Georgios Rousos

Since science was of interest to me, I first studied Material Science and then moved to Mechanical Engineering as my love of kinematics and materials was strong. However, in the last year of my Mechanical Engineering degree, I fell in love with fluid dynamics and computational modelling. A strong influence came from the professor and lecturer who taught me computational fluid dynamics, where I also learnt new techniques and used software such as SIEMENS STAR CCM+ to solve analytical problems that revolved around CFD models. CFD modules intrigued me so much that I based my thesis around this topic. It was a great experience capturing data and a great feeling when my thesis was completed. Since the completion of this piece of work, a future in the field of fluid dynamics or numerical solutions around fluids is something I am extremely looking forward to.

Numerical Analysis of Heatsinks at rough fins with forced convection

Project introduction

Processors are getting more and more powerful with smaller lithographic nodes. Various studies refer to better thermal management of computer hardware cooling by the development of better heatsinks.

Microchannel extruded fin heatsink efficient convective cooling during high CPU operations without changing the overall heatsink geometrical design.

Previous research studies found that the overall surface roughness of heatsink fins tend to produce more efficient thermal management. In order to design better fins, a few modern techniques and expensive machines were used for the studies

Ventola¹ et al. emphasized a researched study of surface rough components to improve the efficient of electronic devices cooling, and thermal convective heat transfer coefficient overall. However, a more sophisticated and interesting approach was pursued. For this, the researchers applied Direct Metal Laser Sintering (DMLS) at the plane of the fins. By the manufacturability of the DMLS, artificial surface roughness was achieved and studied scientifically. Therefore, numerically, a baseline smooth surface model with two unique modified baseline models were compared based on the DMLS manufacturability.

Motivation

A wide range of applications are available for the products of heat sinks. Recently the most exponential increase is based on the conventional electronic appliances and computer systems. More analytically, a few of the applications based on the electronic and electrical applications are:

Micro heatsinks

• GPUs (Graphical Processing Units)
• CPUs (Central Processing Units)
• Other electronic circuit boards with microchips
• Appliances that require thermal management

Macro heatsinks

• Electrical power poles
• Refrigerator heatsinks
• Any other macro device that requires thermal management

Aims

The overall aim is to produce a more efficient heatsink by comparing different types of modified heatsinks with the same extruded fin design. As microprocessors have been getting hotter in recent years, more efficient heatsinks are needed with new design methods.

Objectives

• To produce rough surface fins and study the flow near the microchannels
• Thermally enhance the overall maximum temperature near the fins
• Design a higher velocity profile flow at the microchannel fins
• Apply different types of rough surfaces on the same heatsink geometry without extrusion of the porosity hole
• Compare the different porosity sites (small, large and variable porous sites) under the different fin conditions for the improvement of thermal transfer rate

Theoretical and numerical models

The general and theoretical formulas can be interpreted into single governing equations derived from numerical turbulent complex mathematical, and physical expressions for the foundational operations of heat transfer and convective heat transfer.

The governing equations for a duct on a microchannel is interpreted numerically as TimeAveraged Navier-Stokes Equation (RANS) based models on rectangular coordinates.

Computational Modelling

• Incompressible air of forced heat convection, with a Ma < 0.3, where by assumption; u = 1m/s
• Steady-state turbulent model (non-time dependent)
• Incompressible flow (Ma ≈ 0.3)
• K-E turbulent model RANS two-equation (time-averaged)
• Forced heat convection Negligible radiation effects due to very small Ma number
• Negligible natural convection effects
• A computational mesh independence modelling was evaluated with a volumetric control to refine the elements at the desired area near the fins

Conclusions

• The three modified baselines (MSB) models obtained a lower temperature overall, with higher velocity flow profiles
• Turbulence Kinetic Energy (TKE) was increased on the MSB models
• Moreover, the models were all compared and evaluated as the SB model was referenced for the decrease in temperature percentage calculation regarding the 3 MSB model types at the side view of the heatsink, front and top view, achieving a lower degree of temperature at all the cases compared to the standard SB model
• The overall efficiency was improved by the following ranking porous sites: variable > small > large porosity sites

Recommendations

• A lot of improvements could have been made to the fins while also eliminating a few mistakes during the CFD procedure setup and job process
• Various other studied recommend the process of rough porous site with extruded holes at the fins rather than non-extruded porosity. The further enhancement and study can be integrated into further optimization study regarding the project
• Furthermore, distinct porous shapes would be ideal to be studied and simulated under the extruded and non-extruded porous fin sites, with a comparison table and results

Download Georgios's project overview poster (PDF) (this pdf document is not fully accessible but the contents within it can be viewed via the content on this page)