

Differences between air and surface temperatures measured and simulated were never above 1.5 ˚C and mean errors reached 0.5 ˚C. Temperature measurements by thermocouples and by thermal cameras have been compared to the models outputs.
USE GPU WITH PARTICLESHOP PATCH
The sun patch has been followed by a camera to validate its calculated position and surface. The in situ experiment has been carried out in one of the BESTlab cells (EDF R&D). The validation of the model has been done through a detailed comparison between model and measurements. The model has been developed for environmental conditions that vary over short time-steps and has integrated the projection of solar radiation through a window onto interior walls : the sun patch. This work describes a numerical model to simulate a single room, using a refined spatial three-dimensional description of heat conduction in the envelope but a single air node is considered.

Therefore, it is important to evaluate the performance of current transient thermal models when adapted to low energy buildings. Low energy building constructions become sensitive to internal gains : any internal heating source has an impact on the envelope. The test cases we used to validate our codes account for the strong potential of GPU LBM solvers in practice. In addition, we outline several extensions to the LBM, which appear essential to perform actual building thermo-aeraulic simulations. These contributions address the issues related to single-GPU implementations of the LBM and the optimisation of memory accesses, as well as multi-GPU implementations and the modelling of inter-GPU and internode communication. The present thesis consists of a collection of nine articles published in international journals and proceedings of international conferences (the last one being under review). For LBM, GPU implementations currently provide performance two orders of magnitude higher than a weakly optimised sequential CPU implementation. Yet, due to numerous hardware induced constraints, GPU programming is quite complex and the possible benefits in performance depend strongly on the algorithmic nature of the targeted application. These massively parallel circuits provide up to now unrivalled performance at a rather moderate cost. The use of graphics processors to perform general purpose computations is increasingly widespread in high performance computing. From an algorithmic standpoint, the LBM is well-suited for parallel implementations. It is therefore an interesting alternative to the direct solving of the Navier-Stokes equations using classic numerical analysis. The lattice Boltzmann method, which is based on a discretised version of the Boltzmann equation, is an explicit approach offering numerous attractive features: accuracy, stability, ability to handle complex geometries, etc. The present research work is devoted to explore the potential of such a strategy. The joint use of innovative approaches such as the lattice Boltzmann method (LBM) and massively parallel computing devices such as graphics processing units (GPUs) could help to overcome these limits. Resorting to computational fluid dynamics seems therefore unavoidable, but the required computational effort is in general prohibitive. However, for the time being, the thermo-aeraulic effects are often taken into account through simplified or even empirical models, which fail to provide the expected accuracy. With the advent of low-energy buildings, the need for accurate building performance simulations has significantly increased.
