CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   FVM vs FEM (http://www.cfd-online.com/Forums/main/13526-fvm-vs-fem.html)

elwolv May 26, 2007 03:55

FVM vs FEM
 
looking for books on CFD using FVM, I found only few available.

in most discussions for CFD, FVM is as capable for handling problems with curved geometry and unstructured grid

however, it seems all academic courses are geared for FEM and all books are on FEM, and most simulations performed with packages using FEM.

Q: my interest is in microfluidic simulation, does that means that FVM is no longer a viable solution to use and only FEM should be the one method to use?

thank you for guidance on that.

elwolv


otd May 26, 2007 10:16

Re: FVM vs FEM
 
"does that means that FVM is no longer a viable solution to use and only FEM should be the one method to use?"

No

elwolv May 26, 2007 12:18

Re: FVM vs FEM
 
tank you for response

i found that FVM is used in microfluidic simulation, example problem at the web site of prof Nenes : nenes.eas.gatech.edu/cfd

however, a review article "toward numerical prottyping of lab-n-chi:..." by David Erickson, microfluid nanofluid (2005) 1:301-318, states in page 302: that FV methods are best suited for problems where viscous terms a absent as opposed to the low Re flows encountered in microfluidic in which these terms are dominant!

Q: how do you interpret the results of prof Nenes example if the should not be used for such scale problem?

Q: just because of viscosity we should not use FVM in a microfluidic simulation?

thank you

rt May 27, 2007 02:05

Re: FVM vs FEM
 
FVM vs. FEM is an old question that usually appeared in this forum, but it is very difficult to compare FVM vs. FEM from accuracy and capability veiwpoints because there are several version of FEM and FVM that can compete with each other. Also under some condition FEM is equivalent with FVM e.g. see these papers:

http://www.ann.jussieu.fr/~vohralik/...ub_CRAS_04.pdf http://www.iam.fmph.uniba.sk/amuc/_c...5/vohralik.pdf

in my opinion there is not any limitation for either FVM or FEM for simulation of various kind of fluid but roughly it can be said that FEM is more easy to programming and implementation (especially for multiphysics phenomena) but more memory and CPU expensive.


elwolv May 28, 2007 01:34

Re: FVM vs FEM
 
thank you for your response - RT

1- I understand that FEM is the most popular in commercial pkgs, and it is easier to program, I thought that FVM is easier.

2- my comment regarding the articles indicating that FVM is for viscosity=0, while my interest in microfluidic is viscous in nature for low Re <1 and <10.

Q: can we consider FEM is method of choice to program curved path for simulation? and the viscosity question becomes irrelevant.

Q: why simpler FDM is not used for micro-simulation? which FDM is more accurate/appropriate for micro-simulation, low Re and viscous effect?

thank you for your responses

elwolv

rt May 28, 2007 16:33

Re: FVM vs FEM
 
>Q: can we consider FEM is method of choice to program >curved path for simulation?

what you mean from curved path?

>and the viscosity question becomes irrelevant.

resolution of viscosity effect in low Renold numbers (as you have 1-10) is simple and i think doesn't made concerrn for you, without attention to FVM, FEM or FDM

>Q: why simpler FDM is not used for micro-simulation?

probably due to limitation to capturing curved bc, or multi-physics simulation or elegance

>which FDM is more accurate/appropriate for micro >simulation, low Re and viscous effect?

i don't konw what is important in your simulation, but based on your requirement (and required LOD) you should select method

for solution of FDM problem related to curved boundaries several methods under concept of cut-cell or immersed boundary are suggested, also sometimes quadtree/octree refinement is included. But they are not usually suitable for viscose simulation.


elwolv May 29, 2007 00:12

Re: FVM vs FEM
 
thank you for your response.

curved path is is the flow micro-channel form a non straight pathway to allow constant flow for long distance in small space, called serpentine. the dimensions are in the range of 100-500 micron.

elwolv


Sandeep June 5, 2007 14:23

Re: FVM vs FEM
 
IntelliSense has a FVM/VOF based commercial microfluidic simulator. One of the nice things about VOF is when dealing with Fluid Structure Interaction you do not have to constantly regrid/remesh. Another advantage is the computational time is significantly less than FEM due to small matrix sizes.

-Sandeep

Sandeep June 5, 2007 14:23

Re: FVM vs FEM
 
Here's a link to the webpage

<http://intellisensesoftware.com/modu...ofluidics.html>

-S


All times are GMT -4. The time now is 05:26.