CHT Analysis of a Graphics Card using ANSYS Fluent

Objective:

To perform a CHT analysis on a Graphics Card and determine maximum temperature attained by the processor.

Introduction:

The graphics card is responsible for rendering an image to the monitor by converting data into signals. A better and smoother image can be produced with the help of a graphics card.

The graphics card is primarily designed to remove the graphical processing tasks from the processor or RAM.The dedicated GPU and a dedicated RAM that helps processing graphical data quickly. The Graphics card has a dedicated heat sink to keep the heat out of the GPU.

Types of Graphics Cards:

Integrated:

Graphics cards are built into the motherboard and no add in cards are used.

Discrete:

The add in graphics card is installed on the mother board as an extra component and the card can be upgraded as required.

CHT analysis is carried out when there is thermal interaction between solid and fluids. A fluid is convected across the unit. Heat is conducted through the fin cooling unit placed on the processor and base. The fluid flowing across the unit carries the heat away by convection.

Geometry:

The geometry of the graphics card is as seen in the figure below. It consists of a Base on which the GPU is mounted. On top of the GPU is a cooling unit with fins. An enclosure is created around the unit in order to create the computational domain for the fluid flow.

The card base and thecooling unit can be seen below:

The processor placed on the base unit can be seen below:

The different components are allowed to interact with each other by using the share option from the workbench tab in spaceclaim and the share topology is also activated for all interacting edges in order to generate a conformal mesh.

Meshing:

A coarse base mesh is first generated to run the base case. The simulation is run on the base case and the mesh is further refined.

The details of the two mesh are shown in the table below:

General Element Size (mm)	Number of elements
4 (Coarse)	116625
2 (Fine)	414841

The base mesh generated can be seen in the figure below:

For generation of the refinements in the mesh, along with changing the mesh elements, the multizone meshing method is employed on the Base as it is almost a rectangular structure and hex elements can be used. The refined mesh can be seen below.

From the observations taken on from the base case the fin portion is also refined by applying the body size option with the element size being 0.5mm while the external mesh element size is 2mm.

The named selection created are as follows:

Face Selections:

1. Inlet
2. Outlet
3. Symmetry
4. Processor_Walls
5. PCB_Walls
6. Fin_Walls

The named selections created are shown in the figure below:

Named selctions are also created for different bodies for easy identification in the case setup. They are as follows:

1. Processor_Body
2. Fin_Body
3. PCB_Body
4. Fluid_Enclosure

Case Setup:

A steady state case is setup in ANSYS Fluent. The different materials are assigned to the different bodies named.

The Fluid enclosure is selected to be fluid and the rest of the bodies are assigned as solid elements.

The materials and the details are tabulated below:

Name	Component	Density	Specific Heat	Thermal Conductivity
Air	Fluid Enclosure	1.225	1006.25	0.0242
FR4	PCB_Base	1910	600	0.35
Gold	Processor_Body	19320	12981	297.73
Aluminium	Fin_Body	2719	871	202.4

FR4 is a material that is commonly used in PCB\'s. It is made up of a thin layer of copper foil laminated on bith sides or on one side.

A heat generation source term is activated for the processor_Body.

The Volume of the processor is :

`V=8 times 8times1=64 mm^3=64 times10^-9 m^3`

If a 15W processor is chosen, the heat generated in the source term is computed as:

`H=15/(64 times 10^-9)=234375000 W/m^3`

The boundary conditions are selected as follows:

Inlet: Velocity Inlet V=10m/s

Outlet: Pressure Outlet

Symmetry: Symmetry

The rest of the named selections are given wall boundary conditions and the coupled thermal condition is used.

With the help of the surface definitions, Area weighted average is taken for the outlet velocity.

Initialisation Type: hybrid

Various contours are setup and animations are recorded for the contours. the simulation is carries out for 600 iterations.

The inlet velocity is then varied in steps and the change in Wall heat flux is also observed.

Results:

The residuals of the simulation are checked in order to check for convergence.

From the above plot it can be observed that all the residuals are below 1e-4 and are almost constant hence concluding that the simulation has converged.

The temperature contours for the coarse mesh can be observed in the figure below:

maximum heat is generated in the processor. This heat is conducted by the fin plate and a small portion is conducted through the FR4 Base. however since Aluminium is a better heat conductor most of the heat is conducted through the Aluminium cooling fins.

The temperature contours of the cooling fins can be seen below.

it can be observed that the portion of the fin that is in contact with the processor has the maximum temperature and the heat is then being conducted through the rest of the cooling fins. The heat from these fins, the base and the processor is also carried away by the citculating Ait. The fins are given to increase the heat transfer area between the solid and the fluids for better heat transfer. the contours of temeprature along the central plane of the computational domain can be seen below:

The contours can be seen to be highly diffusive near the walls. the following image shows the velocity contours along the midplane.

The velocity can contours are also observed to be diffusive.

The mesh is further refined as described in the meshing process. the mesh on the base of the PCB and the fins regions is refined for better results.

The temperature contours on the graphics card for the refined mesh can be observed in the figure below.

The heat generation through the processor anf the conduction through the PCB base can be seen above. The following image shows the temeprature contours generated on the fins.

Similar to the coarse mesh condition it can be observed that the portion of the fin that is in contact with the processor has highest temperature and that heat is being conducted through the entire fin element.This heat is in turn carried away by the flowing fluid.

On closer comparison of the fin regions for the coarse mesh in the following Image with the first figure of the fine mesh and the second figure of the coarse mesh:

It can be seen that the contours on the fine mesh fins are smoother in contrast to the abrupty changing contours obtained by using the coarse mesh.

The wall heat flux obtained for the coarse mesh is 183250 and for the fine mesh is197128 W/m^2.

The following figures show the temperature and velocity contours along the midplane.

The temperature are maximum for the processor and the cooling fin unit. The temperature of the base can be observed to be changing very slowly due to its low conductivity. the temperature downstream of the fin unit can be seen to higher than the surrounding temperature as the heat from the graphics card is carried away by the air.

The following contour shows the back portion of the PCB base in order to show the conduction of the heat.

It is observed that the region in contact with the processor has the highest temperature and the conduction of heat through the solid can also be observed.

The following image shows the velocity contours along the midplane

From the velocity contours it can be observed that most of the area in general has a uniform velocity. The regions adjacent to the graphics card walls have zero velocity. The velocity downstream of the graphics card is lower than the genral velocity due to the obstruction vreated by the graphics card for the fluid flow through the wind tunnel. The air downstream of the graphics card seems to have a higher velocity than the inlet as the density and viscosity of the air are lower than the ambient due the heating up of the air while taking away the heat from the graphics card. this hot air mixes with the incoming air leading to slightly higher temperatures of the air.

The inlet velocity of the air is varied and the effect of the flow velocity on the wall heat flux is observed and tabulated in the figure below for the same heat generation rate of the processor.

Fluid Flow Velocity	Wall Heat Flux (W/m2)
1	202127
2.5	202715
5	210005
7.5	198734
10	197128

It can be observed that the increasing fluid velocity causes a reduction in the heat flux as the convective heat transfer increases with increasing in velocity of forced convection. The outlet velocities of the different simulations can be seen below:

The velocities initialy jump to a higher value ans then attain a steady value.

The velocity contours alog the midplane for the different inlet velocities can be observed in the contour plots below:

Velocity Contour Inlet 1m/s

Velocity Contour Inlet 2.5m/s

Velocity Contour Inlet 5m/s

Velocity Contour Inlet 7.5m/s

Conclusion:

CHT analysis is carried out for the graphics card for two different mesh refinement levels.

The analysis is carried out for different flow velocities of the circulating fluid anf its influence on the heat flux is noted for the same heat generation rate of the processor.