Conjugate Heat Transfer Analysis of Graphic card

Aim:

Conjugate Heat Transfer Analysis of Graphic card

Objectives:

• To perform steady state conjugate heat transfer analysis of graphic card
• To understand the effects of air velocity on heat transfer
• To compare coarse and refined mesh results

Introduction:

In the graphic card problem, there is processor which uses the electrical power and convert some part of it in heat during the running. This heat may cause the problem so that the fin is provided at top of processor which transfer heat from processor to surrounding and reduces heating problem. There is also one base provided to support processor. In this simulation, enclosure is provided surrounding graphic card through which air flows. The air come in contact with fin and take heat from fin. In this study, the effects of air velocity on heat transfer will be studied.

Simulation Procedure:

Step: 1 Geometry Preparation

Open the Spaceclaim and select Share to share properties in model.

Figure: Share in Graphic Card Model

Step: 2 Meshing

In this study, there are two types of meshing such as Baseline and Refined are used. Both meshing are given in table.

1. Baseline Mesh:

In this meshing, named selections are given and base mesh of size 0.0015 m is generated which are displayed below:

Figure: Enclosure mesh

Figure: Graphic card mesh

Figure: Cross section of model

2. Refined Mesh:

In this meshing, individual body sizing method is selected and different sizing is given to individual components. The details of various sizing methods are given in table. The edge sizing is also given to base cylinders. Inflation is not used in this because it is not properly created because of complex model. If inflation is given to some individual components then also it is somehow improper or has very aspect ratio. Hence, inflation is not used in this study.

Figure: Enclosure mesh

Figure: Graphic card mesh

Figure: Cross section of model

Type of Mesh	Enclosure Size	Processor Size	Fin Size	Base Size	No. of Elements
Baseline	0.0015 m	-	-	-	575767
Refined	0.004 m	0.00025 m	0.0008 m	0.0008 m	537249

Step: 3 Set-up

In this model, there is a processor which is a source term because it generated heat continuously. If we consider at 64 mm3 by volume compared to generally available ones which are 300-400 mm3 we only consider 1/5th the power they consume. A normal high end GPU draws anywhere between 40 – 60 Watts for our case we have considered a power consumption of 9.6 Watts. Hence, the Energy generated with no losses comes out: 9.6 / 64*10-9 = 1.5 * 10^8 W/m3.

In this simulation, the following parameters are selected:

Type: Steady-state simulation

Solver: Pressure based

Velocity Formulation: Absolute Velocity

Energy Equation: Selected

Turbulence model: k-omega model (SST)

Cell zone conditions:

Processor: Gold (material) Source: 1.5* 10^8 W

Fin: Copper (material)

Base: Aluminum (material)

Inlet velocity: 1, 2 and 4 m/s

Inlet Temperature: 300 K

Initialization: Hybrid

Step: 4 Results

In this study, simulation run for air velocity of 1 m/s for baseline mesh and for air velocity of 1, 2 and 3 m/s for refined mesh. There are various plots such as residual plot, graphic card surface temperature plot and temperature plot on model, etc. are shown below and also compared to each other.

Baseline Mesh:

Air Velocity = 1 m/s

Figure: Residual Change

Figure: Temperature Change on graphic card

Figure: Y plus plot

Figure: Temperature contour on graphic card

Figure: Temperature Plot on plane

Figure: Velocity streamlines on plane

Refined Mesh:

Case: 1

Air Velocity = 1 m/s

Figure: Residual Change

Figure: Temperature Change on graphic card

Figure: Y plus plot

Figure: Temperature contour on graphic card

Figure: Temperature Plot on plane

Figure: Velocity streamlines in plane

Case: 2

Air Velocity = 2 m/s

Figure: Residual Change

Figure: Temperature Change on graphic card

Figure: Y plus Plot

Figure: Temperature contour on graphic card

Figure: Temperature Plot on plane

Figure: Velocity streamlines on plane

Case: 3

Air Velocity = 4 m/s

Figure: Residual Change

Figure: Temperature Change on graphic card

Figure: Y plus plot

Figure: Temperature contour on graphic card

Figure: Temperature Plot on plane

Figure: Velocity streamlines on plane

Some important parameters are obtained from study are given in below table:

	Baseline Mesh	Refined Mesh
		Case 1	Case 2	Case 3
Avg. Temperature	433.81 K	431.39 K	380.83 K	349.31 K
Avg. total Temperature	422.78 K	425.78 K	375.35 K	344.17 K
Heat transfer coefficient of processor	307.48 W/m^2 K	778.892 W/m^2 K	788.468 W/m^2 K	783.132 W/m^2 K
Heat transfer coefficient of fin	98.98 W/m^2 K	188.602 W/m^2 K	190.169 W/m^2 K	194.144 W/m^2 K
Heat transfer coefficient of base	110.69 W/m^2 K	217.989 W/m^2 K	216.904 W/m^2 K	222.292 W/m^2 K
Max. Temp. of Processor	438.47 K	437.206	388.353 K	354.502 K

All the plots which are displayed above are all converged.

Based on above plots and data table, baseline mesh is not giving good temperature distribution because it is not capable to capture good physics at interface zone which we can observe from above table.

We can see that when velocity increases from 1 to 3 m/s then it there will be slight increase in fin and base heat transfer coefficient. The processor lies between two components so that it will not cause major change in processor side heat transfer coefficient. But, fin side heat transfer is increasing which will transfer more from processor to surrounding because of increase in air velocity. Hence, the temperature of processor is decreasing for increase in velocity because of increase in air-fin and air-base heat transfer rate.

In this model, fin and parts of base at which capacitors are connected are most important parts because they are exposed to air and cooling of processor mainly depends upon these parts.

Conclusion:

Graphic card model is very complex so that inflation cannot be provided. Hence, it must be solved by the giving different body sizing to different components.

Baseline mesh is not capable to capture good physics which will give poor results and it can not be useful as per industry standards.

The fin material is selected in such way that it will provide high heat rate to surrounding. Suppose copper is used as fin material in this study.

The velocity change directly affects heat transfer rate in graphic card model. The velocity increase will increase heat transfer rate and cooling of processor and reduces its temperature.