Sources of Bias in Particle-Mesh Methods for PDF Models for Turbulent Flows
Cornell University Ithaca United States
Pagination or Media Count:
Numerical errors in particular bias in PDF-based particle mesh methods for turbulence modeling have been explored It is shown that bias decreases linearly with the increase of the number of particles but increases with grid refinement The fluctuations in mean fields which are fed back into the coefficients of stochastic differential equations are attributed to be the sources of bias. The Frozen Coefficient approach has been proposed and adopted to pinpoint the sources of bias in detail. These results provide guidelines for improving the numerical accuracy of PDF models for turbulent flows.