False diffusion

January 23, 1999, 19:36

Hi I find a problem about the concept of False diffusion in CFD analysis. If any one know about this concept I will be appreciated. Thanks in Advance

January 23, 1999, 21:05

I am not too sure but I think you mean the numerical diffusion. If you try to solve a simple transport equation with numerical method such finite difference you will find that under some conditions related to the adimensional Courant number the transport term behaves like the diffusive term. You can see this effect by solving the simple monodimensional eqation of wave propagation with constant velocity c: df/dt=-c*df/dx if you use a simple forward differencing in time and backward in space (upwind) so that f(n+1)i=f(n)i-c*((f(n)i-f(n)i-1)/Dx)*Dt

you will find with that a simple triangular wave is translated but at the same time smoothed by a numerical diffusion effect that occours when the courant number Cr=dt*c/dx is less than 1 for grater values the scheme is unsteady for Cr=1 there is no diffusion. Bye

January 25, 1999, 10:34

I believe the area you are interested in is mostly call numerical diffusion. Many peopled have looked at these numerical problems and proposed solutions for them. The following article is a one of the many references in this area:

"Why you should not use 'Hybrid' 'Power-Law' or Related Exponential Schemes for Convective Modelling-There are Much Better Alternatives" Internationa Journal for Numerical Methods in fluids, Vol 20, 421-442 1995

drishty · August 7, 2009, 13:43

False diffusion oftens arises when you are using an upwind differencing method to discretise equations. Though upwind differencing provides results closer to analytical ones, when the flow is not aligned with grid lines, this differencing method causes the transported properties to be smeared. the resulting error has a diffusion like appearance and is referred to as false diffusion.

If you need any further inormation pls ask

Drishtysingh

mahdi_kh8 · August 7, 2013, 23:23

False diffusion is defined as an error having a diffusion-like appearance, obtained when the upwind scheme is used in multidimensional cases to solve for the distribution of transported properties flowing non-orthogonally to one or more of the system's major axes. The error is absent when the flow is orthogonal or parallel to each major axis.

January 23, 1999, 19:36	False diffusion	#1
Ehab Guest Posts: n/a	Hi I find a problem about the concept of False diffusion in CFD analysis. If any one know about this concept I will be appreciated. Thanks in Advance

January 23, 1999, 21:05	Re: False diffusion	#2
Luca Liberti Guest Posts: n/a	I am not too sure but I think you mean the numerical diffusion. If you try to solve a simple transport equation with numerical method such finite difference you will find that under some conditions related to the adimensional Courant number the transport term behaves like the diffusive term. You can see this effect by solving the simple monodimensional eqation of wave propagation with constant velocity c: df/dt=-cdf/dx if you use a simple forward differencing in time and backward in space (upwind) so that f(n+1)i=f(n)i-c((f(n)i-f(n)i-1)/Dx)Dt you will find with that a simple triangular wave is translated but at the same time smoothed by a numerical diffusion effect that occours when the courant number Cr=dtc/dx is less than 1 for grater values the scheme is unsteady for Cr=1 there is no diffusion. Bye

January 25, 1999, 10:34	Re: False diffusion	#3
Jeff Franklin Guest Posts: n/a	I believe the area you are interested in is mostly call numerical diffusion. Many peopled have looked at these numerical problems and proposed solutions for them. The following article is a one of the many references in this area: "Why you should not use 'Hybrid' 'Power-Law' or Related Exponential Schemes for Convective Modelling-There are Much Better Alternatives" Internationa Journal for Numerical Methods in fluids, Vol 20, 421-442 1995

August 7, 2009, 13:43	false diffusion	#4
drishty New Member drishty Join Date: Aug 2009 Posts: 1 Rep Power: 0	False diffusion oftens arises when you are using an upwind differencing method to discretise equations. Though upwind differencing provides results closer to analytical ones, when the flow is not aligned with grid lines, this differencing method causes the transported properties to be smeared. the resulting error has a diffusion like appearance and is referred to as false diffusion. If you need any further inormation pls ask Drishtysingh mahdi_kh8 and Teresa.Z like this. Last edited by drishty; August 7, 2009 at 15:16.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Moving mesh	Niklas Wikstrom (Wikstrom)	OpenFOAM Running, Solving & CFD	122	June 15, 2014 07:20
?? zero diffusion coefficient ??	ronen	FLUENT	0	April 11, 2011 14:02
implementation of diffusion coefficients and use of diffusion coefficients	stephan	OpenFOAM Running, Solving & CFD	0	February 28, 2010 15:30
False Diffusion	jj	Main CFD Forum	2	September 29, 2006 01:44
CFX4.3 -build analysis form	Chie Min	CFX	5	July 13, 2001 00:19

August 7, 2013, 23:23		#5
mahdi_kh8 New Member Mehdi Join Date: Aug 2013 Location: Australia Posts: 22 Rep Power: 13	False diffusion is defined as an error having a diffusion-like appearance, obtained when the upwind scheme is used in multidimensional cases to solve for the distribution of transported properties flowing non-orthogonally to one or more of the system's major axes. The error is absent when the flow is orthogonal or parallel to each major axis.