CONJUGATE HEAT TRANSFER ANALYSIS USING GRAPHIC CARD

Aim:

The Aim of the experiment is to develop a Conjugate heat transfer Analysis using a Graphic Card

Conjugate Heat Transfer:

Conjugate heat transfer is a behavioural analysis where convective (transfer between fluid) and conductive (transfer between solids) heat transfer takes place in a system. This combination of heat transfer has forced or natural convection where fluid is dominated by convection and solid is dominated by conduction.

Governing Equation:

Heat Transfer in Solids

When there is only conduction taking place in solid heat transfer it is defined by Fourier’s law which is given by

Where q is the conductive heat flux proportional to Temperature gradient.

In terms of transient Analysis, the Heat Equation is given by

Heat Transfer in Fluids

There are three main contributing factors for the heat transfer in fluids due to its moments, which are the density of the fluid, the viscosity of the fluid and the transport energy of the fluid. The temperature field for fluid flow (transient) which involves the above-mentioned parameters including the conductive term is given by

Objective:

• Carry out mesh Independent study
• Stimulate CHT using two different processor materials
• Carry out the analysis at different velocities
• To determine the hot spots and Heat transfer co-efficient at different velocities

Graphic card:

A graphic card is a device which is used to render the image onto to the monitor using the inputs generated. This converts the input data into signals which the monitor understands to display the output. The better the graphic card the better the quality of the images which we obtain the Monitor screen. There are two types of Graphic cards namely integrated and Discrete. The integrated is built into the motherboard whereas the discrete is installed as a separate component in the motherboard. The components of the graphic card are shown below in the figure. The fins used in the graphic card to conduct the heat generated from the processor as the air flow from the CPU’s fan would not alone would not help the heat generated to cool down.

In this experiment we would be studying how the heat generated is cooled down by distributing the heat from the processor to the fins and increasing the air flow of the fans.

Figure1: Graphic card

Experimental Setup:

The graphic card used for this simulation can be seen in the figure 2 below. A bounding box representing the fluid flow is created using space claim which is shown in the figure 3. The boundary conditions used are represented in the figure3.

Figure2: The Cad model of a Graphic card

Figure3: Fluid domain with Boundary conditions

Mesh Independent Study:

To carry out the mesh independent study, two types of meshes were created which are described in the table below.

Base mesh	Overall element size:0.002m	No of elements:252558
Refined Mesh	Overall element size:0.002m Face mesh at the Processor:0.001m	No of elements:331713

In order to reduce the interpolation errors and computational time a conformal mesh is generated. This is obtained by using the share topology in the space claim which would enable the formation of conformal mesh.

Figure4: Section view of the Conformal Mesh

Fig 5: Final Mesh output

Solver Setting:

Analysis type	Steady state
Solver	Pressure-based
Model	Energy and K-omega SST
Material	Aluminium
Initialization	Standard for the base mesh and hybrid for the refined mesh
No of iterations	Base mesh: 500 Refined mesh: 900

Apart from the settings mentioned above the heat is produced by the processor in the graphic card. So, a source term is used in the processor cell zone to input the heat generated. The heat generated by the processor is calculated by dividing the Thermal heat dissipation power (TDP) by the volume of the processor. From data from the website it is know that the normal processor TDP is 15 watts and the volume of the processor is 0.008 x 0.008 x 0.001 = 0.00000006 m^3. So, the heat generated by the Processor is calculated to be 2.5 e^8. This value is given as the source term for the processor.

Results of the Mesh Independent Study:

Figure6: Heat Distribution of the base mesh

Figure7: Heat distribution using Refined mesh

From the mesh independent study, we can conclude that the refined mesh shows more appropriate results. The heat distribution in the graphic card can been seen more precisely using the refined mesh setup, especially in the Fin base the distribution of the heat generated by the processor is more accuractely defined when compared to the base mesh.

Processor Materials:

Two different materials are used. The first being the aluminium and the second selected to be Titanium. The base mesh setting is applied and the results can be seen below

Result using Titanium as the processor Material:

Figure8: Plot showing the convergence and the temperature plot for the Titanium Processor

Figure9: Plot showing the convergence and the temperature plot for the Aluminium Processor

The results show that the heat generated by both processor material are almost same and since Aluminium is a cheaper material comparing titanium, the best material to proceed with the manufacturing would be Aluminium. Since the choosen material is Aluminium, other following simulation is processed with aluminium as the processor material.

CHT analysis on different velocities:

Since aluminium is the cost-effective material, we carry out the analysis with the same setting with aluminium as the material and velocities at 1m/s, 5m/s and 10m/s. The following table shows the number of iterations for the analysis to converge.

Velocity	No of Iterations
1 m/s	900
5 m/s and 10m/s	700

Results using different velocities:

Heat Distribution at a Velocity of 1m/s

Figure10: Convergence of the residuals

Figure11: Heat transfer co-efficient

Figure12: The temperature and the Velocity plot for 1m/s

Heat Distribution at a Velocity of 5m/s

Figure13: Convergence of the residuals

Figure14: Heat transfer co-efficient

Figure15: The temperature and the Velocity plot for 5m/s

Heat Distribution at a Velocity of 10m/s

Figure16: Convergence of the residuals

Figure17: Heat transfer co-efficient

Figure18: The temperature and the Velocity plot for 10m/s

Temperature and Heat Transfer co-efficient:

Velocity of Atmospheric air(m/s)	Maximum Temperature (K)	Heat Transfer co-efficient (w/m^2 K)
1	523.88	1433.57
5	376.61	1433.73
10	355.77	1434.79

From the above table we could see that the heat transfer taking place between the fluid and the solid model in the three cases is almost same i.e) the rate at which the heat from the solid is given to fluid to push it across the cpu outlet. In order to run the CPU with optimun energy conservation, the velocity inlet for this cases would 5m/s. The velocity of 1m/s is not considered as it generates more heat and at the same time for 10m/s the heat generation is lesser when compared to 5m/s, but to run the fan at 10m/s we need more energy when compared to 5m/s. This concludes that 5m/s would be the optimun inlet atmospheric air velocity.

Discussion on the Results

• From the results of the three velocities, we could see that the hot spot in the graphic card is the processor since its is the part of the graphic card where the heat is generated.
• As the velocity of the atmospheric air increases the overal heat generation of the graphic card decreases, concluding that for this model a higher velocity is needed for the efficient running of the system when this graphic card is used.
• The refined mesh gives more accurated results when compared to the base mesh.
• The Heat-transfer co-efficient is almost same for all the three cases.