Numerical oscillations Roe solver shallow water equations

RicardoGoncalves · February 21, 2012, 05:31

Dear all,
I am using a finite volume shallow water model which uses the Roe Riemann solver. The scheme is only of 1st order and is applied to a structured rectangular grid, but I am still getting some nasty oscillations when simulating flows over a rather complex topography.
As far as I understand, this first order scheme should be stable (although highly diffusive); so my first guess for the cause of the oscillations is the treatment of the source terms (in this case bed topography).
Does anyone have a better guess/solution for this problem?
Thanks

Fabian82 · March 5, 2012, 03:03

Dear RicardoGoncalves

Your problem is most likely be related to the source term. When the discretization errors of the Roe solver and the source term are not compatible, errors and noise may grow in time without any physical forcing. Your solver will therefore not be able to maintain a 'quiet' water state over a complex topography. A possible solution to this problem are so-called 'well-balanced schemes' which can be found in numerous formulations in literature, e.g. by modifying the interface reconstruction or by upwinding the source term at the interfaces.

Best wishes
Fabian

February 21, 2012, 05:31	Numerical oscillations Roe solver shallow water equations	#1
RicardoGoncalves New Member Join Date: Jul 2010 Posts: 19 Rep Power: 15	Dear all, I am using a finite volume shallow water model which uses the Roe Riemann solver. The scheme is only of 1st order and is applied to a structured rectangular grid, but I am still getting some nasty oscillations when simulating flows over a rather complex topography. As far as I understand, this first order scheme should be stable (although highly diffusive); so my first guess for the cause of the oscillations is the treatment of the source terms (in this case bed topography). Does anyone have a better guess/solution for this problem? Thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Numerical oscillations in LES due to mesh refinement	Soder	OpenFOAM	0	June 14, 2011 10:10
Vorticity in inviscid shallow water equations	sonsiest	Main CFD Forum	0	June 7, 2011 14:10
roe solver with entropy fix	mehdi	Main CFD Forum	1	February 26, 2006 05:36
Navier-stokes equations and iterative solver??	wuliang	Main CFD Forum	2	January 13, 2003 22:28
multi-layer shallow water or Sigma coordinate?	Hadian	Main CFD Forum	0	August 8, 2002 09:53

March 5, 2012, 03:03		#2
Fabian82 New Member Join Date: Mar 2012 Posts: 8 Rep Power: 14	Dear RicardoGoncalves Your problem is most likely be related to the source term. When the discretization errors of the Roe solver and the source term are not compatible, errors and noise may grow in time without any physical forcing. Your solver will therefore not be able to maintain a 'quiet' water state over a complex topography. A possible solution to this problem are so-called 'well-balanced schemes' which can be found in numerous formulations in literature, e.g. by modifying the interface reconstruction or by upwinding the source term at the interfaces. Best wishes Fabian