[CvGmt News] [Notiziario] [Settimanale] Avviso di Seminario corretto

[Giulia Curciarello]

Si avvisa che Mercoledi' 3 dicembre alle ore 15 30, presso la sala
seminari del  Dipartimento di Matematica Applicata (V. Bonanno)

il Prof. Volker John (Otto-von-Guericke-Universitaet, Magdeburg)

terra' un seminario


Titolo: Analytical and numerical aspects in the simulation of turbulent
flows with Large Eddy Simulation and Variational MultiScale methods


Abstract:
The space averaged Navier-Stokes equations are the basic equations of LES.
In the first part of the talk, analytical studies of a so-called
commutation error will be presented. This error occurs in the derivation
of the space averaged Navier-Stokes equations in a bounded domain. We will
present models for the Reynolds stress tensor which are based on an
approximation of the Fourier transform of the filter function.  Next, the
discretization of the LES models with second order implicit
discretizations in time and with finite element discretizations in space
is described. The solution of the discrete problems requires the
solution of large  linear saddle point equations. We will describe a
multilevel approach which uses stable non-conforming finite element
discretizations on the coarser levels of the grid hierarchy. A numerical
study of the considered LES  models will be presented.

The problems of classical LES in a bounded domain  motivate a different
definition of the large scales, namely be projection into subspaces as in
the variational multiscale method (VMS). The second part of the talk
presents a new type of VMS where the fluctuations (small scales)
can cross mesh cell boundaries. A finite element error analysis of this
method is given, its implementation is discussed and a numerical study
will be presented

_______________________________________________
Settimanale mailing list
Settimanale at mail.dm.unipi.it
https://mail.dm.unipi.it/mailman/listinfo/settimanale
_______________________________________________
Notiziario mailing list
Notiziario at mail.dm.unipi.it
https://mail.dm.unipi.it/mailman/listinfo/notiziario