Heat transport in turbulent Rayleigh-Benard convection: Effect of finite top- and bottom-plate conductivities

We describe three apparatus, known as the large, medium, and small apparatus, used for high-precision measurements of the Nusselt number N as a function of the Rayleigh number R for cylindrical samples of fluid and present results illustrating the influence of the finite conductivity of the top and bottom plates on the heat transport in the fluid. We used water samples at a mean temperature of 40 degrees C (Prandtl number sigma = 4.4). The samples in the large apparatus had a diameter D of 49.69 cm and heights L similar or equal to 116.33, 74.42, 50.61, and 16.52 cm. For the medium apparatus we had D = 24.81 cm, and L = 90.20 and 24.76 cm. The small apparatus contained a sample with D = 9.21 cm and L = 9.52 cm. For each aspect ratio Gamma equivalent to D/L the data covered a range of a little over a decade of R. The maximum R similar or equal to 1 x 10(12) with Nusselt number N similar or equal to 600 was reached for Gamma = 0.43. Measurements were made with both aluminum (conductivity lambda(p) = 161 W/m K) and copper (lambda(p) = 391 W/m K) top and bottom plates of nominally identical size and shape. For the large and medium apparatus the results with aluminum plates fall below those obtained with copper plates, thus confirming qualitatively the prediction by [Verzicco, "Effects of nonperfect thermal sources in turbulent thermal convection," Phys. Fluids 16, 1965 (2004)] that plates of finite conductivity diminish the heat transport in the fluid. The Nusselt number N-infinity for plates with infinite conductivity was estimated by fitting simultaneously aluminum- and copper-plate data sets to an effective power law for N-infinity multiplied by a correction factor f(X) = 1-exp[-(alpha X)(b)] that depends on the ratio X of the thermal resistance of the fluid to that of the plates, as suggested by Verzicco. Within their uncertainties the parameters a and b were independent of Gamma for the large apparatus and showed a small Gamma dependence for the medium apparatus. The correction was larger for the large, smaller for the medium, and negligible for the small apparatus. (c) 2005 American Institute of Physics.