that minimize the number of entropy generation do not necessarily represent the geometry with a more effective heat exchange, since the latter takes into account only heat transfer aspects (1st-law criterion), whereas the former considers the heat exchanger hydraulic characteristics (2nd-law criterion) as well. It should be mention that the design which led to the minimum entropy generation has Nlo 2, Ntr 10, Nf z 180, and Dt z 11 mm, for Re z 4000, providing a heat exchanger effectiveness of 0.45. Finally, a heat exchanger shall be designed to accomplish a certain heat transfer duty, subjected to some constraints (mass ow rates, inlet temperatures) which depend upon the application. Bearing this in mind, and looking at the results achieved from the simulations, the following practical guidelines for tube-n condenser design can be summarized:  A high effectiveness heat exchanger has not necessarily a good thermal-hydraulic design, as the effectiveness does not account for the fan pumping power. A minimum entropy generation criterion should be adopted instead.  There does exist a Reynolds number which minimizes the rate of entropy generation for any geometric conguration, so that the heat exchanger shall be designed for such a condition.

 There is a combination of number of ns and tube diameter that balances the heat transfer surface and free ow passage trade-offs.  Heat exchangers with larger face area and fewer number of tube rows (slender designs) lead to lower entropy generation. Acknowledgements The authors duly acknowledge the Polo Laboratories, Federal University of Santa Catarina for providing them with the experimental data used for the model validation exercise. References
[1] C. Melo, L.W. Silva, A Perspective on Energy Savings in Household Refrigerators, Sustainable Refrigeration and Heat Pump Technology Conference, Stockholm, Sweden (2010). [2] M. Waltrich, C.J.L. Hermes, C. Melo, Simulation-based design and optimization of refrigeration cassettes, Appl. Energ. 88 (2011) 4756e4765. [3] A. Bejan, Method of entropy generation minimization or modeling and optimization based on combined heat transfer and thermodynamics, Rev. Gen. Therm. 35 (1996) 637e646. [4] J.Y. San, C.L. Jan, Second-law analysis of a wet crossow heat exchanger, Energy 25 (2000) 939e955. [5] M. Yilmaz, O.N. Sara, S. Karsli, Performance evaluation criteria for heat exchangers based on second law analysis, Exergy 1 (2001) 278e294. [6] P.P.P.M. Lerou, T.T. Veenstra, J.F. Burger, H.J.M. ter Brake, H. Rogalla, Optimization of counterow heat exchanger geometry through minimization of entropy generation, Cryogenics 45 (2005) 659e669. [7] M. Mishra, P.K. Das, S. Sarangi, Second law based optimisation of crossow plate-n heat exchanger design using genetic algorithm, Appl. Therm. Eng. 29 (2009) 2983e2989. [8] I. Kotcioglu, S. Caliskan, A. Cansiz, S. Baskaya, Second law analysis and heat transfer in a cross-ow heat exchanger with a new winglet-type vortex generator, Energy 35 (2010) 3686e3695. [9] R.V. Rao, V.K. Patel, Thermodynamic optimization of cross ow plate-n heat exchanger using a particle swarm optimization algorithm, Int. J. Therm. Sci. 49 (2010) 1712e1721. [10] P. Saechan, S. Wongwises, Optimal conguration of cross ow plate nned tube condenser based on the second law of thermodynamics, Int. J. Therm. Sci. 47 (2008) 1473e1481. [11] B.F. Pussoli, J.R. Barbosa, L.W. Silva, M. Kaviani, Air-side Heat Transfer and Pressure Drop Characteristics of Peripheral Fin Heat Exchangers, Int. Heat Transfer Conference, Washington-DC, USA, IHTC14e23037 (2010). [12] M. Waltrich, C.J.L. Hermes, J.M. Gonalves, C. Melo, A rst-principles simulation model for the thermo-hydraulic performance of fan supplied tube-n heat exchangers, Appl. Therm. Eng. 30 (2010) 2011e2018. [13] J.M. Gonalves, C. Melo, C.J.L. Hermes, A semi-empirical model for steady-state simulation of household refrigerators, Appl. Therm. Eng. 29 (2009) 1622e1630. [14] W.M. Kays, A.L. London, Compact Heat Exchangers. McGraw-Hill, New York, 1984. [15] C.C. Wang, K.Y. Chi, C.J. Chang, Heat transfer and friction characteristics of plain n-and-tube exchangers, Part II: correlation, Int. J. Heat Mass Trans. 43 (2000) 2693e2700. [16] F.W. Dittus, L.M.K. Boelter, Heat transfer in automobile radiators of the tubular type, in: , Publications in Engineering, vol. 2. University of California, Berkeley-CA, 1930, p. 443. [17] S.A. Klein, Alvarado FL, Engineering Equation Solver Users Manual, F-Chart Software, Middleton-WI, USA (2004). [18] J.R. Barbosa, C.J.L. Hermes, C. Melo, CFD analysis of tube-n no-frost evaporators, J. Braz. Soc. Mech. Sci. Eng. XXXII (4) (2010) 445e453.