Experiments were performed in which melting of a phase-change medium occurred in a closed vertical tube which was rotated about a vertical axis colinear with that of the tube. The melting was initiated and maintained by a step-change increase in the wall temperature of the containment tube. During the course of the experiments, parametric variations were made in the rotational speed, in the temperature difference which drives the melting, and in the duration of the melting period. The phase-change medium was 99% pure n-eicosane paraffin with a melting temperature of 36.3°C. It was found that rotation gave rise to considerably more rapid melting than that for no rotation, with the time required to achieve a given amount of melting being halved due to rotation. The rate at which energy could be stored was also significantly increased by rotation. Furthermore, at any duration of the melting period, the shape of the unmelted solid differed markedly in the presence or absence of rotation, being either straight-sided or sloped-sided. The melted mass results for all of the investigated conditions were very tightly correlated in terms of the Froude, Stefan, Grashof, and Fourier numbers.

Experiments were performed to study the fluid flow and heat transfer characteristics for turbulent airflow in a tube in which there is a decaying axisymmetric swirl. Measurements were made of the local swirl angle at the tube wall and of the local Nusselt number and friction factor, all as a function of position along the length of the tube. Supplementary flow visualization experiments were performed to establish the axisymmetry of the swirl and to explore the pattern of fluid flow at the inlet of the tube. The swirl angle was found to decay exponentially along the tube, with the decay being more rapid at lower Reynolds numbers. The swirl gave rise to substantial heat transfer enhancement in the initial portion of the tube. The enhancement prevailed over a greater length of the tube at higher Reynolds numbers than at lower Reynolds numbers. Compared with the enhancements encountered in the conventional thermal entrance region in a nonswirling pipe flow, those associated with swirl are substantially greater and longer lived.

Quasi-local heat transfer coefficients were measured on a cylinder on which a circular jet impinged in crossflow. The jet diameter and the distance between the initiation of the jet and the cylinder surface were varied parametrically, as was the jet Reynolds number. The measurements showed that the axial distribution of the heat transfer coefficient peaked at the impingement point. For a fixed jet diameter and Reynolds number, the peak heat transfer coefficient increased as the distance between the initiation of the jet and the cylinder surface decreased. The peak coefficient also increased with decreasing jet diameter at fixed values of the jet initiation distance and the Reynolds number. User-oriented correlations of the peak Nusselt number with the geometrical parameters and with the Reynolds number were derived. The dropoff of the heat transfer coefficient with increasing axial distance from the impingement point was more rapid for smaller jet initiation distances (at a fixed jet diameter and Reynolds number) and for smaller jet diameters (at a fixed initiation distance and Reynolds number). The impingement pattern was examined by flow visualization.

A multi-faceted experimental investigation has been carried out to study heat transfer and pressure drop for airflow in arrays of heat generating rectangular modules deployed along one wall of a flat rectangular duct. Experiments were performed with fully populated arrays, arrays in which there are missing modules, arrays where barriers are implanted to obtain heat transfer enhancement, and arrays in which there is both a missing module and a barrier. For the fully populated array without barriers, row-independent (fully developed) heat transfer coefficients were encountered for the 5th and all subsequent rows. When there is a missing module in the array, the heat transfer coefficients at neighboring modules are increased, with the greatest enhancement (about 40%) occurring when the missing module is just upstream of the module of interest. The enhancement due to side-by-side pairs of missing modules differs very little from that induced by a single missing module. The implantation of a barrier in the array is shown to be an effective enhancement device, with the greatest effect (about a factor of two) being felt in the 2nd row downstream of the barrier but with residual enhancement persisting considerably farther downstream. Under some conditions, the enhancing effects of a missing module and a barrier were found to be mutually reinforcing. Pressure distributions were measured in arrays with and without barriers, and the barrier-induced pressure losses identified.

Heat transfer coefficients were determined experimentally for the cylindrical and base surfaces of a cup-like rotating cavity. The experiments encompassed three cavity length-diameter ratios and rotational Reynolds numbers which covered a fifteen-fold range. Two variants of the thermal boundary conditions were investigated. In one, both the cylindrical and base surfaces were thermally active, while in the other, heat transfer was confined to the cylindrical surface. The heat transfer coefficients for the cylindrical surface decrease with increasing cavity depth, the extent of the decrease being greater when there is no heat transfer at the base. The base-surface coefficients were found to be independent of cavity depth at the higher Reynolds numbers but decreased for deeper cavities at lower Reynolds numbers. In addition, at high Reynolds numbers, the base-surface coefficients were in excellent accord with those for an unshrouded rotating disk while at low Reynolds numbers they fell below the unshrouded-disk values. Excellent correlations of the data were achieved with power-law representations. Supplementary experiments involving a rotating disk showed that the accuracy of the data was in the 2–3% range or better.