The natural convection heat transfer across a cubical cavity capable of arbitrary orientation has been measured experimentally using a heat balance technique. This technique employs a highly sensitive heat flux meter for measuring the average heat transfer and a pressure/density method for varying the scaling parameter, the Rayleigh number. A full description of the technique, from a theoretical treatment to the design and calibration, and the development of a physical model for the cavity are presented.

With two opposing walls being differentially heated and with four other sidewalls having high conductivity to give a linear temperature profile, the cubical cavity has well defined thermal boundary conditions which allow nearly perfect matching between experimental and numerical thermal boundary conditions. By using air as the fluid, the Rayleigh number can be easily varied without varying the temperatures and length scales, and, because of widely available and relatively accurate air properties, the results from the numerical simulations should contain less error attributable to the uncertainties of air properties.

This thesis contains a total of 18 sets of experiments covering six Rayleigh numbers, viz. Ra = 1×10⁴, 4×10⁴, 1×10⁵, 1×10⁶, 1×10⁷ and 1×10⁸, and three inclination angles, viz. = 0°, 45° and 90°. Each set of experiments consists of six cases of different conditions from the combinations of three temperature differences, ΔT, and two mean temperature levels, T_m. The natural convection heat transfer was measured at the cold wall. These precise benchmark experiments are essential for validations of present and future computational-fluid-dynamics (CFD) codes which simulate natural convection flows. In this study, a commercial Finite-Volume based CFD code was used to simulate several experimental cases at a low Rayleigh number. The comparisons between the numerical and experimental results was done to validate the measurement system.