In this study, the probabilistic, data driven nature of the generative adversarial neural networks (GANs)was utilized to conduct virtual spray simulations for conditions relevant to aero engine combustors. Themodel co...In this study, the probabilistic, data driven nature of the generative adversarial neural networks (GANs)was utilized to conduct virtual spray simulations for conditions relevant to aero engine combustors. Themodel consists of two sub-modules: (i) an autoencoder converting the variable length droplet trajectories intofixed length, lower dimensional representations and (ii) a Wasserstein GAN that learns to mimic the latentrepresentations of the evaporating droplets along their lifetime. The GAN module was also conditioned withthe injection location and the diameters of the droplets to increase the generalizability of the whole framework.The training data was provided from highly resolved 3D, transient Eulerian–Lagrangian, large eddy simulationsconducted with OpenFOAM. Neural network models were created and trained within the open source machinelearning framework of PyTorch. Predictive capabilities of the proposed method was discussed with respect tospray statistics and evaporation dynamics. Results show that conditioned GAN models offer a great potentialas low order model approximations with high computational efficiency. Nonetheless, the capabilities of theautoencoder module to preserve local dependencies should be improved to realize this potential. For the currentcase study, the custom model architecture was capable of conducting the simulation in the order of secondsafter a day of training, which had taken one week on HPC with the conventional CFD approach for the samenumber of droplets (200,000 trajectories).展开更多
文摘In this study, the probabilistic, data driven nature of the generative adversarial neural networks (GANs)was utilized to conduct virtual spray simulations for conditions relevant to aero engine combustors. Themodel consists of two sub-modules: (i) an autoencoder converting the variable length droplet trajectories intofixed length, lower dimensional representations and (ii) a Wasserstein GAN that learns to mimic the latentrepresentations of the evaporating droplets along their lifetime. The GAN module was also conditioned withthe injection location and the diameters of the droplets to increase the generalizability of the whole framework.The training data was provided from highly resolved 3D, transient Eulerian–Lagrangian, large eddy simulationsconducted with OpenFOAM. Neural network models were created and trained within the open source machinelearning framework of PyTorch. Predictive capabilities of the proposed method was discussed with respect tospray statistics and evaporation dynamics. Results show that conditioned GAN models offer a great potentialas low order model approximations with high computational efficiency. Nonetheless, the capabilities of theautoencoder module to preserve local dependencies should be improved to realize this potential. For the currentcase study, the custom model architecture was capable of conducting the simulation in the order of secondsafter a day of training, which had taken one week on HPC with the conventional CFD approach for the samenumber of droplets (200,000 trajectories).
