the ability of electronic devices to dissipate heat is an important aspect in electronics. electronic cooling represents a commercial sector where a variety of solutions has been explored [1]. skipping the very little performative solutions based on natural convection, thermal management of most of electronic devices, desktop and notebook computers to name a few, is based on forced convection of air as refrigerant, and usually rely on chip-attached or adhesively bonded extruded aluminum heat sinks [1, 2].
concerning cooling of high power devices as high performance supercomputers, high thermal fluxes have already driven the development of liquid-based cooling loops. however, despite higher heat fluxes expected by these solutions, it is difficult to imagine a wide spread of the latter technology in notebook computers, which will remain dominated by cooling based on forced air convection. thermal performance of the air cooled heat sinks must be further improved due to ever-increasing thermal challenges arising in next-generation electronics systems [3]. hence, in the present paper, we focused on forced (single-phase) convection only.
in particular we investigated the capability of direct metal laser sintering (dmls) manufacturing technique to produce micro-structured rough surface heat sinks [4]. usually heat sinks have been produced by milling manufacturing technique; surface roughness of those heat sinks is very low, around 1 micron. through dmls we are able to produce heat sinks whose surface roughness can vary over a wide range of values. this point is very important because the roughness strongly influences the convective heat transfer. the ability of roughness to enhance heat transfer in fully developed flow has been studied for many years. theory and experimental correlations have been developed for this phenomenon. early works focused on close packed, granular surface roughness [5, 6]. on the other hand, artificial-roughness-based solutions designed to enhance heat transfer (e.g. repeated ribs) have been studied [7, 8, 9]. in the latter cases, usually geometrical structured patterns are designed on the surface experiencing convective heat transfer. recently it has become evident that the roughness has an even greater impact at the leading edge (or entrance region) of heat transfer surfaces, where the thermo-fluid dynamic structures (namely velocity and thermal boundary layer) start growing and they are very thin; thus they are more sensitive to structures that modify the flow field. this is the case of electronic cooling applications, because in this field geometric dimensions involved are usually very small and therefore the fraction of heat transfer at leading edge is far from negligible. therefore we expect heat sinks produced by dmls have better performance in heat dissipation than traditional those produced by milling.
in this work, through an experimental analysis in a wind tunnel, we compared the thermal performance of surfaces obtained by traditional milling to those obtained by dmls. in particular two geometry will be tested, a flat heat sink and a finned heat sink (see figure 1). for each setup, the average convective heat transfer coefficient ? of two samples (with regards to the bare surface exposed to the air) was measured: a milled copper sample and an aluminum alloy (alsi10mg) sample produced by dmls.
click the more button for the entire aricle at iop science: More >>
|