Seven decades of supercritical heat transfer research have not seen an attempt to characterize turbulent thermal transport in internal flows subject to asymmetric heating. Existing heat transfer models therefore do not account for phenomena that dictates thermal transport, in the proximity of the pseudo-critical point, for non-uniform heating. Additionally, past investigations almost exclusively employed macroscale circular test sections and no systematic investigations were conducted to assess the influence of channel geometry on supercritical heat transfer. This dissertation addressed these shortcomings.
Thermal-hydraulic performance of supercritical carbon dioxide was characterized in non-circular microscale geometries ( 525.2 ≤ D_h ≤ 923 µm) for a wide range of experimental variables, including heat flux (5 ≤ q''≤50 W cm-2), mass flux (315 ≤G ≤2000 kg m- 2 s -1), and reduced pressure (1.03 ≤PR ≤1.1). In the two phases of the current investigation, the degree of asymmetry in applied heating was varied. In phase---I experiments, three non-uniformly heated stainless steel test sections, with different channel geometries, were employed. A data analysis technique, employing 2-D and 3-D heat transfer models of the test section, was developed to calculate the average heat transfer coefficients. Limiting the applied heating to the bottom wall of a microchannel based test section was the chief aim for phase---II experiments. This was achieved by using a combination of materials, Torlon 4203, Inconel-718 and stainless steel to build the test section. Infrared thermography was used to obtain local heat transfer data. For non-uniformly heated stainless steel test sections, using micro-pin based geometry and increasing channel aspect ratio led to higher heat transfer coefficients but with a commensurate penalty in pressure drop. Some existing heat transfer correlations were able to predict the heat transfer data from phase---I experiments with a MAPE of under 46%. However, when the applied heating was limited to the bottom wall, in phase---II, existing correlations proved to be inadequate. To address this, a reduced order predictive heat transfer model was developed. The model accounted for variable thermophysical property variations, flow acceleration, and unstable stratification effects. This model was able to predict 87.5% of the experimental data with a MAPE of less than 20 %.The tools and methodology developed in this dissertation can be used to design devices that will employ supercritical working fluids and will be subject to non-uniform heating.