Thermal Analysis of An Aluminum Heat Sink

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 7

Thermal Analysis of an Aluminum Heat Sink

Brice Lu, Ben Stassen


Department of Mechanical Engineering, University of Saint Thomas

1. Introduction

The purpose of this study is to model the temperature of the fin tips of the X0952 aluminum heat sink. This particular heat sink is to be mounted directly on the center of a Northbridge computer chipset that operates at 40C. The heat sink is constructed from an aluminum alloy and there are no fans in this location of the enclosure so natural convection at room temperature is assumed. Due to the sensitive wiring located near the heat sink, the goal is to keep the tips of the heat sink below 30C. To determine if this part meets the specifications, hand calculations will be completed and then compared to FEA results from ANSYS to verify and improve predicted temperature values.

2. Variable Values

Aluminum Heat Sink Base Temperature (Tb): 40C Thermal Conductivity (k): 237.5 W/MC Fin Length (L): 2.464 cm Distance to Tip (x):2.464 cm Thickness (t): .165 cm

Air Properties Air Temperature (Tf): 22C Film Coefficient (h): 50W/m^2C

Chip Properties Chip Temperature (Tc): 40C

Figure 1: Geometric Variables in Respect to Geometry

3. Hand Calculations

To set a baseline range for expected numbers from the ANSYS simulation, hand calculations using the formula for a rectangular fin were used. It is important to note that these calculations make several assumptions about the geometry of the object. First, an even temperature distribution is assumed across the bottom of the heat sink. Second, the hand calculation models only one fin at a time. Third, the geometry of the fin is simplified, and treated as a rectangular fin.

( ) ( )

( ( )

Figure 2: Equation for Tip Temperature of a Rectangular Fin

( ) ( )

( ( )( (

) ) ) )(

Figure 3: Solved Equation

In this situation, the hand equation provides a reasonably accurate estimate of the actual temperature value. When compared to the value provided by the ANSYS simulation, it is demonstrated that the margin of error is approximately 0.65%. A percentage of error this small demonstrates that the excess surface area added by simplifications in the hand equation does not have a significant effect on the final temperature. In addition, the small margin of error provides support for our assumption that the fins located directly over the computer chip should be chosen as locations to poll for the highest temperatures.

4. ANSYS Simulation Results

The ANSYS simulation of the heat sink was designed to produce more refined results than those of the hand calculations. To make the simulation as realistic as possible, the geometry of the heat sink was fully modeled, including the spherical fin tips. In addition, the simulation contained an imprinted surface representing both the size and location of the chip in an actual

situation. Creating a model in this manner would allow for more accurate results and show more realistic temperature gradients across the entire model.

Figure 4: Imprinted face and resulting temperature gradient.

When viewing the overall temperature values across the heat sink, it can be observed that the largest fin temperatures occur at the tips of the fins that are directly above the chip itself. Therefore, in the processing of determining maximum tip temperature, we selected one of the four fins that lie directly above the hottest zone of the heat sink and applied a temperature probe to that location. This temperature value is what was considered to be the value for the max tip temperature, and the value for which mesh independency was computed. The most accurate temperature value that was computed at this location had a magnitude of 38.439C. This value is both in agreement with the value produced from the hand equations, and logically makes sense as this is an object with a high thermal conductivity and relatively short length, so temperature variance from tip to base should be relatively small.
4

Figure 5: Temperature Distribution across the Heat Sink

5. Mesh Independence

In order to verify the ANSYS results, a mesh independence chart was created to demonstrate how the accuracy of the results would improve with more complex meshing techniques. As seen in Figure 6, the default ANSYS mesh produced a result that was accurate to one tenth of a degree Celsius while only using about five thousand elements. By increasing element count accuracy, the magnitude of the temperature was able to be obtained to approximately one hundredth of a degree Celsius. This action came at the expense of the additional time needed to process the large amount of elements used in the mesh. Between 64,000 and 97,000 elements, the slope of the graph becomes very small, indicating that the model is nearing mesh independence. Unfortunately, there was difficulty generating any higher resolution than 97,000 elements due to technical constraints.
5

Figure 6: Final Mesh Profile

Mesh Independence
38.475 38.47 Max Tip Temperature C 38.465 38.46 38.455 38.45 38.445 38.44 38.435 0 20000 40000 60000 Element Count 80000 100000 120000

Figure 7: Mesh Independence of Tip Temperature

6. Conclusion

Based upon the results from the ANSYS model, this particular heat sink would not be suitable for use in this situation. Initial requirements stated that fin tip temperature could not exceed 30C due to the thermal sensitivity of adjacent wiring in this location, but the max fin temperature under these circumstances is over 8C higher than what is acceptable. A possible fix for the high temperature values may be to increase the film coefficient over the heat sink by increasing the flow rate of the fan. Other options may include using a less conductive material, or a longer fin, but these would respectively decrease the amount of heat energy transferred away from the chip or greatly increase the physical area needed to house the chip and heat sink.

You might also like