Forced convection cooling of a simulated array of card-mounted electronic components has been investigated. An important feature of the simulated components is their relatively low profile (height/length = 0.058). Laboratory measurements of heat transfer rates resulting from convective air flow through a low aspect ratio channel are reported. The effect of variations in array position, channel spacing and flow rate is discussed. In the flow range considered laminar, transitional and turbulent heat transfer behavior have been observed. The behavior due to variations in flow rate and channel spacing is well correlated using a Reynolds number based on component length.

Forced convection air cooling of an array of low profile, card-mounted components has been investigated. A simulated array is attached to one wall of a low aspect ratio duct. This is the second half of a two-part study. In this second part the presence of a longitudinally finned heat sink is considered. The heat sink is a thermally passive “flow disturbance”. Laboratory measurements of the heat transfer rates downstream of the heat sink are reported and compared with the measured values which occur when no heat sinks are present. Data are presented for three heat sink geometries subject to variations in channel spacing and flow rate. In the flow range considered laminar, transitional and turbulent heat transfer behavior has been observed. The presence of a heat sink appears to “trip” the start of transition at lower Reynolds numbers than when no heat sinks are present. A Reynolds number based on component length provides a good correlation of the heat transfer behavior due to variations in flow rate and channel spacing. Heat transfer is most strongly effected by flow rate and position relative to the heat sink. Depending on the flow regime (laminar or turbulent) both relative enhancement and reductions in the component Nusselt number have been observed. The impact of introducing a heat sink is greatest for flow rates corresponding to transitional behavior.

An experimental study of forced convection heat transfer is reported. Direct air cooling of an electronics packaging system is modeled by a channel flow, with an array of uniformly sized and spaced elements attached to one channel wall. The presence of a single or complete row of longitudinally finned heat sinks creates a modified flow pattern. Convective heat transfer rates at downstream positions are measured and compared to that of a plain array (no heat sinks). Heat transfer rates are described in terms of adiabatic heat transfer coefficients and thermal wake functions. Empirical correlations are presented for both variations in Reynolds number (5000 < Re < 20,000) and heat sink geometry. It is found that the presence of a heat sink can both enhance and degrade the heat transfer coefficient at downstream locations, depending on the relative position.

The effect of variations in stream-wise spacing and component length on convection from rectangular, surface mounted components in a channel flow are reported. Component dimensions are the same order of magnitude as the channel wall-to-wall spacing. The channel Reynolds number, with air as the coolant, ranged from 670 to 3000. Flow visualization showed that under the above conditions the channel flow is transitional. The effect of variations in component stream-wise spacing on the level of turbulence in the channel and on the interaction between the core of the channel flow and the recirculating flow in cavities between components is discussed. Pressure drop measurements show that the dominant loss mechanism is due to form drag caused by the components. Local heat transfer measurements are made using an interferometer. Analysis of the results shows that the overall heat transfer is properly correlated in terms of a flow Reynolds number based on the component length. At small component Reynolds number, the overall conductance tends towards the laminar smooth wall value. An overall correlation is proposed which includes the effect of component Reynolds number, channel wall-to-wall spacing, and component stream-wise spacing.