CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Numerical Method for groundwater flow in 3D (http://www.cfd-online.com/Forums/main/5738-numerical-method-groundwater-flow-3d.html)

Linda February 13, 2003 19:55

Numerical Method for groundwater flow in 3D
 
I'm a grad student in geology. I'm studying groundwater flow and solutet transport in 3D heterogeneous porous media. Using an experimental deposits reconstruted in 3D, my study will need to be computed with a very large grid (up to 60 million nodes if I incorporate all the data points). Thus it will require parallel computing to solve for the head field and velocity, and then particle tracking. However, I have no supercomputing exeperience (I have since modeled systems with at most half of million elements), and now start to invesigate my options. Someone at the University computing center told me that I may start with OpenMP for my large model since it is easiler to learn and implement; but others told me I would need to learn MPI for such an endeavor and OpenMP may not be suffcient for too large a problem. So, I'm quite undecided as to what to do. Any suggestions from anyone doing supercomputing in flow models?

I'm also deciding on the numerical method to use. I've so far used finite difference and finite element method, but learned that Finte Volume Method might be a best approach (flexbility in geometry, local mass balance via velocity continuity). However, I'm not sure if there are libraries solvers or grid partioning software written for FVM in supercomputing. I know for sure that for FEM, there are such solvers and mesh partitioning software. Any info on this is much much appreciated!


All times are GMT -4. The time now is 21:40.