CFD Project Lead - Chip Cooling

ME123 Computational Engineering, Stanford University

April 2019 - June 2019

As part of the computational engineering course, our team designed a water channel to cool two semiconductor computer chips. We selected geometry and inlet pressure to minimize the temperature of the chips. This final project combined material, thermal, and fluid simulation elements in Ansys AIM to optimize the channel geometry, and the results were validated with hand calculations.

The script for parameterized optimization included the following parameters:

1) Channel Cross Section Radius

The cross-sectional radius determined the surface area for heat transfer between the water and the surrounding metal. We wanted to minimize energy loss from surface friction, especially from any uneven geometries within the channel.

2) Channel Curvature Length

While having a straight channel is the simplest solution, curving the channel allows for more direct heat transfer from the high temperature computer chips to the cooling water. The final channel curve slightly overextends past the chips to increase the contact of the fluid over the points of highest temperature.

3) Inlet pressure

The inlet pressure affects the flow rate and therefore the amount of heat carried away by the liquid. The pressure drop across the channel had to maintain a high enough outlet pressure to flow through the connecting system.

4) Geometry Meshing

While the mesh resolution does not affect the results after convergence is achieved, the mesh size for different areas of the part greatly affected the simulation run time. The initial converging mesh size cost 12 hours to run whereas the final mesh ran at 30 minutes each. We adjusted boundary conditions and reduced the mesh resolution in less significant areas.

# computational fluid dynamics, thermofluids, ansys AIM