Hydrothermal and entropy generation specifications of a hybrid ferronanofluid in microchannel heat sink embedded in CPUs

Amin Shahsavar ,Majid Jafari ,Pouyan Talebizadehsardari,Davood Toghraie

1 Department of Mechanical Engineering,Kermanshah University of Technology,Kermanshah,Iran

2 Metamaterials for Mechanical,Biomechanical and Multiphysical Applications Research Group,Ton Duc Thang University,Ho Chi Minh City,Vietnam

3 Faculty of Applied Sciences,Ton Duc Thang University,Ho Chi Minh City,Vietnam

4 Department of Mechanical Engineering,Khomeinishahr Branch,Islamic Azad University,Khomeinishahr,Iran

Keywords:Liquid-cooled heatsink Hydrothermal aspects Irreversibility Hybrid nanofluid Carbon nanotube Fe3O4

ABSTRACT The objective of this numerical work is to evaluate the first law and second law performances of a hybrid nanofluid flowing through a liquid-cooled microchannel heatsink.The water-based hybrid nanofluid includes the Fe3O4 and carbon nanotubes (CNTs) nanoparticles.The heatsink includes a microchannel configuration for the flow field to gain heat from a processor placed on the bottom of the heatsink.The effects of Fe3O4 concentration (),CNT concentration (φCNT) and Reynolds number (Re) on the convective heat transfer coefficient,CPU surface temperature,thermal resistance,pumping power,as well as the rate of entropy generation due to the heat transfer and fluid friction is examined.The results indicated higher values of convective heat transfer coefficient,pumping power,and frictional entropy generation rate for higher values of Re, and φCNT.By increasing Re,and φCNT,the CPU surface temperature and the thermal resistance decrease,and the temperature distribution at the CPU surface became more uniform.To achieve the maximum performance of the studied heatsink,applying the hybrid nanofluid with low and φCNT was suggested,while the minimum entropy generation was achieved with the application of nanofluid with high and φCNT.

1.Introduction

The problem of effective cooling of electronic processors,fuel cells,and solar cells has been studied for a long time by many researchers.Due to a large amount of heat and insufficient heat reduction of air-cooled systems,liquid-cooled heatsinks have attracted significant attention in recent decades especially for electronic devices with a high rate of heat generation [1,2].Therefore,efforts have-been taken into account to modify the characteristics of heatsinks including temperature distribution,pumping power,and rate of heat removal [3].Different techniques have been considered including configuration improvement of the heatsinks[3–6],the use of microchannels ([7,8],the use of membranes [9–11],as well as the use of highly conductive nanoparticles [12–16] to increase the performance of liquid-cooled heatsinks.

Microchannels are used significantly in different applications due to the lower hydraulic diameter and,consequently,a higher area of heat transfer[17,18].The idea of using microchannel heatsink (MHS) was firstly proposed by Tuckerman and Pease [18] for heat removal from the electronic devices.MHSs have been preferred due to their advantageous characteristics including higher surface area for heat transfer,lower coolant requirement,compactness,and high aspect ratio channels [19].Zhang et al.[20] experimentally investigated a finned liquid-cooled MHS.An analytical model was developed to estimate the thermal resistance and the pressure drop for the developing flow in the heatsink.Biswall et al.[21] suggested a model to optimize the liquid-cooled MHSs.They showed that the use of higher conductive materials causes lower required flow rate and pressure drop.Furthermore,a higher number of channels resulted in lower thermal resistance and a lower pressure drop.Ramos-Alvarado et al.[4] investigated eight different configurations of MHS numerically.They compared the systems according to the uniformity of the temperature distribution,the pressure drop,and the pumping power and showed the advantageous effects of novel liquid-cooled heatsinks in comparison with the conventional ones.Vinodhan and Rajan[7]examined four novels MHS configurations with separate inlet and outlet locations.The examined MHSs showed higher heat transfer rates and lower temperature gradients than the conventional systems due to better recirculation of the fluid in the microchannels.Li et al.[22] performed an optimization analysis of a water-cooled heatsink equipped with dimple and pin–fin.They found a reduction in the hot spot regions and also temperature gradient aiding with the optimization technique.

Heat transfer enhancement from the heatsink has been always seeking to have a lower surface temperature.Low thermal conductivity is the main limitation of conventional heat transfer fluids(HTF) which reduces the performance of the system [23–25].Recently,nanofluids have been used widely as a promising method to modify heat transfer in different energy systems[26–38].There have been several studies in the literature on the application of nanofluid in liquid-cooled heatsinks due to the advantageous effect of nanoparticles in heat transfer phenomena [39–41].However,very few of the surveys from the second law of thermodynamic and exergy generation[42].Sohel et al.[43]experimentally studied the effect of water-Al2O3nanofluid as an HTF in a mini-channel heatsink for an electronic device.They presented that employing nanofluid reduces the heatsink temperature and the thermalentropy generation compared with the pure water.They showed an insignificant intensification in the pressure drop and the frictional entropy generation rate.Bahiraei and Heshmatian [42]investigated a liquid-cooled heatsink for electronic processors using a kind of biological nanofluid according to the first and second laws of thermodynamics.They showed that more uniform temperature distribution,higher heat transfer coefficient,lower surface temperature,and lower irreversibility are achieved for higher φ and Re.Thansekhar and Anbumeenakshi[44]experimentally studied the effect of water-Al2O3nanofluid on the performance of a MHS.They examined various φ and presented the significant enchantment in heat transfer performance of the system compared with the pure water.They showed better heat transfer performance at higher φ which depends also on Re.

Recently,hybrid nanofluids have been utilized as promising novel nanofluids to enhance the heat transfer performance which is produced by mixing different kinds of nanoparticles to capture better characteristics from the nanofluids [45–57].There have been very few studies in the literature on the applications of hybrid nanofluids in liquid-cooled heatsinks especially regarding the second law of thermodynamics.Kumar and Sarkar [58] performed a numerical simulation with experimental validation on the forced convection flow of water-Al2O3and water-CNT/Al2O3nanofluids.They revealed that the maximum heat transfer coefficient is achieved for the hybrid nanofluid with almost no increase in the pressure drop.Arabpour et al.[59]studied the effect of novel nanofluid named kerosene/multi-walled carbon nanotubes (MWCNTs)in two layers MHS regarding the first-law of thermodynamics.They examined a sinusoidal oscillating heat flux with a slip boundary condition on the heat transfer performance of hybrid nanofluid.

The CNT/Fe3O4hybrid nanofluid benefits from both the significant thermal conductivity of CNT and the high magnetic characteristics of Fe3O4[60–62].The application of this hybrid nanofluid is newly and rarely presented in the literature [60–62].Shahsavar et al.[63] conducted an experimental investigation to examine the cooling performance of water-CNT/Fe3O4hybrid nanofluid inside a heated pipe in both the absence and presence of an external magnetic field.The outcomes revealed that the Nusselt number of this hybrid nanofluid in either absence or presence of the magnetic field is better than that of the pure water and water-Fe3O4nanofluid.

To the best of our knowledge,the effects of hybrid CNT/Fe3O4nanofluid on the second-law performance of liquid-cooled heatsinks have not been evaluated so far.The objective of the present study is to assess the hydrothermal and entropy generation aspects of a liquid-cooled heatsink containing the water-CNT/Fe3O4hybrid nanofluid.The studied heatsink includes symmetric bifurcation flow distributors with parallel flow channels and therefore,has improved performance compared with conventional heatsinks[4].

2.Properties of Nanofluid

The water-based hybrid nanofluid containing tetramethylammonium hydroxide (TMAH) coated Fe3O4nanoparticles and gum arabic (GA) coated CNTs were synthesized by mixing the required amount of water-Fe3O4and water-CNT nanofluids,followed by mixture sonication for 5 min.The details of the preparation and characterization of this hybrid nanofluid can be found in the author’s previous work [63–68].Fe3O4and CNT nanoparticles are stuck together due to the interaction between the TMAH and GA molecules.

After careful preparation and characterization,a series of experiments was carried out to obtain the thermophysical properties of the hybrid nanofluids containing different φFeand CNT nanoparticles.The concentrations of employed nanoparticles and the properties of prepared nanofluid samples are presented in Table 1.A liquid density gravity meter(DA-130N,KEN,Japan)and a differential scanning calorimeter (DSC,Q20,TA Instruments,USA) were used to measure the nanofluid density (ρ) and specific heat (cp).Also,the viscosity(?)and the thermal conductivity(k)of the nanofluid samples were measured using a Paar Physica MCR 300 parallel disk rheometer (with an accuracy of 0.5%) and a KD2-pro instrument (with an accuracy of 5%).

Table 1 The density,specific heat,viscosity,and thermal conductivity of the studied nanofluid at different concentrations of CNT and Fe3O4 nanoparticles [64–65,67]

3.System Description

3.1.Geometry,dimensions,and boundary conditions

The considered heatsink in this work includes parallel symmetric bifurcation flow distributors displayed in Fig.1.The dimensions of the considered flow channels in the heatsink are reported in Table 2.In the inlet,the flow is distributed by multiple levels of flow channel bifurcations and in the outlet,the flow is distributed by multiple levels of mirrored flow channel convergences.Due to the repetition of the bifurcation structure,the inlet flow is distributed to twenty small channels.After 4 levels of bifurcation,when the velocity and flow rate are relatively low,it is expected that the inlet fluid will be divided equally between the different channels.

Fig.1.The studied flow field configuration for a heatsink.

Table 2 Dimensions of the channels for flow field configuration

Fig.2(a) demonstrates the heatsink placed on the CPU made from aluminum.All surfaces are adiabatic expect the bottom surface.The heat flux of q′′=80000 W﹒m-2is applied from the bottom surface (backside surface shown in Fig.2).An insulated cap is placed at the heatsink top surface[see Fig.2(b)]where the inlet and outlet tubes are also placed.The thickness of the heatsink bottom surface and cover plate are 2.2 mm and 1.0 mm,respectively.The inlet temperature of the nanofluid has been considered as 293 K.

3.2.Governing equations

The conservation equations of mass,momentum,and energy for a Newtonian incompressible nanofluid are given as [69]:

(i) Conservation of mass,

(ii) Conservation of momentum in the x-direction,

(iii) Conservation of momentum in the y-direction,

(iv) Conservation of momentum in the z-direction,

(v) Conservation of energy,

Fig.2.(a) Image and (b) dimensions of the base plate (in centimeters) that the flow channels are machined/cast.

where vx,vyand vzare the x,y,and z components of velocity,respectively,p is the pressure,and T is the temperature.

Re is defined based on the inlet velocity(vin)and inlet diameter(D) as:

The difference between the maximum and minimum CPU surface temperatures over the applied heat flux is employed as the temperature uniformity criterion [1]:

where TCPU,maxand TCPU,minare maximum and minimum CPU surface temperatures,respectively.A lower θ shows a more uniform temperature distribution.

The uniformity of CPU surface temperature is also assessed by thermal resistance definition in addition to θ as [1]:

where TCPU,meanand Tinare the mean CPU surface and fluid inlet temperatures,respectively,and R is the thermal resistance.A lower R shows a better heat transfer.

Fig.3.h as a function of Re with the effect of(a) at φCNT=0.1%,and(b)φCNT at =0.9%.

The average convective heat transfer coefficient is determined as [1]:

The overall performance of a nanofluid is assessed by the Performance Evaluation Criterion(PEC)showing the relative improvement in h to Δp as [1]:

The subscripts nf and w respectively refer to the nanofluid and water.

3.3.Entropy generation rates

The entropy generation includes the thermal()and frictional() terms which are defined locally as [1]:

Fig.4.Average surface temperature of the CPU as a function of Re with the effect of(a) at φCNT=0.1%,and (b)φCNT at =0.9%.

4.Numerical Procedure

ANSYS FLUENT software is utilized to solve the governing equations by employing the SIMPLE algorithm for the pressure and velocity coupling using the second-order upwind.The convergence criterion is set as 10-6.

For the grid independency analysis,Table 3 schematizes the difference between the inlet and outlet temperatures as well as the pressure drop for the heatsink containing 0.9%FF+1.35%CNT nanofluids at Re=1500.As presented,the difference between the results of the simulations performed with a grid with 267,455 nodes and finer grids is negligible and therefore,267,455 is selected as the element number.Note that a smaller size of elements is considered near the walls.

For the validation,the thermal resistance and pressure drop computed in the current contribution are compared with the results of Ramos-Alvarado et al.[4] for the flow of pure water through the heatsink at different flow rates.As presented in Table 4,good agreement can be seen between the results.

Fig.5.Heatsink temperature (K) contours at y=1.6 mm for 0.1%FF+0.1%CNT,0.9%FF+0.1%CNT and 0.9%FF+1.35%CNT.

Table 3 Details of mesh study (q′′=80000 W﹒m-2,Re=1500).

Table 4 Details of validation

5.Results and Discussion

In the current investigation,analysis is performed to examine the impacts ofand Re on the hydrothermal and entropy generation characteristics of the water-CNT/Fe3O4hybrid nanofluid flowing through an MHS.The numerical simulations are performed atrange of 0.1%–0.9%,φCNTrange of 0.1%–1.35%,and Re range of 500–1500.The inlet temperature of the nanofluid is considered as 293 K.

Fig.3 displays h in terms ofand φCNTat various values of Re.At a constantand φCNT,h augments with increasing Re.To justify these results,the following points should be taken into account:

Fig.6.Nanofluid temperature (K) contours at y=1.6 mm for 0.1%FF+0.1%CNT,0.9%FF+0.1%CNT and 0.9%FF+1.35%.

· The velocity boundary layer thickness is only a function of Re and declines by boosting Re.

· The Prandtl number (Pr=?Cp/k) is the ratio of the velocity boundary layer thickness to the thermal boundary layer thickness.

· h is directly related to the thermal conductivity and the thermal boundary layer thickness.

Figs.5 and 6 illustrate the temperature contours of heatsink and nanofluid at the cross-section through the center of the flow channel (y=1.6 mm),and Fig.7 displays the contour plot of the CPU surface temperature for the hybrid nanofluids of 0.1%FF+0.1%CNT,0.9%FF+0.1%CNT and 0.9%FF+1.35%CNT.The temperature of the heatsink,nanofluid,and CPU surface is lower near the entrance and gradually increases toward the output.In addition,increasingand φCNTas well as Re leads to the decline of the maximum temperature of the heatsink,nanofluid,and CPU surface.Moreover,Fig.7 reveals that a more uniform temperature distribution at the CPU surface happens for higher Re,and φCNT.The reduction in the maximum temperature of the CPU reduces the possibility of hot spot formation.

Fig.7.CPU surface temperature (K) contours for 0.1%FF+0.1%CNT,0.9%FF+0.1%CNT and 0.9%FF+1.35%CNT at different Re.

For proper operation and longer lifetime of electronic devices,their temperature must not exceed a certain limit.The effects ofand φCNTon TCPU,maxare depicted in Fig.8 at different Re.TCPU,maxreduces by increasing,φCNTand Re.Therefore,the possibility of hotspot formation is decreased due to a higher rate of heat transfer by increasing,φCNTand Re,which leads to a lower CPU surface temperature.

Fig.8.Maximum temperature of the CPU surface as a function of Re with the effect of (a) at φCNT=0.1%,and (b)φCNT at =0.9%.

R and θ are two main parameters in assessing the performance of heatsinks.Figs.10 and 11 respectively display the influences ofand φCNTon R and θ at different Re.The results reveal that both the parameters of R and θ decrease with increasing Re,and φCNT.These observations are consistent with the findings of Babar and Ali [70] and Balaji et al.[71].According to Eq.(7),due to the constant value of the heat flux and the inlet temperature of the nanofluid and the reduction of TCPU,maxat higher,φCNTand Re,the reduction of the parameter R is expected.The reduction of the parameter θ also indicates that the decrease in the TCPU,maxis higher than that of the TCPU,min,and the difference between them reduces with increasing,φCNTand Re.

According to the above discussion,it can be concluded that increasingand φCNT,on one hand,leads to a higher h,which is desirable,but on the other hand,increases,which is undesirable.Therefore,to better study the hydrothermal performance of nanofluid in the heatsink,the parameter‘‘PEC”should be investigated.Table 5 presents the impacts ofand φCNTon PECat differentRe.As is seen,only three nanofluid samples (i.e.,0.1%FF+0.1%CNT,0.3%FF+0.1%CNT,and 0.1%FF+0.7%CNT)have a better overall hydrothermal performance than the pure water and their PEC is larger than unity.According to the results,the highest PEC is 1.390,which belongs to 0.1%FF+0.1%CNT nanofluid.Furthermore,the results demonstrate that PEC reduces with increasingand φCNT,while the effect of Re on PEC is different for various nanofluids.For example,PEC of 0.1%FF+0.1%CNT reduces by increasingRe,while it is vice versa for 0.7FF%+0.1%CNT.

Table 5 PEC of the studied hybrid nanofluid samples at different Re

Fig.9.pump as a function of Re with the effect of(a)at φCNT=0.1%,and(b)φCNT at =0.9%.

Fig.10.R as a function of Re with the effect of (a) at φCNT=0.1%,and (b)φCNT at =0.9%.

Fig.12 presents the global thermal entropy generation rate in the entire heatsink(walls of heatsink+fluid)for the studied nanofluid samples as well as the pure water at different Re.Increasingand φCNT,at a constant Re,leads to a reduction in the global thermal entropy generation rate.By increasingand φCNT,at a constant Re,the thermal conductivity of the nanofluid samples increases,and the average temperature of the nanofluid decreases,both of which contribute to a higher global thermal entropy generation rate.On the other hand,increasing the thermal conductivity leads to a lower fluid temperature gradient and consequently,reduces the thermal entropy generation rate.Fig.12 reveals that the effect ofand φCNTincrement on the temperature gradient overcomes the effect of thermal conductivity increment and average temperature reduction.Hence,the global thermal entropy generation rate in the nanofluid reduces by increasingand φCNT.Additionally,cooling enhances with the intensification ofand φCNTand therefore,the temperature gradient in the solid decreases,which reduces the rate of thermal entropy generation in the walls of the heatsink.Moreover,it can be seen in Fig.12 that the global thermal entropy generation rate of the considered nanofluid samples is higher than that of pure water which is a desirable finding.Furthermore,it is clear that increasing Re,at constantand φCNT,leads to a lower global thermal entropy generation rate.

Fig.13 displays the global frictional entropy generation rate in the nanofluid at different Re.The results show that the global frictional entropy generation rate augments with increasing,φCNTand Re.Increasingand φCNT,at a constant Re,leads to a higher nanofluid viscosity and a lower average temperature of the nanofluid.Both of these effects result in a higher rate of global frictional entropy generation in the studied hybrid nanofluid samples.Also,Fig.13 depicts that the global frictional entropy generation rate of the pure water is higher than that of the nanofluid.It is noteworthy that since the values of the thermal entropy generation rate at different Re,and φCNTare much larger compared with the frictional entropy generation rate,the variation of the global total entropy generation rate is generally similar to the global thermal entropy generation rate.

Fig.11.θ as a function of Re with the effect of (a) at φCNT=0.1%,and (b)φCNT at =0.9%.

Fig.12.Global thermal entropy generation rate as a function of Re with the effect of(a) at φCNT=0.1%,and (b)φCNT at =0.9%.

The data presented in this section indicate that,from the first law point of view,the merit of using the proposed water-CNT/Fe3O4hybrid nanofluid in the studied heatsink is greater at lower,φCNTand Re.On the other hand,from the second law point of view,the results reveal that the benefits of using such a hybrid nanofluid are greater at higher,φCNTand Re.Hence,the decision about suitableand φCNTis made by the designer based on the relative importance of the hydrothermal and entropy generation characteristics.

6.Conclusions

The hydrothermal and entropy generation characteristics of the water-CNT/Fe3O4hybrid nanofluid in a liquid-cooled MHS were examined.The heatsink includes a microchannel configuration with parallel symmetric bifurcation flow distributors for CPU cooling.Numerical simulations were performed to assess the influences of,φCNTand Re on the convective heat transfer coefficient,CPU surface temperature,thermal resistance,pumping power,as well as the thermal,frictional,and total entropy generation rates.The outcomes revealed that the intensification of Re,and φCNTcauses an increase in h,and frictional entropy generation rate,while the opposite is true about the CPU surface temperature,thermal resistance,temperature uniformity of CPU surface,and thermal entropy generation rate.Moreover,it was found that the hydrodynamic performance of water-CNT/Fe3O4hybrid nanofluid inside the considered heatsink at low concentrations is better than high concentrations while the entropy generation is lower at high nanofluid concentrations.

Fig.13.Global frictional-entropy generation rate as a function of Re with the effect of (a) at φCNT=0.1%,and (b)φCNT at =0.9%.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.