Conjugate Heat Transfer Analysis on a graphics card.

Conjugate heat transfer analysis on a graphics card

Aim:-

To perform a steady state conjugate heat transfer analysis on a model of graphic card.to calculate the maximum temperature and heat transfer coefficient attend by the processor got four different inlet velocity and comparing their result.

Introduction:-

Technology is advancing day by day, from the days when we need a large PCB board to account all our components, now that has been replaced by smart PCB design with complex circuit and microprocessor. Modern age is demanding for electronic device that have more computational power with compact design. Engineers are developing design for more compacting of electric component, but as the size gets reduce and their processing power increasing, they are generating more heat and have a huge chance to get over heat and burn the system. So need of effective cooling system is persist to protect electric component to get over heat and increase their life. More the current consumed by the component the greater the heat will be generated by them. In most of complicated system forced convection is implement to remove heat generation since the greater the surface area greater will heat removed for this purpose heat sink or fins are designed with special material which have high surface hear transfer coefficient. Some of the best examples we see in a CPU processor, mobile phones, air conditioning placed inside VMC an CNC machine. Many of the highly computational machine have in build air conditioning to remove heat generated by component.

Theory:-

When current flows through a conductor, heat energy is generated in the conductor .The heating effect of electric current depends upon the resistance, the time t for which current flow and the amount of current .The higher the current the higher the heat generated. This heat is a waste of energy. It can damage the insulation and other electric components, or even cause a fire. It reduce the life of several component in the circuit.an additional cooling system needs to be added to the circuit to counter the heating effect. This add to the cost. For complicated circuit there is highly need for successful cooling system like in computer processors. They are small electric component and do lot of work due to their small size and high computational power their consume huge amount of current and in return generate heat. If this heat is not removed it can damage the processor due to over heat. Here the need of effective cooling system has arrived . in many electric components air conditioning system is implement in which there be room to install it. But as the system get compact in size alternate source is needed for cooling like for CNC and VMC machine we use AC compartment for cooling electric components, but for CPU processor we can not provide AC source we have to develop which is compact and perform the operation well. So we choose a small fan system which automatically on when the heat of processor reach at specific value. But accurate ventilation is required to remove heat accurately from all components because fan suck air from one direction and blow it.

In his paper we present a Computer aided simulation for heat transfer from processor to air by forced convection. A fan is used to provide a forced convection effect. The graphic card is the processor which hold many small electric components which generate a hear and can be represent as a heat source. Forced convection is a mechanism or type of transport in which fluid motion is generated by an external source like a pump, fan, suction device etc. Alongside natural convection , thermal radiation and thermal conduction it is one of the methods of heat transfer and allows significant amount of heat energy to be transported very efficiently. Conjugate heat transfer corresponds with the combination of heat transfer in solids and heat transfer in fluids. In solids, conduction often dominates whereas in fluids, convection usually dominates. Conjugate heat transfer is observed in many situations. For example, heat sinks are optimized to combine heat transfer by conduction in the heat sink with the convection in the surrounding fluid.

Modelling Methodology:-

For our present model a graphic card geometry has been prepared in ANSYS Spaceclaim. After geometry preparation we have to make a fluid volume so be design a enclosure around the graphic card. Then for mesh interference option we enable share topology option and enable share option show interfernece during meshing ,mesh generation is adequate between fluid and solid volume. As shown in Figure 1. After preparing geometry we update details to meshing. In meshing first we provide a name selection to all solid volume. For bottom fluid wall we named as wall and all other fluid wall as symmetry. we needed as shown in figure. Then select a body sizing for fluid volume we choose 3 mm, then for board we choose a body sizing of 0.5 mm then for processor or heat source we choose a body sizing of 0.3 mm, then for fins or heat sink we choose a body sizing of 0.3mm. edge sizing is given on cylindrical component on board with number of division 20. As shown in figure 3 and 4. Finer element size is needed at heat source and heat sink so that minute change in heat can be captured .Method is automatic so we can see all tri element. From this meshing setting we get a number of node equal to 184818 and number of element equal to 895082.

Figure 1, Geometry Figure 2, Name selction of Graphic card

Figure 3, Meshing on graphic card Figure 4, Fluid domain meshing

Solver setup:-

Now we select a optimum solve setting based o our requirement. In Fluent window first here we choose a pressure based, steady state analysis is selected with gravity enable in -Y direction with constant 9.81 m/s. Turbulent model is selected as K-omega SST model. Next step is to define the material, here we choose total four material one for fluid, then one for solid board then one for heat sink and heat source. All material property are defined in Table 1, below.

Material	Density (kg/m^3)	Specific heat (J/kg-k)	Thermal conductivity (W/m-k)
Air	1.225	1006.43	0.0242
Copper	8978	381	387.6
FR-4	1250	1300	0.35
Component	1900	795	10

Table 1,

Next we have to define cell zone correctly here are four cell zone which we have created four different material. For fluid we choose a air. For solid board we choose material FR-4, for heat sink we choose a copper and last for heat source we choose a material component. For heat source we can add one parameter make it’s a heat source which continuously generate heat so by enabling source term in cell zone condition for heat source and provide a value of 1e^07 (W/m^3).

In boundary condition we are performing the simulation for four different inlet velocity cases.

CASE 1:- Inlet is provided as velocity inlet at 1 m/s. and outlet as pressure outlet as atmospheric pressure. All other boundary condition remains the same .

CASE 2:- Inlet is provided as velocity inlet at 3 m/s. and outlet as pressure outlet as atmospheric pressure. All other boundary condition remains the same .

CASE 3:- Inlet is provided as velocity inlet at 5 m/s. and outlet as pressure outlet as atmospheric pressure. All other boundary condition remains the same .

CASE 4:- Inlet is provided as velocity inlet at 7 m/s. and outlet as pressure outlet as atmospheric pressure. All other boundary condition remains the same .

Bottom of fluid wall is assumed as wall and all other three fluid wall selected as symmetry. Method is chosen as Coupled with pressure as body force weightage because it account for density of air and buoyancy effect. Pseudo transient is checked in report definition we create new volume report of temperature on heat source. In monitor check for convergence is disabled so that we can analysis up to given iterations .Hybrid initialization has been done and solution is calculated for 300 Iteration.

Results And Discussions :-

The results obtained by CFD simulations are displayed with the help of velocity and temperature contours. Convergence plot as shown in Figure 5, we can sure that solution is successfully converge and we can validate our result. all variables in convergence plot first start to decrease and then oscillate but after some time or when it cross 250 iteration its oscillation amplitude decrease and become more constant and then there be no drastic change in its value from that we can conclude that our simulation has reaches in steady state so we can stop calculation and go for validating result. For Case 1 with inlet velocity of air at 1m/s, as expected when our heat source generate heat all adjacent wall to its applied with coupled condition so they start heating. Fins absorb most of the heat and due to increased surface area and its thermal conductivity most of the heat is dicipetate by it. When air at velocity strike to the fin surface heat transfer phenomena is occurred ,air heats up and absorb heat from the fins . so in this way the generated heat is removed by fins. Now from Figure 6 we see that the wall adjacent near to hear source are heated in same way and temperature range from 300 k to 318.5 k . It’s the maximum temperature attend by processor then due to force convection heat is removed. Also in Figure 7, we plot a temperature VS iteration plot to show how the maximum temperature is attend by processor. Velocity contour is represented in Figure 8. From that we find there is some region of very low velocity when air pass through first fins. Air start to recirculated in that portion .Therefore most of the heat is transferred by the fins placed in first position. Fins at the centre will not transfer heat efficiently as they are blocked by the direct flow of air by successive fins. Heat transfer coefficient of the Fins is calculated as 11.82 W/m^2-k and Wall adjacent heat transfer coefficient is 268.57 W/m^2-k.

Figure 5, Convergence plot for Case 1. Figure 6, Temperature contour for Case 1.

Figure 7, Temperature plot on processor for Case 1 Figure 8, Velocity contour for Case 1.

For Case 2 with inlet velocity increased to 3 m/s we observe the same convergence of solution as show in Figure 9, and temperature contour show the significant reduction in maximum temperature attend by the processor from 300 k to 309.36 k ,which is 9% lesser than velocity at 1 m/s. as shown in Figure 10 and 11, front fins have low heat on surface because they in direct contact with air, but as move to backward their capacity to transfer heat is decreased ,last fins have more temperature than front fins. Velocity contour also shows the same result ,air is recirculated at the centre portion of fins so the heat transfer is less at mid and last fins as shown in Figure 12. The maximum surface heat transfer coefficient attend by fins is 17.3 W/m^2-k. so there is increase in surface heat transfer coefficient from previews value. So as the inlet velocity increase the heat transfer coefficient value is also increasing. Also wall adjacent heat transfer coefficient as 269.94 V W/m^2-k. same situation is we getting as we move to Case 3 with inlet velocity as 5 m/s and Case 4 with inlet velocity as 7 m/s. Both the result are shown in figure . The above result is summarised in table 2 below.

Figure 9, Convergence plot for Case 2 Figure 10, Temperature contour for Case 2.

Figure 11, Temperature plot on processor for Case 2 Figure 12, Velocity contour for Case 2.

Figure 13, Convergence plot for Case 3 Figure 14, Temperature contour for Case 3.

Figure 15, Temperature plot on processor for Case 3 Figure 16, Velocity contour for Case 3.

Figure 17, Convergence plot for Case 4. Figure 18, Temperature contour for Case 4.

Figure 19, Temperature plot on processor for Case 4. Figure 20, Velocity contour for Case 4.

Case	Maximum temperature of processor	Surface heat transfer coefficient	Wall adjacent heat transfer coefficient
Case 1	318.5 k	11.8 W/m^2-k	268.57 W/m^2-k
Case 2	309.37 k	17.3 W/m^2-k	269.94 W/m^2-k
Case 3	306.9 k	19.7 W/m^2-k	273.38 W/m^2-k
Case 4	305.7 k	21.18 W/m^2-k	279.4 W/m^2-k

Table 2,

Same results are also verified for Case 3 and 4 respectively and in case we get tahe maximum Surface heat transfer coefficient. From figure vector contour of velocity its is clearly shown that the air is recirculated on the mid and last section of fins.. as flow strike to the first fins its getting heated and moves up due to buoyancy force. More heat is transferred from the front fins than the middle and last fins. So the temperature is more at middle and last fins.

Figure 21, Velocity vector contour on Case 4.

Conclusion:-

The steady state conjugate heat transfer analysis is performed on a graphic card. From the above displayed result we vary inlet velocity form 1 m/s to 7 m/s and found that as the inlet velocity increase the surface heat transfer coefficient increase ,and the maximum temperature attend by the processor is lowest and by this we full fill the main purpose of our simulation. T velocity of 7m/s we get maximum temperature attend by processor will be is 305.7 k and surface heat transfer coefficient is 21.18 W/m^2-k . From velocity vector we found that the air is getting recirculated at the middle portion of fins due that the fins at middle and last portion will not efficiently transfer heat. The temperature on that is higher than the front fins.