Wind tunnel experiments were performed to determine the local response of the heat transfer at the outer surface of a longitudinal cylinder to geometry-related differences in the pattern of fluid flow. Among the three investigated configurations, one was a cylinder with an open bore through which fluid could pass, while in the second the bore was closed at its downstream end, creating an upstream-open-ended cavity. In the third configuration, the upstream face of the cylinder was impenetrable and blunt. The Reynolds number ranged from 7700 to 47,000. For all cases, the axial distribution of the Nusselt number was characterized by an initial increase, followed by the attainment of a maximum and a monotonic decrease, reflecting the occurrence of flow separation and of post-reattachment boundary layer development. The magnitude and location of the maximum were configuration dependent, with that for the open configuration being highest and occurring first, and that for the blunt face configuration being lowest and occurring last; the cavity configuration gave intermediate results. Upstream of the maxima, the Nusselt numbers were arranged in the same order as at the maximum, with a configuration-related spread of 50 percent. Well downstream of the maxima, the ordering was reversed and the spread was in the 5 percent range. Tight, configuration-independent correlations were achieved both for the maximum Nusselt number and for the Nusselt numbers in the downstream region.

The streamwise distribution of the Nusselt number within and just downstream of a region of flow separation displays a maximum whose location, relative to the point of reattachment of the flow, was investigated here. Wind tunnel experiments were performed in which a circular cylinder was oriented longitudinal to a uniform freestream. Due to flow separation at the outer rim of the forward face of the cylinder, the upstream portion of the cylindrical surface was washed by a zone of recirculating fluid. Three configurations of the longitudinal cylinder were investigated : one with a solid, blunt forward face, a second with a hollow bore open at both its upstream and downstream ends, and a third in which the hollow bore was closed at its downstream end. The blunt-face case was also solved numerically. Additional numerical solutions were carried out for the contrastingly different case of an abrupt enlargement in a parallel-plate channel. Both the experimental and numerical results provided conclusive evidence that the commonly assumed equality of the points of flow reattachment and maximum heat transfer coefficient was, at best, a special case. For most of the cases investigated here, the heat transfer maximum occurred upstream of the reattachment point. Factors influencing the relative positions of the maximum and the reattachment were identified.

A numerical investigation was performed to determine the heat /mass transfer from upstream-facing blunt faces of axisymmetric and plane two-dimensional bodies, which are situated such that a uniform flow approaches normal to the respective blunt faces. Face-average and stagnation-point Nusselt and Sherwood numbers, velocity and temperature profiles, and streamline maps are presented over the Reynolds number range from 5 × 103 to5 × 104. Average Nusselt number predictions for Pr = 0.7 (heat transfer in air) and average Sherwood numbers for Sc = 2.5 (naphthalene sublimation in air) agreed well with experiment. These results are well represented by a 0.5-power Reynolds number and 0.4-power Prandtl number fit. Stagnation-point Nusselt and Sherwood number predictions showed excellent agreement with boundary layer results using experimentally measured pressure distributions as input. Boundary layer results based on potential flow velocities as input overpredicted both the present numerical results and experimental results from the literature. The streamline maps and velocity profiles indicated that the boundary layer region on the blunt face was sensitive to the Reynolds number, while the outer region was not. A thinning of the boundary layer occurred with increasing distance from the stagnation point, which contrasts with the thickening that is characteristic of conventional stagnation flows. The thermal boundary layer thickness decreased with increases of both the Reynolds and Prandtl numbers.

The two-dimensional (i.e. radial and circumferential) heat transfer and fluid flow problem for a fluid-carrying, insulated horizontal cylinder which loses heat to air by natural convection was analyzed by solving the differential form of the conservation laws. The resulting conjugate problem encompassed conduction in the insulation layer and natural convection in the ambient air. A one-dimensional, radial heat flow model of the problem was also investigated in detail, and the circumferential-average natural convection heat transfer coefficients needed for its evaluation were respectively taken from the commonly used correlations of McAdams, Morgan, and Churchill and Chu. It was found that the correlation-related spread of the heat transfer results from the one-dimensional model was greater than the differences between the one- and two-dimensional results. The use of the Morgan correlation gave the most accurate set of one-dimensional heat transfer results (i.e. best agreement with the two-dimensional results). For the critical radius, the standard h0r∗kins = 1 criterion led to significant errors and should no longer be used. The critical radius results from the one-dimensional model, although correlation-dependent and deviant from the two-dimensional results, can be calculated efficiently and accurately from the criterion h0r∗kins = 3n(1 + n), where n is an exponent which can be determined for each specific Nusselt-Rayleigh correlation.

Heat transfer and pressure drop experiments were performed for cross-flow tube banks in which the individual tubes were equipped with longitudinal fins. The investigated geometrical parameters included the placement of the fins (at the front of the tube, at the rear, and at the front and rear), the fin tip shape (blunt or contoured), and the fin thickness. For each tube bank geometry, the Reynolds number was varied by nearly an order of magnitude. The results showed that a high degree of heat transfer enhancement can be obtained by finning, and the enhancements for the various tube bank geometries were compared at fixed pumping power, fixed pressure drop, and fixed mass flow. The finning-related enhancements were also compared with those attainable by the use of increased diameter unfinned tubes. Finning was found to be especially advantageous when the comparison is made at fixed pressure drop. For an array with fixed tube centers, finning permits greater additions of surface area than are possible by increases in the tube diameter.

In concentrated solar technologies, the use of volumetric receivers instead of surface receivers may help to achieve higher solar collection efficiencies. This paper reports a simplified model that computes the concentrated solar radiation absorption and energy transport by advection in the volumetric receiver of a linear Fresnel collector. The proposed linear Fresnel collector is intended for use in industrial solar heat processes. The receiver consists of a rectangular channel whereby flows a nanofluid composed of thermal oil and graphite nanoparticles. The bottom wall of the channel is a glass, while the upper wall is a highly reflective surface that is well insulated. The light absorption into the nanofluid was solved using the Rayleigh dispersion, while the convection was solved with a simplified velocity profile for turbulent flow (power law) and incorporating a radiative source term in the energy equation. All equations were discretized by using finite volumes and were coded in Python language. Receiver efficiencies in the range of 92–96% were found for temperatures of 403–343K. The results show that the volumetric fraction (optical depth) drives the way the radiation is absorpted, and in consequence, the temperature profile and the receiver efficiency. For low optical depths, the solar absorption is low, resulting in high reflected radiation losses. Contrarily, high optical depths mean high solar absorption, consequently, the temperature close to the glass increases, and the convection losses are higher. The results suggest that values of optical depth close to 2.3 are optimum to obtain the highest receiver efficiency.