Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
7156149 | Computers & Fluids | 2018 | 6 Pages |
Abstract
This work introduces a new error indicator which can be used to determine areas of insufficient numerical resolution in unfiltered finite difference simulations. The background behind the methodology is that smaller scales (i.e. the flow features with higher wave numbers) are physically characterised by a smaller energy content in comparison with larger scales. This energy should decrease with increasing wavenumber at a minimum rate; if this rate is not attained it likely means that the smaller scales are not being properly resolved on the computational grid of solution points. An approach using spectral techniques is used to formulate two varieties of the error indicator - one integer-valued and one floating point-valued. These values are computed at a finite number of 'blocks' which span the domain. The indicator is implemented within the OpenSBLI finite difference-based modelling framework, and evaluated in the context of a three-dimensional Taylor-Green vortex problem and flow past a V2C laminar flow aerofoil.
Related Topics
Physical Sciences and Engineering
Engineering
Computational Mechanics
Authors
Christian T. Jacobs, Markus Zauner, Nicola De Tullio, Satya P. Jammy, David J. Lusher, Neil D. Sandham,