|
[Sponsors] |
Scheme for partial differential equation: dissipativeness |
![]() |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
![]() |
![]() |
#1 |
New Member
Join Date: May 2022
Posts: 18
Rep Power: 5 ![]() |
Hi everyone,
I would like to investigate the numerical scheme effects for the discretization of Navier-Stokes equations with relative coarse grids. I often heard that the upwind schemes are more dissipative than centered schemes. Could you explain me why ? Is a 4th order centered scheme less dissipative than a 2nd order centered scheme ? Thanks a lot for your answers ! Boone |
|
![]() |
![]() |
![]() |
![]() |
#2 | |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,989
Rep Power: 73 ![]() ![]() ![]() |
Quote:
What you are asking is one of the basic paragraph in any CFD textbook. Have a look here https://www.researchgate.net/profile...+dynamics+.pdf |
||
![]() |
![]() |
![]() |
![]() |
#3 |
Senior Member
|
I'll play the devil against the Filippo effort to guide you into searching yourself the answer and just give it here for you to consume it. I assume we are talking about convection here:
- Upwind schemes are dissipative while, for uniform grids, centered schemes are not (like not at all). The reason is in the truncation error of the two schemes and the symmetry breaking of the upwind one. In practice, the truncation error of the upwind scheme doesn't cancel certain derivatives, which reappear as numerical diffusion in the equation you are actually solving. Symmetry of the central scheme is such that the same terms instead disappear and so does numerical diffusion, but dispersion exists for both (yet the diffusion of upwind kills that). - For uniform grids 4th and 2nd order central schemes have both the same, null, numerical diffusion. The 4th order scheme has less dispersion error than the 2nd order one. Just for completeness, the dispersion error is the one causing different wavelengths of the solution to travel at different, numerically induced, speeds. |
|
![]() |
![]() |
![]() |
![]() |
#4 |
New Member
Join Date: May 2022
Posts: 18
Rep Power: 5 ![]() |
Thank you very much ! I more clear now !
|
|
![]() |
![]() |
![]() |
Tags |
2nd order discretization, discretization scheme, navier stokes equation, upwind schemes |
Thread Tools | Search this Thread |
Display Modes | |
|
|
![]() |
||||
Thread | Thread Starter | Forum | Replies | Last Post |
Benchmark Problems to test a new scheme for Convection-diffusion equation | alibaig1991 | Main CFD Forum | 1 | June 16, 2018 04:30 |
mass flow in is not equal to mass flow out | saii | CFX | 12 | March 19, 2018 05:21 |
Problem with Velocity Poisson Equation and Vector Potential Poisson Equation | mykkujinu2201 | Main CFD Forum | 1 | August 12, 2017 13:15 |
solving a differential equation with NEWTON.... | myway | Main CFD Forum | 0 | May 15, 2006 02:05 |
Expression of 2ndary differential equation term | J.W.Ryu | FLUENT | 0 | June 17, 2002 23:50 |