CFD analyses of large-scale tests highlighting the effects of thermal radiation in steam-containing atmospheres

Abstract

In simulations of heat transfer in large gaseous volumes under moderate temperatures, thermal radiation is traditionally neglected to simplify the computations. In recent years however, some small as well as large-scale experiments have demonstrated that thermal radiation may have a strong effect on the temperature field if the atmosphere contains an infrared absorbing gas such as vapor, even in small concentrations. This has in particular been shown in the H2P2 series of tests performed in the large-scale PANDA facility at the Paul Scherrer Institute. For these tests an air-helium-steam mixture was compressed with a helium injection, which caused the formation of a hot bubble. The initial vapor volume fraction varied between 0.1 % and 60%. In this paper, we report the results of CFD simulations conducted for the five H2P2 tests. The turbulence model k-ω SST was used, while thermal radiation was considered using the simple P1 model as a default, in conjunction with the Weighted Sum of Gray Gases Model (WSGGM) to specify steam absorptivity. For comparison purposes we also made use of the more advanced Monte-Carlo model for selected experiments. CFD Best Practice Guidelines were employed, in particular through preliminary grid and time-step sensitivity studies.The temperature and helium concentration fields matched the experimental data quite accurately, except for the test with very low vapor content (H2P2_1_2, 0.1% vol. steam), where the temperature was somewhat over-estimated. This may be due to the inadequacy of the radiation models at these extreme conditions. Two major conclusions can be drawn from this study: 1) ignoring thermal radiation can lead to large over-predictions of temperature, even when the steam content is low (order of 1%) and 2) the simple P1 model provides sufficient accuracy. Use of the more sophisticated Monte-Carlo model yielded similar and slightly more accurate results, albeit at a considerably larger CPU expenditure (5–7 times).