An adaptive multilevel wavelet framework for scale‐selective WENO reconstruction schemes

Abstract

Summary We put forth a dynamic computing framework for scale‐selective adaptation of weighted essential nonoscillatory (WENO) schemes for the simulation of hyperbolic conservation laws exhibiting strong discontinuities. A multilevel wavelet‐based multiresolution procedure, embedded in a conservative finite volume formulation, is used for a twofold purpose. (i) a dynamic grid adaptation of the solution field for redistributing grid points optimally (in some sense) according to the underlying flow structures, and (ii) a dynamic minimization of the in built artificial dissipation of WENO schemes. Taking advantage of the structure detection properties of this multiresolution algorithm, the nonlinear weights of the conventional WENO implementation are selectively modified to ensure lower dissipation in smoother areas. This modification is implemented through a linear transition from the fifth‐order upwind stencil at the coarsest regions of the adaptive grid to a fully nonlinear fifth‐order WENO scheme at areas of high irregularity. Therefore, our computing algorithm consists of a dynamic grid adaptation strategy, a scale‐selective state reconstruction, a conservative flux calculation, and a total variation diminishing Runge‐Kutta scheme for time advancement. Results are presented for canonical examples drawn from the inviscid Burgers, shallow water, Euler, and magnetohydrodynamic equations. Our findings represent a novel direction for providing a scale‐selective dissipation process without a compromise on shock capturing behavior for conservation laws, which would be a strong contender for dynamic implicit large eddy simulation approaches.

Hybrid soft sensor modeling for bisphenol‐A synthesis reaction process

“…As long as the estimated initial concentrations and reaction rate constants are obtained, the forth‐order Runge–Kutta method is employed to solve the mechanism model of Equation (13). Thus, the hybrid model of the industrial BPA synthesis process is developed and can be adopted to estimate the outlet concentrations of key components.…”

A modular massively parallel computing environment for three-dimensional multiresolution simulations of compressible flows

“…Harten's algorithm has been further augmented to also save memory, e. g. , by the introduction of flux functions operating on non-equidistant points [17] or by introducing graded-tree concepts for hyperbolic [18] and parabolic equations [19]. For the latter approaches, various modifications have been presented, e. g., using a binary tree structure for n-dimensional meshes [20] or linking of reconstruction-stencil weights to the compression parameters [21]. Within the MR algorithm, further compression can be achieved via LTS schemes.…”

A localised dynamic closure model for Euler turbulence

“…In order to do the regularization in simulations of shock dominated flows, a certain amount of artificial dissipation is added in the numerical scheme which should be sufficient enough to capture shocks but not too much dissipative to damp the small-scale turbulent eddies. For last few decades, lots of successful numerical techniques have been developed with a virtue of adding the artificial dissipation locally near the region of discontinuity while preserving the turbulent characteristics in smooth flow regions (e.g., see Maulik, San, and Behera (2018) and references therein).…”