Periodic Reporting for period 1 - URBANRAD (Radiative transfer effects on air pollution dispersion in urban areas: from the street scale to the neighbourhood scale)
Reporting period: 2015-06-01 to 2017-05-31
This project aims at understanding the effects of radiative transfer on air pollution dispersion in urban areas at both the street scale and the neighbourhood scale. Numerical models have been developed for that purpose, that will help design and manage our cities, buildings and traffic systems in order to produce sustainable, safer, healthier, and more comfortable urban environments. Radiation modelling have been coupled with fluid dynamics, pollutant transport and environmental conditions to produce the first model able to take into account radiation transport effects at the street scale. Numerical simulations at the street scale have been carried out and have shown substantial effects of radiative transfer in low-wind conditions, for a mid-latitude summer atmosphere. In addition, a novel strategy has been developed to improve the computational efficiency of thermal radiation transport calculations. It consists in optimising the angular resolution in space and for each absorption coefficient class to get the minimum error on the radiative source term. The use of such adaptive method will be key to perform numerical simulations at the neighbourhood scale.
In order to enhance the computational efficiency of radiative transfer calculations, a goal-based angular adaptivity method has been developed, which is suitable for non grey media and when the radiation field is coupled with an unsteady flow. The method optimises the angular resolution according to the radiative source term (the goal) which is the key physical quantity for the coupling between flow and thermal radiation. The angular resolution is allowed to vary anisotropically in space and for each absorption coefficient class built from a global model of the radiative properties of the medium. The angular discretisation is based on a Haar wavelet expansion, which is a hierarchical version of the discrete ordinates expansion and does not require any angular interpolation in space when adapting. The method has been tested on a coupled flow-radiation on the 2D street canyon problem from a coupled instantaneous temperature field. Compared to a uniform Haar wavelet resolution, a goal-based adapted resolution requires 5 times less angular basis functions and 6.5 less CPU time to reach the same level of accuracy in the radiative source term. In addition, the adapted resolution proved to be accurate when used in unsteady coupled calculations. The gain in computational time will help to investigate larger computational domains and more realistic flow regimes that involves a larger range of temporal and spatial scales. This work has been accepted for publication in JQSRT and has also been presented in a thermal radiation symposium in Marseille.
The angular adaptivity method developed in this project is expected to have a strong impact in the field of radiative transfer modelling for both research and industry. Compared to the state of the art finite volume/discrete ordinates method, adaptivity is potentially much more efficient in terms of computational time and memory requirements for a given accuracy. Many application areas are concerned, like atmospheric sciences but also nuclear engineering or combustion. In future work, we plan to further enhance the computational efficiency of the adapted calculations by performing load balancing in parallel after adapting, to ensure an even distribution of work. Another perspective would be to extend the application of the method to heterogeneous media that are encountered for instance in combustion processes.