Finite Element Analysis of Mixed Convection in a Rectangular Cavity with a Heat-Conducting Horizontal Circular Cylinder

Abstract. Combined free and forced convection in a two dimensional rec tangular cavity with a uniform heat source applied on the right vertical wall is studied numerically. A circular heat conducting horizontal cylinder is placed som ewhere within the cavity. The present study simulates a practical system, such as a conduc tive material in an inert atmosphere inside a furnace with a constant flow of gas from ou tside. Importance is placed on the influences of the configurations and physical properties of the cavity. The development mathematical model is governed by t he coupled equations of continuity, momentum and energy and is solved by employin g Galerkin weighted residual finite element method. In this paper, a finite elemen t formulation for steadystate incompressible conjugate mixed convection and condu ction flow is developed. The computations are carried out for wide ranges of the governin g parameters, Reynolds number ( Re), Richardson number ( Ri), Prandtl number ( Pr) and some physical parameters. The results indicate that both the heat transfe r rate from the heated wall and the dimensionless temperature in the cavity strongly depen d on the governing parameters and configurations of the system studied, such as size, locat ion, thermal conductivity of the cylinder and the location of the inflow and outflow opening . Detailed results of the interaction between forced airstreams and the buoyancy-dr iven flow by the heat source are demonstrated by the distributions of streamlines, isother ms and heat transfer coefficient.


Introduction
The studies of buoyancy driven flow characteristics in cavities are received considerable attention in recent years due to its extensive applications in the field of engineering, for example cooling of electronic devices, furnaces, lubrication technologies, chemical processing equipment, drying technologies etc. Analysis of above phenomena incorporating a solid heat conducting obstruction extends its usability to various other practical situations.Particularly a conductive material in an inert atmosphere inside a furnace with a constant flow of gas from outside constitutes a practical application for the present simulation.Many authors have recently studied heat transfer in enclosures with partitions, which influence the convection flow phenomenon.Omri and Nasrallah [1] studied mixed convection in an air-cooled cavity with differentially heated vertical isothermal sidewalls having inlet and exit ports by a control volume finite element method.They investigated two different placement configurations of the inlet and exit ports on the sidewalls.Best configuration was selected analyzing the cooling effectiveness of the cavity, which suggested that injecting air through the cold wall was more effective in heat removal and placing inlet near the bottom and exit near the top produce effective cooling.Later on, Singh and Sharif [2] extended their works by considering six placement configurations of the inlet and exit ports of a differentially heated rectangular enclosure whereas the previous work was limited to only two different configurations of inlet and exit port.At the same time, a numerical analysis of laminar mixed convection in an open cavity with a heated wall bounded by a horizontally insulated plate was presented by Manca et al. [3], where three heating modes were considered: assisting flow, opposing flow and heating from below.Results were reported for Richardson number from 0.1 to 100, Reynolds numbers from 100 to 1000 and aspect ratio in the range 0.1-1.5.They showed that the maximum temperature values decrease as the Reynolds and the Richardson numbers increase.The effect of the ratio of channel height to the cavity height was found to play a significant role on streamline and isotherm patterns for different heating configurations.The investigation also indicated that opposing forced flow configuration has the highest thermal performance, in terms of both maximum temperature and average Nusselt number.Later, similar problem for the case of assisting forced flow configuration was tested experimentally by Manca et al. [4] and based on the flow visualization results, they pointed out that for Re = 1000 there were two nearly distinct fluid motions: a parallel forced flow in the channel and a recirculation flow inside the cavity and for Re = 100 the effect of a stronger buoyancy determined a penetration of thermal plume from the heated plate wall into the upper channel.Very recently Manca et al. [5] experimentally analyzed opposing flow in mixed convection in a channel with an open cavity below.Recently Rahman et al. [6] studied numerically the opposing mixed convection in a vented enclosure.They found that with the increase of Reynolds and Richardson numbers the convective heat transfer becomes predominant over the conduction heat transfer and the rate of heat transfer from the heated wall is significantly depended on the position of the inlet port.
However, many authors have studied heat transfer in enclosures with heat-conducting body obstruction, thereby influencing the convective flow phenomenon.Shuja et al. [7] investigated the effect of exit port locations and aspect ratio of the heat generating body on the heat transfer characteristics and irreversibility generation in a square cavity.They found that the overall normalized Nusselt number as well as irreversibility was strongly affected by both of the location of exit port and aspect ratios.Papanicolaou and Jaluria [8] studied mixed convection from an isolated heat sources in a rectangular enclosure.Later on, Papanicolaou and Jaluria [9] performed computations on mixed convection from a localized heat source in a cavity with conducting walls and two openings for application of electronic equipment cooling.Hsu et al. [10] numerically investigated mixed convection in a partially divided rectangular enclosure.They considered the divider as a baffle inside the enclosure with two different orientations and indicated that the average Nusselt number and the dimensionless surface temperature dependent on the locations and height of the baffle.Lee et al. [11] considered the problem of natural convection in a horizontal enclosure with a square body.Natural convection in a horizontal layer of fluid with a periodic array of square cylinder in the interior were conducted by Ha et al. [12], in which they concluded that the transition of the flow from quasi-steady up to unsteady convection depends on the presence of bodies and aspect ratio effect of the cell.However, in the previous literature the body was considered as a rigid wall but internal heat transfer was not calculated.Few numerical studies taking into account heat transfer in the interior of the body have been conducted over the past couple of decades.One of the systematic numerical investigations of this problem was conducted by House et al. [13], who considered natural convection in a vertical square cavity with heat conducting body, placed on center in order to understand the effect of the heat conducting body on the heat transfer process in the cavity.They showed that for given Ra and P r an existence of conducting body with thermal conductivity ratio less than unity leads to heat transfer enhancement.
As the first step toward accurate flow solutions using the adaptive meshing technique, this paper develops a finite element formulation suitable for analysis of steady-state conjugate mixed convection and conduction problems.The paper starts from the Navier-Stokes equations together with the energy equations to derive the corresponding finite element model.The computational procedure used in the development of the computer program is described.The finite element equations derived and then the computer program developed are then evaluated by example of mixed convection in a rectangular cavity with heat conducting horizontal circular cylinder.
The objective of the present study is to investigate the effect of a heat conducting solid cylinder, which may increase or decrease the heat transfer on mixed convection in a rectangular vented cavity.Numerical solutions are obtained over a wide range of Richardson number, Reynolds number, Prandtl number and various physical parameters.The dependence of the thermal and flow fields on the sizes, locations and thermal conductivity of the cylinder is studied in detail.

Model specification
The physical model considered here is shown in Fig. 1 along with the important geometric parameters.A cartesian co-ordinate system is used with origin at the lower left corner of the computational domain.It consists of rectangular cavities with a heat conducting horizontal circular solid cylinder, whose right wall is subjected to hot with T h temperature while the other sidewalls are kept adiabatic.The cavity dimensions are defined by height H and width L. The inflow opening located on the left adiabatic vertical wall and the outflow opening on the opposite heated vertical wall is arranged as shown in the schematic figures and may vary in location, placed either at the top or bottom position.The cavity presented in Fig. 1(a) is subjected to an external flow that enters via the bottom of the insulated vertical wall and leaves via the bottom of the opposite heated vertical wall.For reasons of brevity, this case will be referred to as BB configuration from now.When the horizontal cold jet enters into the cavity from the bottom of the insulated wall and leaves from the top of the opposite vertical one is shown in Fig. 1(b), this case will be referred as BT configuration.Similarly, Fig. 1(c) and 1(d) are referred to as TB and TT configurations respectively.For simplicity, the heights of the two openings are set equal to the one-tenth of the enclosure height.It is assumed that the incoming fluid flow through the inlet at a uniform velocity, u i at the ambient temperature T i and the outgoing flow is assumed to have zero diffusion flux for all variables i.e. convective boundary conditions (CBC).All solid boundaries are assumed to be rigid no-slip walls.

Mathematical formulation
The flow within the cavity is assumed to be two-dimensional, steady, laminar, incompressible and the fluid properties are to be constant.The radiation effects are taken as negligible and the Boussinesq approximation is used.The dimensionless equations describing the flow are as follows: For solid cylinder the energy equation is where Re = u i L/v, Gr = gβ∆T L 3 /v 2 , P r = v/α, Ri = Gr/Re 2 , and K = k s /k (∆T = T h − T i and α = k/ρc p are the temperature difference and thermal diffusivity of the fluid respectively) are the Reynolds number, Grashof number, Prandtl number, Richardson number and solid fluid thermal conductivity ratio respectively.The above equations were non dimensionalized by using the following dimensionless dependent and independent variables where X and Y are the coordinates varying along horizontal and vertical directions respectively, U and V are the velocity components in the X and Y directions respectively, θ is the dimensionless temperature and P is the dimensionless pressure.Non-dimensional forms of the boundary conditions for the present problem are specified as follows: At the inlet: At the outlet: convective boundary condition (CBC), P = 0.At all solid boundaries: U = 0, V = 0.
At the heated right vertical wall: θ = 1.
At the left, top and bottom walls: ∂θ ∂N = 0.At the fluid-solid interface: Where N is the non-dimensional distances either X or Y direction acting normal to the surface and K is the dimensionless ratio of the thermal conductivity (k s /k).
The average Nusselt number at the heated wall is calculated by Nu = 1 dy and the bulk average temperature is defined as θ av = θ dV /V , where L h and h(y) are the length and the local convection heat transfer coefficient of the heated wall respectively, V is the cavity volume and θ av should be minimized.

Numerical analysis
The governing equations along with the boundary conditions are solved numerically by employing Galerkin weighted residual finite element techniques.The finite element formulation and computational procedure are discussed detail in Appendix.

Grid independence test
Geometry studied in this paper is an obstructed vented cavity; therefore several grid size sensitivity tests were conducted in this geometry to determine the sufficiency of the mesh scheme and to ensure that the solutions are grid independent.This is obtained when numerical results of the average Nusselt number Nu, average temperature θ av and solution time become grid size independent, although we continue the refinement of the mesh grid.Five different non-uniform grids with the following number of nodes and elements were considered for the grid refinement tests: 24545 nodes, 3788 elements; 29321 nodes, 4556 elements; 37787 nodes, 5900 elements; 38163 nodes, 5962 elements and 48030 nodes, 7516 elements as shown in Table 1.From these values, 38163 nodes and 5962 elements can be chosen throughout the simulation to optimize the relation between the accuracy required and the computing time.

Code validation
The present code was extensively validated based on the problem of House et al. [13].
We present here some results obtained by our code in comparison with those reported in House et al. [13] for Ra = 0.0, 10 5 and two values of K = 0.2 and 5.0.The physical problem studied by House et al. [13] was a vertical square enclosure with sides of length L. The vertical walls were isothermal and differentially heated, where as the bottom and top walls were adiabatic.A square heat conducting body with sides of length equal to L/2 was placed at the center of the enclosure.For the same parameters used in House et al. [13]; the average Nusselt number (at the hot wall) comparison is shown in Table 2.The present results have an excellent agreement with the results obtained by House et al. [13].

Results and discussion
Numerical results have been presented in order to determine the effects of the presence of dimensionless parameters in a rectangular cavity.The dimensionless governing parameters that must be specified for the system are Reynolds number (Re), Richardson number (Ri), Prandtl number (P r) and the physical parameters in the system are the cylinder diameter (D), solid fluid thermal conductivity ratio (K), location of inlet and outlet openings of the cavity, location of cylinder in the cavity and cavity aspect ratio (AR).Since so many basic dimensionless parameters are required to characterize a system, a comprehensive analysis of all combinations of these parameters is not practical.The numerical results have been used to explain the effect of several parameters at a small fraction of the possible situations by simplifying the configuration.The presentations of the results have been started with the streamline and isotherm patterns in the cavity.Representative distributions of average Nusselt number at the heated wall and average temperature of the fluid in the cavity have also been presented.(i)-(iv), it is noticed that the isothermal lines are more uniformly distributed in the cavity for the BB and TB configurations.On the other hand, it has been observed that for TB configuration the high temperature region is more concentrated near the hot wall and the distribution of the isothermal lines is non-uniform in the cavity.It has also been observed that the isothermal lines are more vertically concentrated around the heat source for the TT configuration, which is similar to conduction-like distribution.It is also seen that the thermal boundary layer near the heated wall in the cavity is developed for BT and TT configurations, where as the thermal boundary layer in the cavity is absent for the BB and TB configurations.The average Nusselt number (Nu) at the hot wall and the bulk average temperature (θ av ) in the cavity have been compared in Fig. 3 for the above four configurations.From this figure it is clear that the Nu is highest for BT configuration.The reason for this is the fresh fluid entering in the bottom of the left wall and travels comparatively long distance and taking heat away from the hot wall before exit the cavity.It can also be seen that the average temperature of the fluid in the cavity is the lowest for the BT configuration in the forced convection dominated region and for the TT configuration in the free convection dominated region.

Effect of the cylinder diameter
The effects of the heat conducting cylinder on the flow and thermal fields for the BT configuration at AR = 1.0,Re = 100, Ri = 1.0,P r = 0.71, L x = L y = 0.5 and K = 5.0 have been presented in Fig. 4. As compared to Fig. 4a(i), the solid cylinder in the cavity reduces the strength of the recirculation cell induced by the heat source.In Fig. 4a(ii), which is for cylinder of diameter D = 0.1, only small differences in the streamlines have been observed when compared with Fig. 4a(i).This is the evidence of no significance influence of a small size solid cylinder on the convective flow of the cavity.On the other hand, as the size of the cylinder increases, the space available for the buoyancy-induced recirculating flow decreases.From the isotherms shown in Figs.4(i)-(iv), it has been observed that the high-temperature zone is confined to a region close to the hot surface for all cases and the lines are uniformly distributed in the cavity.The last line from the heated wall is the line with θ 1 = 0.05 for all the cases presented in this figure .A closer examination shows that the area between the heated wall and the line θ 1 = 0.05 slightly increase with the increase of the cylinder diameter (D).It has been observed that Nu is the highest for the large cylinder diameter (D = 0.4) at Ri ≤ 6.0 and beyond this values of Ri the cylinder diameter has insignificant effect on the average Nusselt number at the hot wall.The effect of cylinder diameter on average temperature of the fluid in the cavity is also shown in Fig. 5. From this figure it has been seen that cylinder diameter has little effect on the average temperature (θ av ) of the fluid in the cavity.A closer examination of Fig. 5 has revealed that the values of θ av decreases at Ri ≤ 0.5 and beyond this value of Ri, θ av increases sharply with increasing Ri for all values of D. On the other hand, θ av is the lowest for D = 0.2 at Ri ≤ 2.0 and beyond this value of Ri, θ av is the lowest for D = 10.0.

Effect of thermal conductivity ratio
The effect of the thermal conductivity ratio of the solid and fluid has also been computed numerically and have shown in Figs. 6 for the BT configuration at AR = 1.0,Re = 100, Ri = 1.0,P r = 0.71, D = 0.2 and L x = L y = 0.5.It has been found that the different heat transfer properties of the cylinder have small effect on the heat transfer in the cavity.The streamlines for these cases appear to be almost identical as shown in the Fig. 6a(i)-(iv).This is because thermal conductivity ratio has no influence on the velocity distribution.The effect of thermal conductivity ratio on the isotherms has been presented in the Figs.6b(i)-(iv).From these figures it has been seen easily that a concentrated thermal boundary layer near the heated surface has developed for all the cases and the isothermal lines moves away from the centre of the heat conducting cylinder with increasing values of the thermal conductivity ratio.
Average Nusselt number at the hot wall and average temperature of the fluid in the cavity as a function of Richardson number have been shown in Fig. 7 for the BT configuration at AR = 1.0,Re = 100, P r = 0.71, L x = L y = 0.5, D = 0.4 and K = 0.2, 1.0, 5.0 and 10.0.The average Nusselt number at the heated surface is found to be the highest for a relatively low thermal conductivity ratio K = 0.2, which is due to the cylinder with low thermal conductivity acts as an insulator and prevents heat transfer between the hot and cold fluid streams.Hence the heat transfer in this case is mainly by mixed convection.The average temperature of the fluid in the cavity decreases for all the cases in the forced convection dominated region and are increases sharply with increasing Ri in the free convection dominated region.On the other hand, the values of θ av is found to be the lowest for K = 0.2 at Ri ≤ 2.0 and beyond this value of Ri, θ av is the lowest for K = 10.0.

Effect of Reynolds number
The particular Reynolds numbers (shown in Fig. 9).From this figure it has been observed that for a fixed values of Ri, the average Nusselt number at the hot wall is the highest and average temperature of the fluid in the cavity is the lowest for large values of Re = 200.This is due to more heat has been carried away from the heat source and dissipated through the out flow opening for the large values of Re.For Ri = 0.0, the major incoming flow is symmetric about the diagonal joining from the inlet to the exit port and a small vortex is developed near the left insulated wall starting from just above the inlet port, due to the domination of forced convection as shown in Fig. 10a(i).At Ri = 2.5, the size of the vortex is increased dramatically and changes its pattern from a uni-cellular vortex to a bi-cellular vortices, which occupies much of the cavity as shown in Fig. 10a(ii).This is because the buoyancy force dominates the forced flow in the cavity.As Ri increases to 5.0, the bi-cellular vortices merge into a single vortex and become slightly large as presented in Fig. 10a(iii).Further increase of Ri at 10.0, the patterns of the streamlines are about the same as those for Ri = 5.0, but, a careful observation indicates that the inner vortex become larger slightly in size and stronger in strength compared this with the upper one, because the effect of free convection on heat transfer and flow increases with increasing Ri.From Figs. 10b(i)-(iv) it has been seen that the isothermal lines are nearly parallel to the vertical heated wall for Ri = 0.0, this indicating a dominant heat conduction mechanism.For the larger Ri (Ri = 2.5, 5.0, 10.0) the high temperature region become more concentrated and thin near the hot wall, and the other isothermal lines uniformly distributed in the remaining parts of the cavity.

Effect of Prandtl number
The influence of Prandtl number on streamlines as well as isotherms for the BT configuration at AR = 1.0,Re = 100, Ri = 1.0,D = 0.2, L x = L y = 0.5 and K = 5.0 has been demonstrated in Fig. 11.The flow with small Prandtl number (P r = 0.71) has been affected by the buoyancy force, thus creating a CCW recirculation region near the left top corner in the cavity as shown in Fig. 11a(i).This recirculation region decreases with increasing Prandtl numbers as shown in Figs.11a(ii)-(iv).The isotherms illustrate the temperature field in the separated flow region has been shown in Fig. 11b(i)-(iv).The last line from the heated wall is the line with θ 1 = 0.05 for all the Prandtl numbers presented in this figure.The most significant information in these plots is the shifting of the θ 1 line for the different Prandtl numbers.For the case of P r = 0.71 this line moves under the cylinder placed at the center in the cavity.With increasing Prandtl numbers, that line moves towards the heated wall.The area enclosed between the θ 1 line and the heated wall can be considered as the thermally influenced region of the fluid by the heated wall.
The large region that is associated with a smaller Prandtl number indicates the relatively strong thermal conduction component in these fluids.The variation of average Nusselt number (Nu) at the heated wall and average temperature of the fluid in the cavity along with Richardson number for different Prandtl numbers has been presented in Fig. 13.From this figure it is clearly seen that for a particular values of Ri the average Nusselt number is the highest and average temperature is the lowest for the large Prandtl number P r = 7.1.This is because, the fluid with the highest Prandtl number is capable to carried more heat away from the heat source and dissipated through the out flow opening in the cavity.

Effect of cylinder locations
The effect of the cylinder location on the thermal transport has great importance and has been shown in Figs. 12 and 14.Streamlines and isothermal lines for various cylinder locations have been shown in Fig. 12 for AR = 1.0,Re = 100, Ri = 1.0,D = 0.2, P r = 0.71 and K = 5.0.As the cylinder moves closer to the left insulated wall along the midhorizontal plane and closer to the top insulated wall along the mid-vertical plane, a large circulation cell with inner vortex has confined at the left top portion in the cavity as shown in Figs.12a(i) and 12a(iv) respectively, and concentrated thermal layer has developed around the heat source as shown in Figs.12b(i) and 12b(iv) respectively.Further, if the cylinder moves near the heat source along the mid-horizontal plane, the recirculating cell reduces as shown in Fig. 12a(ii) and concentrated isotherms become vertical at the heat source.However, the size of the recirculating cell reduces dramatically if the solid body is located lower in the cavity as shown in Fig. 12a(iii) and the isothermal lines are more vertically concentrated around the heat source as exposed in Fig. 12b(iii), which is similar to conduction-like mechanism.
The average Nusselt numbers at the heated surface and the average temperatures in the cavity are plotted against Richardson numbers for four different cylinder locations have been shown in Fig. 14.From this figure it is seen that the Nu is the highest when the cylinder moves closer to the left insulated wall along the mid-horizontal plane and closer

Effect of cavity aspect ratio
The results presented in the preceding are for a square cavity for which the aspect ratio AR is 1.In order to investigate the convective heat transfer behavior at other aspect ratios, computations have also been done for the BT configuration at three additional aspect ratios of 0.5, 1.From Figs. 15(i)-(iv), it has been seen that for AR = 0.5 a small recirculation cell has developed just above the inlet position.This recirculation cell gradually increases with the increase of the value of AR, due to increasing the available space for the fluid in the cavity.On the other hand, another recirculation cell of the same size has also been located near the top surface for AR = 1.0 and it is reduces in size and the cell near the let wall become increases in size for AR = 1.5 and 2.0.Isotherms for these cases have been shown in Figs.16(i)-(iv).It has been seen from these figures that the isothermal lines are nonlinear and occupy most of the part of the cavity for the case of AR = 0.5, the isothermal lines becomes linear and concentrated near the hot wall in the cavity with the increasing values of AR, because of the distance between the hot wall and the inlet, through which fresh cold fluid enter in the cavity.
The average Nusselt number (Nu) at the heat source and the average temperature of the fluid in the cavity for AR = 0.5, 1.0, 1.5 and 2.0 have been shown in Fig. 17.It has been seen that for a particular value of Ri the average Nusselt number at the hot wall and the average temperature of the fluid in the cavity decrease with increasing values of AR.This is because the increasing value of AR increases the available space for the fluid in the cavity.

Conclusion
A finite element method for steady-state incompressible conjugate effect of mixed convection and conduction has been presented.The finite element equations have been derived from the governing equations that consist of the conservation of mass, momentum and energy equations.The derived finite element equations are nonlinear requiring an iterative technique solver.The Newton-Raphson iteration method has been applied to solve these nonlinear equations for solutions of the nodal velocity components, temperatures and pressures.The present study demonstrates the capability of the finite element formulation that can provide insight to steady-state incompressible conjugate effect of mixed convection and conduction problem.The qualitative and quantitative understanding of the influences of conjugate conduction-convection heat exchange has also been presented in this study.Attention is focused on identifying the optimum placement of inlet and outlet port for the best cooling effectiveness and on the effects of the Reynolds number, Richardson number, Prandtl number, cylinder diameter, solid-fluid thermal conductivity ratio, location of the cylinder in the cavity and the aspect ratio of the cavity.The major results have been drawn as follows: • Cavity orientation has a great influence on the streamlines and isotherms distributions.Comparatively large buoyancy induced vortex is located for BB configuration and relatively small buoyancy induced vortex is located for TT configuration.The thermal influenced region is comparatively bulky for BB configuration and slim for TT configuration.The average Nusselt number at the hot wall has been used to compare the heat transfer rate among different configurations.Results show that the configuration BT has the highest heat transfer rates, whereas configuration BB has the lowest effective heat transfer rate.
• Diameter of cylinder affects strongly the streamline distribution in the cavity.As a result, buoyancy-induced circulation cell reduces with increasing cylinder diameter.Comparatively small effect on the isotherms is observed for different cylinder diameter.The highest heat transfer is observed for the large cylinder diameter D = 0.4 at Ri ≤ 5.0, but after this the cylinder diameters have negligible effect on the heat transfer.
• Material properties (K) have insignificant effect on the flow field and have significant effect on the thermal fields.An unexpected result is found for the dependence of thermal transport on the ratio (K) of the thermal conductivity of the solid cylinder to that of the fluid.Negligibly small effect on the thermal phenomenon is observed for small cylinder size.But for large cylinder size, the variation of average Nusselt number at the hot wall is significantly influenced by the ratio K. Enhancement in the heat transfer is observed for the lowest thermal conductivity ratio.
• The forced convection parameter Re has a significant effect on the flow and temperature fields.Buoyancy-induced vortex in the streamlines increased and thermal layer near the heated surface become thin and concentrated with increasing values of Re.The average Nusselt numbers at the heated surface is always the highest and the average temperature of the fluid in the cavity is the lowest for the large value of Re.
• Mixed convection parameter Ri affects strongly on the flow and temperature fields.The recirculation cell due to heat source in the streamlines plot become large and the concentrated thermal layer near the heated surface become thin with increasing Ri.
• The influence of Prandtl number on streamlines and isotherms are remarkable for the different values of P r.Increasing the Prandtl number increase the average Nusselt number at the hot wall and decrease the average temperature of the fluid in the cavity.
• Locations of the cylinder have significant effect on the flow and thermal fields.The value of average Nusselt number (Nu) at the hot wall and the average temperature of the fluid in the cavity vary non-monotonically with the cylinder location in the cavity.
• The cavity aspect ratio has significant effect on the flow and temperature fields.The buoyancy induced recirculation cell increase and the heat transfer become conduction dominated with increasing the cavity aspect ratio.The average Nusselt number at the hot wall is always the highest for AR = 0.5 and the average temperature of the fluid in the cavity is the lowest for AR = 2.0.

Four
different cavity configurations have been investigated for the mixed convection problem in order to compare the behavior of convective heat transfer for different relative inlet and outlet locations.The streamlines corresponding to the four different inlet and outlet positions namely BB, BT, TB and TT with AR = 1.0,Re = 100, Ri = 1.0,P r = 0.71, L x = L y = 0.5, D = 0.2 and K = 5.0 have been shown in Figs.2a(i)-(iv).It has been observed that a large counter-clockwise (CCW) recirculation cell is formed above the main fluid stream for the BB (injection at the bottom of the left insulated wall and exit from the bottom of the heated wall) configuration and it occupies the maximum space of the cavity.This is because the fresh fluid entering the cavity travels the shortest possible distance before leaving the cavity and cannot come into intimate mixing with the hotter fluids.As the outlet port moved along the heated wall at the top corner and keeping the inlet position unchanged, i.e. for BT configuration the CCW recirculation cell reduces in size and is divided into two relatively small vortices, which are located at the left top corner in the cavity.However, for the TB configuration the flow changes its pattern from two recirculation vortices to a single vortex and shifted from left top corner to the right top corner in the cavity.On the other hand, a clock-wise (CW) small eddy is developed near the left insulated wall starting from just below the inlet position whereas the external flow increases its passage region and finally occupies almost the cavity for the TT configuration, which is due to the entering fresh fluids come into intimate mixing with the hotter fluids in the cavity.The contours of the dimensionless temperature (θ) corresponding to the above mentioned four cases with AR = 1.0,Re = 100, Ri = 1.0,P r = 0.71, L x = L y = 0.5, D = 0.2 and K = 5.0 have been presented in Figs.2b(i)-(iv).The value of θ on the heated wall is 1, whereas the value of θ of the fluid entering the cavity is zero and the contour values are incremented by 0.05.From the isotherms shown in Figs.2b
Further, in order to evaluate how the presence of the cylinder affects the heat transfer rate along the hot wall, average Nusselt number (Nu) has been plotted as a function of Richardson number (Ri) for four different cylinder diameters (D = 0, D = 0.1, D = 0.2 and D = 0.4) shown in Fig. 5.

Fig. 10
Fig. 10 has been indicated the dynamic and thermal field for the BT configuration at AR = 1.0,Re = 100, D = 0.2, L x = L y = 0.5, K = 5.0 and different Ri in terms of the streamlines and isotherms.The streamlines shown in Fig.10a(i)-(iv) describe the interaction of forced and natural convection under various convection regimes.For Ri = 0.0, the major incoming flow is symmetric about the diagonal joining from the inlet to the exit port and a small vortex is developed near the left insulated wall starting from just above the inlet port, due to the domination of forced convection as shown in Fig.10a(i).At Ri = 2.5, the size of the vortex is increased dramatically and changes its pattern from a uni-cellular vortex to a bi-cellular vortices, which occupies much of the cavity as shown in Fig.10a(ii).This is because the buoyancy force dominates the forced flow in the cavity.As Ri increases to 5.0, the bi-cellular vortices merge into a single vortex and become slightly large as presented in Fig.10a(iii).Further increase of Ri at 10.0, the patterns of the streamlines are about the same as those for Ri = 5.0, but, a careful observation indicates that the inner vortex become larger slightly in size and stronger in strength compared this with the upper one, because the effect of free convection on heat transfer and flow increases with increasing Ri.From Figs. 10b(i)-(iv) it has been seen that the isothermal lines are nearly parallel to the vertical heated wall for Ri = 0.0, this indicating a dominant heat conduction mechanism.For the larger Ri (Ri = 2.5, 5.0, 10.0) the high temperature region become more concentrated and thin near the hot wall, and the other isothermal lines uniformly distributed in the remaining parts of the cavity.

Fig. 14 .
Fig. 14.Effect of cylinder locations on average Nusselt number and average temperature for the BT configuration while AR = 1.0,Re = 100, P r = 0.71, D = 0.2 and K = 5.0.

Fig. 17 .
Fig. 17.Effect of cavity aspect ratio on average Nusselt number and average temperature for the BT configuration at while Re = 100, P r = 0.71, D = 0.2, Lx = Ly = 0.5 and K = 5.0.

Table 2 .
Nusselt number comparison for P r = 0.71