Abstract Details
Abstracts
Author: Diego Del-Castillo-Negrete
Requested Type: Consider for Invited
Submitted: 2024-04-02 16:52:56
Co-authors: B. Clavier, D. Zarzoso, E. Frénod
Contact Info:
Oak Ridge National Laboratory
PO Box 2008, MS6305
Oak Ridge, TN 37831
USA
Abstract Text:
Turbulent transport plays a key role in confinement degradation limiting the performance of current and future fusion devices. Modelling turbulent transport requires long- time simulations which are limited by the computational resources available. One way to overcome this shortcoming is by using surrogate models that are computationally cheaper to evaluate. In this presentation we apply generative artificial intelligence (AI) methods to construct a surrogate model of plasma edge turbulence described by the HW (Hasegawa-Wakatani) model and use the model to perform fast, long-time turbulent transport computations. The proposed GAIT (Generative Artificial Intelligence Turbulence) model is based on the combination of a convolutional variational autoencoder and a recurrent neural network (RNN). A convolutional network is used to encode snapshots of computed HW turbulence states into a reduce latent space, and a RNN is trained to reproduce the time evolution of turbulence in the latent space. Once the autoencoder is trained, new turbulence states are obtained by decoding the latent space dynamics generated by the RNN. The AI generation process is about 500 times faster than the direct numerical integration of the HW model. To test the fidelity of the method we use Eulerian and Lagrangian metrics. Good agreement is found between the GAIT and the HW models in the spatial and temporal Eulerian turbulence Fourier spectra. A distinctive feature of 2D turbulence is the formation and resilience of coherent structures e.g., vortices and zonal flows. The spatiotemporal multiscale properties of these structures can be analyzed using Proper Orthogonal Decomposition methods (POD), and evidence of the fidelity of the GAIT model is presented by comparing the POD spectra. In the Lagrangian setting, the statistical moments and probability distribution of particle displacements are compared, and agreement is found in the effective turbulent diffusivity in the GAIT and the HW models.
Comments: