ENTROPY-BASED CORRECTION OF FINITE-DIFFERENCE PREDICTIONS

Abstract
A method of improving the accuracy of finite-difference techniques using on entropy-based correction is presented. This method uses the results of a (non-entropy-based) finite-difference prediction to characterize the differential entropy transfers and generation. A system of second-law equations is subsequently solved to improve the initial finite-difference approximation. The increase in accuracy is particularly dramatic when fluctuations in the thermodynamic properties of the substances being considered are a dominant cause of discretization error. Entropy-based correction eliminates nonphysical solutions by ensuring that the nodal entropy generation rate is positive. This article describes entropy-based correction and presents two examples that illustrate the technique and demonstrate its effectiveness.

This publication has 1 reference indexed in Scilit: