Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 57, 1, pp. 221-233, Warsaw 2019 DOI: 10.15632/jtam-pl.57.1.221 NATURAL AND MIXED CONVECTION OF A NANOFLUID IN POROUS CAVITIES: CRITICAL ANALYSIS USING BUONGIORNO’S MODEL Iman Zahmatkesh, Mohammad Reza Habibi Department of Mechanical Engineering, Mashhad Branch, Islamic Azad University, Mashhad, Iran e-mail: zahmatkesh5310@mshdiau.ac.ir In this paper, Buongiorno’s mathematical model is adopted to simulate both natural con- vection and mixed convection of a nanofluid in square porous cavities. The model takes into account the Brownian diffusion and thermophoresis effects. Both constant and varia- ble temperatures are prescribed at the side walls while the remaining walls are maintained adiabatic. Moreover, all boundaries are assumed to be impermeable to the base fluid and the nanoparticles. The governing equations are transformed to a form of dimensionless equ- ations and then solved numerically using the finite-volume method. Thereafter, effects of the Brownian diffusion parameter, the thermophoresis number, and the buoyancy ratio on the flow strength and the averageNusselt number as well as distributions of isocontours of the stream function, temperature, and nanoparticles fraction are presented and discussed. Keywords: natural convection, mixed convection, nanofluid, porous media, Buongiorno’s model 1. Introduction Heat transfer is a significant and widely explored engineering problem that, due to the lack of energy resources, has become truly important. One efficient way to improve heat transfer and to reduce energy consumption goes back to the use of porousmedia. This occurs since a porous medium provides a large surface area for heat exchange. On the other hand, the flow field in a porousmedium is completely three-dimensional and irregular, which intensifies fluidmixing. Further attempt to achieve higher heat transfer rates has led to adding nanoparticles to working liquids and producing nanofluids. The added nanoparticles are usually made up of metals ormetal oxideswithhigh thermal conductivity. So, the resultingfluidhasabetter thermal efficiency than the base liquid.Going into the literature, onemayfind that heat transfer analysis of nanofluid flows has been a hot topic among scientists over the past decade. Examples include the studies of Ali et al. (2014), Ghasemi et al. (2016), and Rostamzadeh et al. (2016). Different mathematical models have been adopted to describe heat transfer in nanofluids. The simplest method with the least computational burden is the homogenous model. In this model, the concentration of nanoparticles is taken constant over the entire flow field. It is also assumed that the base liquid and the nanoparticles are in local equilibrium andmove with the same velocity and temperature. In spite of previous achievements of this model, some studies have proved that more complex models provide better agreement with experimental data (e.g., Behroyan et al. (2016); Torshizi and Zahmatkesh, 2016). Buongiorno (2006) introduced seven transportmechanisms which cause relative velocity be- tween thenanoparticles and thebase liquid, namely, inertia,Browniandiffusion, thermophoresis, diffusiophoresis, Magnus effect, fluid drainage, and gravitational settling. After comparing the diffusion time scales of these mechanisms he drew a conclusion that, in the absence of flow tur- bulence, the Brownian diffusion and thermophoresis are the most important effects. Based on this finding, he then developed a non-homogeneous but an equilibriummodel for nanofluid flow 222 I. Zahmatkesh,M.R. Habibi and heat transfer that incorporates the effects of the Brownian diffusion and thermophoresis. The Brownian diffusion occurs due to random variations in the bombardment of the base fluid molecules against the particles. The thermophoresis phenomenon appears as a net force acting in the opposite direction to the gradient of temperature and is a direct result of the differential bombardment of the base fluidmolecules in the vicinity of the particles (Zahmatkesh, 2008b). There are many recent papers that deal with Buongiorno’s mathematical model (e.g., She- ikholeslami et al., 2016; Kefayati, 2017a; Mustafa, 2017) but to the best of our knowledge, there are nopreviousworks in literaturewhich compare the role of parameters appearing in thismodel for both natural convection andmixed convection in porousmedia. Some simulation studies for natural convection of nanofluids in porous cavities based on Buongiorno’s model are discussed below. Sheremet et al. (2014) simulated natural convection of nanofluids in shallow and slender porous cavities. Their results demonstrated that an inverse relation existed between the ave- rage Nusselt number and the buoyancy ratio. Conjugate natural convection of nanofluid in a square porous cavity was discussed by Sheremet and Pop (2014a). They found that the Nus- slet number was an increasing function of the buoyancy ratio and a decreasing function of the thermophoresis number and the Lewis number. A simulation study of natural convection of a nanofluid in a right-angle triangular porous cavity was reported by Sheremet and Pop (2015a). That investigation showed that the average Nusselt number increased with the enhancement of the Lewis number but any rise in the Brownian diffusion parameter, the buoyancy ratio or the thermophoresis number made it lower. Sheremet et al. (2015) discussed natural convection heat transfer of a nanofluid in a three-dimensional porous cavity. Their results led to the conc- lusion that the average Nusselt number increased with the Brownian diffusion parameter and decreased with the buoyancy ratio and the thermophoresis number. Sheremet and Pop (2015b) analyzed natural convection of a nanofluid in a porous annulus. They indicated that an incre- ase in the thermophoresis number and the buoyancy ratio led to deterioration in the average Nusselt numberwhile theBrownian diffusion parameter contributed neutrally. Ghalambaz et al. (2016) investigated the influence of viscous dissipation and radiation on natural convection of a nanofluid in a porous cavity and concluded that an increase in the Lewis number improved the heat transfer but augmentation of the buoyancy ratio and the thermophoresis number decreased it. More recently, natural convection and entropy generation of a non-Newtonian nanofluid in a porous cavity was pointed out by Kefayati (2017b). The results demonstrated that rise of the Lewis number, the thermophoresis number, and the Brownian diffusion parameter declined the average Nusselt number, but the augmentation of the buoyancy ratio enhanced it. The role of parameters appearing inBuongiorno’smathematical modelmay depend on ther- mal boundary conditions of the cavity. In this context, Sheremet and Pop (2014b) discussed how imposition of a sinusoidal temperature distribution on the side walls may affect natural convection of the nanofluid in a square porous cavity. They found that the average Nusselt number was an increasing function of the buoyancy ratio and the thermophoresis number but a decreasing function of the Lewis number and the Brownian diffusion parameter. More recently, they extended their work to a wavy porous cavity (Sheremet et al., 2017). They found that the dependence of the average Nusselt number to the pertinent parameterswas similar to the square cavity, which was in contrast to the aforesaid findings in uniformly heated/cooled cavities. The current research deals with heat transfer of a nanofluid in square porous cavities. The main purpose of this article is to analyze the effects of the buoyancy ratio, the Brownian dif- fusion parameter and the thermophoresis number on the flow strength and the average Nusselt number as well as developments of streamlines, isotherms and isoconcentrations. To provide a critical analysis, computations are undertaken for various cases in natural convection andmixed convection environments with both uniform and non-uniformwall temperatures. Natural and mixed convection of a nanofluid in porous cavities... 223 2. Problem definition and mathematical formulation Both natural and mixed convection heat transfer in a square cavity filled with a nanofluid- -saturated porousmediumare analyzed in this study.A schematic diagram of the flowproblems is shown in Fig. 1, where x and y are the Cartesian coordinates and L is the size of the cavity. Here, allwalls are assumed tobe impermeable tomass transfer.Thehorizontalwalls are assumed adiabatic while two different conditions are imposed on the side walls. In the first case, the sidewalls are considered to be heated/cooled uniformly while in the second case, the sidewalls are influenced by the existence of a sinusoidal temperature variation. In this paper, Buongiorno’smathematical model is used. Thanks to this approach, the nano- fluid is considered as a two-component dilutemixture. The porousmedium is assumed isotropic andhomogenouswhile the established flow is concerned to be steady, incompressible,Newtonian and laminar. TheDarcymodel is employed for themomentum equation.Moreover, a local ther- mal equilibrium is assumed between the nanoparticles, the base fluid and the porous medium. The Boussinesq approximation is adopted to determine the variations of density in the body force termwithin themomentumequation.Meanwhile, viscous dissipation, theworkdoneby the pressure change and radiation heat transfer are neglected. Additionally, the thermophoresis and Brownian transport coefficients are assumed temperature-independent. On these assumptions, the conservation equations for mass, momentum, energy and flow concentration are (Nield and Bejan, 2013) ∇· (V)= 0 0=−∇P − µ K V+[Cρp+(1−C)ρf(1−β(T −Tc))]g V ·∇T = αm∇ 2T + δ ( DB∇C ·∇T + DT Tc ∇T ·∇T ) ρp ε V ·∇C =−∇· jp (2.1) Here, V is the Darcy velocity vector, P is pressure, T is temperature, C is the nanoparticles fraction, g is gravitational acceleration (g = −gj), ρ is density, µ is dynamic viscosity, β is the volumetric expansion coefficient, α is thermal diffusivity, ε is medium porosity, and K is permeability of the porousmedium.Meanwhile, DB is the Brownian diffusion coefficient, DT is the thermophoretic diffusion coefficient, and δ is a parameter definedby δ = ε(ρcp)p/(ρcp)f with cp being the specific heat. Moreover, the subscripts p, f, and m correspond to the nanparticles, the base fluid and effective values, respectively. InEq. (2.1)4, jp is thenanoparticlesmass flux.Based onBuongiorno’smodel, thenanopartic- les mass flux is made up of two parts, namely, Brownian diffusion jp,B, and thermophoresis jp,T . Thus jp = jp,B + jp,T =−ρpDB∇C −ρpDT ∇T T (2.2) After using the Boussinesq approximation and taking the nanofluid as a dilutemixture, one arrives at the following form of the momentum equation (Nield and Kuznetsov, 2009) 0=−∇P − µ K V+[C(ρp−ρf0)+ρf0(1−β(T −Tc)(1−C0))]g (2.3) with subscript 0 standing for reference values. To simplify this vector equation, cross-differentiation is adopted, which eliminates the pres- sure term. So, the governing equations become ∂u ∂x + ∂v ∂y =0 (2.4) 224 I. Zahmatkesh,M.R. Habibi and 0=− µ K (∂u ∂y − ∂v ∂x ) +g(ρp−ρf0) ∂C ∂x − (1−C0)ρf0βg ∂T ∂x u ∂T ∂x +v ∂T ∂y = αm (∂2T ∂x2 + ∂2T ∂y2 ) + δ { DB (∂C ∂x ∂T ∂x + ∂C ∂y ∂T ∂y ) + DT Tc [(∂T ∂x )2 + (∂T ∂y )2]} 1 ε ( u ∂C ∂x +v ∂C ∂y ) = DB (∂2C ∂x2 + ∂2C ∂y2 ) + DT Tc (∂2T ∂x2 + ∂2T ∂y2 ) (2.5) where u and v denote the velocity components in the x and y directions, respectively. After introducing the stream function by u = ∂ψ/∂y and v = −∂ψ/∂x, the continuity equation will be satisfied. Moreover, Eqs. (2.5) lead to ∂2ψ ∂x2 + ∂2ψ ∂y2 =− (1−C0)ρf0gKβ µ ∂T ∂x + ρp−ρf0 µ gK ∂C ∂x ∂ψ ∂y ∂T ∂x − ∂ψ ∂x ∂T ∂y = αm (∂2T ∂x2 + ∂2T ∂y2 ) + δ { DB (∂C ∂x ∂T ∂x + ∂C ∂y ∂T ∂y ) + DT Tc [(∂T ∂x )2 + (∂T ∂y )2]} 1 ε (∂ψ ∂y ∂C ∂x − ∂ψ ∂x ∂C ∂y ) = DB (∂2C ∂x2 + ∂2C ∂y2 ) + DT Tc (∂2T ∂x2 + ∂2T ∂y2 ) (2.6) We now define the following parameters to make the above equations dimensionless X = x L Y = y L Ψ = ψ αm θ = T −Tc Th−Tc Φ = C C0 Ra= (1−C0)gKρf0β(Th−Tc)L αmµ Pe= V0L αm Le= αm εDB Nb = δDBC0 αm Nr = (ρp−ρf0)C0 ρf0β(Th−Tc)(1−C0) Nt = δDT(Th−Tc) αmTc (2.7) Here, Ra is the Rayleigh number, Pe is the Peclet number (with V0 being the inlet velocity), Le is the Lewis number, Nb is the Brownian diffusion parameter, Nr is the buoyancy ratio, and Nt is the thermophoresis number. Substituting the dimensionless parameters into the governing equations yields ∂2Ψ ∂X2 + ∂2Ψ ∂Y 2 =−Ra ( ∂θ ∂X −Nr ∂Φ ∂X ) ∂Ψ ∂Y ∂θ ∂X − ∂Ψ ∂X ∂θ ∂Y = ∂2θ ∂X2 + ∂2θ ∂Y 2 +Nb (∂Φ ∂X ∂θ ∂X + ∂Φ ∂Y ∂θ ∂Y ) +Nt [( ∂θ ∂X )2 +( ∂θ ∂Y )2] Le (∂Ψ ∂Y ∂Φ ∂X − ∂Ψ ∂X ∂Φ ∂Y ) = ∂2Φ ∂X2 + ∂2Φ ∂Y 2 + Nt Nb ( ∂2θ ∂X2 + ∂2θ ∂Y 2 ) (2.8) Notice that the governing equations reduce to those of a regular fluid if one chooses Nb = Nr = Nt =0. Natural and mixed convection of a nanofluid in porous cavities... 225 The boundary conditions for the flow problems are: Case I:Natural convection with a constant temperature at the side walls Left wall: Ψ =0 θ =1 jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Right wall: Ψ =0 θ =0 jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Horizontal walls: Ψ =0 ∂θ ∂Y =0 ∂Φ ∂Y =0 (2.9) Case II:Natural convection with a sinusoidal temperature distribution at the side walls Left wall: Ψ =0 θ =sin(2πY ) jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Right wall: Ψ =0 θ =sin(2πY ) jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Horizontal walls: Ψ =0 ∂θ ∂Y =0 ∂Φ ∂Y =0 (2.10) Case III:Mixed convection with a constant temperature at the side walls Left wall: Ψ =0.1Pe θ =1 jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Right wall: Ψ =0 θ =0 jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Inlet: Ψ = YPe θ =0 Φ =1 Bottom wall: Ψ =0 ∂θ ∂Y =0 ∂Φ ∂Y =0 Upper wall: Ψ =0.1Pe ∂θ ∂Y =0 ∂Φ ∂Y =0 Outlet: ∂Ψ ∂X =0 ∂θ ∂X =0 ∂Φ ∂X =0 (2.11) Case IV:Mixed convection with a sinusoidal temperature distribution at the side walls Left wall: Ψ =0.1Pe θ =sin(2πY ) jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Right wall: Ψ =0 θ =sin(2πY ) jp =0 ( or,Nb ∂Φ ∂X +Nt ∂θ ∂X =0 ) Inlet: Ψ = YPe θ =0 Φ =1 Bottom wall: Ψ =0 ∂θ ∂Y =0 ∂Φ ∂Y =0 Upper wall: Ψ =0.1Pe ∂θ ∂Y =0 ∂Φ ∂Y =0 Outlet: ∂Ψ ∂X =0 ∂θ ∂X =0 ∂Φ ∂X =0 (2.12) The physical quantities related to the problems are the local and average Nusselt numbers (Nu,Nu) and the local and average Sherwood numbers (Sh,Sh) (Sheremet and Pop, 2014b) 226 I. Zahmatkesh,M.R. Habibi Nu=− ∂θ ∂X ∣ ∣ ∣ ∣ ∣ X=0 Nu= 1 0.9 1 ∫ 0.1 Nu dY Sh=− ∂Φ ∂X ∣ ∣ ∣ ∣ ∣ X=0 Sh= 1 0.9 1 ∫ 0.1 Sh dY (2.13) In the flow problems, we limit our attention to the Nusselt number since, at the side walls, we have Sh=−(Nt/Nb)Nu. 3. Solution method The governing equations constitute a system of nonlinear partial differential equations. In order todiscretize them, thefinite-volumeapproach is adopted.By integrating thegoverning equations over each control volume, a systemof algebraic equations is produced,which is solved by theTri- -DiagonalMatrixAlgorithm(TDMA).Appropriate relaxation is chosenon thebasis of numerical experiments. The iteration is terminated when changes between two consecutive iterations get smaller than 10−5. The solution method has been implemented in FORTRAN software. For the purpose of acquiring an acceptable grid for each current case, four different grid independence tests have been carried out. The results indicated that the suitable grid systems are 200×200 (Case I), 400×400 (Case II and Case III), and 300×300 (Case IV). The employed Fortran code is essentially a modified version of a code built and validated in the previous works (Zahmatkesh, 2008a, 2015; Zahmatkesh and Naghedifar, 2017). In order to evaluate the accuracy of this code for simulation of nanofluid-saturated porous cavities with Buongiorno’smodel, the corresponding results have been comparedwith those of Sheremet et al. (2014) in Table 1. Here, the average Nusselt numbers in a square porous cavity with isothermal vertical walls and adiabatic horizontal walls saturated with the nanofluid are presented. The compared results belong to Ra = 100, Le = 1, 10, 100, Nr = 0.1, 0.4, and Nb = Nt = 0.4. Notice that there is a trustworthy similarity with that study. This assured us that our results are reliable. So, we have applied the code to analyze the flow problems depicted in Fig. 1. Table 1.Comparison of the present results with those of previous works at Ra=100 Le Nr Nu Sheremet et al. (2014) Current study 1 0.1 3.8387 3.8108 0.4 2.7791 2.7617 10 0.1 4.6270 4.5575 0.4 4.0088 3.9637 100 0.1 4.6252 4.4401 0.4 4.3049 4.1542 4. Simulation results In this Section, simulation results for bothnatural convection andmixed convection heat transfer of the nanofluid are presented. The results are discussed for the following values of the pertinent parameters: the Rayleigh number (Ra=30, 100, 300), the Peclet number (Pe=25), the Lewis number (Le= 25), the buoyancy ratio (Nr =0.05, 0.1, 0.5), the Brownian diffusion parameter (Nb =0.05, 0.1, 0.5) and the thermophoresis number (Nt =0.05, 0.1, 0.5). Natural and mixed convection of a nanofluid in porous cavities... 227 Fig. 1. Physical models of the flow problem: (a) Case I: Natural convection with a constant temperature at the side walls; (b) Case II: Natural convection with a sinusoidal temperature distribution at the side walls; (c) Case III: Mixed convection with a constant temperature at the side walls; (d) Case IV: Mixed convection with a sinusoidal temperature distribution at the side walls Tables 2, 3, and 4 illustrate the numerical values of |Ψmax| andNu for the four configurations at Ra= 30, 100, 300, respectively. Here, |Ψmax| provides a measure of the convection vigor. In a general way, the imposition of the sinusoidal temperature distribution on the sidewalls leads to heat transfer enhancement both in the natural and mixed convection environments within the current range of the Rayleigh number. This imposition also intensifies the flow strength in themixed convection case. In the natural convection problem, however, depending on the value of Ra, the sinusoidal wall temperature may enhance or deteriorate the flow strength. Table 2.Numerical values of |Ψmax| and Nu for the flow at Ra=30 Ra=30 Nb Nr Nt Nb = Nr = Nt 0.05 0.5 0.05 0.5 0.05 0.5 0.1 Case I |Ψmax| 1.909 1.918 1.916 1.894 1.917 1.889 1.914 Nu 1.421 1.424 1.423 1.417 1.455 1.186 1.422 Case II |Ψmax| 0.750 0.750 0.750 0.750 0.736 0.878 0.750 Nu 3.577 3.576 3.576 3.576 3.575 3.722 3.576 Case III |Ψmax| 2.499 2.499 2.499 2.499 2.499 2.499 2.499 Nu 3.570 3.542 3.541 3.610 3.584 3.381 3.549 Case IV |Ψmax| 3.370 3.362 3.364 3.377 3.351 3.462 3.366 Nu 4.613 4.610 4.611 4.616 4.590 4.914 4.611 Inspectionof thenumerical values of |Ψmax| in the conduction–dominated regime (i.e.,Table 2 with Ra = 30) indicates that in Case I, increasing Nb leads to an insignificant growth in the flow strength (maximum0.47%) but rising Nr or Nt causes a slight drop in it (maximum1.15% and 1.46%, respectively). The results of Case II show that an increment in the thermophoresis number from 0.05 to 0.5 intensifies the flow strength to about 19.29%. Nb and Nr, however, contributes neutrally there. The results of Case III indicate that the value of |Ψmax| is not dependent in this case to Nb, Nr, and Nt. Meanwhile, notice that all current parameters are influential to the flow strength in Case IV. This is similar to Case I but the trends of the 228 I. Zahmatkesh,M.R. Habibi Table 3.Numerical values of |Ψmax| and Nu for the flow at Ra=100 Ra=100 Nb Nr Nt Nb = Nr = Nt 0.05 0.5 0.05 0.5 0.05 0.5 0.1 Case I |Ψmax| 4.707 4.725 4.722 4.676 4.722 4.668 4.717 Nu 2.932 2.936 2.935 2.925 2.999 2.450 2.934 Case II |Ψmax| 2.618 2.622 2.622 2.608 2.579 2.942 2.621 Nu 4.265 4.252 4.254 4.254 4.247 4.536 4.258 Case III |Ψmax| 3.264 3.240 3.245 3.288 3.237 3.342 3.251 Nu 4.975 4.932 4.940 5.017 5.003 4.518 4.950 Case IV |Ψmax| 5.435 5.459 5.456 5.398 5.412 5.728 5.449 Nu 5.512 5.500 5.503 5.530 5.468 5.987 5.506 Table 4.Numerical values of |Ψmax| and Nu for the flow at Ra=300 Ra=300 Nb Nr Nt Nb = Nr = Nt 0.05 0.5 0.05 0.5 0.05 0.5 0.1 Case I |Ψmax| 9.813 9.847 9.841 9.749 9.843 9.709 9.831 Nu 6.082 6.083 6.083 6.073 6.214 5.111 6.083 Case II |Ψmax| 6.270 6.286 6.284 6.181 6.222 6.756 6.282 Nu 7.553 7.534 7.535 7.585 7.525 7.933 7.537 Case III |Ψmax| 8.02 7.952 7.958 8.094 7.953 8.094 7.981 Nu 8.181 8.061 8.100 8.262 8.168 7.488 8.128 Case IV |Ψmax| 8.898 8.951 8.918 8.903 8.898 9.470 8.916 Nu 8.800 8.904 8.831 8.981 8.799 9.721 8.830 variations are quite distinct. Evidently,with the increase in Nb, a slight drop in the flow strength appears (maximum 0.24%), but with an elevation in Nr or Nt, insignificant increases occur in it (maximum 0.39% and 3.31%, respectively). Scrutiny of the Nu values in Table 2 demonstrates that Nb and Nr possess a minor impact on the average Nusselt number in all current cases. Notice thatmaximumdeviations of Nu, as a result of the tenfold increase in Nb and Nr, may not exceed 0.78% and 1.95%, respectively. The pattern is completely different when we go to Nt, since this parameter affects the heat transfer rate in all configurations. Specifically, a rise in Nt from 0.05 to 0.5 increases Nu to 4.11% and 7.06% inCase II andCase IVwith sinusoidal wall temperatures, but decreases it to 18.49% and 5.66% in Case I and Case III with constant wall temperatures, respectively. This controversy in the effect of the thermophoresis number on the heat transfer of cavities with uniform wall temperatures and those with non-uniformwall temperatures is in agreement with the previous findings in the natural convection environment, as pointed out previously. The results presented inTable 3belonging toRa=100 indicate that, inCase I, thevariations of |Ψmax| with Nb, Nr and Nt are similar to those of Ra = 30. The corresponding deviations are +0.38%, −0.97%, and −1.14%, respectively. In Case II and III, the consequences of the pertinent parameters on the flow strength are no longer negligible at this Rayleigh number. In Case II, the tenfold increase in Nb, Nt, and Nr leads to 0.15% and 14.07% growths and a 0.53% drop in |Ψmax|, respectively. The deviations are −0.74%, +3.24%, and +1.33% in Case III and +0.44%, +5.84%, and−1.06% in Case IV, respectively. Analysis of the average Nusselt number is also interesting. Similarly to what appeared at Ra=30, it is evident that Nb and Nr are not so influential on Nu prediction at Ra=100.Maximum changes of Nu by increasing Nb and Nr are+0.14% and−0.34% inCase I,−0.30% and 0% inCase II,−0.86% and+4.93% inCase III, and−0.22% and+0.49% inCase IV, respectively. The effect of the theromophoresis number on Natural and mixed convection of a nanofluid in porous cavities... 229 the heat transfer rate is more remarkable. Specifically, a rise in Nt from 0.05 to 0.5 increases Nu to 6.80% and 9.49% in Case II and IV but decreases it to 18.31% and 9.69%Case I and III, respectively. Table 4 indicates thatwhenRa=300, then Nb, Nr, and Nt affect the value |Ψmax| in all the problems.Notice thatmaximumvariations of |Ψmax|as a result of the increase inNbare+0.35%, +0.26%, −0.85% and +0.60% in Case I, II, III, and IV, respectively. The corresponding values due to a rise in Nr are −0.93%, −1.64%, +1.71%, and −0.17%. The results also demonstrates that a rise in Nt from0.05 to 0.5 results in 8.58%, 1.77% and 6.43% increase in the flow strength in Case II, III, and IV, respectively, but decreases the value |Ψmax| to 1.36% in Case I. Numerical values of the average Nusselt number in Table 4 indicate that an increase in Nb from 0.05 to 0.5 leads to +0.02%, +0.01%, −1.47%, and +1.18% deviations in heat transfer in Case I, II, III, and IV, respectively. The corresponding changes due to a rise in Nr are−0.16%, +0.66%,+2.0%, and+1.70%. The alternation of Nt also brings−17.75%, +5.42%, -8.33%, and +10.48% variations in the average Nusselt number. The thermophoresis parameter is found to be the most effective coefficient in the current cases. In order to provide a better picture about the consequences of this parameter on di- stributions of isocontours of the stream function, temperature and nanoparticles fraction, the corresponding contours for theflowproblemsareprovided inFigs. 2-5,whichbelong toRa=100 with both Nt =0.05 and Nt =0.5. Fig. 2. Isocontours of the stream function, temperature and nanoparticles fraction for Case I at Ra=100, Le= 25 with different values of Nt (up: Nt =0.05; down: Nt =0.5) Figure 2 shows the isocontours of the stream function (left), temperature (middle) and nanoparticles fraction (right) for Case I. Regardless of the value of Nt, a single convective cell appears inside the cavitywith an ascendingflownear the leftwall and adescendingflownear the right wall. It is evident that a growth in Nt does not have a significant effect on the streamlines and isothermal lines butmakes the nanoparticles distribution more non-homogeneous. The streamlines, isotherms and isoconcentrations of Case II are provided in Fig. 3. Obvio- usly, four convective cells appear here within the cavity. The convective cells located in the 230 I. Zahmatkesh,M.R. Habibi Fig. 3. Isocontours of the stream function, temperature and nanoparticles fraction for Case II at Ra=100, Le= 25 with different values of Nt (up: Nt =0.05; down: Nt =0.5) bottom-left/top-right parts of the cavity are rotating clockwise but those located in the bottom- -right/top-left parts are counter-clockwise vortices. The appearance of these circulations is at- tributed to the imposition of the sinusoidal temperature distribution on the side walls in this case. Cores of the convective cells are located close to the side walls due to large temperature gradients there. The distributions of Ψ and θ are symmetric with respect to X =0.5. It is evi- dent that the Nt promotion leads to variations in all characteristics noticeably. Obviously, the streamlines pattern is changed in a way of growing the two top convective cells. Moreover, the bottom half of the cavity experiences more intensive heating while the opposite side transfers less heat. The main variations with Nt are related to the isoconcentrations. The Nt elevation causes a more non-homogeneous nanoparticles distribution. This is similar to Case I, but the effect is more remarkable here. Figures 4 and 5 depict the isocontours of the stream function, temperature andnanoparticles fraction forCase III and IV, respectively. They correspond to themixed convection environment. Theeffect of theNtpromotionon thedistributionof thecontourplotsbearsa strongresemblance to what is observed in the natural convection cases. 5. Concluding remarks A critical analysis of natural andmixed convection of a nanofluid in square porous cavities has been presented here using Buongiorno’s mathematical model. The findings of this study can be summarized as: (1) Imposition of a sinusoidal temperature distribution on the sidewalls leads to heat transfer improvement both in the natural andmixed convection environments. (2) The consequence of the thermophoresis number on the flow strength and the average Nusselt number is more prominent than the Brownian diffusion parameter and the ther- mophoresis number. Natural and mixed convection of a nanofluid in porous cavities... 231 Fig. 4. Isocontours of stream the function, temperature and nanoparticles fraction for Case III at Ra=100, Pe=25, Le=25 with different values of Nt (up: Nt=0.05; down: Nt=0.5) Fig. 5. Isocontours of the stream function, temperature and nanoparticles fraction for Case IV at Ra=100, Pe=25, Le=25 with different values of Nt (up: Nt=0.05; down: Nt=0.5) 232 I. Zahmatkesh,M.R. Habibi (3) With an increase in the thermophoresis number, progressive changes occur in the isoconto- urs of the stream function, temperature and nanoparticles fraction, and the nanoparticles distribution becomes more non-homogeneous. (4) With the sinusoidal wall temperatures, the heat transfer rate is an increasing function of the thermophoresis number, but in a cavity with uniform wall temperatures, depending on the value of the Rayleigh number, an increase in Nt may enhance or deteriorate the average Nusselt number. (5) The Brownian diffusion parameter and the buoyancy ratio have almost no effect onNu in the natural convection butwith an increase inRa, they become graduallymore influential in the mixed convection. References 1. Ali K., Iqbai M.F., Akbar Z., Ashraf M., 2014, Numerical simulation of unsteady water- -based nanofluid flow and heat transfer between two orthogonally moving porous coaxial disks, Journal of Theoretical and Applied Mechanics, 52, 1033-1046 2. Behroyan I., Vanaki S.M., Ganesan P., Saidur R., 2016, A comprehensive comparison of various CFD models for convective heat transfer of Al2O3 nanofluid inside a heated tube, Inter- national Communications in Heat and Mass Transfer, 70, 27-37 3. Buongiorno J., 2006,Convective transport in nanofluids,Journal of Heat Transfer, 128, 240-250 4. Ghalambaz M., Sabour M., Pop I., 2016, Free convection in a square cavity filled by a porous medium saturated by a nanofluid: Viscous dissipation and radiation effects, Engineering Science and Technology, 19, 1244-1253 5. Ghasemi S.E., Hatami M., Salarian A., Domairry G., 2016, Thermal and fluid analysis on effects of a nanofluid outside a stretching cylinder with magnetic field using the differential quadraturemethod, Journal of Theoretical and Applied Mechanics, 54, 517-528 6. Kefayati G.H.R., 2017a, Mixed convection of non-Newtonian nanofluid in an enclosure using Buongiorno’s mathematical model, International Journal of Heat and Mass Transfer, 108, 1481-1500 7. Kefayati G.H.R., 2017b, Simulation of natural convection and entropy generation of non- -Newtonian nanofluid in a porous cavity using Buongiorno’s mathematical model, International Journal of Heat and Mass Transfer, 112, 709-744 8. Mustafa M., 2017,MHD nanofluid flow over a rotating disk with partial slip effects: Buongiorno model, International Journal of Heat and Mass Transfer, 108, 1910-1916 9. Nield D.A., Bejan A., 2013,Convection in Porous Media, Springer, NewYork 10. Nield D.A., Kuznetsov A.V., 2009, The Cheng-Minkowycz problem for natural convective boundary-layer flow in a porous medium saturated by a nanofluid, International Journal of Heat and Mass Transfer, 52, 5792-5795 11. Rostamzadeh A., Jafarpur K., Goshtasbi Rad E., 2016, Numerical investigation of pool nucleate boiling in nanofluid with lattice Boltzmann method, Journal of Theoretical and Applied Mechanics, 54, 3, 811-825 12. SheikholeslamiM., Ganji D.D., RashidiM.M., 2016,Magnetic field effect on unsteady nano- fluidflowandheat transferusingBuongiornomodel,Journal ofMagnetism andMagneticMaterials, 416, 164-173 13. SheremetM.A.,CimpeanD.S., Pop I., 2017,Free convection in a partially heatedwavyporous cavity filled with a nanofluid under the effects of Brownian diffusion and thermophoresis,Applied Thermal Engineering, 113, 413-418 Natural and mixed convection of a nanofluid in porous cavities... 233 14. SheremetM.A.,GrosanT.,Pop I., 2014,Free convection in shallowandslenderporous cavities filled by a nanofluid using Buongiorno’s model, Journal of Heat Transfer, 136, 082501-1 15. Sheremet M.A., Pop I., 2014a, Conjugate natural convection in a square porous cavity filled by a nanofluid using Buongiorno’s mathematical model, International Journal of Heat and Mass Transfer, 79, 137-145 16. Sheremet M.A., Pop I., 2014b, Natural convection in a square porous cavity with sinusoidal temperature distributions on both side walls filled with a nanofluid: Buongiorno’s mathematical model,Transport in Porous Media, 105, 411-429 17. SheremetM.A.,Pop I., 2015a,Free convection in a triangular cavityfilledwith aporousmedium saturated by a nanofluid Buongiorno’s mathematical model, International Journal of Numerical Methods for Heat and Fluid Flow, 25, 1138-1161 18. Sheremet M.A., Pop I., 2015b, Free convection in a porous horizontal cylindrical annulus with a nanofluid using Buongiorno’smodel,Computers and Fluids, 118, 182-190 19. SheremetM.A.,Pop I.,RahmanM.M., 2015,Three-dimensionalnatural convection inaporous enclosure filled with a nanofluid using Buongiorno’smathematical model, International Journal of Heat and Mass Transfer, 82, 396-405 20. Torshizi E., Zahmatkesh I., 2016, Comparison between single-phase, two-phase mixture, and Eulerian-Eulerian models for the simulation of jet impingement of nanofluids, Journal of Applied and Computational Sciences in Mechanics, 27, 55-70 21. Zahmatkesh I., 2008a,Onthe importance of thermalboundaryconditions inheat transfer anden- tropy generation for natural convection inside a porous enclosure, International Journal of Thermal Sciences, 47, 339-346 22. Zahmatkesh I., 2008b, On the importance of thermophoresis and Brownian diffusion for the deposition ofmicro- and nanoparticles, International Communications in Heat andMass Transfer, 35, 369-375 23. Zahmatkesh I., 2015,Heatline visualization for buoyancy-drivenflow inside a nanofluid-saturated porous enclosure, Jordan Journal of Mechanical and Industrial Engineering, 9, 149-157 24. Zahmatkesh I.,NaghedifarS.A., 2017,Oscillatorymixedconvection in jet impingementcooling of ahorizontal surface immersed inananofluid-saturatedporousmedium,NumericalHeatTransfer, Part A, 72, 401-416 Manuscript received August 19, 2017; accepted for print September 18, 2018