Tensor Networking for PDF Simulation
Please login to view abstract download link
The PDF methodology [1] has proven very effective for RANS and LES of chemically reactive turbulent flows. Due to their inherent capabilities, the equation governing the transport of PDF is high dimensional. It is shown that “tensor networks” (TN) can deal with such a large dimensional transport in a very effective manner. These networks are employed here to tackle the Fokker-Planck form of the PDF instead of the Langevin form as is typically done via Monte Carlo methodologies. Demonstrative simulations are conducted of a chemically reactive flow, governed by a time-dependent five-dimensional modelled Fokker-Planck equation. [1] C. Dopazo, “Non-Isothermal Turbulent Reactive Turbulent Reactive Flows: Stochastic Approaches,” Ph.D. Thesis, Department of Mechanical Engineering, State University of New York, Stony Brook, NY (1973).
