Radiative heat transfer in a rectangular enclosure with a two-layer thermal structure is analyzed using a neural-network-based exact approach. Results show that the heat transfer is sensitive to the three-dimensional geometry, the inhomogeneous and nonisothermal mixture conditions within the enclosure, and the nongray spectroscopic properties of the medium. The effect of system parameters such as geometric dimensions, mixture concentration, and temperature are highly nonlinear and cannot be readily correlated by simple closed form analytical models. An existing radiation solver used in a two-zone compartment model, CFAST, is shown to be highly inaccurate; relative errors are shown to be greater than 100% for some cases.
Analysis of Radiative Heat Transfer in Inhomogeneous Nonisothermal Media Using Neural Networks
Journal of Thermophysics and Heat Transfer ; 30 , 4 ; 897-911
2016-10-01
Article (Journal)
Electronic Resource
English