Finite Difference vs. FEM: FLAC3D 1

[flamby]

I have come accross a program for geotechical engineering called FLAC3D (

http://www.itascacg.com/flac3d.html).

It is said to be a 'Finite Difference' program. How a Finite Difference program is different from Finite Element programs that we use? Is there any special advantage of using such programs for geotechnical problems?

Any insight will be helpful.

Thanks and regards to all folks.

 

[corus]

Finite Difference methods apply a grid over the region and solve the Partial Differential Equation by approximating the derivatives via the Taylor series expansion and using differences as an approximation. For this method it is important that a uniform grid is applied over the region to reduce the errors by the differencing method. Finite Differencing methods are thus less robust for irregular shaped bodies than finite element methods which divide the region into separate elements to fit the region and use a variational approach to solving the PDE.
The benefits of FD methods are that it is easy to understand, easy to explain , easy to program, meshing is simple, and the error is known in terms of the remainder from the Taylor series expansion of the derivatives. It used to be commonly used in fluid dynamic methods mainly because of its stability, I believe, but FE methods are now used.
What the benefits of the software you mention are, I don't know, as all of the features mentioned are covered in most FE codes, unless it's modelling some phenomena not covered by FE methods which I can't see.

 

[flamby]

Thanks corus. The very good explanation from you clarifies my doubts to great extent.

Thanks to Eng-Tips too that such things are possible in no time.

Regards.

 

[Cmfg]

For Dynamic analysis, I could say that 99% of programs are a mix of FE to dicretize the space and FD to discretize the time (marching time).
Some people at U of Illinois are working in a true FE dynamic analysis discretizing space and time (kind a 4D analysis).