The Cavity Width Effect on Immersion Cooling of Discrete Flush-Heaters on One Vertical Wall of an Enclosure Cooled from the Top

Abstract

The width effect on natural convection heat transfer due to discrete flush-heated sections of equal height in an enclosure cooled from the top is experimentally investigated. Five heated sections are uniformly distributed along a vertical side wall, where the height of the unheated sections is equal to that of the heated sections. All other vertical surfaces and the bottom plate are insulated. The experiments are conducted for six values of cavity width resulting in a variation in the cavity height-to-width ratio (aspect ratio) of 3.67 to 12.22. Ethylene glycol and FC-75 (a dielectric fluid) are used as the convective media. The flow visualization results with ethylene glycol reveal a fairly inactive core flow at low power inputs and small cavity width. Higher values of the power input and cavity width transforms a rather well structured core flow into a time dependent one with higher horizontal velocities toward the “hot” wall. For a given cavity width, it is found that the heat transfer results for all the heated sections can be unified and presented by a single correlation through use of a local height as the length scale. The local height of a given heated section is measured from the bottom of the cavity to the mid-height of that section. Based on the local height length scale the data for all the cavity widths are correlated and an explicit relation for the aspect ratio effect on local Nusselt number is reported. The data indicate that an increase in the cavity width results in an increase in the heat transfer coefficients of the heated sections. A comparison between ethylene glycol and FC-75 revealed no appreciable Prandtl number effect, which is in agreement with a previously reported prediction. A heat transfer coefficient for the top surface is defined based on the total convective heat flux at this surface and an average temperature difference between the heated sections and the top plate. The results show that for a given heat flux at the top, the highest heat transfer coefficient on this surface is obtained with the lowest cavity width.