The use of natural convection for achieving enhanced circulation (and hence more uniform heating) in enclosed spaces, such as exist in ovens and furnaces, or for cooling electronic components, reduces the necessity either for fans or recirculating pumps, which are expensive, require power inputs and increase maintenance. The primary aim of this investigation is to determine, by experimental means, the steady-state rate of heat transfer to an inner body and the associated temperature distributions within a natural-convection tunnel oven, with protruding heat sources.

Studies in raising the thermal performances and permissible rates of stock passing through ovens and furnaces have recommended enhancing the thermal insulation, (e.g. by reducing the surface emissivities) of the walls, as well as improving the heater configuration employed. Scheitlin and DeWitt found that lowering the emissivity of an oven’s walls increased significantly the efficiency of heating a standard test-block contained therein. However, when the findings were applied to a cubical oven, as tested by Scarisbrick et al., the quality of the finished product was unsatisfactory because hot-spots were produced on the surface of the test block. Kay concluded that, in general, heat tends to be transferred more uniformly (to the product being heated) by convection. This is the reason for trying to augment the proportion of the total heat-transfer occurring by natural convection in ovens. Appropriate developments have ensued as a result of the efforts of Beuker and Hurts and Brown for baking ovens, Royce for lehrs and industrial ovens and Read for furnaces.

Fundamental natural-convection investigations have been undertaken for simple geometrical configurations, which include uniform wall-heating, flush-mounted heaters and line heat-sources in enclosures, with or without an inner body present. For a uniformly-heated vertical wall of a rectangular enclosure, Hamady et al. correlated the results for the steady-state convective heat-transfer across the cavity by the equation Nu=0.175Ra^0.275 for 10⁴ < Ra < 10⁶. A numerical study by Chu et al. for a concentrated heat-source, flush-mounted horizontally along a vertical wall, concluded that the optimal configuration for maximising the natural-convection rate was with the flush-mounted heater located at 35% of the height of the vertical wall above its base. The local variation of Nu was then in the range 0.7 < Nu < 2.6 (for 10³ < Ra < 10⁵). Ekundayo subsequently corroborated that the optimal configuration to achieve the maximum steady-state rate of convection was with the heating element located in the lower half of the enclosure. In the experimental studies of Warrington and Crupper, the analysis of the heat transfer, from a set of four symmetrical, electrically-heated cylindrical pipes (as the inner bodies), into the enclosure was facilitated by evacuating the test rig to less than 20 torr, thereby obtaining the proportion of the heat transferred by solid conduction and radiation. The resulting correlation for the convective–conductive component through the air of the steady-state heat transferred was Nu=0.498Ra^0.245. However, as Zhao et al. subsequently established, this correlation was of limited applicability.

This consisted of the inner body mounted symmetrically within the enclosure, which had been designed to withstand evacuation. The enclosure could be fitted with heat-sources, located slightly proud of the vertical walls and the lower horizontal wall. The ensuing steady-state measurements were used to assess the steady-state convective heat-transfers for the three considered heater configurations, and to develop easily applicable heat-transfer correlations. The effects upon the inner body of the temperature distributions over the inner surfaces of the enclosure walls were also examined.

The 1700 mm long, square-sectioned enclosure was fabricated by fillet-welding four 350 mm× 1700 mm stainless-steel plates (type 304L), and, when fitted with the inner body of 175 mm× 175 mm square section, provided a length/annular gap aspect-ratio of 19. The design process, was guided by the analytical studies of Timoshenko on plates and shells, the commentary on the design guidance for non-circular cross-section pressure vessels, and BS5500 concerning the acceptable pressures for vacuum vessels. The enclosure plates were reinforced by various types of stiffeners. These permitted the use of a plate thickness of only 6.35 mm for the enclosure. However, mild-steel was utilised for the inner-body sidewalls and so, their thickness was increased to 12 mm in order to conform with the requirement for a mandatory corrosion-thickness. The enclosure ends were provided with bolt-on mild-steel cover plates.

The solid inner-body was maintained isothermal by circulating water (mixed with a corrosion inhibitor) through a series of pipes and baffles within it. The inlet and outlet “double-strength” steel pipes also supported the weight of the inner body via the enclosure supporting/carrier plates. A compressed-fibre sheet (170 mm× 145 mm× 5 mm) was glued to each end of the inner body, so that the inner body assembly almost touched the mild-steel enclosure end-plates, thus preventing longitudinal air circulation at the ends. Further details of the apparatus have been described by Ekundayo. The 1400 mm long heaters were suspended at their ends via sindanyo tubes, which acted both as electrical and heat insulators, thereby inhibiting significantly the rate of losses via solid conduction.

The enclosure was insulated externally with two layers of 50 mm thick Rockwool blanket (of effective thermal conductivity 0.058 (±0.020) W m^?1 K^?1 throughout the range 323 K <T < 523 K). The outer surface of the insulant was further covered in reflective aluminium-foil to inhibit radiative heat-losses.

The chiller closed-circuit consists of a refrigeration unit with a built-in process pump, which delivered cooled water to the inner body at 1.7 bar. Its temperature-control system was maintained within ±0.75 K of the set water-circulating temperature (321 K). This kept the inner-body’s surface temperature constant (at 323 K). Preliminary tests were carried out to investigate (i) the isothermality of the inner body, (ii) the uniformity of each heating-element’s surface temperature, (iii) the outputs of the thermo-junctions, (iv) the variation of wall emissivities with temperature, and (v) the effectiveness of the applied insulation. The horizontal offset of each sidewall-mounted heater from its vertical wall was 15 mm, measured from the closest part of the periphery of the cylindrical heater, whereas the vertical offset from the base of a base-mounted heater was 30 mm. The main tests were carried out with configurations designated as C1, C2 and C3: these are depicted with their respective flow-visualisations. The steady-state values were attained most rapidly with heater configuration C3, followed by C1 and most slowly by configuration C2.

Type-T thermo-junctions indicated the temperature in the water circuit, whereas type-K thermo-junctions were employed on the heaters and walls and in the air spaces. A Fluke 2201A data-logger recorded the outputs from the type-K thermo-junctions, a multimeter processed the water-turbine flow-meter observations, and a Newport TC2 data-logger handled the type-T thermo-junction signals. Regulated power at 235 (±0.02) V was supplied, via individually-dedicated variable transformers and wattmeters, to each of the heaters.

The degree of isothermality of the inner body was ascertained from 10 strategically placed thermo-junctions. Each of the enclosure walls was divided into 9 × 2 symmetrical rectangular zones, thermo-junctions being positioned, at their approximate centroids, between the steel enclosure and the insulant. The thermo-junctions installed corresponded in distribution with the embedded thermo-junctions, but their signals were given precedence because spurious readings occurred occasionally from those installed internally: 140 thermo-junction signals were recorded for each configuration, 81 of these were for the air-space temperatures. Each thermo-junction was spot-welded to give head diameters of less than 0.5 mm, and not shielded. A total of 20 thermo-junctions were installed on the traversing-thermo-junction rack to measure the air-space temperature distributions. The thermo-junctions fixed to the internal surfaces were attached by reflective aluminium tape.